Compare commits

..
15 Commits
Author SHA1 Message Date
mathias 18d01b64d7 chore(context): restore the root rules in AGENTS.md and system-prompt.txt (#200) (#201)
CI / Lint / Test / Vet (push) Successful in 1m0s
CI / Mirror to GitHub (push) Successful in 4s
Co-authored-by: Mathias <mthbqv@gmail.com>
2026-10-09 06:02:43 +00:00
mathiasandClaude Sonnet 5.5 5dfd2c0f7e docs: main is protected, ship by PR (local-dev#42)
CI / Lint / Test / Vet (push) Successful in 1m28s
CI / Mirror to GitHub (push) Successful in 3s
Gitea will refuse direct pushes to main in this repo. Rule 5's reviewed lane was enforced on the merge tools
(tbd_ship / pr_merge, gitea-mcp v0.17.0/v0.18.0) but not on a direct commit; this overrides rule 5's "risk/low straight
to main" for this repo and says why. Derived adapter files regenerated where the repo has them.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01SZkhi49hfeSqBp5aZJNXsM
2026-10-08 08:34:50 +02:00
mathiasandClaude Opus 5.5 d6ddac4a03 fix(metrics): label an unmatched request "unmatched", never its raw path (#185)
CI / Mirror to GitHub (push) Successful in 5s
CI / Lint / Test / Vet (push) Successful in 1m4s
An authenticated REST request to an unregistered path reached the inner
mux, which left Pattern empty, and metrics.Middleware fell back to the
raw URL path: any token holder could create one Prometheus series per
unique path. The fallback is now the fixed label "unmatched"; real routes
keep their pattern. Tests: buildMux with an authenticated random path,
and the middleware unit test; the timing test now routes through a
ServeMux as production does.

Fixes #185.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-05 23:42:45 +02:00
mathiasandClaude Opus 5.5 4422dd1f69 test(symlinkguard): synchronise the alert-queue tests on observable completion (#193)
CI / Lint / Test / Vet (push) Successful in 1m3s
CI / Mirror to GitHub (push) Successful in 4s
Three Watch tests flaked under -race and CPU load (3/8 of 8 parallel
x10 runs on main):
- QueueIsBounded slept 100ms before looking for the summary; it now
  waits for the summary, then checks the bound.
- DeliversEveryQueuedAlert asserted FIFO order, but the outage ends
  asynchronously, possibly mid-flush, so a later item can send first. It
  now asserts each removal is delivered exactly once (order is not what
  it is named for).
- AccountsForEveryRemoval slept 50ms after the first summary, counted
  only "removed 1" and took the last summary's count. It now sums every
  "removed N" and every suppression count and waits until they account
  for all plants.

Stress: 0/8 failures in three rounds of 8 parallel x -test.count=10 under
-race. Flips still red: unbounded queue, dropped retries, suppressions
not counted.

Fixes #193.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-05 23:33:55 +02:00
mathiasandClaude Opus 5.5 f65f0b7826 fix(audit): each ntfy alert source sends its own title (#163)
CI / Lint / Test / Vet (push) Successful in 1m13s
CI / Mirror to GitHub (push) Successful in 4s
Every alert was titled "brain-capture audit degraded", including the
symlink tripwire's and governance-edit alerts. NtfyNotifier.WithTitle
returns a titled copy (nil stays nil); buildNotifier(getenv, title) is
called per source: capture audit (unchanged title), "brain symlink
tripwire", "brain governance edit". Tests assert the Title header.

Fixes #163.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-05 23:28:38 +02:00
mathiasandClaude Opus 5.5 f4636c7ef0 fix: creating and generated writes go through the brain root (#188, #158 part A) (#189)
CI / Lint / Test / Vet (push) Failing after 1m2s
CI / Mirror to GitHub (push) Skipped
New brainfs PutRegular (create or replace), CreateRegular (create only), AppendRegular, Remove and RealParent: missing parents created one component at a time through the root, every existing component a real directory, new files created with O_EXCL and confirmed before writing. brain_write, promote/reject (O_EXCL create, source kept on refusal), StampProvenanceFile, BuildWingIndex, RebuildIndex, AppendLog and the tunnel candidate log use them; provenance.ApplyFile removed.

Fixes #188. Refs #158. Adversarial review: BLOCKED, BLOCKED, PASS.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Mathias <mthbqv@gmail.com>
2026-10-05 21:25:54 +00:00
mathiasandClaude Opus 5.5 bd989f887f fix(linkcheck): keep the write when the link index cannot be built (#178) (#191)
CI / Lint / Test / Vet (push) Successful in 1m4s
CI / Mirror to GitHub (push) Successful in 4s
StripUnresolvedLinks logs a warning and returns the content unchanged when the link index cannot be built (e.g. an unreadable subdirectory), instead of failing every brain_write/brain_update and losing the learning. Nothing is unwrapped on a partial index.

Fixes #178.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Mathias <mthbqv@gmail.com>
2026-10-05 21:16:04 +00:00
mathiasandClaude Opus 5.5 4770f97ffd fix(symlinkguard): report and remove FIFOs, sockets and devices too (#179) (#190)
CI / Lint / Test / Vet (push) Successful in 1m5s
CI / Mirror to GitHub (push) Successful in 4s
Scan reports special files (not regular, directory or symlink) in Result.Special, so the startup check refuses with a message naming them; Watch removes them through an os.Root after re-checking the type, and reports any it cannot remove (never as removed).

Fixes #179. Adversarial review: BLOCKED, BLOCKED, PASS. Flake on main tracked in #193.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Mathias <mthbqv@gmail.com>
2026-10-05 21:15:57 +00:00
mathiasandClaude Opus 5.5 ddd4bdc2ac fix(brainfs): refuse a file behind a symlinked directory for reads and writes (#173) (#187)
CI / Lint / Test / Vet (push) Successful in 1m2s
CI / Mirror to GitHub (push) Successful in 3s
ReadRegular, ReadRegularMax and WriteRegular require a clean, local path and a real directory at every level above the file, so an in-brain directory symlink (or a ".." after one) can no longer redirect brain_update, brain_tunnel, brain_read/get or the other readers elsewhere in the brain. session.Append opens through an os.Root on the brain dir, refuses a non-directory sessions dir, and writes only after confirming the log path is the regular file it opened.

Fixes #173. Adversarial review: BLOCKED, PASS. Follow-up: #188 (brain_write create path).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Mathias <mthbqv@gmail.com>
2026-10-05 20:17:37 +00:00
mathiasandClaude Opus 5.5 1fff54769d fix(tunnel): DetectTunnels walks and reads through brainfs (#174, #157 slice 5b) (#186)
CI / Lint / Test / Vet (push) Successful in 1m2s
CI / Mirror to GitHub (push) Successful in 4s
DetectTunnels (inside every brain_write/brain_update via AutoTunnel) walks with brainfs.WalkRegular and reads titles with ReadRegular: a FIFO under wiki/ no longer hangs the write, and a symlinked note's title no longer becomes a tunnel candidate. A wiki/ that is not a real directory yields no candidates; other walk errors still abort.

Fixes #174. Refs #157 (last reader slice). Adversarial review: BLOCKED, PASS.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Mathias <mthbqv@gmail.com>
2026-10-05 20:12:44 +00:00
mathias 0c3e82ca0a test(brainfs): per-caller in-brain symlink targets for slices 1-3 (#171) (#172)
CI / Lint / Test / Vet (push) Successful in 1m2s
CI / Mirror to GitHub (push) Successful in 4s
Co-authored-by: Mathias <mthbqv@gmail.com>
2026-10-05 19:36:31 +00:00
mathiasandClaude Opus 5.5 eedb110be5 fix(pipeline,classification): read the brain through brainfs (#157, slice 6) (#176)
CI / Lint / Test / Vet (push) Successful in 1m1s
CI / Mirror to GitHub (push) Successful in 4s
Ingest pipeline readers (readSourcePage and loadSchema feed the LLM prompt; reviseOrCreate; injectSourceRefs; BackfillRefs list and reads) and classification.Load read through brainfs: symlinked or special files no longer reach the prompt, written pages or the I1 sovereignty config, and a FIFO no longer hangs ingest or startup. A symlinked/special classification.yaml is a hard error; an absent one still yields defaults.

Refs #157. Adversarial review: BLOCKED, BLOCKED, PASS. Follow-ups: #158 (write paths), #179 (FIFO tripwire).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Mathias <mthbqv@gmail.com>
2026-10-05 18:54:42 +00:00
mathiasandClaude Opus 5.5 febab3e45d fix: brain_tunnel, brain_update and session_log confine caller-supplied paths (#169) (#170)
CI / Lint / Test / Vet (push) Successful in 1m2s
CI / Mirror to GitHub (push) Successful in 4s
brain_tunnel accepts only clean wiki/.../*.md paths and reads/rewrites both notes through brainfs; new brainfs.WriteRegular never creates, never writes through a final symlink (Lstat + O_NOFOLLOW|O_NONBLOCK + os.SameFile on the handle, since os.Root follows in-root final symlinks even with O_NOFOLLOW). brain_update resolves like brain_edit (knowledge/ and wiki/ notes only) and reads/rewrites through brainfs. session_log requires a plain-name session id and opens with O_NOFOLLOW|O_NONBLOCK.

Fixes #169. Adversarial review: BLOCKED, BLOCKED, PASS (round 3 by the estate session's reviewer). Follow-ups: #173, #174.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Mathias <mthbqv@gmail.com>
2026-10-05 18:47:07 +00:00
mathias bdc1e51a1a fix(ingestion): build /mcp's bearer and the metrics wrap inside buildMux (#154) (#183)
CI / Lint / Test / Vet (push) Successful in 1m1s
CI / Mirror to GitHub (push) Successful in 4s
Co-authored-by: Mathias <mthbqv@gmail.com>
2026-10-05 17:48:44 +00:00
mathias 8c48e2e343 fix(symlinkguard): deliver alerts beside the scan, not inside it (#164) (#180)
CI / Lint / Test / Vet (push) Successful in 54s
CI / Mirror to GitHub (push) Successful in 4s
Co-authored-by: Mathias <mthbqv@gmail.com>
2026-10-05 14:48:16 +00:00
53 changed files with 2946 additions and 258 deletions
+12
View File
@@ -84,3 +84,15 @@ When acting as a coding agent on this project:
4. Never modify files outside the project root without explicit permission 4. Never modify files outside the project root without explicit permission
5. When adding a dependency, explain why in the commit message 5. When adding a dependency, explain why in the commit message
6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM 6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM
## `main` is protected: ship by PR (local-dev#42, decided 2026-10-08)
Gitea refuses direct pushes to `main` in this repo, and a PR merges only after its CI status (`CI / Lint / Test / Vet (pull_request)`) passes.
**Why:** local-dev rule 5's reviewed lane (risk/medium+ is reviewed by a human) was enforced on the merge tools
(`tbd_ship`, `pr_merge`: gitea-mcp v0.17.0/v0.18.0, merge only for an issue labelled exactly `risk/low` with green CI)
but not on a direct commit to `main`, which bypassed it. This overrides rule 5's "risk/low commits straight to main"
for this repo; control code is where the lane matters.
**How:** `tbd_ship` for a one-file change, or branch -> PR -> `pr_merge` (name the `risk/low` issue); above `risk/low`
open the PR and a human merges it in the Gitea UI after review. Tags are not protected. CD jobs push image tags to
the **infra** repo with deploy keys, never to this repo's `main`. Undo (Mathias only): Settings -> Branches, or
`PATCH /repos/mathias/hyperguild/branch_protections/main {"enable_push": true}`.
+81 -17
View File
@@ -164,6 +164,63 @@ These rules apply to every task across every project, regardless of harness.
the thing it was named for was exploitable. If you cannot name the mechanism, the thing it was named for was exploitable. If you cannot name the mechanism,
you do not yet know what the test observes. you do not yet know what the test observes.
10. **Short sessions, one rc per repo, edit only in a locked worktree of the repo
you are changing** (ADR-0002, accepted 2026-10-06).
- One session per issue or task, ending with rule 6's close ritual. Related
issues may share a session only when they share a worktree and an owner.
State that must outlive the session goes in an issue, not the chat.
- Start `claude rc` in the repo you will edit (`infra`, `trucker-ecommerce`,
`hyperguild`, ...), not in `~/dev`: the rc worktree is a worktree of the
launch directory's repo, so launched from `~/dev` it protects nothing you
edit (measured: 73% of edits landed in shared checkouts). Start it only
while working there.
- Keep at most one thin `~/dev` coordinator session. It files issues and hands
out tasks; it reads and edits no sub-repo code.
- Split cross-repo work per repo. The contract between the halves lives in an
issue on each repo.
- Edit in a worktree of that repo, locked at creation:
`scripts/wt.sh add <repo> <task>` does it (fetch, `wt/<task>` from origin/main, lock);
`scripts/wt.sh done <repo>-<task>` removes it after the push and refuses dirty or
unpushed work. By hand: `git worktree add ~/worktrees/<name> -b <branch> origin/main`
then `git worktree lock --reason "manual: <purpose>" ~/worktrees/<name>`.
- Before working a repo another session may touch, announce your issue list to
that session (`SendMessage`) and ask for theirs.
- Auto-memory is keyed by cwd, so a repo rc starts without it. Cross-repo
working rules live in brain (`wiki/homelab/decisions/2026-10-06-062416-auto.md`, wing homelab, hall
decisions) and the section below, not in per-directory memory.
## Standing footguns
Each cost real time once. Detail and evidence: brain `wiki/homelab/decisions/2026-10-06-062416-auto.md` (title standing-working-rules).
- **Gate then commit with `&&`, never `;`.** "Green" means exit code 0 and
`grep -c FAIL` is 0, not a filtered grep.
- **Stage by path, never `git add -A` or `git add .`** (they stage from the repo
root and swept in other sessions' work twice). Check `git status --porcelain` first.
- **Before committing in a shared checkout:** `git branch --show-current`.
- **Before tagging:** `git fetch origin --tags`, then `git describe --tags --abbrev=0 <sha>`
and `git tag --sort=-v:refname | head`; go above both. Never take the last tag
from a report.
- **Run only what the project's gate runs.** No repo-wide formatter unless the gate
runs it.
- **Reproduce a CI red at the cited commit before applying an issue's prescribed fix.**
- **Read an issue's comments (`issue_list_comments`), not just its body, before working it.**
- **A comment naming a control ("behind branch protection") is not evidence; query
the control.** A working detector can hide a missing preventer.
- **On koala the MCP servers are `mcp__plugin_skills_{brain,gitea,infra}__*`.** The
per-turn auth reminder is not evidence; call the tool or run `claude mcp list`.
Brain writes always pass `wing` + `hall`.
- **`cat /etc/hostname` before assuming which host you are on.**
- **Never run bare `git remote -v` or `git config --list`** (url.insteadOf expands
into a live token); use `git config --get remote.origin.url`.
- **On koala `sudo` is denied outright; k8s exec, env dumps and flux reconcile need
per-action approval; decrypting a SOPS file and printing in one Bash call is
hook-blocked.** Do the repo-side work and hand the host-side step over in an issue.
- **`set -o pipefail` plus an optional `grep | head | awk` dies on no match:** append `|| true`.
- **Gitea `file_write_branch` takes plain text, not base64.**
- **`OP_SERVICE_ACCOUNT_TOKEN` in the shell is read-only for `op`.** Operational
secrets go in the AgentSecrets vault via `set -a; . ~/.op-agent-write.env; set +a`.
## CAD sprint patterns ## CAD sprint patterns
Two CAD sprint archetypes (defs: brain `wiki/homelab/decisions/continuous-agentic-development-cad-concept-2026-06-16`): Two CAD sprint archetypes (defs: brain `wiki/homelab/decisions/continuous-agentic-development-cad-concept-2026-06-16`):
@@ -244,25 +301,20 @@ Three machines on Tailscale:
All development repos live at `~/dev/` (softlink from `~/Documents/local-dev/`). All development repos live at `~/dev/` (softlink from `~/Documents/local-dev/`).
Organized in thematic folders: Organized by **role**, because the role decides which rules apply (ADR-0003):
| Folder | Focus | Count | | Folder | Role | Rules that key off it |
|--------|-------|-------| |--------|------|-----------------------|
| `GO/` | Go web frameworks, API integrations, learning projects | ~10 | | `platform/` | infra, brain, hyperguild, gitea-mcp, dispatch, shared tooling | control code (rule 8), GitOps review lane |
| `AI/` | ML research, AI frameworks (FinRL, DSPy, crawl4ai) | ~6 | | `products/` | tapir, cobalt-dingo, openbanking-api, ... | ordinary trunk-based work |
| `AGENTS/` | Autonomous agents, coding agents, MCP servers, infra | ~15 | | `clients/` | client work | local models only, never cloud APIs (Client work rules) |
| `QKX/` | Invoice processing, financial automation, payment systems | ~13 | | `spikes/` | throwaway | no ceremony |
| `XT/` | Climate data, sustainability (Klimatkollen, Garbo) | ~2 |
See `~/dev/PROJECT_SUMMARY.md` for detailed descriptions of each project. New projects are born into a role folder (`task new-project -- <name> <client|personal>
<role-folder>`); a client tag requires `clients/` and `clients/` requires a client tag.
### Key active projects Existing repos still sit in the old `AGENTS/` and `AI/` folders until they are moved one by
one (ADR-0003 step 3). `~/dev/PROJECT_SUMMARY.md` (generated) lists every repo and where it
- **super-koala** (`AGENTS/`) — multi-component agent stack with LangGraph, DSPy, MCP is. How a session runs on koala (worktrees, rc per repo, the reaper, memory): `docs/working-on-koala.md`.
- **azure-tiger** (`QKX/`) — invoice extraction → ISO 20022 payment instructions
- **gocrwl** (`AGENTS/`) — Go web crawler with containerized deployment
- **koala-ai-stack** (`AGENTS/`) — local AI server infrastructure management
- **klimatkollen** (`XT/`) — Swedish municipal climate data platform
## Knowledge base — actively use it ## Knowledge base — actively use it
@@ -514,4 +566,16 @@ When acting as a coding agent on this project:
5. When adding a dependency, explain why in the commit message 5. When adding a dependency, explain why in the commit message
6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM 6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM
## `main` is protected: ship by PR (local-dev#42, decided 2026-10-08)
Gitea refuses direct pushes to `main` in this repo, and a PR merges only after its CI status (`CI / Lint / Test / Vet (pull_request)`) passes.
**Why:** local-dev rule 5's reviewed lane (risk/medium+ is reviewed by a human) was enforced on the merge tools
(`tbd_ship`, `pr_merge`: gitea-mcp v0.17.0/v0.18.0, merge only for an issue labelled exactly `risk/low` with green CI)
but not on a direct commit to `main`, which bypassed it. This overrides rule 5's "risk/low commits straight to main"
for this repo; control code is where the lane matters.
**How:** `tbd_ship` for a one-file change, or branch -> PR -> `pr_merge` (name the `risk/low` issue); above `risk/low`
open the PR and a human merges it in the Gitea UI after review. Tags are not protected. CD jobs push image tags to
the **infra** repo with deploy keys, never to this repo's `main`. Undo (Mathias only): Settings -> Branches, or
`PATCH /repos/mathias/hyperguild/branch_protections/main {"enable_push": true}`.
--- ---
+81 -17
View File
@@ -159,6 +159,63 @@ These rules apply to every task across every project, regardless of harness.
the thing it was named for was exploitable. If you cannot name the mechanism, the thing it was named for was exploitable. If you cannot name the mechanism,
you do not yet know what the test observes. you do not yet know what the test observes.
10. **Short sessions, one rc per repo, edit only in a locked worktree of the repo
you are changing** (ADR-0002, accepted 2026-10-06).
- One session per issue or task, ending with rule 6's close ritual. Related
issues may share a session only when they share a worktree and an owner.
State that must outlive the session goes in an issue, not the chat.
- Start `claude rc` in the repo you will edit (`infra`, `trucker-ecommerce`,
`hyperguild`, ...), not in `~/dev`: the rc worktree is a worktree of the
launch directory's repo, so launched from `~/dev` it protects nothing you
edit (measured: 73% of edits landed in shared checkouts). Start it only
while working there.
- Keep at most one thin `~/dev` coordinator session. It files issues and hands
out tasks; it reads and edits no sub-repo code.
- Split cross-repo work per repo. The contract between the halves lives in an
issue on each repo.
- Edit in a worktree of that repo, locked at creation:
`scripts/wt.sh add <repo> <task>` does it (fetch, `wt/<task>` from origin/main, lock);
`scripts/wt.sh done <repo>-<task>` removes it after the push and refuses dirty or
unpushed work. By hand: `git worktree add ~/worktrees/<name> -b <branch> origin/main`
then `git worktree lock --reason "manual: <purpose>" ~/worktrees/<name>`.
- Before working a repo another session may touch, announce your issue list to
that session (`SendMessage`) and ask for theirs.
- Auto-memory is keyed by cwd, so a repo rc starts without it. Cross-repo
working rules live in brain (`wiki/homelab/decisions/2026-10-06-062416-auto.md`, wing homelab, hall
decisions) and the section below, not in per-directory memory.
## Standing footguns
Each cost real time once. Detail and evidence: brain `wiki/homelab/decisions/2026-10-06-062416-auto.md` (title standing-working-rules).
- **Gate then commit with `&&`, never `;`.** "Green" means exit code 0 and
`grep -c FAIL` is 0, not a filtered grep.
- **Stage by path, never `git add -A` or `git add .`** (they stage from the repo
root and swept in other sessions' work twice). Check `git status --porcelain` first.
- **Before committing in a shared checkout:** `git branch --show-current`.
- **Before tagging:** `git fetch origin --tags`, then `git describe --tags --abbrev=0 <sha>`
and `git tag --sort=-v:refname | head`; go above both. Never take the last tag
from a report.
- **Run only what the project's gate runs.** No repo-wide formatter unless the gate
runs it.
- **Reproduce a CI red at the cited commit before applying an issue's prescribed fix.**
- **Read an issue's comments (`issue_list_comments`), not just its body, before working it.**
- **A comment naming a control ("behind branch protection") is not evidence; query
the control.** A working detector can hide a missing preventer.
- **On koala the MCP servers are `mcp__plugin_skills_{brain,gitea,infra}__*`.** The
per-turn auth reminder is not evidence; call the tool or run `claude mcp list`.
Brain writes always pass `wing` + `hall`.
- **`cat /etc/hostname` before assuming which host you are on.**
- **Never run bare `git remote -v` or `git config --list`** (url.insteadOf expands
into a live token); use `git config --get remote.origin.url`.
- **On koala `sudo` is denied outright; k8s exec, env dumps and flux reconcile need
per-action approval; decrypting a SOPS file and printing in one Bash call is
hook-blocked.** Do the repo-side work and hand the host-side step over in an issue.
- **`set -o pipefail` plus an optional `grep | head | awk` dies on no match:** append `|| true`.
- **Gitea `file_write_branch` takes plain text, not base64.**
- **`OP_SERVICE_ACCOUNT_TOKEN` in the shell is read-only for `op`.** Operational
secrets go in the AgentSecrets vault via `set -a; . ~/.op-agent-write.env; set +a`.
## CAD sprint patterns ## CAD sprint patterns
Two CAD sprint archetypes (defs: brain `wiki/homelab/decisions/continuous-agentic-development-cad-concept-2026-06-16`): Two CAD sprint archetypes (defs: brain `wiki/homelab/decisions/continuous-agentic-development-cad-concept-2026-06-16`):
@@ -239,25 +296,20 @@ Three machines on Tailscale:
All development repos live at `~/dev/` (softlink from `~/Documents/local-dev/`). All development repos live at `~/dev/` (softlink from `~/Documents/local-dev/`).
Organized in thematic folders: Organized by **role**, because the role decides which rules apply (ADR-0003):
| Folder | Focus | Count | | Folder | Role | Rules that key off it |
|--------|-------|-------| |--------|------|-----------------------|
| `GO/` | Go web frameworks, API integrations, learning projects | ~10 | | `platform/` | infra, brain, hyperguild, gitea-mcp, dispatch, shared tooling | control code (rule 8), GitOps review lane |
| `AI/` | ML research, AI frameworks (FinRL, DSPy, crawl4ai) | ~6 | | `products/` | tapir, cobalt-dingo, openbanking-api, ... | ordinary trunk-based work |
| `AGENTS/` | Autonomous agents, coding agents, MCP servers, infra | ~15 | | `clients/` | client work | local models only, never cloud APIs (Client work rules) |
| `QKX/` | Invoice processing, financial automation, payment systems | ~13 | | `spikes/` | throwaway | no ceremony |
| `XT/` | Climate data, sustainability (Klimatkollen, Garbo) | ~2 |
See `~/dev/PROJECT_SUMMARY.md` for detailed descriptions of each project. New projects are born into a role folder (`task new-project -- <name> <client|personal>
<role-folder>`); a client tag requires `clients/` and `clients/` requires a client tag.
### Key active projects Existing repos still sit in the old `AGENTS/` and `AI/` folders until they are moved one by
one (ADR-0003 step 3). `~/dev/PROJECT_SUMMARY.md` (generated) lists every repo and where it
- **super-koala** (`AGENTS/`) — multi-component agent stack with LangGraph, DSPy, MCP is. How a session runs on koala (worktrees, rc per repo, the reaper, memory): `docs/working-on-koala.md`.
- **azure-tiger** (`QKX/`) — invoice extraction → ISO 20022 payment instructions
- **gocrwl** (`AGENTS/`) — Go web crawler with containerized deployment
- **koala-ai-stack** (`AGENTS/`) — local AI server infrastructure management
- **klimatkollen** (`XT/`) — Swedish municipal climate data platform
## Knowledge base — actively use it ## Knowledge base — actively use it
@@ -508,3 +560,15 @@ When acting as a coding agent on this project:
4. Never modify files outside the project root without explicit permission 4. Never modify files outside the project root without explicit permission
5. When adding a dependency, explain why in the commit message 5. When adding a dependency, explain why in the commit message
6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM 6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM
## `main` is protected: ship by PR (local-dev#42, decided 2026-10-08)
Gitea refuses direct pushes to `main` in this repo, and a PR merges only after its CI status (`CI / Lint / Test / Vet (pull_request)`) passes.
**Why:** local-dev rule 5's reviewed lane (risk/medium+ is reviewed by a human) was enforced on the merge tools
(`tbd_ship`, `pr_merge`: gitea-mcp v0.17.0/v0.18.0, merge only for an issue labelled exactly `risk/low` with green CI)
but not on a direct commit to `main`, which bypassed it. This overrides rule 5's "risk/low commits straight to main"
for this repo; control code is where the lane matters.
**How:** `tbd_ship` for a one-file change, or branch -> PR -> `pr_merge` (name the `risk/low` issue); above `risk/low`
open the PR and a human merges it in the Gitea UI after review. Tags are not protected. CD jobs push image tags to
the **infra** repo with deploy keys, never to this repo's `main`. Undo (Mathias only): Settings -> Branches, or
`PATCH /repos/mathias/hyperguild/branch_protections/main {"enable_push": true}`.
+12
View File
@@ -84,3 +84,15 @@ When acting as a coding agent on this project:
4. Never modify files outside the project root without explicit permission 4. Never modify files outside the project root without explicit permission
5. When adding a dependency, explain why in the commit message 5. When adding a dependency, explain why in the commit message
6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM 6. For client projects: never send code or context to cloud APIs — use local models via LiteLLM
## `main` is protected: ship by PR (local-dev#42, decided 2026-10-08)
Gitea refuses direct pushes to `main` in this repo, and a PR merges only after its CI status (`CI / Lint / Test / Vet (pull_request)`) passes.
**Why:** local-dev rule 5's reviewed lane (risk/medium+ is reviewed by a human) was enforced on the merge tools
(`tbd_ship`, `pr_merge`: gitea-mcp v0.17.0/v0.18.0, merge only for an issue labelled exactly `risk/low` with green CI)
but not on a direct commit to `main`, which bypassed it. This overrides rule 5's "risk/low commits straight to main"
for this repo; control code is where the lane matters.
**How:** `tbd_ship` for a one-file change, or branch -> PR -> `pr_merge` (name the `risk/low` issue); above `risk/low`
open the PR and a human merges it in the Gitea UI after review. Tags are not protected. CD jobs push image tags to
the **infra** repo with deploy keys, never to this repo's `main`. Undo (Mathias only): Settings -> Branches, or
`PATCH /repos/mathias/hyperguild/branch_protections/main {"enable_push": true}`.
+9 -6
View File
@@ -179,9 +179,12 @@ func alertingWarnings(getenv func(string) string) []string {
return out return out
} }
func buildNotifier() audit.Notifier { // buildNotifier returns the ntfy notifier for one alert source, titled so
if n := audit.NewNtfyNotifier(os.Getenv("BRAIN_NTFY_URL"), os.Getenv("BRAIN_NTFY_TOKEN")); n != nil { // the source is recognisable (hyperguild#163), or nil when ntfy is not
return n // configured.
func buildNotifier(getenv func(string) string, title string) audit.Notifier {
if n := audit.NewNtfyNotifier(getenv("BRAIN_NTFY_URL"), getenv("BRAIN_NTFY_TOKEN")); n != nil {
return n.WithTitle(title)
} }
return nil return nil
} }
@@ -203,7 +206,7 @@ func buildAuditSink(ctx context.Context, brainDir string, logger *slog.Logger) c
logger.Error("capture audit buffer init", "err", err) logger.Error("capture audit buffer init", "err", err)
os.Exit(1) os.Exit(1)
} }
notifier := buildNotifier() notifier := buildNotifier(os.Getenv, "brain-capture audit degraded")
reconcileInterval := time.Duration(envInt("BRAIN_AUDIT_RECONCILE_INTERVAL", 60)) * time.Second reconcileInterval := time.Duration(envInt("BRAIN_AUDIT_RECONCILE_INTERVAL", 60)) * time.Second
audit.StartReconcile(ctx, central, buffer, notifier, reconcileInterval) audit.StartReconcile(ctx, central, buffer, notifier, reconcileInterval)
logger.Info("capture audit: loki+buffer sink", "loki", lokiURL, "reconcile_s", int(reconcileInterval.Seconds())) logger.Info("capture audit: loki+buffer sink", "loki", lokiURL, "reconcile_s", int(reconcileInterval.Seconds()))
@@ -366,7 +369,7 @@ func main() {
ctx := context.Background() ctx := context.Background()
// hyperguild#157: report a symlink that appears in the brain while running. // hyperguild#157: report a symlink that appears in the brain while running.
go symlinkguard.Watch(ctx, brainDir, symlinkScanInterval(os.Getenv), buildNotifier()) go symlinkguard.Watch(ctx, brainDir, symlinkScanInterval(os.Getenv), buildNotifier(os.Getenv, "brain symlink tripwire"))
// Never fatal on an unreachable issuer (#83): on a cold cluster boot this // Never fatal on an unreachable issuer (#83): on a cold cluster boot this
// pod can beat dex to serving and get a 502 from ingress, and exiting meant // pod can beat dex to serving and get a 502 from ingress, and exiting meant
@@ -546,7 +549,7 @@ func main() {
// capture already gets, plus an out-of-band alert. Wired here because // capture already gets, plus an out-of-band alert. Wired here because
// brain_edit_root needs the capture plumbing anyway — that is what // brain_edit_root needs the capture plumbing anyway — that is what
// re-derives the static principal its gate checks. // re-derives the static principal its gate checks.
mcpSrv.WithGovernanceObserver(auditSink, buildNotifier()) mcpSrv.WithGovernanceObserver(auditSink, buildNotifier(os.Getenv, "brain governance edit"))
logger.Info("capture enabled (REST + MCP tool)", "sovereign_principals", len(sovereign), logger.Info("capture enabled (REST + MCP tool)", "sovereign_principals", len(sovereign),
"governance_observer", true) "governance_observer", true)
} else { } else {
+13
View File
@@ -159,3 +159,16 @@ func TestBuildMuxObservesRequestsAndServesMetrics(t *testing.T) {
require.Equal(t, http.StatusOK, code) require.Equal(t, http.StatusOK, code)
assert.Contains(t, body, `path="GET /pending"`) assert.Contains(t, body, `path="GET /pending"`)
} }
// hyperguild#185: an authenticated request to an unregistered path reached
// the inner REST mux, which left Pattern empty, and the metrics label fell
// back to the raw URL path: one Prometheus series per unique path any token
// holder sends. It is now labelled with a fixed catch-all.
func TestBuildMuxLabelsAnUnregisteredPathWithAFixedLabel(t *testing.T) {
r := testMux(t)
code, _ := do(t, r, "GET", "/zqrandom-185-segment", "Bearer "+routesTok)
require.Equal(t, http.StatusNotFound, code)
_, body := do(t, r, "GET", "/metrics", "")
assert.NotContains(t, body, "zqrandom-185-segment", "the raw path became a metrics label")
assert.Contains(t, body, `path="unmatched"`)
}
@@ -1,9 +1,13 @@
package main package main
import ( import (
"context"
"net/http"
"net/http/httptest"
"os" "os"
"path/filepath" "path/filepath"
"strings" "strings"
"syscall"
"testing" "testing"
"github.com/stretchr/testify/assert" "github.com/stretchr/testify/assert"
@@ -63,3 +67,26 @@ func TestStartupSymlinkCheckRefusesAnUnlistableBrainRoot(t *testing.T) {
t.Cleanup(func() { _ = os.Chmod(brain, 0o755) }) t.Cleanup(func() { _ = os.Chmod(brain, 0o755) })
assert.Error(t, startupSymlinkCheck(brain)) assert.Error(t, startupSymlinkCheck(brain))
} }
// hyperguild#179: a FIFO in the brain is refused at startup with a clear
// message, instead of hanging or exiting later in a reader.
func TestStartupSymlinkCheckRefusesABrainWithAFIFO(t *testing.T) {
brain := t.TempDir()
require.NoError(t, syscall.Mkfifo(filepath.Join(brain, "classification.yaml"), 0o644))
err := startupSymlinkCheck(brain)
require.Error(t, err)
assert.Contains(t, err.Error(), "classification.yaml")
}
// hyperguild#163: each alert source gets its own ntfy title.
func TestBuildNotifierUsesTheGivenTitle(t *testing.T) {
var got string
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) { got = r.Header.Get("Title") }))
defer srv.Close()
env := map[string]string{"BRAIN_NTFY_URL": srv.URL}
n := buildNotifier(func(k string) string { return env[k] }, "brain symlink tripwire")
require.NotNil(t, n)
require.NoError(t, n.Notify(context.Background(), "m"))
assert.Equal(t, "brain symlink tripwire", got)
assert.Nil(t, buildNotifier(func(string) string { return "" }, "x"), "no ntfy URL, no notifier")
}
+1 -1
View File
@@ -68,7 +68,7 @@ require (
github.com/segmentio/asm v1.2.0 // indirect github.com/segmentio/asm v1.2.0 // indirect
golang.org/x/crypto v0.51.0 // indirect golang.org/x/crypto v0.51.0 // indirect
golang.org/x/sync v0.21.0 // indirect golang.org/x/sync v0.21.0 // indirect
golang.org/x/sys v0.45.0 // indirect golang.org/x/sys v0.45.0
golang.org/x/text v0.39.0 // indirect golang.org/x/text v0.39.0 // indirect
gopkg.in/yaml.v3 v3.0.1 gopkg.in/yaml.v3 v3.0.1
) )
+14 -5
View File
@@ -12,6 +12,7 @@ import (
"time" "time"
"github.com/mathiasbq/hyperguild/ingestion/internal/brain" "github.com/mathiasbq/hyperguild/ingestion/internal/brain"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
"github.com/mathiasbq/hyperguild/ingestion/internal/claim" "github.com/mathiasbq/hyperguild/ingestion/internal/claim"
"github.com/mathiasbq/hyperguild/ingestion/internal/pipeline" "github.com/mathiasbq/hyperguild/ingestion/internal/pipeline"
"github.com/mathiasbq/hyperguild/ingestion/internal/search" "github.com/mathiasbq/hyperguild/ingestion/internal/search"
@@ -227,14 +228,22 @@ func writeHallNote(brainDir string, opts WriteNoteOptions) (string, error) {
// The directory is created only once the note is known to be writable: // The directory is created only once the note is known to be writable:
// a refused write leaves nothing behind, not even an empty hall dir // a refused write leaves nothing behind, not even an empty hall dir
// (brain#21; provenance refusal joins the envelope refusal here). // (brain#21; provenance refusal joins the envelope refusal here).
if err := os.MkdirAll(filepath.Dir(dest), 0o755); err != nil { // Created through the brain root, never through a symlinked wing or
return "", fmt.Errorf("create hall dir: %w", err) // hall directory (hyperguild#188).
rel, err := filepath.Rel(brainDir, dest)
if err != nil {
return "", fmt.Errorf("note path: %w", err)
} }
if err := os.WriteFile(dest, []byte(final), 0o644); err != nil { rel = filepath.ToSlash(rel)
bfs, err := brainfs.Open(brainDir)
if err != nil {
return "", fmt.Errorf("open brain: %w", err)
}
defer func() { _ = bfs.Close() }()
if err := bfs.PutRegular(rel, []byte(final)); err != nil {
return "", fmt.Errorf("write: %w", err) return "", fmt.Errorf("write: %w", err)
} }
rel, _ := filepath.Rel(brainDir, dest) return rel, nil
return filepath.ToSlash(rel), nil
} }
// scalarValues flattens frontmatter fields to key -> scalar value for // scalarValues flattens frontmatter fields to key -> scalar value for
+10 -12
View File
@@ -127,18 +127,18 @@ func PromoteNote(brainDir string, opts PromoteOptions) (string, error) {
} }
fm.set("promoted_at", now) fm.set("promoted_at", now)
if err := os.MkdirAll(filepath.Dir(dest), 0o755); err != nil { // Write-then-delete: the source survives any write failure. Both go
return "", fmt.Errorf("create wing dir: %w", err) // through the brain root (hyperguild#188).
}
// Write-then-delete: the source survives any write failure.
promoted, err := StampProvenance(fm.render()+body, opts.Origin, opts.WrittenBy) promoted, err := StampProvenance(fm.render()+body, opts.Origin, opts.WrittenBy)
if err != nil { if err != nil {
return "", err return "", err
} }
if err := os.WriteFile(dest, []byte(promoted), 0o644); err != nil { // Created with O_EXCL: a note that appeared after the collision check
// above is never overwritten.
if err := createBrainFile(brainDir, dest, []byte(promoted)); err != nil {
return "", fmt.Errorf("write promoted note: %w", err) return "", fmt.Errorf("write promoted note: %w", err)
} }
if err := os.Remove(src); err != nil { if err := removeBrainFile(brainDir, src); err != nil {
return "", fmt.Errorf("promoted note written but source removal failed: %w", err) return "", fmt.Errorf("promoted note written but source removal failed: %w", err)
} }
@@ -196,14 +196,12 @@ func RejectNote(brainDir string, opts RejectOptions) (string, error) {
fm.set("reject_reason", opts.Reason) fm.set("reject_reason", opts.Reason)
} }
if err := os.MkdirAll(filepath.Dir(dest), 0o755); err != nil { // Write-then-delete: the source survives any write failure. Both go
return "", fmt.Errorf("create rejected dir: %w", err) // through the brain root (hyperguild#188).
} if err := createBrainFile(brainDir, dest, []byte(fm.render()+body)); err != nil {
// Write-then-delete: the source survives any write failure.
if err := os.WriteFile(dest, []byte(fm.render()+body), 0o644); err != nil {
return "", fmt.Errorf("write rejected note: %w", err) return "", fmt.Errorf("write rejected note: %w", err)
} }
if err := os.Remove(src); err != nil { if err := removeBrainFile(brainDir, src); err != nil {
return "", fmt.Errorf("rejected note written but source removal failed: %w", err) return "", fmt.Errorf("rejected note written but source removal failed: %w", err)
} }
@@ -0,0 +1,87 @@
package api_test
import (
"os"
"path/filepath"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/mathiasbq/hyperguild/ingestion/internal/api"
)
// hyperguild#188: promote and reject wrote the destination with
// os.MkdirAll + os.WriteFile by path, so a symlinked wing (promote) or a
// symlinked raw/rejected (reject) redirected the note; the raw source was
// then removed.
func TestPromoteAndRejectRefuseASymlinkedDestinationDirectory(t *testing.T) {
brain := t.TempDir()
outside := t.TempDir()
raw := filepath.Join(brain, "raw")
require.NoError(t, os.MkdirAll(raw, 0o755))
require.NoError(t, os.WriteFile(filepath.Join(raw, "2026-10-05-n.md"), []byte("---\ntitle: n\n---\nbody\n"), 0o644))
require.NoError(t, os.MkdirAll(filepath.Join(brain, "wiki"), 0o755))
require.NoError(t, os.Symlink(outside, filepath.Join(brain, "wiki", "lw")))
require.NoError(t, os.Symlink(outside, filepath.Join(raw, "rejected")))
_, err := api.PromoteNote(brain, api.PromoteOptions{Filename: "2026-10-05-n.md", Wing: "lw", Hall: "facts"})
assert.Error(t, err)
_, err = api.RejectNote(brain, api.RejectOptions{Filename: "2026-10-05-n.md"})
assert.Error(t, err)
ents, _ := os.ReadDir(outside)
assert.Empty(t, ents, "a note was written outside the brain")
_, err = os.Stat(filepath.Join(raw, "2026-10-05-n.md"))
assert.NoError(t, err, "the raw source is kept when the write is refused")
require.NoError(t, os.Remove(filepath.Join(raw, "rejected")))
_, err = api.PromoteNote(brain, api.PromoteOptions{Filename: "2026-10-05-n.md", Wing: "w", Hall: "facts"})
require.NoError(t, err, "positive control: promote into a real wing")
}
// StampProvenanceFile read and rewrote the note by path (provenance.ApplyFile).
func TestStampProvenanceFileRefusesANoteBehindASymlinkedDirectory(t *testing.T) {
brain := t.TempDir()
victim := filepath.Join(brain, "raw", "facts", "v.md")
require.NoError(t, os.MkdirAll(filepath.Dir(victim), 0o755))
require.NoError(t, os.WriteFile(victim, []byte("---\ntitle: v\n---\nuntouched\n"), 0o644))
require.NoError(t, os.MkdirAll(filepath.Join(brain, "wiki", "w"), 0o755))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "raw", "facts"), filepath.Join(brain, "wiki", "w", "alias")))
assert.Error(t, api.StampProvenanceFile(brain, "wiki/w/alias/v.md", "static", "x"))
got, _ := os.ReadFile(victim)
assert.Equal(t, "---\ntitle: v\n---\nuntouched\n", string(got))
real := filepath.Join(brain, "wiki", "w", "facts", "r.md")
require.NoError(t, os.MkdirAll(filepath.Dir(real), 0o755))
require.NoError(t, os.WriteFile(real, []byte("---\ntitle: r\n---\nbody\n"), 0o644))
require.NoError(t, api.StampProvenanceFile(brain, "wiki/w/facts/r.md", "static", "x"), "positive control")
got, _ = os.ReadFile(real)
assert.Contains(t, string(got), "written_by: x")
}
// The same through an in-brain symlinked wing and raw/rejected, which
// os.Root alone would follow.
func TestPromoteAndRejectRefuseAnInBrainSymlinkedDestinationDirectory(t *testing.T) {
brain := t.TempDir()
raw := filepath.Join(brain, "raw")
require.NoError(t, os.MkdirAll(raw, 0o755))
require.NoError(t, os.WriteFile(filepath.Join(raw, "2026-10-05-n.md"), []byte("---\ntitle: n\n---\nbody\n"), 0o644))
target := filepath.Join(brain, "archive")
require.NoError(t, os.MkdirAll(target, 0o755))
require.NoError(t, os.MkdirAll(filepath.Join(brain, "wiki"), 0o755))
require.NoError(t, os.Symlink(filepath.Join("..", "archive"), filepath.Join(brain, "wiki", "lw")))
require.NoError(t, os.Symlink(filepath.Join("..", "archive"), filepath.Join(raw, "rejected")))
_, err := api.PromoteNote(brain, api.PromoteOptions{Filename: "2026-10-05-n.md", Wing: "lw", Hall: "facts"})
assert.Error(t, err)
_, err = api.RejectNote(brain, api.RejectOptions{Filename: "2026-10-05-n.md"})
assert.Error(t, err)
ents, _ := os.ReadDir(target)
assert.Empty(t, ents, "a note was written at the in-brain link's target")
_, err = os.Stat(filepath.Join(raw, "2026-10-05-n.md"))
assert.NoError(t, err, "the raw source is kept")
require.NoError(t, os.Remove(filepath.Join(raw, "rejected")))
_, err = api.RejectNote(brain, api.RejectOptions{Filename: "2026-10-05-n.md"})
require.NoError(t, err, "positive control: reject into a real raw/rejected")
}
+57 -3
View File
@@ -1,6 +1,10 @@
package api package api
import ( import (
"fmt"
"path/filepath"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
"github.com/mathiasbq/hyperguild/ingestion/internal/provenance" "github.com/mathiasbq/hyperguild/ingestion/internal/provenance"
) )
@@ -11,11 +15,61 @@ func StampProvenance(content, origin, writtenBy string) (string, error) {
return provenance.Apply(content, provenance.Stamp{Origin: origin, WrittenBy: writtenBy}) return provenance.Apply(content, provenance.Stamp{Origin: origin, WrittenBy: writtenBy})
} }
// StampProvenanceFile applies StampProvenance to a brain-relative note path. // StampProvenanceFile applies StampProvenance to a brain-relative note path,
// read and rewritten through the brain root (hyperguild#158).
func StampProvenanceFile(brainDir, relPath, origin, writtenBy string) error { func StampProvenanceFile(brainDir, relPath, origin, writtenBy string) error {
_, abs, err := resolveWithin(brainDir, relPath) rel, _, err := resolveWithin(brainDir, relPath)
if err != nil { if err != nil {
return err return err
} }
return provenance.ApplyFile(abs, provenance.Stamp{Origin: origin, WrittenBy: writtenBy}) bfs, err := brainfs.Open(brainDir)
if err != nil {
return fmt.Errorf("open brain: %w", err)
}
defer func() { _ = bfs.Close() }()
b, err := bfs.ReadRegular(rel)
if err != nil {
return fmt.Errorf("read for provenance: %w", err)
}
out, err := provenance.Apply(string(b), provenance.Stamp{Origin: origin, WrittenBy: writtenBy})
if err != nil {
return err
}
if out == string(b) {
return nil
}
if err := bfs.WriteRegular(rel, []byte(out)); err != nil {
return fmt.Errorf("write provenance: %w", err)
}
return nil
}
// createBrainFile creates the file at abs (inside brainDir) through the
// brain root and fails if anything is already there.
func createBrainFile(brainDir, abs string, data []byte) error {
rel, err := filepath.Rel(brainDir, abs)
if err != nil {
return err
}
bfs, err := brainfs.Open(brainDir)
if err != nil {
return fmt.Errorf("open brain: %w", err)
}
defer func() { _ = bfs.Close() }()
return bfs.CreateRegular(filepath.ToSlash(rel), data)
}
// removeBrainFile removes the file at abs (inside brainDir) through the
// brain root.
func removeBrainFile(brainDir, abs string) error {
rel, err := filepath.Rel(brainDir, abs)
if err != nil {
return err
}
bfs, err := brainfs.Open(brainDir)
if err != nil {
return fmt.Errorf("open brain: %w", err)
}
defer func() { _ = bfs.Close() }()
return bfs.Remove(filepath.ToSlash(rel))
} }
@@ -23,6 +23,13 @@ func TestReadNoteRefusesSymlinksAndSurvivesSwapRaces(t *testing.T) {
_, body, _, err := api.ReadNote(brain, "wiki/w/facts/leak.md") _, body, _, err := api.ReadNote(brain, "wiki/w/facts/leak.md")
assert.Error(t, err) assert.Error(t, err)
assert.NotContains(t, body, brainfstest.Sentinel) assert.NotContains(t, body, brainfstest.Sentinel)
// A link to a file elsewhere in the brain: os.Root allows it, only the
// regular-file check refuses it (hyperguild#171).
brainfstest.WriteSentinelAt(t, filepath.Join(brain, "raw", "s.md"))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "..", "raw", "s.md"), filepath.Join(facts, "inlink.md")))
_, body, _, err = api.ReadNote(brain, "wiki/w/facts/inlink.md")
assert.Error(t, err)
assert.NotContains(t, body, brainfstest.Sentinel)
p := filepath.Join(facts, "r.md") p := filepath.Join(facts, "r.md")
require.NoError(t, os.WriteFile(p, []byte("---\nx: y\n---\ninside"), 0o644)) require.NoError(t, os.WriteFile(p, []byte("---\nx: y\n---\ninside"), 0o644))
@@ -39,5 +46,7 @@ func TestReadNoteRefusesSymlinksAndSurvivesSwapRaces(t *testing.T) {
return s return s
} }
brainfstest.AssertNoLeak(t, p, "---\nx: y\n---\ninside", 1500*time.Millisecond, read) brainfstest.AssertNoLeak(t, p, "---\nx: y\n---\ninside", 1500*time.Millisecond, read)
inBrain := brainfstest.WriteSentinelAt(t, filepath.Join(brain, "raw", "s.md"))
brainfstest.AssertNoLeakFrom(t, p, "---\nx: y\n---\ninside", inBrain, 1500*time.Millisecond, read)
brainfstest.AssertNoHang(t, p, "---\nx: y\n---\ninside", 1500*time.Millisecond, 6*time.Second, func() { _ = read() }) brainfstest.AssertNoHang(t, p, "---\nx: y\n---\ninside", 1500*time.Millisecond, 6*time.Second, func() { _ = read() })
} }
+18 -6
View File
@@ -3,15 +3,18 @@ package api
import ( import (
"crypto/sha256" "crypto/sha256"
"encoding/hex" "encoding/hex"
"errors"
"fmt" "fmt"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs" "io/fs"
"os" "os"
"path/filepath" "path/filepath"
"strings" "strings"
"time" "time"
"github.com/mathiasbq/hyperguild/ingestion/internal/brain" "github.com/mathiasbq/hyperguild/ingestion/internal/brain"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
"github.com/mathiasbq/hyperguild/ingestion/internal/claim" "github.com/mathiasbq/hyperguild/ingestion/internal/claim"
"github.com/mathiasbq/hyperguild/ingestion/internal/notes"
) )
// ContentHash returns the lowercase hex sha256 of b. It is the note's // ContentHash returns the lowercase hex sha256 of b. It is the note's
@@ -91,14 +94,23 @@ func UpdateNote(brainDir string, opts UpdateNoteOptions) (relPath, contentHash,
rel = filepath.ToSlash(rel) rel = filepath.ToSlash(rel)
} }
rel, abs, err := resolveWithin(brainDir, rel) // The path is caller-supplied (brain_update): resolve it as brain_edit
// does, knowledge/ and wiki/ notes only, and read and rewrite it through
// the brain root, never through a symlink (hyperguild#169).
t, err := notes.Resolve(brainDir, rel, notes.ScopeNotes)
if err != nil { if err != nil {
return "", "", "", err return "", "", "", err
} }
rel = t.Rel
prior, err := os.ReadFile(abs) bfs, err := brainfs.Open(brainDir)
if err != nil { if err != nil {
if os.IsNotExist(err) { return "", "", "", fmt.Errorf("open brain: %w", err)
}
defer func() { _ = bfs.Close() }()
prior, err := bfs.ReadRegular(rel)
if err != nil {
if errors.Is(err, fs.ErrNotExist) {
return "", "", "", fmt.Errorf("note %q does not exist: use brain_write to create", rel) return "", "", "", fmt.Errorf("note %q does not exist: use brain_write to create", rel)
} }
return "", "", "", fmt.Errorf("read target: %w", err) return "", "", "", fmt.Errorf("read target: %w", err)
@@ -164,7 +176,7 @@ func UpdateNote(brainDir string, opts UpdateNoteOptions) (relPath, contentHash,
return "", "", "", err return "", "", "", err
} }
out := []byte(stamped) out := []byte(stamped)
if err := os.WriteFile(abs, out, 0o644); err != nil { if err := bfs.WriteRegular(rel, out); err != nil {
return "", "", "", fmt.Errorf("write: %w", err) return "", "", "", fmt.Errorf("write: %w", err)
} }
return rel, ContentHash(out), priorHash, nil return rel, ContentHash(out), priorHash, nil
@@ -0,0 +1,36 @@
package api_test
import (
"os"
"path/filepath"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/mathiasbq/hyperguild/ingestion/internal/api"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
)
// hyperguild#173: a directory symlink inside wiki/ that points elsewhere in
// the brain redirected brain_update's write (os.Root follows in-root
// intermediate symlinks). Every directory above the note must be real.
func TestUpdateNoteRefusesANoteBehindASymlinkedDirectory(t *testing.T) {
brain := t.TempDir()
victim := filepath.Join(brain, "raw", "facts", "v.md")
require.NoError(t, os.MkdirAll(filepath.Dir(victim), 0o755))
require.NoError(t, os.WriteFile(victim, []byte("---\ntitle: v\n---\nuntouched\n"), 0o644))
require.NoError(t, os.MkdirAll(filepath.Join(brain, "wiki", "w"), 0o755))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "raw", "facts"), filepath.Join(brain, "wiki", "w", "alias")))
_, _, _, err := api.UpdateNote(brain, api.UpdateNoteOptions{Path: "wiki/w/alias/v.md", Content: "replaced"})
assert.ErrorIs(t, err, brainfs.ErrNotRegular)
got, _ := os.ReadFile(victim)
assert.Contains(t, string(got), "untouched")
real := filepath.Join(brain, "wiki", "w", "facts", "r.md")
require.NoError(t, os.MkdirAll(filepath.Dir(real), 0o755))
require.NoError(t, os.WriteFile(real, []byte("---\ntitle: r\n---\nold\n"), 0o644))
_, _, _, err = api.UpdateNote(brain, api.UpdateNoteOptions{Path: "wiki/w/facts/r.md", Content: "new"})
require.NoError(t, err, "positive control: a note in a real directory is updated")
}
+112
View File
@@ -0,0 +1,112 @@
package api_test
import (
"os"
"path/filepath"
"sync"
"sync/atomic"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/mathiasbq/hyperguild/ingestion/internal/api"
)
// brain_update passed the caller's path through resolveWithin, a lexical
// containment check only: any file in the brain could be superseded
// (classification.yaml, raw/, sessions/), and a final symlink was written
// through. It now resolves like brain_edit: knowledge/ and wiki/ notes only,
// read and rewritten through the brain root.
func TestUpdateNoteRefusesPathsOutsideKnowledgeAndWikiNotes(t *testing.T) {
brain := t.TempDir()
files := map[string]string{
"classification.yaml": "levels: {}\n",
"raw/r.md": "raw\n",
"sessions/s.jsonl": "{}\n",
"wiki/w/facts/ok.md": "---\ntitle: ok\n---\nold\n",
"knowledge/k.md": "---\ntitle: k\n---\nold\n",
"wiki/w/facts/x.txt": "txt\n",
}
for rel, body := range files {
p := filepath.Join(brain, filepath.FromSlash(rel))
require.NoError(t, os.MkdirAll(filepath.Dir(p), 0o755))
require.NoError(t, os.WriteFile(p, []byte(body), 0o644))
}
for _, rel := range []string{"classification.yaml", "raw/r.md", "sessions/s.jsonl", "wiki/w/facts/x.txt", "wiki/w/../../raw/r.md"} {
_, _, _, err := api.UpdateNote(brain, api.UpdateNoteOptions{Path: rel, Content: "replaced"})
assert.Error(t, err, rel)
}
for rel, body := range files {
if rel == "wiki/w/facts/ok.md" || rel == "knowledge/k.md" {
continue
}
got, err := os.ReadFile(filepath.Join(brain, filepath.FromSlash(rel)))
require.NoError(t, err)
assert.Equal(t, body, string(got), rel)
}
for _, rel := range []string{"wiki/w/facts/ok.md", "knowledge/k.md"} {
_, _, _, err := api.UpdateNote(brain, api.UpdateNoteOptions{Path: rel, Content: "new body"})
require.NoError(t, err, "positive control: %s", rel)
got, _ := os.ReadFile(filepath.Join(brain, filepath.FromSlash(rel)))
assert.Contains(t, string(got), "new body")
}
}
func TestUpdateNoteDoesNotWriteThroughASymlinkedNote(t *testing.T) {
brain := t.TempDir()
facts := filepath.Join(brain, "wiki", "w", "facts")
require.NoError(t, os.MkdirAll(facts, 0o755))
victim := filepath.Join(brain, "raw", "victim.md")
require.NoError(t, os.MkdirAll(filepath.Dir(victim), 0o755))
require.NoError(t, os.WriteFile(victim, []byte("untouched\n"), 0o644))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "..", "raw", "victim.md"), filepath.Join(facts, "l.md")))
_, _, _, err := api.UpdateNote(brain, api.UpdateNoteOptions{Path: "wiki/w/facts/l.md", Content: "replaced"})
assert.Error(t, err)
got, _ := os.ReadFile(victim)
assert.Equal(t, "untouched\n", string(got))
}
// The note is swapped between a regular file and an in-brain symlink to a
// raw/ file while UpdateNote runs: the read and the write each refuse a
// symlink, so the raw/ file is never rewritten.
func TestUpdateNoteNeverRewritesASymlinkSwappedInForTheNote(t *testing.T) {
brain := t.TempDir()
facts := filepath.Join(brain, "wiki", "w", "facts")
require.NoError(t, os.MkdirAll(facts, 0o755))
victim := filepath.Join(brain, "raw", "victim.md")
require.NoError(t, os.MkdirAll(filepath.Dir(victim), 0o755))
require.NoError(t, os.WriteFile(victim, []byte("untouched\n"), 0o644))
note := filepath.Join(facts, "n.md")
body := "---\ntitle: n\n---\nold\n"
require.NoError(t, os.WriteFile(note, []byte(body), 0o644))
var stop atomic.Bool
var wg sync.WaitGroup
wg.Add(1)
go func() {
defer wg.Done()
tmp := note + ".swap"
for i := 0; !stop.Load(); i++ {
_ = os.Remove(tmp)
if i%2 == 0 {
_ = os.WriteFile(tmp, []byte(body), 0o644)
} else {
_ = os.Symlink(filepath.Join("..", "..", "..", "raw", "victim.md"), tmp)
}
_ = os.Rename(tmp, note)
}
}()
ok := 0
for deadline := time.Now().Add(1500 * time.Millisecond); time.Now().Before(deadline); {
if _, _, _, err := api.UpdateNote(brain, api.UpdateNoteOptions{Path: "wiki/w/facts/n.md", Content: "new"}); err == nil {
ok++
}
}
stop.Store(true)
wg.Wait()
got, _ := os.ReadFile(victim)
assert.Equal(t, "untouched\n", string(got))
assert.Positive(t, ok, "positive control: some updates reach the regular note")
}
@@ -0,0 +1,34 @@
package api_test
import (
"os"
"path/filepath"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/mathiasbq/hyperguild/ingestion/internal/api"
)
// hyperguild#178: an unreadable directory under wiki/ made the link check
// fail and brain_write refuse the note, losing the learning. The note is now
// written with its links as given.
func TestWriteNoteKeepsTheLearningWhenTheLinkCheckCannotIndex(t *testing.T) {
if os.Geteuid() == 0 {
t.Skip("root ignores directory permissions")
}
brain := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(brain, "wiki", "w", "facts"), 0o755))
require.NoError(t, os.WriteFile(filepath.Join(brain, "wiki", "w", "facts", "real.md"), []byte("x"), 0o644))
locked := filepath.Join(brain, "wiki", "other", "locked")
require.NoError(t, os.MkdirAll(locked, 0o755))
require.NoError(t, os.Chmod(locked, 0o000))
t.Cleanup(func() { _ = os.Chmod(locked, 0o755) })
rel, err := api.WriteNote(brain, api.WriteNoteOptions{Wing: "w", Hall: "facts", Filename: "n", Content: "learning [[real]] and [[gone]]"})
require.NoError(t, err)
got, err := os.ReadFile(filepath.Join(brain, filepath.FromSlash(rel)))
require.NoError(t, err)
assert.Contains(t, string(got), "[[gone]]", "links kept as given when the index is partial")
}
@@ -0,0 +1,43 @@
package api_test
import (
"os"
"path/filepath"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/mathiasbq/hyperguild/ingestion/internal/api"
)
// hyperguild#188: brain_write created a new note with os.MkdirAll +
// os.WriteFile by path, so a symlinked wing directory redirected the note
// elsewhere in the brain or outside it.
func TestWriteNoteRefusesASymlinkedWing(t *testing.T) {
for name, target := range map[string]func(t *testing.T, brain string) string{
"in-brain": func(t *testing.T, brain string) string {
require.NoError(t, os.MkdirAll(filepath.Join(brain, "raw"), 0o755))
return filepath.Join("..", "raw")
},
"outside": func(t *testing.T, _ string) string { return t.TempDir() },
} {
t.Run(name, func(t *testing.T) {
brain := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(brain, "wiki"), 0o755))
link := target(t, brain)
require.NoError(t, os.Symlink(link, filepath.Join(brain, "wiki", "lw")))
_, err := api.WriteNote(brain, api.WriteNoteOptions{Wing: "lw", Hall: "facts", Filename: "n", Content: "learning"})
assert.Error(t, err)
resolved := link
if !filepath.IsAbs(resolved) {
resolved = filepath.Join(brain, "wiki", link)
}
_, serr := os.Stat(filepath.Join(resolved, "facts"))
assert.True(t, os.IsNotExist(serr), "a hall was created at the link's target")
_, err = api.WriteNote(brain, api.WriteNoteOptions{Wing: "w", Hall: "facts", Filename: "n", Content: "learning"})
require.NoError(t, err, "positive control: a real wing is written")
})
}
}
+20 -1
View File
@@ -13,9 +13,14 @@ import (
type NtfyNotifier struct { type NtfyNotifier struct {
topicURL string topicURL string
token string token string
title string
http *http.Client http *http.Client
} }
// defaultNtfyTitle is the title of a notifier built without WithTitle: the
// capture audit's degraded-state alert, its first user.
const defaultNtfyTitle = "brain-capture audit degraded"
// NewNtfyNotifier constructs a notifier for the given ntfy topic URL // NewNtfyNotifier constructs a notifier for the given ntfy topic URL
// (e.g. https://ntfy.sh/my-topic). token is an optional bearer for // (e.g. https://ntfy.sh/my-topic). token is an optional bearer for
// protected ntfy instances; it is held here and only sent in the // protected ntfy instances; it is held here and only sent in the
@@ -27,17 +32,31 @@ func NewNtfyNotifier(topicURL, token string) *NtfyNotifier {
return &NtfyNotifier{ return &NtfyNotifier{
topicURL: strings.TrimRight(topicURL, "/"), topicURL: strings.TrimRight(topicURL, "/"),
token: token, token: token,
title: defaultNtfyTitle,
http: &http.Client{Timeout: 10 * time.Second}, http: &http.Client{Timeout: 10 * time.Second},
} }
} }
// WithTitle returns a copy that sends title as the ntfy Title header, so each
// alert source (the capture audit, the symlink tripwire, governance edits)
// is recognisable on the phone (hyperguild#163). A nil notifier (ntfy not
// configured) stays nil.
func (n *NtfyNotifier) WithTitle(title string) *NtfyNotifier {
if n == nil {
return nil
}
c := *n
c.title = title
return &c
}
// Notify posts a message to the ntfy topic. // Notify posts a message to the ntfy topic.
func (n *NtfyNotifier) Notify(ctx context.Context, msg string) error { func (n *NtfyNotifier) Notify(ctx context.Context, msg string) error {
req, err := http.NewRequestWithContext(ctx, http.MethodPost, n.topicURL, strings.NewReader(msg)) req, err := http.NewRequestWithContext(ctx, http.MethodPost, n.topicURL, strings.NewReader(msg))
if err != nil { if err != nil {
return err return err
} }
req.Header.Set("Title", "brain-capture audit degraded") req.Header.Set("Title", n.title)
req.Header.Set("Priority", "high") req.Header.Set("Priority", "high")
req.Header.Set("Tags", "warning,brain") req.Header.Set("Tags", "warning,brain")
if n.token != "" { if n.token != "" {
@@ -0,0 +1,29 @@
package audit
import (
"context"
"net/http"
"net/http/httptest"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
// hyperguild#163: every alert went out titled "brain-capture audit
// degraded", including the symlink tripwire's and governance alerts.
func TestNtfyNotifierSendsItsOwnTitle(t *testing.T) {
var got []string
srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
got = append(got, r.Header.Get("Title"))
}))
defer srv.Close()
base := NewNtfyNotifier(srv.URL, "")
require.NoError(t, base.Notify(context.Background(), "m"))
require.NoError(t, base.WithTitle("brain symlink tripwire").Notify(context.Background(), "m"))
require.NoError(t, base.Notify(context.Background(), "m"))
assert.Equal(t, []string{"brain-capture audit degraded", "brain symlink tripwire", "brain-capture audit degraded"}, got,
"WithTitle returns a copy; the base keeps its default title")
var nilN *NtfyNotifier
assert.Nil(t, nilN.WithTitle("x"), "nil (ntfy not configured) stays nil")
}
@@ -0,0 +1,125 @@
package brain_test
import (
"fmt"
"os"
"path/filepath"
"syscall"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/mathiasbq/hyperguild/ingestion/internal/brain"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs/brainfstest"
)
// hyperguild#174 / #157 slice 5b: DetectTunnels walked wiki/ and opened
// every note with a blocking os.Open, so a FIFO named *.md anywhere under
// wiki/ hung every brain_write and brain_update (wikiUpkeep -> AutoTunnel),
// and a symlinked note's title became a tunnel candidate.
func TestDetectTunnelsReturnsPromptlyWithAFIFOUnderWikiWithAndWithoutAWriter(t *testing.T) {
for _, withWriter := range []bool{false, true} {
t.Run(fmt.Sprintf("writer=%v", withWriter), func(t *testing.T) {
b := t.TempDir()
facts := filepath.Join(b, "wiki", "w", "facts")
require.NoError(t, os.MkdirAll(facts, 0o755))
require.NoError(t, os.WriteFile(filepath.Join(facts, "ok.md"), []byte("---\ntitle: Zqinside Phrase\n---\n"), 0o644))
fifo := filepath.Join(facts, "f.md")
require.NoError(t, syscall.Mkfifo(fifo, 0o644))
if withWriter {
// A writer held open: a reader that opens the FIFO gets no
// EOF and parks, unless the FIFO is never opened.
w, err := os.OpenFile(fifo, os.O_RDWR, 0)
require.NoError(t, err)
t.Cleanup(func() { _ = w.Close() })
}
done := make(chan []brain.TunnelCandidate, 1)
go func() {
cs, _ := brain.DetectTunnels(b, "text mentioning zqinside phrase here")
done <- cs
}()
select {
case cs := <-done:
assert.Contains(t, fmt.Sprintf("%+v", cs), "Zqinside Phrase", "positive control")
case <-time.After(3 * time.Second):
if r, err := os.OpenFile(fifo, os.O_WRONLY|syscall.O_NONBLOCK, 0); err == nil {
_ = r.Close()
}
t.Fatal("DetectTunnels blocked on a FIFO")
}
})
}
}
func TestDetectTunnelsRefusesSymlinkedNotesAndSurvivesSwapRaces(t *testing.T) {
b := t.TempDir()
facts := filepath.Join(b, "wiki", "w", "facts")
require.NoError(t, os.MkdirAll(facts, 0o755))
// The static note's title differs from the raced note's, so the
// harness's "inside was read" floor can only be met by reading r.md.
require.NoError(t, os.WriteFile(filepath.Join(facts, "ok.md"), []byte("---\ntitle: Zqstatic Words\n---\n"), 0o644))
require.NoError(t, os.Symlink(brainfstest.Outside(t), filepath.Join(facts, "out.md")))
inBrain := brainfstest.WriteSentinelAt(t, filepath.Join(b, "raw", "s.md"))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "..", "raw", "s.md"), filepath.Join(facts, "in.md")))
content := "zqstatic words, zqinside phrase and " + brainfstest.Sentinel
cs, err := brain.DetectTunnels(b, content)
require.NoError(t, err)
got := fmt.Sprintf("%+v", cs)
assert.Contains(t, got, "Zqstatic Words", "positive control")
assert.NotContains(t, got, brainfstest.Sentinel)
p := filepath.Join(facts, "r.md")
read := func() string {
cs, _ := brain.DetectTunnels(b, content)
return fmt.Sprintf("%+v", cs)
}
brainfstest.AssertNoLeak(t, p, "---\ntitle: Zqinside Phrase\n---\n", 1500*time.Millisecond, read)
brainfstest.AssertNoLeakFrom(t, p, "---\ntitle: Zqinside Phrase\n---\n", inBrain, 1500*time.Millisecond, read)
}
// A wiki/ that is not a real directory yields no candidates and no error:
// AutoTunnel runs inside brain_write.
func TestDetectTunnelsSkipsAWikiThatIsNotARealDirectory(t *testing.T) {
b := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(b, "raw", "w", "facts"), 0o755))
require.NoError(t, os.WriteFile(filepath.Join(b, "raw", "w", "facts", "r.md"), []byte("---\ntitle: Zqrawonly Phrase\n---\n"), 0o644))
require.NoError(t, os.Symlink("raw", filepath.Join(b, "wiki")))
cs, err := brain.DetectTunnels(b, "zqrawonly phrase")
require.NoError(t, err)
assert.Empty(t, cs)
}
// An unreadable directory under wiki/ is a walk error and DetectTunnels
// reports it, as filepath.WalkDir did; it is not silently dropped.
func TestDetectTunnelsReportsAnUnreadableDirectoryUnderWiki(t *testing.T) {
if os.Geteuid() == 0 {
t.Skip("root ignores directory permissions")
}
b := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(b, "wiki", "w", "facts"), 0o755))
require.NoError(t, os.WriteFile(filepath.Join(b, "wiki", "w", "facts", "ok.md"), []byte("---\ntitle: Zqinside Phrase\n---\n"), 0o644))
locked := filepath.Join(b, "wiki", "other", "locked")
require.NoError(t, os.MkdirAll(locked, 0o755))
require.NoError(t, os.Chmod(locked, 0o000))
t.Cleanup(func() { _ = os.Chmod(locked, 0o755) })
_, err := brain.DetectTunnels(b, "zqinside phrase")
assert.ErrorIs(t, err, os.ErrPermission)
}
// _index.md files are generated, not notes: their title never becomes a
// candidate.
func TestDetectTunnelsSkipsIndexFiles(t *testing.T) {
b := t.TempDir()
w := filepath.Join(b, "wiki", "w")
require.NoError(t, os.MkdirAll(filepath.Join(w, "facts"), 0o755))
require.NoError(t, os.WriteFile(filepath.Join(w, "_index.md"), []byte("---\ntitle: Zqindex Words\n---\n"), 0o644))
require.NoError(t, os.WriteFile(filepath.Join(w, "facts", "ok.md"), []byte("---\ntitle: Zqinside Phrase\n---\n"), 0o644))
cs, err := brain.DetectTunnels(b, "zqindex words and zqinside phrase")
require.NoError(t, err)
got := fmt.Sprintf("%+v", cs)
assert.Contains(t, got, "Zqinside Phrase", "positive control")
assert.NotContains(t, got, "Zqindex")
}
+3 -22
View File
@@ -7,8 +7,6 @@ import (
"fmt" "fmt"
"io" "io"
"io/fs" "io/fs"
"os"
"path/filepath"
"sort" "sort"
"strings" "strings"
"time" "time"
@@ -32,7 +30,6 @@ func BuildWingIndex(brainDir, wing string) error {
if w == "" { if w == "" {
return fmt.Errorf("invalid wing %q", wing) return fmt.Errorf("invalid wing %q", wing)
} }
wingDir := filepath.Join(brainDir, "wiki", w)
// Listed and read through the brain root (hyperguild#157): a wing or // Listed and read through the brain root (hyperguild#157): a wing or
// hall that is a symlink is not followed, and notes are regular files. // hall that is a symlink is not followed, and notes are regular files.
bfs, err := brainfs.Open(brainDir) bfs, err := brainfs.Open(brainDir)
@@ -63,8 +60,9 @@ func BuildWingIndex(brainDir, wing string) error {
fmt.Fprintf(&b, "| %s | [%s](%s/%s.md) | %s |\n", e.Hall, e.Title, e.Hall, e.Slug, e.Created) fmt.Fprintf(&b, "| %s | [%s](%s/%s.md) | %s |\n", e.Hall, e.Title, e.Hall, e.Slug, e.Created)
} }
dest := filepath.Join(wingDir, "_index.md") // Written through the brain root, never through a symlinked _index.md
return os.WriteFile(dest, []byte(b.String()), 0o644) // (hyperguild#158).
return bfs.PutRegular("wiki/"+w+"/_index.md", []byte(b.String()))
} }
// BuildAllWingIndexes regenerates _index.md for every wing under brain/wiki/. // BuildAllWingIndexes regenerates _index.md for every wing under brain/wiki/.
@@ -116,23 +114,6 @@ func collectWingEntries(bfs *brainfs.Brain, wingRel string) ([]noteEntry, error)
return out, nil return out, nil
} }
// readTitleAndCreated reads YAML frontmatter for title + created_at; falls
// back to slug and file mtime when absent.
func readTitleAndCreated(path, slug string) (string, string) {
f, err := os.Open(path)
if err != nil {
return slug, ""
}
defer func() { _ = f.Close() }()
title, created := parseTitleAndCreated(f)
if created == "" {
if info, err := os.Stat(path); err == nil {
created = info.ModTime().UTC().Format("2006-01-02")
}
}
return titleOrSlug(title, slug), created
}
// readTitleAndCreatedIn is readTitleAndCreated through the brain root, // readTitleAndCreatedIn is readTitleAndCreated through the brain root,
// regular files only (hyperguild#157). // regular files only (hyperguild#157).
func readTitleAndCreatedIn(bfs *brainfs.Brain, rel, slug string) (string, string) { func readTitleAndCreatedIn(bfs *brainfs.Brain, rel, slug string) (string, string) {
+6 -1
View File
@@ -47,7 +47,12 @@ var (
func StripUnresolvedLinks(brainDir, content string) (string, []string, error) { func StripUnresolvedLinks(brainDir, content string) (string, []string, error) {
index, err := entryIndex(brainDir) index, err := entryIndex(brainDir)
if err != nil { if err != nil {
return "", nil, err // A partial index cannot tell an invented link from one into the
// part it could not read, so unwrap nothing, and do not refuse the
// write either: this runs before every brain_write, and a refused
// write is a lost learning (hyperguild#178).
slog.Warn("link check skipped: cannot index the brain", "err", err)
return content, nil, nil
} }
if len(index) == 0 { if len(index) == 0 {
return content, nil, nil return content, nil, nil
@@ -0,0 +1,41 @@
package brain_test
import (
"os"
"path/filepath"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/mathiasbq/hyperguild/ingestion/internal/brain"
)
// hyperguild#178: StripUnresolvedLinks runs before every brain_write and
// brain_update. An unreadable directory under knowledge/ or wiki/ made it
// error, the write was refused and the learning lost. When the link index
// cannot be built, the content is now returned unchanged (no link is
// unwrapped, none is invented) and no error stops the write.
func TestStripUnresolvedLinksKeepsContentWhenTheIndexCannotBeBuilt(t *testing.T) {
if os.Geteuid() == 0 {
t.Skip("root ignores directory permissions")
}
b := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(b, "wiki", "w", "facts"), 0o755))
require.NoError(t, os.WriteFile(filepath.Join(b, "wiki", "w", "facts", "real.md"), []byte("x"), 0o644))
locked := filepath.Join(b, "wiki", "other", "locked")
require.NoError(t, os.MkdirAll(locked, 0o755))
require.NoError(t, os.Chmod(locked, 0o000))
t.Cleanup(func() { _ = os.Chmod(locked, 0o755) })
in := "learning [[real]] and [[gone]]"
out, unlinked, err := brain.StripUnresolvedLinks(b, in)
require.NoError(t, err)
assert.Equal(t, in, out, "content unchanged: no link unwrapped on a partial index")
assert.Empty(t, unlinked)
require.NoError(t, os.Chmod(locked, 0o755))
out, _, err = brain.StripUnresolvedLinks(b, in)
require.NoError(t, err)
assert.NotContains(t, out, "[[gone]]", "positive control: with a readable tree the strip runs")
}
+96 -60
View File
@@ -2,11 +2,15 @@ package brain
import ( import (
"bufio" "bufio"
"errors"
"fmt" "fmt"
"os" "io/fs"
"path"
"path/filepath" "path/filepath"
"strings" "strings"
"time" "time"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
) )
// seeAlsoHeader is the markdown heading used to group cross-wing links. // seeAlsoHeader is the markdown heading used to group cross-wing links.
@@ -36,24 +40,29 @@ type TunnelCandidate struct {
// A note's title is read from YAML frontmatter `title:`; failing that, // A note's title is read from YAML frontmatter `title:`; failing that,
// the filename slug (sans `.md`, hyphens → spaces) is used. // the filename slug (sans `.md`, hyphens → spaces) is used.
func DetectTunnels(brainDir, content string) ([]TunnelCandidate, error) { func DetectTunnels(brainDir, content string) ([]TunnelCandidate, error) {
wikiDir := filepath.Join(brainDir, "wiki") // Walked and read through the brain root, regular files only
if _, err := os.Stat(wikiDir); os.IsNotExist(err) { // (hyperguild#174, #157): this runs inside every brain_write, so a FIFO
// must not block it and a symlinked note's title must not become a
// candidate.
bfs, err := brainfs.Open(brainDir)
if err != nil {
return nil, fmt.Errorf("open brain: %w", err)
}
defer func() { _ = bfs.Close() }()
if _, err := bfs.Lstat("wiki"); errors.Is(err, fs.ErrNotExist) {
return nil, nil return nil, nil
} else if err != nil {
return nil, fmt.Errorf("stat wiki: %w", err)
} }
lowerContent := strings.ToLower(content) lowerContent := strings.ToLower(content)
var out []TunnelCandidate var out []TunnelCandidate
err := filepath.WalkDir(wikiDir, func(path string, d os.DirEntry, err error) error { var walkErr error
if err != nil { err = bfs.WalkRegular("wiki", func(rel string) error {
return err name := path.Base(rel)
} if !strings.HasSuffix(name, ".md") || name == "_index.md" {
if d.IsDir() || !strings.HasSuffix(path, ".md") || d.Name() == "_index.md" {
return nil return nil
} }
title, _ := readTitleAndCreated(path, strings.TrimSuffix(d.Name(), ".md")) title, _ := readTitleAndCreatedIn(bfs, rel, strings.TrimSuffix(name, ".md"))
title = strings.TrimSpace(title) title = strings.TrimSpace(title)
if title == "" { if title == "" {
return nil return nil
@@ -75,17 +84,25 @@ func DetectTunnels(brainDir, content string) ([]TunnelCandidate, error) {
if idx == -1 { if idx == -1 {
return nil return nil
} }
rel, err := filepath.Rel(brainDir, path)
if err != nil {
return err
}
out = append(out, TunnelCandidate{ out = append(out, TunnelCandidate{
TargetPath: filepath.ToSlash(rel), TargetPath: rel,
MatchedTerm: title, MatchedTerm: title,
Exact: isWholeWord(haystack, idx, len(needle)), Exact: isWholeWord(haystack, idx, len(needle)),
}) })
return nil return nil
}, func(rel string, err error) {
// A wiki/ that is not a real directory yields no candidates, as
// filepath.WalkDir did; other walk errors still abort.
if rel == "wiki" && errors.Is(err, brainfs.ErrNotRegular) {
return
}
if walkErr == nil {
walkErr = fmt.Errorf("%s: %w", rel, err)
}
}) })
if err == nil {
err = walkErr
}
if err != nil { if err != nil {
return nil, err return nil, err
} }
@@ -158,35 +175,23 @@ func logFuzzyCandidates(brainDir, sourcePath string, cs []TunnelCandidate) error
if len(cs) == 0 { if len(cs) == 0 {
return nil return nil
} }
rawDir := filepath.Join(brainDir, "raw") // Appended through the brain root, never through a symlinked raw/ or
if err := os.MkdirAll(rawDir, 0o755); err != nil { // log file (hyperguild#158).
return err bfs, err := brainfs.Open(brainDir)
}
stamp := time.Now().UTC().Format("2006-01-02")
path := filepath.Join(rawDir, "tunnel-candidates-"+stamp+".md")
existed := fileExists(path)
f, err := os.OpenFile(path, os.O_APPEND|os.O_CREATE|os.O_WRONLY, 0o644)
if err != nil { if err != nil {
return err return fmt.Errorf("open brain: %w", err)
} }
defer func() { _ = f.Close() }() defer func() { _ = bfs.Close() }()
if !existed { stamp := time.Now().UTC().Format("2006-01-02")
if _, err := f.WriteString("# Tunnel candidates " + stamp + "\n\nFuzzy cross-wing matches surfaced by AutoTunnel. Review and promote to a tunnel via `brain_tunnel` if relevant.\n\n"); err != nil { rel := "raw/tunnel-candidates-" + stamp + ".md"
return err var buf strings.Builder
} if _, err := bfs.Lstat(rel); errors.Is(err, fs.ErrNotExist) {
buf.WriteString("# Tunnel candidates " + stamp + "\n\nFuzzy cross-wing matches surfaced by AutoTunnel. Review and promote to a tunnel via `brain_tunnel` if relevant.\n\n")
} }
for _, c := range cs { for _, c := range cs {
line := fmt.Sprintf("- `%s` ↔ `%s` (term: %q)\n", sourcePath, c.TargetPath, c.MatchedTerm) fmt.Fprintf(&buf, "- `%s` ↔ `%s` (term: %q)\n", sourcePath, c.TargetPath, c.MatchedTerm)
if _, err := f.WriteString(line); err != nil {
return err
}
} }
return nil return bfs.AppendRegular(rel, []byte(buf.String()))
}
func fileExists(p string) bool {
_, err := os.Stat(p)
return err == nil
} }
// WriteTunnel appends a bidirectional wikilink between sourcePath and // WriteTunnel appends a bidirectional wikilink between sourcePath and
@@ -197,6 +202,12 @@ func fileExists(p string) bool {
// section headers. Rejects same-wing pairs (a tunnel is by definition // section headers. Rejects same-wing pairs (a tunnel is by definition
// cross-wing) and missing notes. // cross-wing) and missing notes.
func WriteTunnel(brainDir, sourcePath, targetPath string) error { func WriteTunnel(brainDir, sourcePath, targetPath string) error {
if err := checkNotePath(sourcePath); err != nil {
return fmt.Errorf("source: %w", err)
}
if err := checkNotePath(targetPath); err != nil {
return fmt.Errorf("target: %w", err)
}
srcWing, err := wingOf(sourcePath) srcWing, err := wingOf(sourcePath)
if err != nil { if err != nil {
return fmt.Errorf("source: %w", err) return fmt.Errorf("source: %w", err)
@@ -209,20 +220,50 @@ func WriteTunnel(brainDir, sourcePath, targetPath string) error {
return fmt.Errorf("tunnel must cross wings; got both in %q", srcWing) return fmt.Errorf("tunnel must cross wings; got both in %q", srcWing)
} }
srcFull := filepath.Join(brainDir, filepath.FromSlash(sourcePath)) // Both notes are read and rewritten through the brain root, regular
tgtFull := filepath.Join(brainDir, filepath.FromSlash(targetPath)) // files only: brain_tunnel passes caller-supplied paths, and the brain is
if _, err := os.Stat(srcFull); err != nil { // a Syncthing folder that can hold symlinks.
bfs, err := brainfs.Open(brainDir)
if err != nil {
return fmt.Errorf("open brain: %w", err)
}
defer func() { _ = bfs.Close() }()
srcContent, err := bfs.ReadRegular(sourcePath)
if err != nil {
return fmt.Errorf("source note: %w", err) return fmt.Errorf("source note: %w", err)
} }
if _, err := os.Stat(tgtFull); err != nil { tgtContent, err := bfs.ReadRegular(targetPath)
if err != nil {
return fmt.Errorf("target note: %w", err) return fmt.Errorf("target note: %w", err)
} }
if err := appendSeeAlso(srcFull, wikilinkOf(targetPath)); err != nil { if out, changed, err := withSeeAlso(srcContent, wikilinkOf(targetPath)); err != nil {
return fmt.Errorf("update source: %w", err) return fmt.Errorf("update source: %w", err)
} else if changed {
if err := bfs.WriteRegular(sourcePath, out); err != nil {
return fmt.Errorf("update source: %w", err)
}
} }
if err := appendSeeAlso(tgtFull, wikilinkOf(sourcePath)); err != nil { if out, changed, err := withSeeAlso(tgtContent, wikilinkOf(sourcePath)); err != nil {
return fmt.Errorf("update target: %w", err) return fmt.Errorf("update target: %w", err)
} else if changed {
if err := bfs.WriteRegular(targetPath, out); err != nil {
return fmt.Errorf("update target: %w", err)
}
}
return nil
}
// checkNotePath accepts only a clean, local wiki/<wing>/<hall>/.../<slug>.md
// path: no "." or ".." segments, no empty segments, no leading slash.
func checkNotePath(p string) error {
if p == "" || path.Clean(p) != p || !filepath.IsLocal(p) || !strings.HasSuffix(p, ".md") {
return fmt.Errorf("not a clean wiki note path: %q", p)
}
for _, part := range strings.Split(p, "/") {
if part == "" || part == "." || part == ".." {
return fmt.Errorf("not a clean wiki note path: %q", p)
}
} }
return nil return nil
} }
@@ -248,16 +289,13 @@ func wingOf(relPath string) (string, error) {
return parts[1], nil return parts[1], nil
} }
// appendSeeAlso inserts `- [[link]]` under the file's See also section, // withSeeAlso returns content with `- [[link]]` under its See also section,
// creating the section if absent. No-op when the link is already present. // creating the section if absent. changed is false when the link is already
func appendSeeAlso(filePath, link string) error { // present.
content, err := os.ReadFile(filePath) func withSeeAlso(content []byte, link string) (out []byte, changed bool, err error) {
if err != nil {
return err
}
wikilink := "[[" + link + "]]" wikilink := "[[" + link + "]]"
if strings.Contains(string(content), wikilink) { if strings.Contains(string(content), wikilink) {
return nil return content, false, nil
} }
bullet := "- " + wikilink bullet := "- " + wikilink
@@ -266,8 +304,7 @@ func appendSeeAlso(filePath, link string) error {
// No section yet — append a fresh one. Always emit a trailing // No section yet — append a fresh one. Always emit a trailing
// newline so subsequent appends don't merge into the previous line. // newline so subsequent appends don't merge into the previous line.
trimmed := strings.TrimRight(string(content), "\n") trimmed := strings.TrimRight(string(content), "\n")
out := trimmed + "\n\n" + seeAlsoHeader + "\n\n" + bullet + "\n" return []byte(trimmed + "\n\n" + seeAlsoHeader + "\n\n" + bullet + "\n"), true, nil
return os.WriteFile(filePath, []byte(out), 0o644)
} }
// Section exists — splice the bullet in just before the next `## ` // Section exists — splice the bullet in just before the next `## `
@@ -293,12 +330,11 @@ func appendSeeAlso(filePath, link string) error {
b.WriteByte('\n') b.WriteByte('\n')
} }
if err := scanner.Err(); err != nil { if err := scanner.Err(); err != nil {
return err return nil, false, err
} }
if !inserted { if !inserted {
// section was the last thing in the file — just append bullet // section was the last thing in the file — just append bullet
out := strings.TrimRight(b.String(), "\n") + "\n" + bullet + "\n" return []byte(strings.TrimRight(b.String(), "\n") + "\n" + bullet + "\n"), true, nil
return os.WriteFile(filePath, []byte(out), 0o644)
} }
return os.WriteFile(filePath, []byte(b.String()), 0o644) return []byte(b.String()), true, nil
} }
@@ -0,0 +1,38 @@
package brain_test
import (
"os"
"path/filepath"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/mathiasbq/hyperguild/ingestion/internal/brain"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
)
// hyperguild#173: brain_tunnel's write followed a directory symlink inside
// wiki/ that points elsewhere in the brain.
func TestWriteTunnelRefusesATargetBehindASymlinkedDirectory(t *testing.T) {
b := t.TempDir()
src := filepath.Join(b, "wiki", "a", "facts", "s.md")
require.NoError(t, os.MkdirAll(filepath.Dir(src), 0o755))
require.NoError(t, os.WriteFile(src, []byte("x\n"), 0o644))
victim := filepath.Join(b, "raw", "facts", "v.md")
require.NoError(t, os.MkdirAll(filepath.Dir(victim), 0o755))
require.NoError(t, os.WriteFile(victim, []byte("untouched\n"), 0o644))
require.NoError(t, os.MkdirAll(filepath.Join(b, "wiki", "b"), 0o755))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "raw", "facts"), filepath.Join(b, "wiki", "b", "facts")))
assert.ErrorIs(t, brain.WriteTunnel(b, "wiki/a/facts/s.md", "wiki/b/facts/v.md"), brainfs.ErrNotRegular)
got, _ := os.ReadFile(victim)
assert.Equal(t, "untouched\n", string(got))
got, _ = os.ReadFile(src)
assert.Equal(t, "x\n", string(got), "the source is not rewritten either")
real := filepath.Join(b, "wiki", "c", "facts", "t.md")
require.NoError(t, os.MkdirAll(filepath.Dir(real), 0o755))
require.NoError(t, os.WriteFile(real, []byte("t\n"), 0o644))
require.NoError(t, brain.WriteTunnel(b, "wiki/a/facts/s.md", "wiki/c/facts/t.md"), "positive control: a tunnel between real directories is written")
}
@@ -0,0 +1,141 @@
package brain_test
import (
"os"
"path/filepath"
"strings"
"sync"
"sync/atomic"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/mathiasbq/hyperguild/ingestion/internal/brain"
)
// brain_tunnel passes caller-supplied source and target paths to WriteTunnel,
// which checked only "wiki/<wing>/..." and then joined them onto the brain
// dir: ".." segments resolved outside the brain and appendSeeAlso rewrote the
// file there.
func TestWriteTunnelRefusesATargetThatTraversesOutOfTheBrain(t *testing.T) {
brain0 := t.TempDir()
src := filepath.Join(brain0, "wiki", "a", "facts", "s.md")
require.NoError(t, os.MkdirAll(filepath.Dir(src), 0o755))
require.NoError(t, os.WriteFile(src, []byte("---\ntitle: S\n---\nbody\n"), 0o644))
victim := filepath.Join(t.TempDir(), "victim.md")
require.NoError(t, os.WriteFile(victim, []byte("untouched\n"), 0o644))
up := strings.Repeat("../", strings.Count(filepath.Clean(brain0), "/")+3)
target := "wiki/b/x/" + up + strings.TrimPrefix(victim, "/")
err := brain.WriteTunnel(brain0, "wiki/a/facts/s.md", target)
assert.Error(t, err)
got, rerr := os.ReadFile(victim)
require.NoError(t, rerr)
assert.Equal(t, "untouched\n", string(got), "a file outside the brain was modified")
}
func TestWriteTunnelRefusesUncleanOrNonWikiPaths(t *testing.T) {
brain0 := t.TempDir()
for _, p := range []string{"wiki/a/facts/s.md", "wiki/b/facts/t.md", "wiki/b/facts/t.txt", "raw/r.md"} {
require.NoError(t, os.MkdirAll(filepath.Dir(filepath.Join(brain0, p)), 0o755))
require.NoError(t, os.WriteFile(filepath.Join(brain0, p), []byte("x\n"), 0o644))
}
for _, target := range []string{
"wiki/b/facts/../../../raw/r.md",
"wiki/b/./facts/t.md",
"wiki/b//facts/t.md",
"/wiki/b/facts/t.md",
"wiki/b/facts/t.txt",
"wiki/b/facts/",
} {
assert.Error(t, brain.WriteTunnel(brain0, "wiki/a/facts/s.md", target), target)
}
for _, p := range []string{"raw/r.md", "wiki/b/facts/t.txt"} {
got, err := os.ReadFile(filepath.Join(brain0, p))
require.NoError(t, err)
assert.Equal(t, "x\n", string(got), p)
}
require.NoError(t, brain.WriteTunnel(brain0, "wiki/a/facts/s.md", "wiki/b/facts/t.md"), "positive control")
got, err := os.ReadFile(filepath.Join(brain0, "wiki", "b", "facts", "t.md"))
require.NoError(t, err)
assert.Contains(t, string(got), "[[a/facts/s]]")
}
func TestWriteTunnelDoesNotWriteThroughASymlinkedNote(t *testing.T) {
brain0 := t.TempDir()
src := filepath.Join(brain0, "wiki", "a", "facts", "s.md")
require.NoError(t, os.MkdirAll(filepath.Dir(src), 0o755))
require.NoError(t, os.WriteFile(src, []byte("x\n"), 0o644))
tdir := filepath.Join(brain0, "wiki", "b", "facts")
require.NoError(t, os.MkdirAll(tdir, 0o755))
victim := filepath.Join(t.TempDir(), "victim.md")
require.NoError(t, os.WriteFile(victim, []byte("untouched\n"), 0o644))
require.NoError(t, os.Symlink(victim, filepath.Join(tdir, "t.md")))
assert.Error(t, brain.WriteTunnel(brain0, "wiki/a/facts/s.md", "wiki/b/facts/t.md"))
got, err := os.ReadFile(victim)
require.NoError(t, err)
assert.Equal(t, "untouched\n", string(got))
// The same with a link to a file elsewhere in the brain, which os.Root
// would follow.
inVictim := filepath.Join(brain0, "raw", "v.md")
require.NoError(t, os.MkdirAll(filepath.Dir(inVictim), 0o755))
require.NoError(t, os.WriteFile(inVictim, []byte("untouched\n"), 0o644))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "..", "raw", "v.md"), filepath.Join(tdir, "u.md")))
assert.Error(t, brain.WriteTunnel(brain0, "wiki/a/facts/s.md", "wiki/b/facts/u.md"))
got, err = os.ReadFile(inVictim)
require.NoError(t, err)
assert.Equal(t, "untouched\n", string(got))
got, err = os.ReadFile(src)
require.NoError(t, err)
assert.Equal(t, "x\n", string(got), "the source is not touched when the target is refused")
}
// The target is swapped between a regular note and a symlink to an in-brain
// file outside wiki/ while WriteTunnel runs: the read and the write each
// refuse a symlink, so the other file is never rewritten.
func TestWriteTunnelNeverRewritesASymlinkSwappedInForTheTarget(t *testing.T) {
brain0 := t.TempDir()
src := filepath.Join(brain0, "wiki", "a", "facts", "s.md")
require.NoError(t, os.MkdirAll(filepath.Dir(src), 0o755))
require.NoError(t, os.WriteFile(src, []byte("x\n"), 0o644))
tdir := filepath.Join(brain0, "wiki", "b", "facts")
require.NoError(t, os.MkdirAll(tdir, 0o755))
victim := filepath.Join(brain0, "raw", "victim.md")
require.NoError(t, os.MkdirAll(filepath.Dir(victim), 0o755))
require.NoError(t, os.WriteFile(victim, []byte("untouched\n"), 0o644))
tgt := filepath.Join(tdir, "t.md")
require.NoError(t, os.WriteFile(tgt, []byte("t\n"), 0o644))
var stop atomic.Bool
var wg sync.WaitGroup
wg.Add(1)
go func() {
defer wg.Done()
tmp := tgt + ".swap"
for i := 0; !stop.Load(); i++ {
_ = os.Remove(tmp)
if i%2 == 0 {
_ = os.WriteFile(tmp, []byte("t\n"), 0o644)
} else {
_ = os.Symlink(filepath.Join("..", "..", "..", "raw", "victim.md"), tmp)
}
_ = os.Rename(tmp, tgt)
}
}()
ok := 0
for deadline := time.Now().Add(1500 * time.Millisecond); time.Now().Before(deadline); {
if brain.WriteTunnel(brain0, "wiki/a/facts/s.md", "wiki/b/facts/t.md") == nil {
ok++
}
_ = os.WriteFile(src, []byte("x\n"), 0o644)
}
stop.Store(true)
wg.Wait()
got, err := os.ReadFile(victim)
require.NoError(t, err)
assert.Equal(t, "untouched\n", string(got), "the swapped-in symlink's target was rewritten")
assert.Positive(t, ok, "positive control: some tunnels were written to the regular target")
}
@@ -0,0 +1,52 @@
package brain_test
import (
"os"
"path/filepath"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/mathiasbq/hyperguild/ingestion/internal/brain"
)
// hyperguild#158: BuildWingIndex wrote wiki/<wing>/_index.md with
// os.WriteFile, so a symlinked _index.md was written through.
func TestBuildWingIndexDoesNotWriteThroughASymlinkedIndex(t *testing.T) {
b := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(b, "wiki", "w", "facts"), 0o755))
require.NoError(t, os.WriteFile(filepath.Join(b, "wiki", "w", "facts", "n.md"), []byte("---\ntitle: zqn\n---\n"), 0o644))
victim := filepath.Join(b, "raw", "v.md")
require.NoError(t, os.MkdirAll(filepath.Dir(victim), 0o755))
require.NoError(t, os.WriteFile(victim, []byte("untouched"), 0o644))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "raw", "v.md"), filepath.Join(b, "wiki", "w", "_index.md")))
assert.Error(t, brain.BuildWingIndex(b, "w"))
got, _ := os.ReadFile(victim)
assert.Equal(t, "untouched", string(got))
require.NoError(t, os.Remove(filepath.Join(b, "wiki", "w", "_index.md")))
require.NoError(t, brain.BuildWingIndex(b, "w"), "positive control")
idx, _ := os.ReadFile(filepath.Join(b, "wiki", "w", "_index.md"))
assert.Contains(t, string(idx), "zqn")
}
// AutoTunnel's fuzzy-candidate log appended to raw/tunnel-candidates-*.md
// with os.OpenFile, so a symlinked raw/ (or log file) was written through.
func TestAutoTunnelDoesNotAppendTheCandidateLogThroughASymlinkedRaw(t *testing.T) {
b := t.TempDir()
outside := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(b, "wiki", "a", "facts"), 0o755))
require.NoError(t, os.MkdirAll(filepath.Join(b, "wiki", "b", "facts"), 0o755))
// "zq widget" matches as a substring of "zq widgets", a fuzzy candidate.
require.NoError(t, os.WriteFile(filepath.Join(b, "wiki", "b", "facts", "t.md"), []byte("---\ntitle: Zq Widget\n---\n"), 0o644))
require.NoError(t, os.Symlink(outside, filepath.Join(b, "raw")))
_ = brain.AutoTunnel(b, "wiki/a/facts/s.md", "about zq widgets")
ents, _ := os.ReadDir(outside)
assert.Empty(t, ents, "the candidate log was written outside the brain")
require.NoError(t, os.Remove(filepath.Join(b, "raw")))
require.NoError(t, brain.AutoTunnel(b, "wiki/a/facts/s.md", "about zq widgets"))
logs, _ := filepath.Glob(filepath.Join(b, "raw", "tunnel-candidates-*.md"))
assert.Len(t, logs, 1, "positive control: the fuzzy candidate is logged in the real raw/")
}
+205
View File
@@ -104,6 +104,17 @@ func (b *Brain) ReadDirRegular(relDir string) ([]string, error) {
return names, nil return names, nil
} }
// cleanRel refuses a path that is not clean and local. realDir checks the
// cleaned parent, so the I/O must use that same path: os.Root resolves ".."
// after following a symlink, and a ".." after a symlinked directory would
// otherwise pass the check and redirect the read or write.
func cleanRel(rel string) error {
if rel == "" || path.Clean(rel) != rel || !filepath.IsLocal(rel) || strings.Contains(rel, "\\") {
return fmt.Errorf("%q: not a clean brain-relative path: %w", rel, ErrNotRegular)
}
return nil
}
// ReadDirDirs lists the real subdirectories directly in a brain-relative // ReadDirDirs lists the real subdirectories directly in a brain-relative
// directory, by name. Every component of relDir must be a real directory // directory, by name. Every component of relDir must be a real directory
// (ErrNotRegular otherwise); symlinks to directories are left out. // (ErrNotRegular otherwise); symlinks to directories are left out.
@@ -125,9 +136,149 @@ func (b *Brain) ReadDirDirs(relDir string) ([]string, error) {
return names, nil return names, nil
} }
// PutRegular creates or replaces a regular file at a brain-relative path,
// for creating writes (hyperguild#188, #158). Missing parent directories are
// created one component at a time through the root, and every existing
// component must be a real directory, so a symlinked directory cannot
// redirect the write elsewhere in the brain or outside it. An existing file
// is replaced only if it is regular (WriteRegular's checks); a new file is
// created with O_EXCL and confirmed to be the file that was opened. A
// directory swapped in after the checks is the #167 residual.
func (b *Brain) PutRegular(rel string, data []byte) error {
if err := cleanRel(rel); err != nil {
return err
}
if err := b.mkdirReal(path.Dir(rel)); err != nil {
return err
}
if _, err := b.root.Lstat(rel); err == nil {
return b.WriteRegular(rel, data)
} else if !errors.Is(err, fs.ErrNotExist) {
return err
}
return b.createNew(rel, data)
}
// CreateRegular creates a new regular file at a brain-relative path, with
// PutRegular's parent rules, and fails with fs.ErrExist if anything is
// already there: for writers that must never overwrite (promote, reject).
func (b *Brain) CreateRegular(rel string, data []byte) error {
if err := cleanRel(rel); err != nil {
return err
}
if err := b.mkdirReal(path.Dir(rel)); err != nil {
return err
}
return b.createNew(rel, data)
}
// createNew creates rel with O_EXCL and writes data only after confirming
// the path is the regular file it created.
func (b *Brain) createNew(rel string, data []byte) error {
f, err := b.root.OpenFile(rel, os.O_WRONLY|os.O_CREATE|os.O_EXCL|syscall.O_NOFOLLOW|syscall.O_NONBLOCK, 0o644)
if err != nil {
return err
}
defer func() { _ = f.Close() }()
hfi, herr := f.Stat()
lfi, lerr := b.root.Lstat(rel)
if herr != nil || lerr != nil || !lfi.Mode().IsRegular() || !os.SameFile(lfi, hfi) {
return fmt.Errorf("%s: changed while creating: %w", rel, ErrNotRegular)
}
if _, err := f.Write(data); err != nil {
return err
}
return f.Close()
}
// AppendRegular appends data to a regular file at a brain-relative path,
// creating it (and missing parent directories, as PutRegular does) if it is
// absent. It writes only after confirming the path is the regular file it
// opened, so a final symlink or special file, or one swapped in, is refused.
func (b *Brain) AppendRegular(rel string, data []byte) error {
if err := cleanRel(rel); err != nil {
return err
}
if err := b.mkdirReal(path.Dir(rel)); err != nil {
return err
}
if fi, err := b.root.Lstat(rel); err == nil && !fi.Mode().IsRegular() {
return fmt.Errorf("%s: %w", rel, ErrNotRegular)
} else if err != nil && !errors.Is(err, fs.ErrNotExist) {
return err
}
f, err := b.root.OpenFile(rel, os.O_WRONLY|os.O_APPEND|os.O_CREATE|syscall.O_NOFOLLOW|syscall.O_NONBLOCK, 0o644)
if err != nil {
return err
}
defer func() { _ = f.Close() }()
hfi, herr := f.Stat()
lfi, lerr := b.root.Lstat(rel)
if herr != nil || lerr != nil || !lfi.Mode().IsRegular() || !os.SameFile(lfi, hfi) {
return fmt.Errorf("%s: changed while opening: %w", rel, ErrNotRegular)
}
if _, err := f.Write(data); err != nil {
return err
}
return f.Close()
}
// Remove removes the file at a clean brain-relative path through the root,
// refusing a path behind a symlinked directory. A final symlink is removed
// itself, never its target.
func (b *Brain) Remove(rel string) error {
if err := b.RealParent(rel); err != nil {
return err
}
return b.root.Remove(rel)
}
// RealParent returns ErrNotRegular unless rel is clean and every directory
// above it is a real directory, for writers that do their own I/O (an atomic
// rename) but must not be redirected by a symlinked directory.
func (b *Brain) RealParent(rel string) error {
if err := cleanRel(rel); err != nil {
return err
}
return b.realDir(path.Dir(rel))
}
// mkdirReal makes sure every component of rel exists as a real directory,
// creating missing ones through the root; an existing component that is
// not a real directory is ErrNotRegular.
func (b *Brain) mkdirReal(rel string) error {
if rel == "." {
return nil
}
prefix := ""
for _, part := range strings.Split(rel, "/") {
prefix = path.Join(prefix, part)
fi, err := b.root.Lstat(prefix)
if errors.Is(err, fs.ErrNotExist) {
if err := b.root.Mkdir(prefix, 0o755); err != nil && !errors.Is(err, fs.ErrExist) {
return err
}
fi, err = b.root.Lstat(prefix)
}
if err != nil {
return err
}
if !fi.IsDir() {
return fmt.Errorf("%s: component %s is not a directory: %w", rel, prefix, ErrNotRegular)
}
}
return nil
}
// ReadRegular reads a brain-relative path, refusing anything that is not a // ReadRegular reads a brain-relative path, refusing anything that is not a
// regular file, both before the open and on the opened handle. // regular file, both before the open and on the opened handle.
func (b *Brain) ReadRegular(rel string) ([]byte, error) { func (b *Brain) ReadRegular(rel string) ([]byte, error) {
if err := cleanRel(rel); err != nil {
return nil, err
}
if err := b.realDir(path.Dir(rel)); err != nil {
return nil, err
}
fi, err := b.root.Lstat(rel) fi, err := b.root.Lstat(rel)
if err != nil { if err != nil {
return nil, err return nil, err
@@ -154,6 +305,12 @@ func (b *Brain) readOpened(rel string) ([]byte, error) { return b.readOpenedMax(
// ReadRegularMax is ReadRegular with a size limit, checked on the opened // ReadRegularMax is ReadRegular with a size limit, checked on the opened
// handle before reading (max < 0 means no limit). // handle before reading (max < 0 means no limit).
func (b *Brain) ReadRegularMax(rel string, max int64) ([]byte, error) { func (b *Brain) ReadRegularMax(rel string, max int64) ([]byte, error) {
if err := cleanRel(rel); err != nil {
return nil, err
}
if err := b.realDir(path.Dir(rel)); err != nil {
return nil, err
}
fi, err := b.root.Lstat(rel) fi, err := b.root.Lstat(rel)
if err != nil { if err != nil {
return nil, err return nil, err
@@ -249,3 +406,51 @@ func (b *Brain) WalkRegular(relDir string, fn func(rel string) error, onErr func
return fn(p) return fn(p)
}) })
} }
// WriteRegular replaces the contents of an existing regular file at a
// brain-relative path. It never creates a file and never writes through a
// final symlink: os.Root.OpenFile follows an in-root final symlink even with
// O_NOFOLLOW (verified on go1.27), so the path is Lstat-ed through the root
// first and the opened handle must be that same file (os.SameFile), which
// also catches a swap between the two. O_NONBLOCK keeps a FIFO from blocking
// the open. Every directory above the file must be a real directory
// (hyperguild#173); a directory swapped in after that check is the #167
// residual.
func (b *Brain) WriteRegular(rel string, data []byte) error {
if err := cleanRel(rel); err != nil {
return err
}
// Every directory above the file must be real: os.Root follows an
// in-brain intermediate symlink, which would redirect the write
// elsewhere in the brain (hyperguild#173). A swap after this check is
// the #167 residual.
if err := b.realDir(path.Dir(rel)); err != nil {
return err
}
lfi, err := b.root.Lstat(rel)
if err != nil {
return err
}
if !lfi.Mode().IsRegular() {
return fmt.Errorf("%s: %w", rel, ErrNotRegular)
}
f, err := b.root.OpenFile(rel, os.O_WRONLY|syscall.O_NOFOLLOW|syscall.O_NONBLOCK, 0)
if err != nil {
return err
}
defer func() { _ = f.Close() }()
fi, err := f.Stat()
if err != nil {
return err
}
if !fi.Mode().IsRegular() || !os.SameFile(lfi, fi) {
return fmt.Errorf("%s: changed while opening: %w", rel, ErrNotRegular)
}
if err := f.Truncate(0); err != nil {
return err
}
if _, err := f.Write(data); err != nil {
return err
}
return f.Close()
}
+250 -1
View File
@@ -218,7 +218,7 @@ func TestReadDirRegularListsOnlyRegularFilesAndRefusesSymlinkedComponents(t *tes
// A final component swapped for a symlink to another regular file in the // A final component swapped for a symlink to another regular file in the
// brain between the Lstat and the open: os.Root follows in-root links, so // brain between the Lstat and the open: os.Root follows in-root links, so
// only comparing the two (os.SameFile) refuses it. // only comparing the two (os.SameFile) refuses it.
func TestReadRegularNeverReturnsAnInBrainFileSwappedInBehindASymlink(t *testing.T) { func TestReadRegularDoesNotReturnAnInBrainFileSwappedInBehindASymlink(t *testing.T) {
dir := t.TempDir() dir := t.TempDir()
require.NoError(t, os.WriteFile(filepath.Join(dir, "s.md"), []byte("zqsamefilesentinel"), 0o644)) require.NoError(t, os.WriteFile(filepath.Join(dir, "s.md"), []byte("zqsamefilesentinel"), 0o644))
p := filepath.Join(dir, "r.md") p := filepath.Join(dir, "r.md")
@@ -263,6 +263,52 @@ func TestReadRegularNeverReturnsAnInBrainFileSwappedInBehindASymlink(t *testing.
assert.Positive(t, ok, "positive control: the regular file is read") assert.Positive(t, ok, "positive control: the regular file is read")
} }
func TestWriteRegularRefusesFinalSymlinksFIFOsEscapesAndMissingFilesAndRewritesRegularFiles(t *testing.T) {
dir := t.TempDir()
outside := filepath.Join(t.TempDir(), "o.md")
require.NoError(t, os.WriteFile(outside, []byte("untouched"), 0o644))
require.NoError(t, os.Symlink(outside, filepath.Join(dir, "l.md")))
require.NoError(t, syscall.Mkfifo(filepath.Join(dir, "f.md"), 0o644))
require.NoError(t, os.WriteFile(filepath.Join(dir, "n.md"), []byte("old content"), 0o644))
b, err := Open(dir)
require.NoError(t, err)
defer func() { _ = b.Close() }()
assert.Error(t, b.WriteRegular("l.md", []byte("x")))
got, _ := os.ReadFile(outside)
assert.Equal(t, "untouched", string(got))
// A symlink to another file inside the brain: os.Root would allow it,
// only O_NOFOLLOW refuses it.
require.NoError(t, os.WriteFile(filepath.Join(dir, "in.md"), []byte("inside"), 0o644))
require.NoError(t, os.Symlink("in.md", filepath.Join(dir, "il.md")))
assert.Error(t, b.WriteRegular("il.md", []byte("x")))
got, _ = os.ReadFile(filepath.Join(dir, "in.md"))
assert.Equal(t, "inside", string(got))
// A reader held open, so a non-blocking write open would succeed. The
// Lstat type check refuses the FIFO first; the handle check and
// O_NONBLOCK matter only if it is swapped in after the Lstat.
r, err := os.OpenFile(filepath.Join(dir, "f.md"), os.O_RDONLY|syscall.O_NONBLOCK, 0)
require.NoError(t, err)
defer func() { _ = r.Close() }()
done := make(chan error, 1)
go func() { done <- b.WriteRegular("f.md", []byte("x")) }()
select {
case err := <-done:
assert.Error(t, err)
case <-time.After(3 * time.Second):
t.Fatal("WriteRegular hung on a FIFO")
}
assert.Error(t, b.WriteRegular("../"+filepath.Base(filepath.Dir(outside))+"/o.md", []byte("x")))
assert.Error(t, b.WriteRegular("missing.md", []byte("x")), "does not create files")
require.NoError(t, b.WriteRegular("n.md", []byte("new")))
got, _ = os.ReadFile(filepath.Join(dir, "n.md"))
assert.Equal(t, "new", string(got), "truncated and rewritten")
}
func TestReadDirDirsListsOnlyRealSubdirsAndRefusesSymlinkedComponents(t *testing.T) { func TestReadDirDirsListsOnlyRealSubdirsAndRefusesSymlinkedComponents(t *testing.T) {
dir := t.TempDir() dir := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(dir, "a", "real"), 0o755)) require.NoError(t, os.MkdirAll(filepath.Join(dir, "a", "real"), 0o755))
@@ -280,3 +326,206 @@ func TestReadDirDirsListsOnlyRealSubdirsAndRefusesSymlinkedComponents(t *testing
_, err = b.ReadDirDirs("alias") _, err = b.ReadDirDirs("alias")
assert.ErrorIs(t, err, ErrNotRegular) assert.ErrorIs(t, err, ErrNotRegular)
} }
// hyperguild#173: a file reached through a symlinked directory inside the
// brain is refused for reads and writes, so a caller-supplied path cannot be
// redirected to another part of the brain.
func TestReadAndWriteRegularRefuseAFileBehindASymlinkedDirectory(t *testing.T) {
dir := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(dir, "raw"), 0o755))
require.NoError(t, os.WriteFile(filepath.Join(dir, "raw", "v.md"), []byte("untouched"), 0o644))
require.NoError(t, os.MkdirAll(filepath.Join(dir, "wiki"), 0o755))
require.NoError(t, os.Symlink(filepath.Join("..", "raw"), filepath.Join(dir, "wiki", "alias")))
b, err := Open(dir)
require.NoError(t, err)
defer func() { _ = b.Close() }()
_, err = b.ReadRegular("wiki/alias/v.md")
assert.ErrorIs(t, err, ErrNotRegular)
_, err = b.ReadRegularMax("wiki/alias/v.md", 100)
assert.ErrorIs(t, err, ErrNotRegular)
assert.ErrorIs(t, b.WriteRegular("wiki/alias/v.md", []byte("x")), ErrNotRegular)
got, _ := os.ReadFile(filepath.Join(dir, "raw", "v.md"))
assert.Equal(t, "untouched", string(got))
got, err = b.ReadRegular("raw/v.md")
require.NoError(t, err, "positive control: the real path still reads")
assert.Equal(t, "untouched", string(got))
}
// realDir checks the cleaned parent, so the I/O must use that same clean
// path: os.Root resolves ".." after following a symlink, and a ".." placed
// after a symlinked directory would otherwise pass the check and redirect.
func TestReadAndWriteRegularRefuseAPathThatIsNotClean(t *testing.T) {
dir := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(dir, "raw", "facts"), 0o755))
require.NoError(t, os.WriteFile(filepath.Join(dir, "raw", "v.md"), []byte("untouched"), 0o644))
require.NoError(t, os.MkdirAll(filepath.Join(dir, "wiki"), 0o755))
require.NoError(t, os.Symlink(filepath.Join("..", "raw", "facts"), filepath.Join(dir, "wiki", "alias")))
b, err := Open(dir)
require.NoError(t, err)
defer func() { _ = b.Close() }()
for _, rel := range []string{"wiki/alias/../v.md", "./raw/v.md", "raw//v.md", "/raw/v.md"} {
got, err := b.ReadRegular(rel)
assert.ErrorIs(t, err, ErrNotRegular, rel)
assert.NotContains(t, string(got), "untouched", rel)
_, err = b.ReadRegularMax(rel, 100)
assert.ErrorIs(t, err, ErrNotRegular, rel)
assert.ErrorIs(t, b.WriteRegular(rel, []byte("OVERWRITTEN")), ErrNotRegular, rel)
}
got, _ := os.ReadFile(filepath.Join(dir, "raw", "v.md"))
assert.Equal(t, "untouched", string(got))
}
// hyperguild#188: PutRegular creates or replaces a regular file through the
// root. Missing parent directories are created; an existing component that
// is not a real directory, a final symlink or special file, and an unclean
// path are refused, and nothing is created at a link's target.
func TestPutRegularCreatesOrReplacesOnlyThroughRealDirectories(t *testing.T) {
dir := t.TempDir()
outside := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(dir, "raw"), 0o755))
require.NoError(t, os.MkdirAll(filepath.Join(dir, "wiki"), 0o755))
require.NoError(t, os.Symlink(filepath.Join("..", "raw"), filepath.Join(dir, "wiki", "inwing")))
require.NoError(t, os.Symlink(outside, filepath.Join(dir, "wiki", "outwing")))
require.NoError(t, os.WriteFile(filepath.Join(dir, "raw", "v.md"), []byte("untouched"), 0o644))
require.NoError(t, os.MkdirAll(filepath.Join(dir, "wiki", "w"), 0o755))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "raw", "v.md"), filepath.Join(dir, "wiki", "w", "link.md")))
require.NoError(t, syscall.Mkfifo(filepath.Join(dir, "wiki", "w", "f.md"), 0o644))
b, err := Open(dir)
require.NoError(t, err)
defer func() { _ = b.Close() }()
for _, rel := range []string{"wiki/inwing/facts/n.md", "wiki/outwing/facts/n.md", "wiki/w/link.md", "wiki/w/f.md", "wiki/w/../../raw/v.md", "/abs.md"} {
assert.ErrorIs(t, b.PutRegular(rel, []byte("x")), ErrNotRegular, rel)
}
ents, _ := os.ReadDir(outside)
assert.Empty(t, ents, "nothing created outside the brain")
_, err = os.Stat(filepath.Join(dir, "raw", "facts"))
assert.True(t, os.IsNotExist(err), "nothing created through the in-brain link")
got, _ := os.ReadFile(filepath.Join(dir, "raw", "v.md"))
assert.Equal(t, "untouched", string(got))
require.NoError(t, b.PutRegular("wiki/new/facts/n.md", []byte("one")), "positive control: create with missing parents")
got, _ = os.ReadFile(filepath.Join(dir, "wiki", "new", "facts", "n.md"))
assert.Equal(t, "one", string(got))
require.NoError(t, b.PutRegular("wiki/new/facts/n.md", []byte("two")), "positive control: replace")
got, _ = os.ReadFile(filepath.Join(dir, "wiki", "new", "facts", "n.md"))
assert.Equal(t, "two", string(got))
}
func TestAppendRegularCreatesOrAppendsOnlyThroughRealDirectories(t *testing.T) {
dir := t.TempDir()
outside := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(dir, "raw"), 0o755))
require.NoError(t, os.WriteFile(filepath.Join(dir, "raw", "v.md"), []byte("untouched"), 0o644))
require.NoError(t, os.Symlink(outside, filepath.Join(dir, "outdir")))
require.NoError(t, os.Symlink("raw", filepath.Join(dir, "indir")))
require.NoError(t, os.Symlink(filepath.Join("raw", "v.md"), filepath.Join(dir, "log.md")))
require.NoError(t, syscall.Mkfifo(filepath.Join(dir, "f.md"), 0o644))
b, err := Open(dir)
require.NoError(t, err)
defer func() { _ = b.Close() }()
for _, rel := range []string{"outdir/l.md", "indir/v.md", "log.md", "f.md", "raw/../log.md"} {
assert.ErrorIs(t, b.AppendRegular(rel, []byte("x")), ErrNotRegular, rel)
}
ents, _ := os.ReadDir(outside)
assert.Empty(t, ents)
got, _ := os.ReadFile(filepath.Join(dir, "raw", "v.md"))
assert.Equal(t, "untouched", string(got))
require.NoError(t, b.AppendRegular("logs/a.md", []byte("one\n")), "positive control: create")
require.NoError(t, b.AppendRegular("logs/a.md", []byte("two\n")), "positive control: append")
got, _ = os.ReadFile(filepath.Join(dir, "logs", "a.md"))
assert.Equal(t, "one\ntwo\n", string(got))
}
func TestRealParentRefusesASymlinkedDirectoryOrUncleanPath(t *testing.T) {
dir := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(dir, "raw"), 0o755))
require.NoError(t, os.MkdirAll(filepath.Join(dir, "wiki"), 0o755))
require.NoError(t, os.Symlink(filepath.Join("..", "raw"), filepath.Join(dir, "wiki", "alias")))
b, err := Open(dir)
require.NoError(t, err)
defer func() { _ = b.Close() }()
assert.ErrorIs(t, b.RealParent("wiki/alias/x.md"), ErrNotRegular)
assert.ErrorIs(t, b.RealParent("wiki/../raw/x.md"), ErrNotRegular)
assert.NoError(t, b.RealParent("wiki/x.md"), "positive control")
}
// cleanRel must refuse an unclean path on its own: here the ".." resolves,
// after a symlinked directory, to a real regular file, so only the clean
// check stands between the call and that file.
func TestAppendRegularRefusesAnUncleanPathThatResolvesToARealFile(t *testing.T) {
dir := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(dir, "raw", "keep"), 0o755))
require.NoError(t, os.WriteFile(filepath.Join(dir, "raw", "v.md"), []byte("untouched"), 0o644))
require.NoError(t, os.Symlink(filepath.Join("raw", "keep"), filepath.Join(dir, "indir")))
b, err := Open(dir)
require.NoError(t, err)
defer func() { _ = b.Close() }()
assert.ErrorIs(t, b.AppendRegular("indir/../v.md", []byte("x")), ErrNotRegular)
assert.ErrorIs(t, b.PutRegular("indir/../v.md", []byte("x")), ErrNotRegular)
got, _ := os.ReadFile(filepath.Join(dir, "raw", "v.md"))
assert.Equal(t, "untouched", string(got))
}
func TestRemoveRefusesAPathBehindASymlinkedDirectoryAndRemovesOnlyTheLink(t *testing.T) {
dir := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(dir, "raw"), 0o755))
require.NoError(t, os.WriteFile(filepath.Join(dir, "raw", "v.md"), []byte("keep"), 0o644))
require.NoError(t, os.MkdirAll(filepath.Join(dir, "wiki"), 0o755))
require.NoError(t, os.Symlink(filepath.Join("..", "raw"), filepath.Join(dir, "wiki", "alias")))
require.NoError(t, os.Symlink(filepath.Join("..", "raw", "v.md"), filepath.Join(dir, "wiki", "link.md")))
require.NoError(t, os.WriteFile(filepath.Join(dir, "wiki", "n.md"), []byte("n"), 0o644))
b, err := Open(dir)
require.NoError(t, err)
defer func() { _ = b.Close() }()
assert.ErrorIs(t, b.Remove("wiki/alias/v.md"), ErrNotRegular)
_, err = os.Stat(filepath.Join(dir, "raw", "v.md"))
assert.NoError(t, err, "the file behind the symlinked directory is kept")
require.NoError(t, b.Remove("wiki/link.md"))
_, err = os.Stat(filepath.Join(dir, "raw", "v.md"))
assert.NoError(t, err, "removing a final symlink never removes its target")
require.NoError(t, b.Remove("wiki/n.md"), "positive control")
_, err = os.Stat(filepath.Join(dir, "wiki", "n.md"))
assert.True(t, os.IsNotExist(err))
}
// CreateRegular is the only write path for promote and reject: it creates
// missing parents through the root, refuses an in-brain symlinked parent and
// an unclean path, and never replaces anything already there (fs.ErrExist,
// content unchanged), including a dangling link.
func TestCreateRegularCreatesOnlyNewFilesThroughRealDirectories(t *testing.T) {
dir := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(dir, "raw", "keep"), 0o755))
require.NoError(t, os.MkdirAll(filepath.Join(dir, "wiki"), 0o755))
require.NoError(t, os.Symlink(filepath.Join("..", "raw"), filepath.Join(dir, "wiki", "inwing")))
require.NoError(t, os.WriteFile(filepath.Join(dir, "raw", "v.md"), []byte("untouched"), 0o644))
require.NoError(t, os.Symlink(filepath.Join("raw", "keep"), filepath.Join(dir, "indir")))
require.NoError(t, os.WriteFile(filepath.Join(dir, "wiki", "exists.md"), []byte("original"), 0o644))
require.NoError(t, os.Symlink("nowhere.md", filepath.Join(dir, "wiki", "dangling.md")))
b, err := Open(dir)
require.NoError(t, err)
defer func() { _ = b.Close() }()
assert.ErrorIs(t, b.CreateRegular("wiki/inwing/facts/n.md", []byte("x")), ErrNotRegular)
_, err = os.Stat(filepath.Join(dir, "raw", "facts"))
assert.True(t, os.IsNotExist(err), "nothing created through the in-brain link")
assert.ErrorIs(t, b.CreateRegular("indir/../v.md", []byte("x")), ErrNotRegular)
got, _ := os.ReadFile(filepath.Join(dir, "raw", "v.md"))
assert.Equal(t, "untouched", string(got))
assert.ErrorIs(t, b.CreateRegular("wiki/exists.md", []byte("x")), fs.ErrExist)
got, _ = os.ReadFile(filepath.Join(dir, "wiki", "exists.md"))
assert.Equal(t, "original", string(got))
assert.ErrorIs(t, b.CreateRegular("wiki/dangling.md", []byte("x")), fs.ErrExist)
_, err = os.Stat(filepath.Join(dir, "wiki", "nowhere.md"))
assert.True(t, os.IsNotExist(err), "nothing created at the dangling link's target")
require.NoError(t, b.CreateRegular("wiki/new/facts/n.md", []byte("one")), "positive control: create with missing parents")
got, _ = os.ReadFile(filepath.Join(dir, "wiki", "new", "facts", "n.md"))
assert.Equal(t, "one", string(got))
}
@@ -1,8 +1,10 @@
// Package brainfstest is the shared harness for hyperguild#157's swap-race // Package brainfstest is the shared harness for hyperguild#157's swap-race
// tests: it keeps swapping a regular note, a symlink to an outside file, or a // tests: it keeps swapping a regular note, a symlink to a sentinel target file
// FIFO (with a writer held open) into a path while a reader under test loops. // (outside the brain, or inside it for hyperguild#171), or a FIFO (with a
// It observes two things: no sentinel byte from the outside file reaches the // writer held open) into a path while a reader under test loops.
// reader's output, and the reader returns within a timeout. It does not cover // It observes two things, over a sampled window only: no sentinel byte from
// the target reaches the reader's output, and the reader returns within a
// timeout. It does not cover
// hard links, other paths into the reader, or races it does not generate. // hard links, other paths into the reader, or races it does not generate.
package brainfstest package brainfstest
@@ -18,7 +20,7 @@ import (
"time" "time"
) )
// Sentinel marks the outside file's contents. // Sentinel marks the target file's contents.
const Sentinel = "zqbrainfssentinelonefiftyseven" const Sentinel = "zqbrainfssentinelonefiftyseven"
// outsideContent carries Sentinel in every field a reader might extract: the // outsideContent carries Sentinel in every field a reader might extract: the
@@ -141,7 +143,7 @@ func AssertNoLeakFrom(t testing.TB, path, inside, target string, d time.Duration
} }
} }
if leaks > 0 { if leaks > 0 {
t.Fatalf("outside file leaked %d times in %d iterations", leaks, iters) t.Fatalf("symlink target %s leaked %d times in %d iterations", target, leaks, iters)
} }
if n := swaps.Load(); n == 0 { if n := swaps.Load(); n == 0 {
t.Fatalf("no symlink was swapped into %s in %d iterations: the race did not run", path, iters) t.Fatalf("no symlink was swapped into %s in %d iterations: the race did not run", path, iters)
@@ -16,9 +16,10 @@
package classification package classification
import ( import (
"errors"
"fmt" "fmt"
"os" "github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
"path/filepath" "io/fs"
"strings" "strings"
"gopkg.in/yaml.v3" "gopkg.in/yaml.v3"
@@ -110,9 +111,20 @@ type rawConfig struct {
func Load(brainDir string) (*Config, error) { func Load(brainDir string) (*Config, error) {
cfg := &Config{wings: map[string]Level{}, repos: map[string]Level{}} cfg := &Config{wings: map[string]Level{}, repos: map[string]Level{}}
data, err := os.ReadFile(filepath.Join(brainDir, "classification.yaml")) // Read through the brain root, regular files only (hyperguild#157): a
// symlinked or special file is not trusted as the config, which is a hard
// error like a malformed one.
bfs, err := brainfs.Open(brainDir)
if errors.Is(err, fs.ErrNotExist) {
return cfg, nil
}
if err != nil { if err != nil {
if os.IsNotExist(err) { return nil, fmt.Errorf("open brain: %w", err)
}
defer func() { _ = bfs.Close() }()
data, err := bfs.ReadRegular("classification.yaml")
if err != nil {
if errors.Is(err, fs.ErrNotExist) {
return cfg, nil return cfg, nil
} }
return nil, fmt.Errorf("read classification.yaml: %w", err) return nil, fmt.Errorf("read classification.yaml: %w", err)
@@ -0,0 +1,46 @@
package classification
import (
"os"
"path/filepath"
"syscall"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
// hyperguild#157 slice 6: Load read classification.yaml with os.ReadFile. It
// sets the I1 sovereignty levels; a symlinked file must not be trusted as the
// config (a hard error, as for a malformed file), and a FIFO must not hang
// startup.
func TestLoadRefusesASymlinkedOrFIFOClassificationFile(t *testing.T) {
ok := t.TempDir()
require.NoError(t, os.WriteFile(filepath.Join(ok, "classification.yaml"), []byte("wings:\n w: confidential\n"), 0o644))
_, err := Load(ok)
require.NoError(t, err, "positive control")
linked := t.TempDir()
require.NoError(t, os.WriteFile(filepath.Join(linked, "real.yaml"), []byte("wings:\n w: public\n"), 0o644))
require.NoError(t, os.Symlink("real.yaml", filepath.Join(linked, "classification.yaml")))
_, err = Load(linked)
assert.Error(t, err, "an in-brain symlink is not trusted as the config")
fifo := t.TempDir()
require.NoError(t, syscall.Mkfifo(filepath.Join(fifo, "classification.yaml"), 0o644))
w, err := os.OpenFile(filepath.Join(fifo, "classification.yaml"), os.O_RDWR, 0)
require.NoError(t, err)
defer func() { _ = w.Close() }()
done := make(chan error, 1)
go func() { _, err := Load(fifo); done <- err }()
select {
case err := <-done:
assert.Error(t, err)
case <-time.After(3 * time.Second):
t.Fatal("Load hung on a FIFO")
}
_, err = Load(t.TempDir())
assert.NoError(t, err, "an absent file is still not an error")
}
@@ -21,6 +21,11 @@ func TestReadExcerptRefusesSymlinksAndSurvivesSwapRaces(t *testing.T) {
require.NoError(t, os.MkdirAll(facts, 0o755)) require.NoError(t, os.MkdirAll(facts, 0o755))
require.NoError(t, os.Symlink(brainfstest.Outside(t), filepath.Join(facts, "leak.md"))) require.NoError(t, os.Symlink(brainfstest.Outside(t), filepath.Join(facts, "leak.md")))
assert.NotContains(t, readExcerpt(brain, "wiki/w/facts/leak.md", 500), brainfstest.Sentinel) assert.NotContains(t, readExcerpt(brain, "wiki/w/facts/leak.md", 500), brainfstest.Sentinel)
// A link to a file elsewhere in the brain: os.Root allows it, only the
// regular-file check refuses it (hyperguild#171).
inBrain := brainfstest.WriteSentinelAt(t, filepath.Join(brain, "raw", "s.md"))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "..", "raw", "s.md"), filepath.Join(facts, "inlink.md")))
assert.NotContains(t, readExcerpt(brain, "wiki/w/facts/inlink.md", 500), brainfstest.Sentinel)
p := filepath.Join(facts, "r.md") p := filepath.Join(facts, "r.md")
require.NoError(t, os.WriteFile(p, []byte("---\nx: y\n---\ninside"), 0o644)) require.NoError(t, os.WriteFile(p, []byte("---\nx: y\n---\ninside"), 0o644))
@@ -28,6 +33,9 @@ func TestReadExcerptRefusesSymlinksAndSurvivesSwapRaces(t *testing.T) {
brainfstest.AssertNoLeak(t, p, "---\nx: y\n---\ninside", 1500*time.Millisecond, func() string { brainfstest.AssertNoLeak(t, p, "---\nx: y\n---\ninside", 1500*time.Millisecond, func() string {
return readExcerpt(brain, "wiki/w/facts/r.md", 500) return readExcerpt(brain, "wiki/w/facts/r.md", 500)
}) })
brainfstest.AssertNoLeakFrom(t, p, "---\nx: y\n---\ninside", inBrain, 1500*time.Millisecond, func() string {
return readExcerpt(brain, "wiki/w/facts/r.md", 500)
})
brainfstest.AssertNoHang(t, p, "---\nx: y\n---\ninside", 1500*time.Millisecond, 6*time.Second, func() { brainfstest.AssertNoHang(t, p, "---\nx: y\n---\ninside", 1500*time.Millisecond, 6*time.Second, func() {
_ = readExcerpt(brain, "wiki/w/facts/r.md", 500) _ = readExcerpt(brain, "wiki/w/facts/r.md", 500)
}) })
+4 -3
View File
@@ -83,8 +83,9 @@ func (r *Registry) Observe(path, status string, d time.Duration) {
// Middleware wraps next, observing every request's duration + status. // Middleware wraps next, observing every request's duration + status.
// The metric label `path` uses the request's Pattern (Go 1.22+ ServeMux), // The metric label `path` uses the request's Pattern (Go 1.22+ ServeMux),
// falling back to the URL path if no Pattern is set. Pattern keeps // or the fixed label "unmatched" if no Pattern is set: falling back to the
// cardinality bounded (one series per route, not one per unique URL). // raw URL path gave any caller one series per unique path (hyperguild#185).
// Pattern keeps cardinality bounded (one series per route).
func (r *Registry) Middleware(next http.Handler) http.Handler { func (r *Registry) Middleware(next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, req *http.Request) { return http.HandlerFunc(func(w http.ResponseWriter, req *http.Request) {
rec := &statusRecorder{ResponseWriter: w, code: http.StatusOK} rec := &statusRecorder{ResponseWriter: w, code: http.StatusOK}
@@ -92,7 +93,7 @@ func (r *Registry) Middleware(next http.Handler) http.Handler {
next.ServeHTTP(rec, req) next.ServeHTTP(rec, req)
path := req.Pattern path := req.Pattern
if path == "" { if path == "" {
path = req.URL.Path path = "unmatched"
} }
r.Observe(path, statusClass(rec.code), time.Since(start)) r.Observe(path, statusClass(rec.code), time.Since(start))
}) })
+23 -2
View File
@@ -71,11 +71,16 @@ func TestMiddleware_RecordsTiming(t *testing.T) {
t.Parallel() t.Parallel()
r := New() r := New()
handler := r.Middleware(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) { // Routed through a ServeMux, as in production: the mux sets the
// request's Pattern, which is the label (an unrouted request is
// "unmatched", hyperguild#185).
mux := http.NewServeMux()
mux.HandleFunc("/query", func(w http.ResponseWriter, _ *http.Request) {
time.Sleep(2 * time.Millisecond) time.Sleep(2 * time.Millisecond)
w.WriteHeader(http.StatusOK) w.WriteHeader(http.StatusOK)
_, _ = io.WriteString(w, "ok") _, _ = io.WriteString(w, "ok")
})) })
handler := r.Middleware(mux)
srv := httptest.NewServer(handler) srv := httptest.NewServer(handler)
defer srv.Close() defer srv.Close()
@@ -117,3 +122,19 @@ func TestStatusRecorder_DefaultsTo200(t *testing.T) {
t.Errorf("code %d, want 200", rec.Code) t.Errorf("code %d, want 200", rec.Code)
} }
} }
// A request with no matched Pattern is labelled "unmatched", never with its
// raw URL path (unbounded cardinality, hyperguild#185).
func TestMiddleware_LabelsAnUnmatchedRequestWithAFixedLabel(t *testing.T) {
r := New()
h := r.Middleware(http.NotFoundHandler())
h.ServeHTTP(httptest.NewRecorder(), httptest.NewRequest("GET", "/zqrandom-185-unit", nil))
rec := httptest.NewRecorder()
r.Handler().ServeHTTP(rec, httptest.NewRequest("GET", "/metrics", nil))
if strings.Contains(rec.Body.String(), "zqrandom-185-unit") {
t.Errorf("the raw path became a metrics label:\n%s", rec.Body.String())
}
if !strings.Contains(rec.Body.String(), `path="unmatched"`) {
t.Errorf("no unmatched label:\n%s", rec.Body.String())
}
}
@@ -0,0 +1,36 @@
package notes
import (
"os"
"path/filepath"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
// hyperguild#188/#173: brain_edit's atomic rename must not be redirected by
// a symlinked directory inside the brain.
func TestEditRefusesANoteBehindASymlinkedDirectory(t *testing.T) {
b := t.TempDir()
// Edit refuses a tree that is not the sync source before any path
// check, so the test needs a .git to reach the symlink refusal.
require.NoError(t, os.MkdirAll(filepath.Join(b, ".git"), 0o755))
victim := filepath.Join(b, "raw", "facts", "v.md")
require.NoError(t, os.MkdirAll(filepath.Dir(victim), 0o755))
require.NoError(t, os.WriteFile(victim, []byte("old text\n"), 0o644))
require.NoError(t, os.MkdirAll(filepath.Join(b, "wiki", "w"), 0o755))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "raw", "facts"), filepath.Join(b, "wiki", "w", "alias")))
_, err := Edit(b, "wiki/w/alias/v.md", "old text", "new text", ScopeNotes)
assert.Error(t, err)
got, _ := os.ReadFile(victim)
assert.Equal(t, "old text\n", string(got))
real := filepath.Join(b, "wiki", "w", "facts", "r.md")
require.NoError(t, os.MkdirAll(filepath.Dir(real), 0o755))
require.NoError(t, os.WriteFile(real, []byte("old text\n"), 0o644))
_, err = Edit(b, "wiki/w/facts/r.md", "old text", "new text", ScopeNotes)
require.NoError(t, err, "positive control: a note in a real directory is edited")
got, _ = os.ReadFile(real)
assert.Equal(t, "new text\n", string(got))
}
@@ -38,3 +38,31 @@ func TestReadScopedRefusesANoteOverMaxNoteBytes(t *testing.T) {
_, err := ReadScoped(brain, "wiki/w/facts/big.md", ScopeNotes) _, err := ReadScoped(brain, "wiki/w/facts/big.md", ScopeNotes)
assert.ErrorIs(t, err, ErrTooLarge) assert.ErrorIs(t, err, ErrTooLarge)
} }
// hyperguild#171 (review of #172, finding 1): os.Root alone refuses links that
// leave the brain, so the outside-target race above stays green for a reader
// that skips brainfs's Lstat and SameFile checks. An in-brain target (raw/,
// outside ScopeNotes) is what tells them apart.
func TestReadScopedDoesNotReturnAnInBrainSymlinkTargetStaticallyOrDuringASampledSwapRace(t *testing.T) {
brain := t.TempDir()
facts := filepath.Join(brain, "wiki", "w", "facts")
require.NoError(t, os.MkdirAll(facts, 0o755))
target := brainfstest.WriteSentinelAt(t, filepath.Join(brain, "raw", "s.md"))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "..", "raw", "s.md"), filepath.Join(facts, "inlink.md")))
b, err := ReadScoped(brain, "wiki/w/facts/inlink.md", ScopeNotes)
assert.Error(t, err, "static in-brain link")
assert.NotContains(t, string(b), brainfstest.Sentinel, "static in-brain link")
p := filepath.Join(facts, "r.md")
require.NoError(t, os.WriteFile(p, []byte("inside"), 0o644))
b, err = ReadScoped(brain, "wiki/w/facts/r.md", ScopeNotes)
require.NoError(t, err)
assert.Equal(t, "inside", string(b), "positive control: a regular note reads")
read := func() string {
b, _ := ReadScoped(brain, "wiki/w/facts/r.md", ScopeNotes)
return string(b)
}
brainfstest.AssertNoLeakFrom(t, p, "inside", target, 1500*time.Millisecond, read)
}
+17 -8
View File
@@ -3,7 +3,10 @@ package pipeline
import ( import (
"context" "context"
"errors"
"fmt" "fmt"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
"io/fs"
"os" "os"
"path/filepath" "path/filepath"
"strings" "strings"
@@ -23,10 +26,16 @@ func BackfillRefs(ctx context.Context, brainDir string) (int, error) {
return 0, fmt.Errorf("load inventory: %w", err) return 0, fmt.Errorf("load inventory: %w", err)
} }
sourcesDir := filepath.Join(brainDir, "wiki", "sources") // Listed and read through the brain root, regular files only
entries, err := os.ReadDir(sourcesDir) // (hyperguild#157).
bfs, err := brainfs.Open(brainDir)
if err != nil { if err != nil {
if os.IsNotExist(err) { return 0, fmt.Errorf("open brain: %w", err)
}
defer func() { _ = bfs.Close() }()
names, err := bfs.ReadDirRegular("wiki/sources")
if err != nil {
if errors.Is(err, fs.ErrNotExist) {
return 0, nil return 0, nil
} }
return 0, fmt.Errorf("read sources dir: %w", err) return 0, fmt.Errorf("read sources dir: %w", err)
@@ -37,20 +46,20 @@ func BackfillRefs(ctx context.Context, brainDir string) (int, error) {
// get both refs merged before a single write. // get both refs merged before a single write.
pending := make(map[string]wiki.Page) pending := make(map[string]wiki.Page)
for _, e := range entries { for _, name := range names {
if ctx.Err() != nil { if ctx.Err() != nil {
return 0, ctx.Err() return 0, ctx.Err()
} }
if e.IsDir() || !strings.HasSuffix(e.Name(), ".md") { if !strings.HasSuffix(name, ".md") {
continue continue
} }
b, err := os.ReadFile(filepath.Join(sourcesDir, e.Name())) b, err := bfs.ReadRegular("wiki/sources/" + name)
if err != nil { if err != nil {
continue continue
} }
sourceContent := string(b) sourceContent := string(b)
sourceSlug := strings.TrimSuffix(e.Name(), ".md") sourceSlug := strings.TrimSuffix(name, ".md")
sourceTitle := extractTitle(sourceContent) sourceTitle := extractTitle(sourceContent)
if sourceTitle == "" { if sourceTitle == "" {
sourceTitle = sourceSlug sourceTitle = sourceSlug
@@ -70,7 +79,7 @@ func BackfillRefs(ctx context.Context, brainDir string) (int, error) {
// Start from already-accumulated version if we've seen this page. // Start from already-accumulated version if we've seen this page.
page, seen := pending[relPath] page, seen := pending[relPath]
if !seen { if !seen {
raw, err := os.ReadFile(filepath.Join(brainDir, filepath.FromSlash(relPath))) raw, err := bfs.ReadRegular(relPath)
if err != nil { if err != nil {
continue continue
} }
+22 -6
View File
@@ -5,7 +5,10 @@ import (
"context" "context"
"crypto/sha256" "crypto/sha256"
"encoding/hex" "encoding/hex"
"errors"
"fmt" "fmt"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
"io/fs"
"os" "os"
"path/filepath" "path/filepath"
"strings" "strings"
@@ -129,7 +132,7 @@ func buildAndWrite(ctx context.Context, rawPages []RawPage, sourceSlug, date, br
if err := os.MkdirAll(filepath.Dir(dest), 0o755); err != nil { if err := os.MkdirAll(filepath.Dir(dest), 0o755); err != nil {
return Result{}, fmt.Errorf("mkdir for %s: %w", page.Path, err) return Result{}, fmt.Errorf("mkdir for %s: %w", page.Path, err)
} }
content, conflicts, err := reviseOrCreate(dest, page) content, conflicts, err := reviseOrCreate(brainDir, page)
if err != nil { if err != nil {
return Result{}, err return Result{}, err
} }
@@ -188,14 +191,27 @@ func hashBody(content string) string {
// readSourcePage returns the full on-disk content of the source page at // readSourcePage returns the full on-disk content of the source page at
// wiki/sources/<sourceSlug>.md, or "" when it doesn't exist yet. // wiki/sources/<sourceSlug>.md, or "" when it doesn't exist yet.
// It reads through the brain root, regular files only (hyperguild#157): the
// content goes into the LLM prompt.
func readSourcePage(brainDir, sourceSlug string) string { func readSourcePage(brainDir, sourceSlug string) string {
raw, err := os.ReadFile(filepath.Join(brainDir, "wiki", "sources", sourceSlug+".md")) raw, err := readBrainFile(brainDir, "wiki/sources/"+sourceSlug+".md")
if err != nil { if err != nil {
return "" return ""
} }
return string(raw) return string(raw)
} }
// readBrainFile reads one brain-relative file through the brain root,
// regular files only (hyperguild#157).
func readBrainFile(brainDir, rel string) ([]byte, error) {
bfs, err := brainfs.Open(brainDir)
if err != nil {
return nil, err
}
defer func() { _ = bfs.Close() }()
return bfs.ReadRegular(rel)
}
// storedSourceHash reads source_hash from the source page a previous ingest // storedSourceHash reads source_hash from the source page a previous ingest
// wrote, or "" when there is no such page or no such field. // wrote, or "" when there is no such page or no such field.
// //
@@ -230,10 +246,10 @@ func storedSourceHash(brainDir, sourceSlug string) string {
// its first-ingest date on a revised page. Left deliberately — the alternative // its first-ingest date on a revised page. Left deliberately — the alternative
// is this function editing frontmatter it does not own, and no lint reads that // is this function editing frontmatter it does not own, and no lint reads that
// field today. // field today.
func reviseOrCreate(dest string, page wiki.Page) (string, []string, error) { func reviseOrCreate(brainDir string, page wiki.Page) (string, []string, error) {
existing, err := os.ReadFile(dest) existing, err := readBrainFile(brainDir, page.Path)
switch { switch {
case os.IsNotExist(err): case errors.Is(err, fs.ErrNotExist):
return page.Content, nil, nil return page.Content, nil, nil
case err != nil: case err != nil:
return "", nil, fmt.Errorf("read existing %s: %w", page.Path, err) return "", nil, fmt.Errorf("read existing %s: %w", page.Path, err)
@@ -267,7 +283,7 @@ See brain/schema.md for the full schema.
` `
func loadSchema(brainDir string) string { func loadSchema(brainDir string) string {
b, err := os.ReadFile(filepath.Join(brainDir, "schema.md")) b, err := readBrainFile(brainDir, "schema.md")
if err != nil { if err != nil {
return defaultSchema return defaultSchema
} }
+1 -3
View File
@@ -2,7 +2,6 @@
package pipeline package pipeline
import ( import (
"os"
"path/filepath" "path/filepath"
"regexp" "regexp"
"strings" "strings"
@@ -53,8 +52,7 @@ func injectSourceRefs(pages []wiki.Page, inventory map[wiki.PageType][]wiki.Entr
if !ok { if !ok {
continue continue
} }
diskPath := filepath.Join(brainDir, "wiki", string(pt), slug+".md") b, err := readBrainFile(brainDir, "wiki/"+string(pt)+"/"+slug+".md")
b, err := os.ReadFile(diskPath)
if err != nil { if err != nil {
continue continue
} }
+167
View File
@@ -0,0 +1,167 @@
package pipeline
import (
"context"
"fmt"
"io"
"os"
"path/filepath"
"syscall"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs/brainfstest"
"github.com/mathiasbq/hyperguild/ingestion/internal/wiki"
)
// hyperguild#157 slice 6: the pipeline read the source page, schema.md, the
// pages it revises and the pages it adds source refs to with os.ReadFile, so
// a symlinked file's content reached the LLM prompt and the written pages,
// and a FIFO hung the ingest.
func mkdirs(t *testing.T, brain string, dirs ...string) {
t.Helper()
for _, d := range dirs {
require.NoError(t, os.MkdirAll(filepath.Join(brain, filepath.FromSlash(d)), 0o755))
}
}
func TestReadSourcePageAndSchemaRefuseSymlinksAndSurviveSwapRaces(t *testing.T) {
brain := t.TempDir()
mkdirs(t, brain, "wiki/sources")
inBrain := brainfstest.WriteSentinelAt(t, filepath.Join(brain, "raw", "s.md"))
require.NoError(t, os.WriteFile(filepath.Join(brain, "wiki", "sources", "ok.md"), []byte("zqinside source"), 0o644))
require.NoError(t, os.WriteFile(filepath.Join(brain, "schema.md"), []byte("zqinside schema"), 0o644))
assert.Equal(t, "zqinside source", readSourcePage(brain, "ok"), "positive control")
assert.Equal(t, "zqinside schema", loadSchema(brain), "positive control")
require.NoError(t, os.Symlink(brainfstest.Outside(t), filepath.Join(brain, "wiki", "sources", "out.md")))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "raw", "s.md"), filepath.Join(brain, "wiki", "sources", "in.md")))
assert.NotContains(t, readSourcePage(brain, "out"), brainfstest.Sentinel)
assert.NotContains(t, readSourcePage(brain, "in"), brainfstest.Sentinel)
assert.NotContains(t, storedSourceHash(brain, "in"), brainfstest.Sentinel)
src := filepath.Join(brain, "wiki", "sources", "r.md")
read := func() string { return readSourcePage(brain, "r") }
brainfstest.AssertNoLeak(t, src, "zqinside", 1500*time.Millisecond, read)
brainfstest.AssertNoLeakFrom(t, src, "zqinside", inBrain, 1500*time.Millisecond, read)
brainfstest.AssertNoHang(t, src, "zqinside", 1500*time.Millisecond, 6*time.Second, func() { _ = read() })
schema := filepath.Join(brain, "schema.md")
rs := func() string { return loadSchema(brain) }
brainfstest.AssertNoLeakFrom(t, schema, "zqinside schema", inBrain, 1500*time.Millisecond, rs)
brainfstest.AssertNoHang(t, schema, "zqinside schema", 1500*time.Millisecond, 6*time.Second, func() { _ = rs() })
}
func TestReviseOrCreateRefusesASymlinkedExistingPage(t *testing.T) {
brain := t.TempDir()
mkdirs(t, brain, "wiki/concepts")
brainfstest.WriteSentinelAt(t, filepath.Join(brain, "raw", "s.md"))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "raw", "s.md"), filepath.Join(brain, "wiki", "concepts", "c.md")))
page := wiki.Page{Path: "wiki/concepts/c.md", Content: "---\ntitle: C\n---\nnew\n"}
out, _, err := reviseOrCreate(brain, page)
assert.Error(t, err)
assert.NotContains(t, out, brainfstest.Sentinel)
require.NoError(t, os.WriteFile(filepath.Join(brain, "wiki", "concepts", "d.md"), []byte("---\ntitle: D\n---\nzqold\n"), 0o644))
out, _, err = reviseOrCreate(brain, wiki.Page{Path: "wiki/concepts/d.md", Content: "---\ntitle: D\n---\nnew\n"})
require.NoError(t, err)
assert.Contains(t, out, "zqold", "positive control: the existing page is read")
out, _, err = reviseOrCreate(brain, wiki.Page{Path: "wiki/concepts/new.md", Content: "fresh"})
require.NoError(t, err)
assert.Equal(t, "fresh", out, "positive control: a missing page is created")
}
func TestInjectSourceRefsDoesNotReadASymlinkedPage(t *testing.T) {
brain := t.TempDir()
mkdirs(t, brain, "wiki/concepts")
brainfstest.WriteSentinelAt(t, filepath.Join(brain, "raw", "s.md"))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "raw", "s.md"), filepath.Join(brain, "wiki", "concepts", "c.md")))
require.NoError(t, os.WriteFile(filepath.Join(brain, "wiki", "concepts", "d.md"), []byte("zqinside d"), 0o644))
inv := map[wiki.PageType][]wiki.Entry{wiki.PageTypeConcept: {
{Slug: "c", Type: wiki.PageTypeConcept}, {Slug: "d", Type: wiki.PageTypeConcept},
}}
src := wiki.Page{Path: "wiki/sources/src.md", Content: "---\ntitle: Src\n---\nsee [[c|C]] and [[d|D]]\n"}
out := injectSourceRefs([]wiki.Page{src}, inv, brain)
got := fmt.Sprintf("%+v", out)
assert.Contains(t, got, "zqinside d", "positive control")
assert.NotContains(t, got, brainfstest.Sentinel)
}
func TestBackfillRefsSkipsSymlinkedSourcesAndPages(t *testing.T) {
brain := t.TempDir()
mkdirs(t, brain, "wiki/sources", "wiki/concepts", "raw")
require.NoError(t, os.WriteFile(filepath.Join(brain, "wiki", "concepts", "real.md"), []byte("---\ntitle: R\n---\nbody\n"), 0o644))
require.NoError(t, os.WriteFile(filepath.Join(brain, "wiki", "sources", "ok.md"), []byte("---\ntitle: zqinside\n---\n[[real|R]]\n"), 0o644))
// A source elsewhere in the brain whose title is the sentinel and which
// links the real concept with the [[slug|title]] form the pipeline follows.
require.NoError(t, os.WriteFile(filepath.Join(brain, "raw", "src.md"),
[]byte("---\ntitle: "+brainfstest.Sentinel+"\n---\n[[real|R]]\n"), 0o644))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "raw", "src.md"), filepath.Join(brain, "wiki", "sources", "in.md")))
// A concept page linked by the regular source that is itself a symlink.
require.NoError(t, os.WriteFile(filepath.Join(brain, "wiki", "sources", "ok2.md"), []byte("---\ntitle: Two\n---\n[[lnk|L]]\n"), 0o644))
brainfstest.WriteSentinelAt(t, filepath.Join(brain, "raw", "page.md"))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "raw", "page.md"), filepath.Join(brain, "wiki", "concepts", "lnk.md")))
_, err := BackfillRefs(context.Background(), brain)
require.NoError(t, err)
b, _ := os.ReadFile(filepath.Join(brain, "wiki", "concepts", "real.md"))
assert.Contains(t, string(b), "zqinside", "positive control: the regular source's ref is added")
assert.NotContains(t, string(b), brainfstest.Sentinel, "a symlinked source's title was written as a ref")
raw, _ := os.ReadFile(filepath.Join(brain, "raw", "page.md"))
assert.NotContains(t, string(raw), "[[ok2", "the symlinked concept page's target was rewritten")
}
// A source swapped between a regular file and a symlink to an in-brain source
// whose title is the sentinel: the read refuses the swapped-in link, so the
// sentinel never lands in the concept page as a ref.
func TestBackfillRefsDoesNotWriteARefFromASourceSwappedInBehindASymlink(t *testing.T) {
brain := t.TempDir()
mkdirs(t, brain, "wiki/sources", "wiki/concepts", "raw")
realPage := filepath.Join(brain, "wiki", "concepts", "real.md")
target := filepath.Join(brain, "raw", "src.md")
require.NoError(t, os.WriteFile(target, []byte("---\ntitle: "+brainfstest.Sentinel+"\n---\n[[real|R]]\n"), 0o644))
src := filepath.Join(brain, "wiki", "sources", "r.md")
inside := "---\ntitle: zqinside\n---\n[[real|R]]\n"
read := func() string {
_ = os.WriteFile(realPage, []byte("---\ntitle: R\n---\nbody\n"), 0o644)
_, _ = BackfillRefs(context.Background(), brain)
b, _ := os.ReadFile(realPage)
return string(b)
}
require.NoError(t, os.WriteFile(src, []byte(inside), 0o644))
require.Contains(t, read(), "zqinside", "positive control")
brainfstest.AssertNoLeakFrom(t, src, inside, target, 1500*time.Millisecond, read)
}
// A concept page that is regular when the inventory lists it, then swapped
// for a symlink to an in-brain file before BackfillRefs reads it: only the
// page read's own check refuses it (a static symlink never gets this far,
// because LoadInventory does not list it).
func TestBackfillRefsDoesNotCopyAPageSwappedInBehindASymlink(t *testing.T) {
brain := t.TempDir()
mkdirs(t, brain, "wiki/sources", "wiki/concepts", "raw")
require.NoError(t, os.WriteFile(filepath.Join(brain, "wiki", "sources", "s.md"), []byte("---\ntitle: S\n---\n[[c|C]]\n"), 0o644))
page := filepath.Join(brain, "wiki", "concepts", "c.md")
target := filepath.Join(brain, "raw", "secret.md")
inside := "---\ntitle: C\n---\nzqinside\n"
read := func() string {
// The write path still follows a swapped-in link (#158) and can
// overwrite the target, so restore it each round.
_ = os.WriteFile(target, []byte("---\ntitle: X\n---\n"+brainfstest.Sentinel+"\n"), 0o644)
_, _ = BackfillRefs(context.Background(), brain)
f, err := os.OpenFile(page, os.O_RDONLY|syscall.O_NOFOLLOW|syscall.O_NONBLOCK, 0)
if err != nil {
return ""
}
defer func() { _ = f.Close() }()
b, _ := io.ReadAll(f)
return string(b)
}
require.NoError(t, os.WriteFile(page, []byte(inside), 0o644))
require.Contains(t, read(), "[[s|S]]", "positive control: the regular page gets the source ref")
brainfstest.AssertNoLeakFrom(t, page, inside, target, 3*time.Second, read)
}
@@ -28,7 +28,6 @@ import (
"context" "context"
"errors" "errors"
"fmt" "fmt"
"os"
"regexp" "regexp"
"strings" "strings"
@@ -102,25 +101,6 @@ func Apply(content string, s Stamp) (string, error) {
return out, nil return out, nil
} }
// ApplyFile applies the stamp to the note at absPath in place.
func ApplyFile(absPath string, s Stamp) error {
b, err := os.ReadFile(absPath)
if err != nil {
return fmt.Errorf("read for provenance: %w", err)
}
out, err := Apply(string(b), s)
if err != nil {
return err
}
if out == string(b) {
return nil
}
if err := os.WriteFile(absPath, []byte(out), 0o644); err != nil {
return fmt.Errorf("write provenance: %w", err)
}
return nil
}
// plainGroups keeps the provably plain key groups of a frontmatter block. // plainGroups keeps the provably plain key groups of a frontmatter block.
func plainGroups(block string) string { func plainGroups(block string) string {
lines := strings.Split(strings.TrimSuffix(block, "\n"), "\n") lines := strings.Split(strings.TrimSuffix(block, "\n"), "\n")
@@ -12,6 +12,8 @@ import (
"github.com/stretchr/testify/assert" "github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require" "github.com/stretchr/testify/require"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs/brainfstest"
) )
// hyperguild#157 option 1: search walked and read the brain by path, so a // hyperguild#157 option 1: search walked and read the brain by path, so a
@@ -189,3 +191,49 @@ func TestHydrateRefusesASymlink(t *testing.T) {
assert.Error(t, err) assert.Error(t, err)
assert.NotContains(t, r.Excerpt, sentinel157) assert.NotContains(t, r.Excerpt, sentinel157)
} }
// hyperguild#171: os.Root refuses links that leave the brain on its own, so
// the outside-target cases above would stay green for a caller that read
// without brainfs's Lstat and SameFile checks. These targets are inside the
// brain (raw/, which is not a search root).
func inBrainTarget(t *testing.T, brain string) string {
t.Helper()
return brainfstest.WriteSentinelAt(t, filepath.Join(brain, "raw", "s.md"))
}
func TestQueryDoesNotReturnAnInBrainSymlinkTargetDuringASampledSwapRace(t *testing.T) {
brain, facts, _ := searchBrain(t)
target := inBrainTarget(t, brain)
read := func() string {
// Both terms, so the regular note is a possible hit: a reader that
// never returns the inside note would pass vacuously (#177 floor).
rs, _ := Query(brain, QueryOptions{Query: brainfstest.Sentinel + " zqinside", Limit: 10})
s := ""
for _, r := range rs {
s += r.Title + r.Excerpt
}
return s
}
brainfstest.AssertNoLeakFrom(t, filepath.Join(facts, "r.md"), "# r\n\nzqinside", target, 1500*time.Millisecond, read)
require.NoError(t, os.Symlink(filepath.Join("..", "..", "..", "raw", "s.md"), filepath.Join(facts, "inlink.md")))
assert.NotContains(t, read(), brainfstest.Sentinel, "static in-brain link")
}
func TestHydrateRefusesAnInBrainSymlinkAndDoesNotReturnItsTargetDuringASampledSwapRace(t *testing.T) {
brain, facts, _ := searchBrain(t)
target := inBrainTarget(t, brain)
require.NoError(t, os.Symlink(filepath.Join("..", "..", "..", "raw", "s.md"), filepath.Join(facts, "inlink.md")))
r, err := hydrate(brain, "wiki/w/facts/inlink.md")
assert.Error(t, err)
assert.NotContains(t, r.Excerpt+r.Title, brainfstest.Sentinel)
p := filepath.Join(facts, "r.md")
require.NoError(t, os.WriteFile(p, []byte("# r\n\nzqinside"), 0o644))
r, err = hydrate(brain, "wiki/w/facts/r.md")
require.NoError(t, err)
assert.Contains(t, r.Excerpt, "zqinside", "positive control: a regular note hydrates")
brainfstest.AssertNoLeakFrom(t, p, "# r\n\nzqinside", target, 1500*time.Millisecond, func() string {
r, _ := hydrate(brain, "wiki/w/facts/r.md")
return r.Title + r.Excerpt
})
}
+177
View File
@@ -0,0 +1,177 @@
package session
import (
"errors"
"fmt"
"os"
"path/filepath"
"sync/atomic"
"syscall"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
"golang.org/x/sys/unix"
)
// session_log passed the caller's session_id straight into
// filepath.Join(sessionsDir, id+".jsonl") and opened it O_CREATE|O_APPEND:
// path segments in the id created or appended files anywhere the pod could
// write. The id is now a plain name, and the open does not follow a final
// symlink.
func TestAppendRefusesSessionIDsThatAreNotPlainNames(t *testing.T) {
root := t.TempDir()
dir := filepath.Join(root, "sessions")
for _, id := range []string{"../x", "a/b", "/abs", ".", "..", ".hidden", "", "a\\b", "a\x00b"} {
assert.Error(t, Append(dir, id, Entry{SessionID: id}), "%q", id)
}
ents, _ := os.ReadDir(root)
for _, e := range ents {
assert.Equal(t, "sessions", e.Name(), "nothing created beside the sessions dir")
}
for _, id := range []string{"1557b2e1-bf0d-5c4f-bbe3-0754d087e356", "_routing", "2026-05-05-infra.plan", "0"} {
require.NoError(t, Append(dir, id, Entry{SessionID: id}), "positive control: %q", id)
}
}
func TestAppendDoesNotWriteThroughASymlinkedLog(t *testing.T) {
root := t.TempDir()
dir := filepath.Join(root, "sessions")
require.NoError(t, os.MkdirAll(dir, 0o755))
victim := filepath.Join(root, "victim.md")
require.NoError(t, os.WriteFile(victim, []byte("untouched\n"), 0o644))
require.NoError(t, os.Symlink(filepath.Join("..", "victim.md"), filepath.Join(dir, "s.jsonl")))
assert.Error(t, Append(dir, "s", Entry{SessionID: "s"}))
got, _ := os.ReadFile(victim)
assert.Equal(t, "untouched\n", string(got))
}
func TestReadRefusesSessionIDsThatAreNotPlainNames(t *testing.T) {
root := t.TempDir()
require.NoError(t, os.WriteFile(filepath.Join(root, "x.jsonl"), []byte(`{"session_id":"x"}`+"\n"), 0o644))
_, err := Read(filepath.Join(root, "sessions"), "../x")
assert.Error(t, err)
}
// A name that starts plainly and then climbs out with ".." is exactly what the
// ban on "/" in the id's tail stops: filepath.Join cleans "a/../../x" to a
// path beside the sessions dir without needing sessions/a to exist. The
// assertion is on the refusal itself (ErrInvalidID), not on any error.
func TestAppendRefusesAnIDThatClimbsOutAfterAPlainPrefix(t *testing.T) {
root := t.TempDir()
dir := filepath.Join(root, "sessions")
require.NoError(t, os.MkdirAll(dir, 0o755))
for _, id := range []string{"a/../../escape", "a/../../../escape", "a/../b"} {
err := Append(dir, id, Entry{SessionID: id})
assert.ErrorIs(t, err, ErrInvalidID, "%q", id)
}
_, err := os.Lstat(filepath.Join(root, "escape.jsonl"))
assert.True(t, errors.Is(err, os.ErrNotExist), "nothing created beside the sessions dir")
_, err = os.Lstat(filepath.Join(dir, "b.jsonl"))
assert.True(t, errors.Is(err, os.ErrNotExist), "no in-dir alias created either")
}
// Append does no Lstat type check, so O_NONBLOCK is the only thing that keeps
// a FIFO named <id>.jsonl with no reader from blocking session_log forever.
func TestAppendReturnsPromptlyOnAFIFOLogWithNoReader(t *testing.T) {
dir := filepath.Join(t.TempDir(), "sessions")
require.NoError(t, os.MkdirAll(dir, 0o755))
fifo := filepath.Join(dir, "s.jsonl")
require.NoError(t, syscall.Mkfifo(fifo, 0o644))
done := make(chan error, 1)
go func() { done <- Append(dir, "s", Entry{SessionID: "s"}) }()
select {
case err := <-done:
assert.Error(t, err, "a FIFO with no reader is refused, not written")
case <-time.After(3 * time.Second):
// Unblock the stuck open so the goroutine and TempDir cleanup finish.
if r, err := os.OpenFile(fifo, os.O_RDONLY|syscall.O_NONBLOCK, 0); err == nil {
defer r.Close() //nolint:errcheck
}
<-done
t.Fatal("Append blocked on a FIFO log with no reader")
}
}
// hyperguild#173: session_log followed a symlinked sessions dir.
func TestAppendRefusesASymlinkedSessionsDir(t *testing.T) {
root := t.TempDir()
elsewhere := filepath.Join(root, "elsewhere")
require.NoError(t, os.MkdirAll(elsewhere, 0o755))
dir := filepath.Join(root, "sessions")
require.NoError(t, os.Symlink("elsewhere", dir))
assert.Error(t, Append(dir, "s", Entry{SessionID: "s"}))
ents, _ := os.ReadDir(elsewhere)
assert.Empty(t, ents, "a log was created through the symlinked dir")
}
// The sessions dir is swapped (RENAME_EXCHANGE) with a symlink to a
// directory outside the brain while Append runs. Append opens through an
// os.Root on the sessions dir's parent, so a swap can redirect it only
// inside that parent (the #167 residual), never outside it.
func TestAppendNeverCreatesALogOutsideTheParentUnderADirSwapRace(t *testing.T) {
brain := t.TempDir()
outside := t.TempDir()
dir := filepath.Join(brain, "sessions")
other := filepath.Join(brain, "sessions.link")
require.NoError(t, os.MkdirAll(dir, 0o755))
require.NoError(t, os.Symlink(outside, other))
var stop atomic.Bool
done := make(chan struct{})
go func() {
defer close(done)
for !stop.Load() {
_ = unix.Renameat2(unix.AT_FDCWD, dir, unix.AT_FDCWD, other, unix.RENAME_EXCHANGE)
}
}()
ok := 0
for i := 0; i < 20000; i++ {
if Append(dir, fmt.Sprintf("s%d", i), Entry{SessionID: "s"}) == nil {
ok++
}
}
stop.Store(true)
<-done
ents, _ := os.ReadDir(outside)
assert.Empty(t, ents, "logs were created outside the brain")
assert.Positive(t, ok, "positive control: appends to the real dir succeed")
}
// The log file is swapped between a regular file and a symlink to another
// file in the brain while Append runs: Append writes only after confirming
// the path is the regular file it opened, so the other file never grows.
func TestAppendNeverWritesThroughALogSwappedForASymlink(t *testing.T) {
brain := t.TempDir()
dir := filepath.Join(brain, "sessions")
require.NoError(t, os.MkdirAll(dir, 0o755))
victim := filepath.Join(brain, "victim.md")
require.NoError(t, os.WriteFile(victim, []byte("untouched\n"), 0o644))
log := filepath.Join(dir, "s.jsonl")
var stop atomic.Bool
done := make(chan struct{})
go func() {
defer close(done)
tmp := log + ".swap"
for i := 0; !stop.Load(); i++ {
_ = os.Remove(tmp)
if i%2 == 0 {
_ = os.WriteFile(tmp, nil, 0o644)
} else {
_ = os.Symlink(filepath.Join("..", "victim.md"), tmp)
}
_ = os.Rename(tmp, log)
}
}()
ok := 0
for i := 0; i < 20000; i++ {
if Append(dir, "s", Entry{SessionID: "s"}) == nil {
ok++
}
}
stop.Store(true)
<-done
got, _ := os.ReadFile(victim)
assert.Equal(t, "untouched\n", string(got), "a log line was written through the swapped-in symlink")
assert.Positive(t, ok, "positive control: appends to the regular log succeed")
}
+55 -3
View File
@@ -9,6 +9,8 @@ import (
"io/fs" "io/fs"
"os" "os"
"path/filepath" "path/filepath"
"regexp"
"syscall"
"time" "time"
) )
@@ -42,16 +44,63 @@ type Attempt struct {
RunnerOutput string `json:"runner_output,omitempty"` RunnerOutput string `json:"runner_output,omitempty"`
} }
// idPattern is a plain file name: session_log takes the id from the caller,
// and a path segment in it created or appended files anywhere the pod could
// write (hyperguild#169).
var idPattern = regexp.MustCompile(`^[A-Za-z0-9_][A-Za-z0-9._-]{0,127}$`)
// ErrInvalidID is returned when a session id is not a plain file name.
var ErrInvalidID = errors.New("session id is not a plain name")
func checkID(sessionID string) error {
if !idPattern.MatchString(sessionID) {
return fmt.Errorf("%w: %q", ErrInvalidID, sessionID)
}
return nil
}
// Append writes entry as a single JSON line to sessionsDir/{sessionID}.jsonl. // Append writes entry as a single JSON line to sessionsDir/{sessionID}.jsonl.
// The open does not follow a final symlink.
func Append(sessionsDir, sessionID string, entry Entry) error { func Append(sessionsDir, sessionID string, entry Entry) error {
if err := os.MkdirAll(sessionsDir, 0o755); err != nil { if err := checkID(sessionID); err != nil {
return err
}
// Opened through an os.Root on the sessions dir's parent (the brain dir),
// so a sessions dir swapped for a symlink after the check below can
// redirect the log only inside that parent, never outside it (#173; the
// in-brain swap is the #167 residual).
parentDir, name := filepath.Dir(sessionsDir), filepath.Base(sessionsDir)
if err := os.MkdirAll(parentDir, 0o755); err != nil {
return fmt.Errorf("create sessions parent: %w", err)
}
root, err := os.OpenRoot(parentDir)
if err != nil {
return fmt.Errorf("open sessions parent: %w", err)
}
defer func() { _ = root.Close() }()
if err := root.Mkdir(name, 0o755); err != nil && !errors.Is(err, fs.ErrExist) {
return fmt.Errorf("create sessions dir: %w", err) return fmt.Errorf("create sessions dir: %w", err)
} }
path := filepath.Join(sessionsDir, sessionID+".jsonl") // The sessions dir itself must be a real directory, not a symlink that
f, err := os.OpenFile(path, os.O_CREATE|os.O_APPEND|os.O_WRONLY, 0o644) // redirects the log elsewhere (hyperguild#173).
if fi, err := root.Lstat(name); err != nil {
return fmt.Errorf("stat sessions dir: %w", err)
} else if !fi.IsDir() {
return fmt.Errorf("sessions dir %s is not a real directory", sessionsDir)
}
logRel := name + "/" + sessionID + ".jsonl"
f, err := root.OpenFile(logRel, os.O_CREATE|os.O_APPEND|os.O_WRONLY|syscall.O_NOFOLLOW|syscall.O_NONBLOCK, 0o644)
if err != nil { if err != nil {
return fmt.Errorf("open session log: %w", err) return fmt.Errorf("open session log: %w", err)
} }
// os.Root follows an in-root final symlink even with O_NOFOLLOW, so
// confirm, before writing, that the path is the regular file we opened.
hfi, herr := f.Stat()
lfi, lerr := root.Lstat(logRel)
if herr != nil || lerr != nil || !lfi.Mode().IsRegular() || !os.SameFile(lfi, hfi) {
_ = f.Close()
return fmt.Errorf("session log %s is not a regular file", logRel)
}
line, err := json.Marshal(entry) line, err := json.Marshal(entry)
if err != nil { if err != nil {
@@ -70,6 +119,9 @@ func Append(sessionsDir, sessionID string, entry Entry) error {
// Read returns all entries for sessionID. Returns empty slice if no log exists. // Read returns all entries for sessionID. Returns empty slice if no log exists.
func Read(sessionsDir, sessionID string) ([]Entry, error) { func Read(sessionsDir, sessionID string) ([]Entry, error) {
if err := checkID(sessionID); err != nil {
return nil, err
}
path := filepath.Join(sessionsDir, sessionID+".jsonl") path := filepath.Join(sessionsDir, sessionID+".jsonl")
f, err := os.Open(path) f, err := os.Open(path)
if errors.Is(err, fs.ErrNotExist) { if errors.Is(err, fs.ErrNotExist) {
@@ -0,0 +1,104 @@
package symlinkguard
import (
"context"
"net"
"os"
"path/filepath"
"syscall"
"testing"
"time"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
// hyperguild#179: the tripwire looked only for symlinks, so a FIFO (or
// socket) in the brain passed it. A FIFO at classification.yaml then made
// startup exit; at a note path it could hang readers that skip the handle
// check.
func TestScanReportsFIFOsAndSocketsAsSpecialFiles(t *testing.T) {
brain := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(brain, "wiki", "w"), 0o755))
require.NoError(t, syscall.Mkfifo(filepath.Join(brain, "classification.yaml"), 0o644))
require.NoError(t, syscall.Mkfifo(filepath.Join(brain, "wiki", "w", "f.md"), 0o644))
l, err := net.Listen("unix", filepath.Join(brain, "s.sock"))
require.NoError(t, err)
defer func() { _ = l.Close() }()
require.NoError(t, os.WriteFile(filepath.Join(brain, "wiki", "w", "n.md"), []byte("n"), 0o644))
r, err := Scan(brain)
require.NoError(t, err)
assert.Equal(t, []string{"classification.yaml", "s.sock", "wiki/w/f.md"}, r.Special)
assert.False(t, r.Clean())
assert.Contains(t, r.String(), "special file")
}
// Watch removes a FIFO as it removes symlinks, and leaves a regular file
// (which Scan never reports) alone.
func TestWatchRemovesAFIFOAndLeavesAnUnreportedRegularFile(t *testing.T) {
brain := t.TempDir()
fifo := filepath.Join(brain, "f.md")
require.NoError(t, syscall.Mkfifo(fifo, 0o644))
regular := filepath.Join(brain, "n.md")
require.NoError(t, os.WriteFile(regular, []byte("n"), 0o644))
n := &recNotifier{}
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
go Watch(ctx, brain, 10*time.Millisecond, n)
require.Eventually(t, func() bool {
_, err := os.Lstat(fifo)
return os.IsNotExist(err)
}, 2*time.Second, 10*time.Millisecond, "the FIFO is removed")
require.Eventually(t, func() bool { return n.count() >= 1 }, 2*time.Second, 10*time.Millisecond)
b, err := os.ReadFile(regular)
require.NoError(t, err)
assert.Equal(t, "n", string(b))
n.mu.Lock()
defer n.mu.Unlock()
assert.Contains(t, n.msgs[0], "f.md")
}
// removeSpecial re-checks the type: a path that is a regular file or a
// directory at the re-check is not removed. (A swap between the re-check and
// the unlink can still remove one; Linux has no conditional unlink. The same
// window exists in removeSymlinks.)
func TestRemoveSpecialRefusesRegularFilesAndDirectories(t *testing.T) {
brain := t.TempDir()
require.NoError(t, os.WriteFile(filepath.Join(brain, "n.md"), []byte("n"), 0o644))
require.NoError(t, os.MkdirAll(filepath.Join(brain, "d"), 0o755))
require.NoError(t, syscall.Mkfifo(filepath.Join(brain, "f.md"), 0o644))
removed, failed := removeSpecial(brain, []string{"n.md", "d", "f.md"})
assert.Equal(t, []string{"f.md"}, removed)
assert.Equal(t, []string{"n.md", "d"}, failed)
_, err := os.Stat(filepath.Join(brain, "n.md"))
assert.NoError(t, err)
}
// A special file Watch cannot remove (its directory is read-only) stays
// reported: the persistent-finding alert names it.
func TestWatchReportsASpecialFileItCannotRemove(t *testing.T) {
if os.Geteuid() == 0 {
t.Skip("root ignores directory permissions")
}
brain := t.TempDir()
ro := filepath.Join(brain, "ro")
require.NoError(t, os.MkdirAll(ro, 0o755))
require.NoError(t, syscall.Mkfifo(filepath.Join(ro, "f.md"), 0o644))
require.NoError(t, os.Chmod(ro, 0o555))
t.Cleanup(func() { _ = os.Chmod(ro, 0o755) })
n := &recNotifier{}
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
go Watch(ctx, brain, 10*time.Millisecond, n)
require.Eventually(t, func() bool { return n.count() >= 1 }, 2*time.Second, 10*time.Millisecond)
_, err := os.Lstat(filepath.Join(ro, "f.md"))
require.NoError(t, err, "the FIFO could not be removed")
n.mu.Lock()
defer n.mu.Unlock()
assert.Contains(t, n.msgs[0], "ro/f.md")
assert.Contains(t, n.msgs[0], "special file")
for _, m := range n.msgs {
assert.NotContains(t, m, "removed", "a file it could not remove was reported as removed")
}
}
+153 -7
View File
@@ -18,9 +18,14 @@ import (
"sort" "sort"
"strconv" "strconv"
"strings" "strings"
"sync"
"sync/atomic"
"time" "time"
) )
// shutdownFlushTimeout bounds the final flush when Watch stops.
var shutdownFlushTimeout = 5 * time.Second
// DefaultInterval is used when Watch is given a non-positive interval. // DefaultInterval is used when Watch is given a non-positive interval.
const DefaultInterval = 30 * time.Second const DefaultInterval = 30 * time.Second
@@ -41,10 +46,16 @@ type Notifier interface {
type Result struct { type Result struct {
Symlinks []string Symlinks []string
Unreadable []string Unreadable []string
// Special holds FIFOs, sockets and devices (hyperguild#179): a reader
// that opens one by path can hang, and a FIFO at classification.yaml
// stops startup.
Special []string
} }
// Clean reports whether the scan found nothing to act on. // Clean reports whether the scan found nothing to act on.
func (r Result) Clean() bool { return len(r.Symlinks) == 0 && len(r.Unreadable) == 0 } func (r Result) Clean() bool {
return len(r.Symlinks) == 0 && len(r.Unreadable) == 0 && len(r.Special) == 0
}
// String describes the findings for a log line or alert. Paths are quoted, so // String describes the findings for a log line or alert. Paths are quoted, so
// a name containing a newline cannot blur two findings together. // a name containing a newline cannot blur two findings together.
@@ -54,6 +65,9 @@ func (r Result) String() string {
if len(r.Symlinks) > 0 { if len(r.Symlinks) > 0 {
parts = append(parts, fmt.Sprintf("%d symlink(s): %s", len(r.Symlinks), q(r.Symlinks))) parts = append(parts, fmt.Sprintf("%d symlink(s): %s", len(r.Symlinks), q(r.Symlinks)))
} }
if len(r.Special) > 0 {
parts = append(parts, fmt.Sprintf("%d special file(s) (FIFO, socket or device; delete them): %s", len(r.Special), q(r.Special)))
}
if len(r.Unreadable) > 0 { if len(r.Unreadable) > 0 {
parts = append(parts, fmt.Sprintf("%d unreadable dir(s): %s", len(r.Unreadable), q(r.Unreadable))) parts = append(parts, fmt.Sprintf("%d unreadable dir(s): %s", len(r.Unreadable), q(r.Unreadable)))
} }
@@ -108,12 +122,16 @@ func Scan(brainDir string) (Result, error) {
if d.IsDir() && p == rootGit { if d.IsDir() && p == rootGit {
return filepath.SkipDir return filepath.SkipDir
} }
if d.Type()&os.ModeSymlink != 0 { switch {
case d.Type()&os.ModeSymlink != 0:
r.Symlinks = append(r.Symlinks, rel) r.Symlinks = append(r.Symlinks, rel)
case !d.IsDir() && !d.Type().IsRegular():
r.Special = append(r.Special, rel)
} }
return nil return nil
}) })
sort.Strings(r.Symlinks) sort.Strings(r.Symlinks)
sort.Strings(r.Special)
sort.Strings(r.Unreadable) sort.Strings(r.Unreadable)
if walkErr != nil { if walkErr != nil {
// Only the root's own error reaches here (deeper errors are recorded // Only the root's own error reaches here (deeper errors are recorded
@@ -133,13 +151,32 @@ func Scan(brainDir string) (Result, error) {
// queue until Notify succeeds (see alertQueue): an outage delays them, and // queue until Notify succeeds (see alertQueue): an outage delays them, and
// any that do not fit the bounded queue, or are undeliverable, are counted in // any that do not fit the bounded queue, or are undeliverable, are counted in
// a summary that is never dropped, so every incident is delivered or counted. // a summary that is never dropped, so every incident is delivered or counted.
//
// Delivery runs in its own goroutine (hyperguild#164): a hung notifier, with
// a full queue, blocked one flush for several notify timeouts, and while the
// scan waited on it a planted link stayed readable for several intervals.
// The scan hands alerts over without blocking; any that do not fit are
// counted in the summary. When ctx ends, Watch returns after a final flush
// of what was handed over (bounded by shutdownFlushTimeout); alerts that
// flush cannot deliver are counted in one error log line instead. That
// shutdown path only runs for a caller that cancels ctx: cmd/server passes
// context.Background(), so in production the flush on handoff is what keeps
// alerts from waiting on a restart.
func Watch(ctx context.Context, brainDir string, interval time.Duration, n Notifier) { func Watch(ctx context.Context, brainDir string, interval time.Duration, n Notifier) {
if interval <= 0 { if interval <= 0 {
interval = DefaultInterval interval = DefaultInterval
} }
t := time.NewTicker(interval) t := time.NewTicker(interval)
defer t.Stop() defer t.Stop()
q := &alertQueue{} alerts := make(chan string, maxPending)
var overflow atomic.Int64
var wg sync.WaitGroup
wg.Add(1)
go func() { defer wg.Done(); deliver(ctx, interval, n, alerts, &overflow) }()
// The scan is the only sender: closing alerts tells deliver to drain and
// flush once more, and Watch returns only after it has.
defer func() { close(alerts); wg.Wait() }()
q := alertSender{alerts: alerts, overflow: &overflow}
reported := "" reported := ""
for { for {
select { select {
@@ -158,17 +195,92 @@ func Watch(ctx context.Context, brainDir string, interval time.Duration, n Notif
q.push(fmt.Sprintf("brain symlink tripwire (hyperguild#157): removed %d symlink(s) from the brain dir: %s", q.push(fmt.Sprintf("brain symlink tripwire (hyperguild#157): removed %d symlink(s) from the brain dir: %s",
len(removed), Result{}.quoted(removed))) len(removed), Result{}.quoted(removed)))
} }
key = Result{Symlinks: failed, Unreadable: r.Unreadable}.String() sRemoved, sFailed := removeSpecial(brainDir, r.Special)
if len(sRemoved) > 0 {
slog.Error("brain symlink tripwire removed special files", "count", len(sRemoved))
q.push(fmt.Sprintf("brain symlink tripwire (hyperguild#179): removed %d special file(s) (FIFO, socket or device) from the brain dir: %s",
len(sRemoved), Result{}.quoted(sRemoved)))
}
key = Result{Symlinks: failed, Unreadable: r.Unreadable, Special: sFailed}.String()
} }
if key != reported { if key != reported {
reported = key reported = key
if key != "" { if key != "" {
slog.Error("brain symlink tripwire", "finding", key) slog.Error("brain symlink tripwire", "finding", key)
q.push("brain symlink tripwire (hyperguild#157): " + key + q.push("brain symlink tripwire (hyperguild#157, #179): " + key +
". Readers may follow symlinks out of the brain; fix them.") ". Symlinks can lead readers out of the brain and special files can hang them; remove them.")
} }
} }
q.flush(ctx, n) }
}
// alertSender hands alerts to deliver without ever blocking the scan; an
// alert that does not fit is counted in overflow, which deliver adds to the
// summary.
type alertSender struct {
alerts chan<- string
overflow *atomic.Int64
}
func (s alertSender) push(msg string) {
select {
case s.alerts <- msg:
default:
s.overflow.Add(1)
}
}
// deliver owns the alertQueue. It flushes as soon as an alert is handed
// over, and again every interval while any are pending (outage retries).
// When alerts is closed (Watch is stopping) it drains it and flushes once
// more, bounded by shutdownFlushTimeout, so an alert handed over just before
// shutdown is not lost (PR #180 review). Whatever that flush cannot deliver
// is logged as one error with its counts, never dropped silently.
func deliver(ctx context.Context, interval time.Duration, n Notifier, alerts <-chan string, overflow *atomic.Int64) {
t := time.NewTicker(interval)
defer t.Stop()
q := &alertQueue{}
// drain moves everything handed over into the queue; it reports false
// once alerts is closed.
drain := func() bool {
defer func() { q.suppressed += int(overflow.Swap(0)) }()
for {
select {
case msg, ok := <-alerts:
if !ok {
return false
}
q.push(msg)
default:
return true
}
}
}
for {
select {
case msg, ok := <-alerts:
if ok {
q.push(msg)
ok = drain()
}
if !ok {
q.suppressed += int(overflow.Swap(0))
fctx, cancel := context.WithTimeout(context.WithoutCancel(ctx), shutdownFlushTimeout)
q.flush(fctx, n)
cancel()
if len(q.items) > 0 || q.suppressed > 0 {
slog.Error("symlink tripwire alerts not delivered at shutdown",
"undelivered", len(q.items), "suppressed", q.suppressed)
}
return
}
q.flush(ctx, n)
case <-t.C:
if !drain() {
continue // the closed channel is handled by the case above
}
q.flush(ctx, n)
}
} }
} }
@@ -229,6 +341,40 @@ func permanent(err error) bool {
return errors.As(err, &p) && p.Permanent() return errors.As(err, &p) && p.Permanent()
} }
// removeSpecial unlinks each brain-relative FIFO, socket or device through
// an os.Root on the brain dir, after re-checking that it is still one: a
// path that is a regular file, a directory or a symlink at the re-check is
// left alone and reported as failed. The re-check narrows, but cannot close,
// the window before the unlink (Linux has no conditional unlink); the same
// window exists in removeSymlinks, and only a local writer can use it.
func removeSpecial(brainDir string, rels []string) (removed, failed []string) {
if len(rels) == 0 {
return nil, nil
}
resolved, err := filepath.EvalSymlinks(brainDir)
if err != nil {
return nil, rels
}
root, err := os.OpenRoot(resolved)
if err != nil {
return nil, rels
}
defer func() { _ = root.Close() }()
for _, rel := range rels {
fi, err := root.Lstat(rel)
if err != nil || fi.IsDir() || fi.Mode().IsRegular() || fi.Mode()&os.ModeSymlink != 0 {
failed = append(failed, rel)
continue
}
if err := root.Remove(rel); err != nil {
failed = append(failed, rel)
continue
}
removed = append(removed, rel)
}
return removed, failed
}
// removeSymlinks unlinks each brain-relative symlink through an os.Root on // removeSymlinks unlinks each brain-relative symlink through an os.Root on
// the brain dir, after re-checking it is still a symlink. Removing a symlink // the brain dir, after re-checking it is still a symlink. Removing a symlink
// never touches its target. It returns what was removed and what was not. // never touches its target. It returns what was removed and what was not.
@@ -1,10 +1,12 @@
package symlinkguard package symlinkguard
import ( import (
"bytes"
"context" "context"
"errors" "errors"
"fmt" "fmt"
"os" "os"
"log/slog"
"path/filepath" "path/filepath"
"strconv" "strconv"
"strings" "strings"
@@ -345,13 +347,19 @@ func TestWatchDropsAMessageThatNeverSendsSoLaterAlertsFlow(t *testing.T) {
}, 3*time.Second, 10*time.Millisecond, "a later alert is delivered despite a poison message") }, 3*time.Second, 10*time.Millisecond, "a later alert is delivered despite a poison message")
} }
// toggleNotifier fails while down is set. // toggleNotifier fails while down is set. Like the real audit.NtfyNotifier
// (http.NewRequestWithContext), it also fails, without recording, when its
// ctx is already done (PR #180 round 2: notifiers that ignored ctx let a
// final flush on the cancelled ctx pass).
type toggleNotifier struct { type toggleNotifier struct {
recNotifier recNotifier
down atomic.Bool down atomic.Bool
} }
func (r *toggleNotifier) Notify(ctx context.Context, m string) error { func (r *toggleNotifier) Notify(ctx context.Context, m string) error {
if err := ctx.Err(); err != nil {
return err
}
if r.down.Load() { if r.down.Load() {
return errors.New("ntfy down") return errors.New("ntfy down")
} }
@@ -373,18 +381,19 @@ func TestWatchQueueIsBoundedDuringAnOutage(t *testing.T) {
require.Eventually(t, func() bool { _, err := os.Lstat(link); return os.IsNotExist(err) }, 2*time.Second, 2*time.Millisecond) require.Eventually(t, func() bool { _, err := os.Lstat(link); return os.IsNotExist(err) }, 2*time.Second, 2*time.Millisecond)
} }
n.down.Store(false) n.down.Store(false)
require.Eventually(t, func() bool { return n.count() >= 1 }, 2*time.Second, 5*time.Millisecond) // Synchronise on the summary being delivered (hyperguild#193: a fixed
time.Sleep(100 * time.Millisecond) // sleep flaked under load); once it is, the queue has been flushed.
assert.LessOrEqual(t, n.count(), maxPending+1, "queued alerts plus at most one suppression summary") require.Eventually(t, func() bool {
n.mu.Lock() n.mu.Lock()
defer n.mu.Unlock() defer n.mu.Unlock()
found := false for _, m := range n.msgs {
for _, m := range n.msgs { if strings.Contains(m, "suppressed") {
if strings.Contains(m, "suppressed") { return true
found = true }
} }
} return false
assert.True(t, found, "suppressed alerts are counted in a summary") }, 10*time.Second, 5*time.Millisecond, "suppressed alerts are counted in a summary")
assert.LessOrEqual(t, n.count(), maxPending+1, "queued alerts plus at most one suppression summary")
} }
// Within the bound, every alert queued during an outage is delivered after // Within the bound, every alert queued during an outage is delivered after
@@ -403,11 +412,21 @@ func TestWatchDeliversEveryQueuedAlertAfterAShortOutage(t *testing.T) {
require.Eventually(t, func() bool { _, err := os.Lstat(link); return os.IsNotExist(err) }, 2*time.Second, 2*time.Millisecond) require.Eventually(t, func() bool { _, err := os.Lstat(link); return os.IsNotExist(err) }, 2*time.Second, 2*time.Millisecond)
} }
n.down.Store(false) n.down.Store(false)
require.Eventually(t, func() bool { return n.count() == 3 }, 2*time.Second, 5*time.Millisecond) require.Eventually(t, func() bool { return n.count() >= 3 }, 10*time.Second, 5*time.Millisecond)
n.mu.Lock() n.mu.Lock()
defer n.mu.Unlock() defer n.mu.Unlock()
for i, name := range []string{"qa", "qb", "qc"} { // Each removal delivered exactly once, in any order: the outage ends
assert.Contains(t, n.msgs[i], strconv.Quote(name)) // asynchronously, possibly mid-flush, so a later item can send before an
// earlier one (hyperguild#193). Order is not what this test is about.
assert.Len(t, n.msgs, 3)
for _, name := range []string{"qa", "qb", "qc"} {
hits := 0
for _, m := range n.msgs {
if strings.Contains(m, strconv.Quote(name)) {
hits++
}
}
assert.Equal(t, 1, hits, "%s delivered exactly once", name)
} }
} }
@@ -449,29 +468,30 @@ func TestWatchAccountsForEveryRemovalDuringAnOutage(t *testing.T) {
require.Eventually(t, func() bool { _, err := os.Lstat(link); return os.IsNotExist(err) }, 2*time.Second, 1*time.Millisecond) require.Eventually(t, func() bool { _, err := os.Lstat(link); return os.IsNotExist(err) }, 2*time.Second, 1*time.Millisecond)
} }
n.down.Store(false) n.down.Store(false)
require.Eventually(t, func() bool { // Synchronise on the record being complete, not on a fixed sleep after
// the first summary (hyperguild#193): sum every "removed N" alert and
// every suppression count until they account for all plants.
account := func() (delivered, counted int) {
n.mu.Lock() n.mu.Lock()
defer n.mu.Unlock() defer n.mu.Unlock()
for _, m := range n.msgs { for _, m := range n.msgs {
if strings.Contains(m, "suppressed") { var k int
return true if i := strings.Index(m, "removed "); i >= 0 && strings.Contains(m, "symlink(s)") {
if _, err := fmt.Sscanf(m[i+len("removed "):], "%d", &k); err == nil {
delivered += k
}
}
if i := strings.Index(m, "): "); i >= 0 && strings.Contains(m, "suppressed") {
if _, err := fmt.Sscanf(m[i+3:], "%d", &k); err == nil {
counted += k
}
} }
} }
return false return delivered, counted
}, 2*time.Second, 5*time.Millisecond)
time.Sleep(50 * time.Millisecond)
n.mu.Lock()
defer n.mu.Unlock()
delivered, counted := 0, 0
for _, m := range n.msgs {
if strings.Contains(m, "removed 1 symlink(s)") {
delivered++
}
if i := strings.Index(m, "): "); i >= 0 && strings.Contains(m, "suppressed") {
_, _ = fmt.Sscanf(m[i+3:], "%d", &counted)
}
} }
assert.Equal(t, plants, delivered+counted, "delivered %d + counted %d must equal %d removals", delivered, counted, plants) require.Eventually(t, func() bool { d, c := account(); return d+c >= plants }, 10*time.Second, 5*time.Millisecond)
d, c := account()
assert.Equal(t, plants, d+c, "delivered %d + counted %d must equal %d removals", d, c, plants)
} }
// A poison message is still dropped (others send while it fails), and the // A poison message is still dropped (others send while it fails), and the
@@ -496,3 +516,271 @@ func TestWatchDropsAPoisonMessageAndCountsIt(t *testing.T) {
return false return false
}, 3*time.Second, 5*time.Millisecond, "the dropped poison alert is counted in a summary") }, 3*time.Second, 5*time.Millisecond, "the dropped poison alert is counted in a summary")
} }
// hungNotifier answers every Notify only after latency (or when ctx ends),
// with an outage error: an ntfy that accepts connections but never responds
// until its timeout.
type hungNotifier struct{ latency time.Duration }
func (h hungNotifier) Notify(ctx context.Context, _ string) error {
select {
case <-time.After(h.latency):
case <-ctx.Done():
}
return errors.New("ntfy timeout")
}
func waitGone(t *testing.T, link string) {
t.Helper()
require.Eventually(t, func() bool { _, err := os.Lstat(link); return os.IsNotExist(err) }, 10*time.Second, time.Millisecond)
}
// hyperguild#164: alerts were flushed in the scan goroutine, so with a hung
// ntfy and a full queue one flush (6 x notify timeout) delayed the next scan.
// The notifier here is slower than the interval, as in production (10 s
// timeout vs 30 s interval with 6 messages), so the alert handoff is full:
// a scan that waited on it would keep a link for seconds.
func TestWatchLinkLifetimeStaysNearOneIntervalWithANotifierSlowerThanTheInterval(t *testing.T) {
const interval = 100 * time.Millisecond
brain := t.TempDir()
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
go Watch(ctx, brain, interval, hungNotifier{latency: 400 * time.Millisecond})
for i := 0; i < 2*maxPending+3; i++ { // overfill queue and handoff
link := filepath.Join(brain, "fill"+strconv.Itoa(i))
require.NoError(t, os.Symlink(t.TempDir(), link))
waitGone(t, link)
}
var worst time.Duration
for i := 0; i < 6; i++ {
link := filepath.Join(brain, "probe"+strconv.Itoa(i))
require.NoError(t, os.Symlink(t.TempDir(), link))
planted := time.Now()
waitGone(t, link)
worst = max(worst, time.Since(planted))
}
assert.LessOrEqual(t, worst, interval+interval/2, "worst link lifetime %v with a hung notifier", worst)
}
// Review of PR #180: delivery waited for its own tick, about one interval
// after the removal. An alert is now flushed as soon as it is handed over.
func TestWatchDeliversARemovalAlertWellWithinOneInterval(t *testing.T) {
const interval = 400 * time.Millisecond
brain := t.TempDir()
n := &recNotifier{}
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
go Watch(ctx, brain, interval, n)
link := filepath.Join(brain, "prompt")
require.NoError(t, os.Symlink(t.TempDir(), link))
waitGone(t, link)
removed := time.Now()
require.Eventually(t, func() bool { return n.count() == 1 }, 2*time.Second, time.Millisecond)
assert.Less(t, time.Since(removed), interval/4)
}
// Review of PR #180: on ctx end, deliver returned without draining what the
// scan had handed over, so a removal just before shutdown was neither
// delivered nor counted. Watch now drains and flushes before returning.
func TestWatchDeliversAnAlertHandedOverJustBeforeShutdown(t *testing.T) {
for i := 0; i < 20; i++ {
brain := t.TempDir()
n := &recNotifier{}
ctx, cancel := context.WithCancel(context.Background())
done := make(chan struct{})
go func() { Watch(ctx, brain, 5*time.Millisecond, n); close(done) }()
link := filepath.Join(brain, "late")
require.NoError(t, os.Symlink(t.TempDir(), link))
waitGone(t, link)
cancel()
<-done
require.Equal(t, 1, n.count(), "iteration %d: the removal alert was lost at shutdown", i)
}
}
// blockingNotifier holds every Notify until its ctx ends, tracking calls in
// flight.
type blockingNotifier struct{ inFlight atomic.Int32 }
func (b *blockingNotifier) Notify(ctx context.Context, _ string) error {
b.inFlight.Add(1)
defer b.inFlight.Add(-1)
<-ctx.Done()
return ctx.Err()
}
// Watch does not return while its delivery goroutine is still notifying.
func TestWatchReturnsOnlyAfterDeliveryHasStopped(t *testing.T) {
prev := shutdownFlushTimeout
shutdownFlushTimeout = 50 * time.Millisecond
t.Cleanup(func() { shutdownFlushTimeout = prev })
brain := t.TempDir()
n := &blockingNotifier{}
ctx, cancel := context.WithCancel(context.Background())
done := make(chan struct{})
go func() { Watch(ctx, brain, 5*time.Millisecond, n); close(done) }()
link := filepath.Join(brain, "x")
require.NoError(t, os.Symlink(t.TempDir(), link))
waitGone(t, link)
require.Eventually(t, func() bool { return n.inFlight.Load() == 1 }, 2*time.Second, time.Millisecond)
cancel()
select {
case <-done:
case <-time.After(5 * time.Second):
t.Fatal("Watch did not return after cancel")
}
assert.Zero(t, n.inFlight.Load(), "Notify still running after Watch returned")
}
// slowToggleNotifier fails slowly (a hung ntfy) while down, then records.
type slowToggleNotifier struct {
recNotifier
down atomic.Bool
latency time.Duration
}
func (r *slowToggleNotifier) Notify(ctx context.Context, m string) error {
if r.down.Load() {
select {
case <-time.After(r.latency):
case <-ctx.Done():
}
return errors.New("ntfy timeout")
}
return r.recNotifier.Notify(ctx, m)
}
// Review of PR #180: with a slow notifier the scan's handoff overflows, and
// those alerts exist only in the overflow counter. Every removal is still
// delivered or counted in the summary.
func TestWatchCountsHandoffOverflowWithASlowNotifier(t *testing.T) {
brain := t.TempDir()
n := &slowToggleNotifier{latency: 30 * time.Millisecond}
n.down.Store(true)
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
go Watch(ctx, brain, 2*time.Millisecond, n)
const plants = 40
for i := 0; i < plants; i++ {
link := filepath.Join(brain, "ov"+strconv.Itoa(i))
require.NoError(t, os.Symlink(t.TempDir(), link))
waitGone(t, link)
}
n.down.Store(false)
delivered, counted := 0, 0
require.Eventually(t, func() bool {
n.mu.Lock()
defer n.mu.Unlock()
delivered, counted = 0, 0
for _, m := range n.msgs {
if strings.Contains(m, "removed 1 symlink(s)") {
delivered++
}
if i := strings.Index(m, "): "); i >= 0 && strings.Contains(m, "suppressed") {
var c int
_, _ = fmt.Sscanf(m[i+3:], "%d", &c)
counted += c
}
}
return delivered+counted == plants
}, 5*time.Second, 10*time.Millisecond, "delivered %d + counted %d of %d", delivered, counted, plants)
}
// An alert still queued when Watch stops (its send failed, the notifier has
// since recovered, the retry tick has not come) is flushed once more before
// Watch returns, on a ctx that is not the cancelled one: toggleNotifier fails
// on a done ctx, as the real ntfy notifier does.
func TestWatchFinalFlushSendsAQueuedAlertOnACtxThatOutlivesTheCancel(t *testing.T) {
brain := t.TempDir()
n := &toggleNotifier{}
n.down.Store(true)
ctx, cancel := context.WithCancel(context.Background())
done := make(chan struct{})
go func() { Watch(ctx, brain, 300*time.Millisecond, n); close(done) }()
link := filepath.Join(brain, "queued")
require.NoError(t, os.Symlink(t.TempDir(), link))
waitGone(t, link)
time.Sleep(20 * time.Millisecond) // let the immediate send fail
n.down.Store(false)
cancel()
<-done
assert.Equal(t, 1, n.count(), "the queued alert was not flushed at shutdown")
}
// lockedBuffer is a goroutine-safe log sink: Watch goroutines of earlier
// tests may still log while one test captures.
type lockedBuffer struct {
mu sync.Mutex
buf bytes.Buffer
}
func (b *lockedBuffer) Write(p []byte) (int, error) {
b.mu.Lock()
defer b.mu.Unlock()
return b.buf.Write(p)
}
func (b *lockedBuffer) String() string { b.mu.Lock(); defer b.mu.Unlock(); return b.buf.String() }
func captureLog(t *testing.T) *lockedBuffer {
t.Helper()
b := &lockedBuffer{}
prev := slog.Default()
slog.SetDefault(slog.New(slog.NewTextHandler(b, nil)))
t.Cleanup(func() { slog.SetDefault(prev) })
return b
}
// PR #180 round 2: when the final flush fails (ntfy hung at shutdown) the
// undelivered alerts vanished with only "will retry" warnings. Watch now logs
// one error with both counts.
func TestWatchLogsUndeliveredAndSuppressedCountsWhenTheFinalFlushFails(t *testing.T) {
prev := shutdownFlushTimeout
shutdownFlushTimeout = 50 * time.Millisecond
t.Cleanup(func() { shutdownFlushTimeout = prev })
logs := captureLog(t)
brain := t.TempDir()
ctx, cancel := context.WithCancel(context.Background())
done := make(chan struct{})
go func() { Watch(ctx, brain, 5*time.Millisecond, hungNotifier{latency: time.Hour}); close(done) }()
link := filepath.Join(brain, "lost")
require.NoError(t, os.Symlink(t.TempDir(), link))
waitGone(t, link)
cancel()
<-done
out := logs.String()
assert.Contains(t, out, "level=ERROR")
assert.Contains(t, out, `msg="symlink tripwire alerts not delivered at shutdown"`)
assert.Contains(t, out, "undelivered=1")
assert.Contains(t, out, "suppressed=0")
}
// deliver, given a handoff that already holds alerts and is closed, moves them
// into the queue and runs the final flush, even though the ctx is cancelled:
// the drain-on-close path, without depending on when the close lands.
func TestDeliverFlushesAlertsDrainedFromAClosedHandoffOnACancelledCtx(t *testing.T) {
alerts := make(chan string, 3)
alerts <- "a1"
alerts <- "a2"
close(alerts)
ctx, cancel := context.WithCancel(context.Background())
cancel()
n := &toggleNotifier{}
var overflow atomic.Int64
deliver(ctx, time.Hour, n, alerts, &overflow)
assert.Equal(t, []string{"a1", "a2"}, n.msgs)
}
// PR #180 round 2: when the first receive after close finds the handoff empty,
// overflow counted by the scan was never moved into the summary.
func TestDeliverCountsOverflowWhenTheHandoffIsClosedAndEmpty(t *testing.T) {
alerts := make(chan string)
close(alerts)
n := &recNotifier{}
var overflow atomic.Int64
overflow.Store(3)
deliver(context.Background(), time.Hour, n, alerts, &overflow)
require.Len(t, n.msgs, 1)
assert.Contains(t, n.msgs[0], "3 further alert(s)")
}
@@ -42,6 +42,11 @@ func TestSyncRefusesSymlinkedNotesAndSurvivesSwapRaces(t *testing.T) {
brain := t.TempDir() brain := t.TempDir()
writeNote(t, brain, "wiki/w/facts/ok.md", "inside note") writeNote(t, brain, "wiki/w/facts/ok.md", "inside note")
require.NoError(t, os.Symlink(brainfstest.Outside(t), filepath.Join(brain, "wiki", "w", "facts", "leak.md"))) require.NoError(t, os.Symlink(brainfstest.Outside(t), filepath.Join(brain, "wiki", "w", "facts", "leak.md")))
// A link to a file elsewhere in the brain: os.Root allows it, only the
// regular-file checks refuse it (hyperguild#171). raw/ is not a scan root,
// so its sentinel can only arrive through the link.
inBrain := brainfstest.WriteSentinelAt(t, filepath.Join(brain, "raw", "s.md"))
require.NoError(t, os.Symlink(filepath.Join("..", "..", "..", "raw", "s.md"), filepath.Join(brain, "wiki", "w", "facts", "inlink.md")))
emb := &recordingEmbedder{} emb := &recordingEmbedder{}
store := &stubStore{} store := &stubStore{}
@@ -51,6 +56,7 @@ func TestSyncRefusesSymlinkedNotesAndSurvivesSwapRaces(t *testing.T) {
assert.NotContains(t, emb.all(), brainfstest.Sentinel) assert.NotContains(t, emb.all(), brainfstest.Sentinel)
for p := range store.upserts { for p := range store.upserts {
assert.NotContains(t, p, "leak.md") assert.NotContains(t, p, "leak.md")
assert.NotContains(t, p, "inlink.md")
} }
p := filepath.Join(brain, "knowledge", "r.md") p := filepath.Join(brain, "knowledge", "r.md")
@@ -61,6 +67,7 @@ func TestSyncRefusesSymlinkedNotesAndSurvivesSwapRaces(t *testing.T) {
return e.all() return e.all()
} }
brainfstest.AssertNoLeak(t, p, "inside", 1500*time.Millisecond, sync) brainfstest.AssertNoLeak(t, p, "inside", 1500*time.Millisecond, sync)
brainfstest.AssertNoLeakFrom(t, p, "inside", inBrain, 1500*time.Millisecond, sync)
brainfstest.AssertNoHang(t, p, "inside", 1500*time.Millisecond, 6*time.Second, func() { _ = sync() }) brainfstest.AssertNoHang(t, p, "inside", 1500*time.Millisecond, 6*time.Second, func() { _ = sync() })
} }
+2 -4
View File
@@ -3,8 +3,6 @@ package wiki
import ( import (
"fmt" "fmt"
"os"
"path/filepath"
"strings" "strings"
"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs" "github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
@@ -50,8 +48,8 @@ func RebuildIndex(brainDir, date string) error {
sb.WriteString("\n") sb.WriteString("\n")
} }
dest := filepath.Join(brainDir, "wiki", "index.md") // Written through the brain root, never through a symlink (#158).
return os.WriteFile(dest, []byte(sb.String()), 0o644) return bfs.PutRegular("wiki/index.md", []byte(sb.String()))
} }
// pageFirstSentence reads the page through the brain root, regular files // pageFirstSentence reads the page through the brain root, regular files
+7 -8
View File
@@ -3,8 +3,7 @@ package wiki
import ( import (
"fmt" "fmt"
"os" "github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
"path/filepath"
"strings" "strings"
) )
@@ -27,14 +26,14 @@ func AppendLog(brainDir, source string, pages, warnings []string, date string) e
} }
sb.WriteString("\n") sb.WriteString("\n")
logPath := filepath.Join(brainDir, "log.md") // Appended through the brain root, never through a symlink (#158).
f, err := os.OpenFile(logPath, os.O_APPEND|os.O_CREATE|os.O_WRONLY, 0o644) bfs, err := brainfs.Open(brainDir)
if err != nil { if err != nil {
return fmt.Errorf("open log: %w", err) return fmt.Errorf("open brain: %w", err)
} }
if _, err = f.WriteString(sb.String()); err != nil { defer func() { _ = bfs.Close() }()
f.Close() //nolint:errcheck if err := bfs.AppendRegular("log.md", []byte(sb.String())); err != nil {
return fmt.Errorf("write log: %w", err) return fmt.Errorf("write log: %w", err)
} }
return f.Close() return nil
} }
@@ -0,0 +1,35 @@
package wiki
import (
"os"
"path/filepath"
"testing"
"github.com/stretchr/testify/assert"
"github.com/stretchr/testify/require"
)
// hyperguild#158: RebuildIndex wrote wiki/index.md and AppendLog appended
// log.md with os.WriteFile / os.OpenFile, so a symlinked file was written
// through.
func TestRebuildIndexAndAppendLogDoNotWriteThroughSymlinks(t *testing.T) {
b := t.TempDir()
require.NoError(t, os.MkdirAll(filepath.Join(b, "wiki", "concepts"), 0o755))
victim := filepath.Join(b, "raw", "v.md")
require.NoError(t, os.MkdirAll(filepath.Dir(victim), 0o755))
require.NoError(t, os.WriteFile(victim, []byte("untouched"), 0o644))
require.NoError(t, os.Symlink(filepath.Join("..", "raw", "v.md"), filepath.Join(b, "wiki", "index.md")))
require.NoError(t, os.Symlink(filepath.Join("raw", "v.md"), filepath.Join(b, "log.md")))
assert.Error(t, RebuildIndex(b, "2026-10-05"))
assert.Error(t, AppendLog(b, "src", []string{"p"}, nil, "2026-10-05"))
got, _ := os.ReadFile(victim)
assert.Equal(t, "untouched", string(got))
require.NoError(t, os.Remove(filepath.Join(b, "wiki", "index.md")))
require.NoError(t, os.Remove(filepath.Join(b, "log.md")))
require.NoError(t, RebuildIndex(b, "2026-10-05"), "positive control")
require.NoError(t, AppendLog(b, "src", []string{"p"}, nil, "2026-10-05"), "positive control")
l, _ := os.ReadFile(filepath.Join(b, "log.md"))
assert.Contains(t, string(l), "src")
}