test(api): race the pdf/docmark temp-copy path too

PR #156 round 4: TextIn's pdf/docmark branch copies the file through the
root to a temp file, and pointing the extractor back at the raw/ path by
name stayed green (race tests used only r.md); a stub docmark then got
the outside file's bytes 34k times in 4s.

TestDocmarkExtractionLeaksNothingUnderASymlinkSwapRace races a symlink
into raw/r.docx through ReadIngestFile while a fake docmark counts the
sentinel in every document it receives. Flipped: the by-name revert is
red.

Refs #153

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

ingestion/cmd/server/main.go

@@ -33,7 +33,6 @@ import (
 	"github.com/mathiasbq/hyperguild/ingestion/internal/reranker"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/search"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/startupretry"
-	"github.com/mathiasbq/hyperguild/ingestion/internal/symlinkguard"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/vectorstore"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/watcher"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/webhook"
@@ -174,8 +173,7 @@ func alertingWarnings(getenv func(string) string) []string {
 	var out []string
 	if getenv("BRAIN_NTFY_URL") == "" {
 		out = append(out, "alerting DISABLED: BRAIN_NTFY_URL unset — "+
-			"brain_edit_root governance rewrites, capture audit degradation and the "+
+			"brain_edit_root governance rewrites and capture audit degradation will not notify")
-			"brain symlink tripwire (hyperguild#157) will not notify")
 	}
 	return out
 }
@@ -241,10 +239,6 @@ func main() {
 	}
 
 	brainDir := envOr("INGEST_BRAIN_DIR", "../brain")
-	if err := startupSymlinkCheck(brainDir); err != nil {
-		logger.Error("brain symlink tripwire", "err", err)
-		os.Exit(1)
-	}
 	port := envOr("INGEST_PORT", "3300")
 
 	llmURL := envOr("INGEST_LLM_URL", "http://iguana:4000/v1")
@@ -366,8 +360,6 @@ func main() {
 	}
 
 	ctx := context.Background()
-	// hyperguild#157: report a symlink that appears in the brain while running.
-	go symlinkguard.Watch(ctx, brainDir, symlinkScanInterval(os.Getenv), buildNotifier())
 
 	// Built before the watcher, the graph backfill and session ingestion, and
 	// retried rather than fatal (#83). Two separate reasons:

ingestion/cmd/server/symlink_check.go

@@ -1,39 +0,0 @@
-package main
-
-import (
-	"errors"
-	"fmt"
-	"os"
-	"strconv"
-	"time"
-
-	"github.com/mathiasbq/hyperguild/ingestion/internal/symlinkguard"
-)
-
-// startupSymlinkCheck is the hyperguild#157 tripwire at boot: a symlink in the
-// brain dir would let brain readers serve a file from outside the brain, so
-// the server does not start until it is removed. A directory the scan cannot
-// read could hide one, so it refuses on that too. A missing brain dir (first
-// boot) is the only case it lets pass.
-func startupSymlinkCheck(brainDir string) error {
-	r, err := symlinkguard.Scan(brainDir)
-	if err != nil {
-		if errors.Is(err, os.ErrNotExist) {
-			return nil
-		}
-		return fmt.Errorf("refusing to start: cannot scan the brain dir %s (hyperguild#157): %w", brainDir, err)
-	}
-	if !r.Clean() {
-		return fmt.Errorf("refusing to start: brain dir %s has %s (hyperguild#157); fix them first", brainDir, r)
-	}
-	return nil
-}
-
-// symlinkScanInterval reads BRAIN_SYMLINK_SCAN_INTERVAL (seconds), falling
-// back to the default on an unset, non-numeric or non-positive value.
-func symlinkScanInterval(getenv func(string) string) time.Duration {
-	if s, err := strconv.Atoi(getenv("BRAIN_SYMLINK_SCAN_INTERVAL")); err == nil && s > 0 {
-		return time.Duration(s) * time.Second
-	}
-	return symlinkguard.DefaultInterval
-}

ingestion/cmd/server/symlink_check_test.go

@@ -1,65 +0,0 @@
-package main
-
-import (
-	"os"
-	"path/filepath"
-	"strings"
-	"testing"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-)
-
-// hyperguild#157: the server refuses to start while a symlink exists in the
-// brain dir, or while any part of it cannot be scanned (which could hide
-// one); a missing brain dir is the only thing it lets pass.
-func TestStartupSymlinkCheckRefusesABrainWithASymlink(t *testing.T) {
-	brain := t.TempDir()
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(brain, "planted")))
-	err := startupSymlinkCheck(brain)
-	require.Error(t, err)
-	assert.Contains(t, err.Error(), "planted")
-}
-
-func TestStartupSymlinkCheckRefusesABrainWithAnUnreadableDir(t *testing.T) {
-	if os.Geteuid() == 0 {
-		t.Skip("root reads mode-000 directories")
-	}
-	brain := t.TempDir()
-	locked := filepath.Join(brain, "aaa")
-	require.NoError(t, os.MkdirAll(locked, 0o755))
-	require.NoError(t, os.Chmod(locked, 0o000))
-	t.Cleanup(func() { _ = os.Chmod(locked, 0o755) })
-	err := startupSymlinkCheck(brain)
-	require.Error(t, err)
-	assert.Contains(t, err.Error(), "aaa")
-}
-
-func TestStartupSymlinkCheckPassesACleanOrMissingBrain(t *testing.T) {
-	brain := t.TempDir()
-	require.NoError(t, os.WriteFile(filepath.Join(brain, "n.md"), []byte("x"), 0o644))
-	assert.NoError(t, startupSymlinkCheck(brain))
-	assert.NoError(t, startupSymlinkCheck(filepath.Join(brain, "missing")))
-}
-
-func TestSymlinkScanIntervalFallsBackOnNonPositive(t *testing.T) {
-	for _, v := range []string{"0", "-5", "", "junk"} {
-		got := symlinkScanInterval(func(string) string { return v })
-		assert.Positive(t, got, "env %q", v)
-	}
-}
-
-func TestAlertingWarningsNamesTheSymlinkTripwire(t *testing.T) {
-	w := strings.Join(alertingWarnings(func(string) string { return "" }), " ")
-	assert.Contains(t, w, "symlink tripwire")
-}
-
-func TestStartupSymlinkCheckRefusesAnUnlistableBrainRoot(t *testing.T) {
-	if os.Geteuid() == 0 {
-		t.Skip("root lists mode-0100 directories")
-	}
-	brain := t.TempDir()
-	require.NoError(t, os.Chmod(brain, 0o100))
-	t.Cleanup(func() { _ = os.Chmod(brain, 0o755) })
-	assert.Error(t, startupSymlinkCheck(brain))
-}

ingestion/internal/brainfs/brainfs.go

@@ -1,113 +0,0 @@
-// Package brainfs reads the brain directory through an os.Root (hyperguild#157).
-// The brain dir is a Syncthing sendreceive folder and Syncthing syncs
-// symlinks; a reader that walks and reads by path follows a planted symlink
-// out of the brain, and blocks forever on a planted FIFO. Readers that go
-// through a Brain only ever read regular files inside the brain: os.Root
-// refuses to leave the root even if an entry is swapped mid-read, and every
-// open is non-blocking with the type checked on the opened handle.
-package brainfs
-
-import (
-	"errors"
-	"fmt"
-	"io"
-	"io/fs"
-	"os"
-	"path/filepath"
-	"syscall"
-)
-
-// ErrNotRegular is returned for an entry that is not a regular file.
-var ErrNotRegular = errors.New("not a regular file")
-
-// Brain is an open brain directory.
-type Brain struct {
-	root *os.Root
-	dir  string
-}
-
-// Open opens brainDir (which may itself be a mount symlink) as a root.
-func Open(brainDir string) (*Brain, error) {
-	r, err := os.OpenRoot(brainDir)
-	if err != nil {
-		return nil, fmt.Errorf("open brain root: %w", err)
-	}
-	return &Brain{root: r, dir: brainDir}, nil
-}
-
-// Close releases the root.
-func (b *Brain) Close() error { return b.root.Close() }
-
-// Rel converts a path under the brain dir to a root-relative path.
-func (b *Brain) Rel(path string) (string, error) {
-	rel, err := filepath.Rel(b.dir, path)
-	if err != nil {
-		return "", err
-	}
-	if rel == ".." || len(rel) > 2 && rel[:3] == ".."+string(filepath.Separator) {
-		return "", fmt.Errorf("%s is outside the brain dir", path)
-	}
-	return filepath.ToSlash(rel), nil
-}
-
-// ReadRegular reads a brain-relative path, refusing anything that is not a
-// regular file, both before the open and on the opened handle.
-func (b *Brain) ReadRegular(rel string) ([]byte, error) {
-	fi, err := b.root.Lstat(rel)
-	if err != nil {
-		return nil, err
-	}
-	if !fi.Mode().IsRegular() {
-		return nil, fmt.Errorf("%s: %w", rel, ErrNotRegular)
-	}
-	return b.readOpened(rel)
-}
-
-// readOpened opens rel non-blocking and reads it only if the OPENED handle is
-// a regular file. This is the check that matters under a race: an entry
-// swapped for a FIFO after the Lstat, with a writer holding it open, would
-// otherwise park io.ReadAll forever (PR #161 review: removing it hung search
-// after 16 iterations).
-//
-// Accepted residual, as in api/rawroot.go: a HARD link to an outside file is
-// a regular file and is read. Syncthing syncs contents, not links, so a peer
-// cannot create one; fs.protected_hardlinks=1 and the pod's secrets are on a
-// different filesystem from the brain volume, so a local caller cannot link
-// them in. Refusing Nlink > 1 is the next step if either changes.
-func (b *Brain) readOpened(rel string) ([]byte, error) {
-	f, err := b.root.OpenFile(rel, os.O_RDONLY|syscall.O_NONBLOCK, 0)
-	if err != nil {
-		return nil, err
-	}
-	defer func() { _ = f.Close() }()
-	hfi, err := f.Stat()
-	if err != nil {
-		return nil, err
-	}
-	if !hfi.Mode().IsRegular() {
-		return nil, fmt.Errorf("%s: %w", rel, ErrNotRegular)
-	}
-	return io.ReadAll(f)
-}
-
-// WalkRegular walks a brain-relative directory through the root and calls fn
-// for each regular file (directories are descended; symlinks, FIFOs and
-// devices are skipped, never followed). An entry error is passed to onErr and
-// the walk continues; a nil onErr ignores it.
-func (b *Brain) WalkRegular(relDir string, fn func(rel string) error, onErr func(rel string, err error)) error {
-	return fs.WalkDir(b.root.FS(), filepath.ToSlash(relDir), func(p string, d fs.DirEntry, err error) error {
-		if err != nil {
-			if onErr != nil {
-				onErr(p, err)
-			}
-			if d != nil && d.IsDir() {
-				return fs.SkipDir
-			}
-			return nil
-		}
-		if d.IsDir() || !d.Type().IsRegular() {
-			return nil
-		}
-		return fn(p)
-	})
-}

ingestion/internal/brainfs/brainfs_test.go

@@ -1,92 +0,0 @@
-package brainfs
-
-import (
-	"errors"
-	"os"
-	"path/filepath"
-	"syscall"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-)
-
-// The handle check is what stops a FIFO swapped in after the Lstat. A FIFO
-// with no writer reads EOF at once, so a race test could not see the check
-// missing; here a writer holds the FIFO open, as any local process can, and
-// readOpened must refuse it promptly (PR #161 review).
-func TestReadOpenedRefusesAFIFOHeldOpenByAWriter(t *testing.T) {
-	dir := t.TempDir()
-	fifo := filepath.Join(dir, "p.md")
-	require.NoError(t, syscall.Mkfifo(fifo, 0o644))
-	w, err := os.OpenFile(fifo, os.O_RDWR, 0)
-	require.NoError(t, err)
-	defer func() { _ = w.Close() }()
-
-	b, err := Open(dir)
-	require.NoError(t, err)
-	defer func() { _ = b.Close() }()
-
-	done := make(chan error, 1)
-	go func() { _, err := b.readOpened("p.md"); done <- err }()
-	select {
-	case err := <-done:
-		assert.True(t, errors.Is(err, ErrNotRegular), "got %v", err)
-	case <-time.After(2 * time.Second):
-		t.Fatal("readOpened blocked on a FIFO with a writer")
-	}
-}
-
-func TestReadRegularRefusesSymlinksAndReadsRegularFiles(t *testing.T) {
-	dir := t.TempDir()
-	require.NoError(t, os.WriteFile(filepath.Join(dir, "ok.md"), []byte("ok"), 0o644))
-	outside := filepath.Join(t.TempDir(), "s.md")
-	require.NoError(t, os.WriteFile(outside, []byte("secret"), 0o644))
-	require.NoError(t, os.Symlink(outside, filepath.Join(dir, "link.md")))
-	b, err := Open(dir)
-	require.NoError(t, err)
-	defer func() { _ = b.Close() }()
-	got, err := b.ReadRegular("ok.md")
-	require.NoError(t, err)
-	assert.Equal(t, "ok", string(got))
-	_, err = b.ReadRegular("link.md")
-	assert.True(t, errors.Is(err, ErrNotRegular), "got %v", err)
-	_, err = b.ReadRegular("../" + filepath.Base(outside))
-	assert.Error(t, err)
-}
-
-func TestRelAcceptsPathsInsideAndRefusesEscapes(t *testing.T) {
-	dir := t.TempDir()
-	b, err := Open(dir)
-	require.NoError(t, err)
-	defer func() { _ = b.Close() }()
-	for in, want := range map[string]string{
-		filepath.Join(dir, "wiki", "x.md"): "wiki/x.md",
-		filepath.Join(dir, "..foo", "x"):   "..foo/x",
-		dir:                                ".",
-	} {
-		got, err := b.Rel(in)
-		require.NoError(t, err, in)
-		assert.Equal(t, want, got, in)
-	}
-	for _, bad := range []string{filepath.Join(dir, "..", "x"), dir + "-evil/x", "/etc/passwd", filepath.Join(dir, "wiki", "..", "..", "x")} {
-		_, err := b.Rel(bad)
-		assert.Error(t, err, bad)
-	}
-}
-
-func TestWalkRegularYieldsOnlyRegularFiles(t *testing.T) {
-	dir := t.TempDir()
-	require.NoError(t, os.MkdirAll(filepath.Join(dir, "a"), 0o755))
-	require.NoError(t, os.WriteFile(filepath.Join(dir, "a", "n.md"), []byte("n"), 0o644))
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(dir, "a", "dirlink")))
-	require.NoError(t, os.Symlink(filepath.Join(dir, "a", "n.md"), filepath.Join(dir, "a", "filelink.md")))
-	require.NoError(t, syscall.Mkfifo(filepath.Join(dir, "a", "p.md"), 0o644))
-	b, err := Open(dir)
-	require.NoError(t, err)
-	defer func() { _ = b.Close() }()
-	var got []string
-	require.NoError(t, b.WalkRegular(".", func(rel string) error { got = append(got, rel); return nil }, nil))
-	assert.Equal(t, []string{"a/n.md"}, got)
-}

ingestion/internal/search/brainroot_test.go

@@ -1,149 +0,0 @@
-package search
-
-import (
-	"os"
-	"path/filepath"
-	"strings"
-	"sync"
-	"sync/atomic"
-	"syscall"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-)
-
-// hyperguild#157 option 1: search walked and read the brain by path, so a
-// symlink anywhere in the Syncthing-shared brain dir (wiki/<w>/<h>/leak.md ->
-// /any/file) put the target's contents in brain_query results (reproduced in
-// the PR #156 review). Search now walks and reads through an os.Root on the
-// brain dir and only reads regular files.
-
-const sentinel157 = "zqsentinelonefiftyseven"
-
-func searchBrain(t *testing.T) (brain, facts, outside string) {
-	t.Helper()
-	brain = t.TempDir()
-	facts = filepath.Join(brain, "wiki", "w", "facts")
-	require.NoError(t, os.MkdirAll(facts, 0o755))
-	require.NoError(t, os.MkdirAll(filepath.Join(brain, "knowledge"), 0o755))
-	outside = filepath.Join(t.TempDir(), "secret.md")
-	require.NoError(t, os.WriteFile(outside, []byte("# secret\n\n"+sentinel157), 0o644))
-	return brain, facts, outside
-}
-
-func leaked(rs []Result) int {
-	n := 0
-	for _, r := range rs {
-		n += strings.Count(r.Excerpt+r.Title, sentinel157)
-	}
-	return n
-}
-
-func TestQueryDoesNotFollowASymlinkInTheWiki(t *testing.T) {
-	brain, facts, outside := searchBrain(t)
-	require.NoError(t, os.Symlink(outside, filepath.Join(facts, "leak.md")))
-	require.NoError(t, os.WriteFile(filepath.Join(facts, "real.md"), []byte("# real\n\nzqrealnote"), 0o644))
-	rs, err := Query(brain, QueryOptions{Query: sentinel157, Limit: 10})
-	require.NoError(t, err)
-	assert.Zero(t, leaked(rs))
-	rs, err = Query(brain, QueryOptions{Query: "zqrealnote", Limit: 10})
-	require.NoError(t, err)
-	assert.Len(t, rs, 1, "regular notes are still found")
-}
-
-func TestQueryDoesNotFollowASymlinkedDirectory(t *testing.T) {
-	brain, _, outside := searchBrain(t)
-	require.NoError(t, os.Symlink(filepath.Dir(outside), filepath.Join(brain, "wiki", "linked")))
-	rs, err := Query(brain, QueryOptions{Query: sentinel157, Limit: 10})
-	require.NoError(t, err)
-	assert.Zero(t, leaked(rs))
-}
-
-// swapInto keeps renaming a regular note, and either a symlink to outside or
-// a FIFO, into path. Each FIFO has a writer held open (as any local process
-// can): a FIFO with no writer reads EOF at once, which would let a reader
-// that skips the handle check pass (PR #161 round 2: with a held writer, a
-// bypass of readOpened hung Query after 5 iterations).
-// startSwap runs swapInto until the test ends and waits for it before
-// t.TempDir's cleanup (hyperguild#162).
-func startSwap(t *testing.T, path, outside string, fifo bool) {
-	t.Helper()
-	var stop atomic.Bool
-	var wg sync.WaitGroup
-	wg.Add(1)
-	go func() { defer wg.Done(); swapInto(path, outside, fifo, &stop) }()
-	t.Cleanup(func() { stop.Store(true); wg.Wait() })
-}
-
-func swapInto(path, outside string, fifo bool, stop *atomic.Bool) {
-	tmp := path + ".swap"
-	// Writers stay open until the swapper exits, so a reader parked on a
-	// FIFO stays parked (closing one would hand the reader EOF and hide the
-	// hang). Past the fd limit, new FIFOs simply have no writer.
-	var writers []*os.File
-	defer func() {
-		for _, w := range writers {
-			_ = w.Close()
-		}
-	}()
-	for i := 0; !stop.Load(); i++ {
-		_ = os.Remove(tmp)
-		switch {
-		case i%2 == 0:
-			_ = os.WriteFile(tmp, []byte("# r\n\nzqinside "+"zqracenote"), 0o644)
-		case fifo:
-			if err := syscall.Mkfifo(tmp, 0o644); err == nil {
-				if w, err := os.OpenFile(tmp, os.O_RDWR, 0); err == nil {
-					writers = append(writers, w)
-				}
-			}
-		default:
-			_ = os.Symlink(outside, tmp)
-		}
-		_ = os.Rename(tmp, path)
-	}
-}
-
-func TestQueryLeaksNothingUnderASymlinkSwapRace(t *testing.T) {
-	brain, facts, outside := searchBrain(t)
-	startSwap(t, filepath.Join(facts, "r.md"), outside, false)
-	leaks, iters := 0, 0
-	for deadline := time.Now().Add(1500 * time.Millisecond); time.Now().Before(deadline); iters++ {
-		rs, _ := Query(brain, QueryOptions{Query: sentinel157, Limit: 10})
-		leaks += leaked(rs)
-	}
-	if leaks > 0 {
-		t.Fatalf("outside file leaked %d times in %d iterations", leaks, iters)
-	}
-}
-
-func TestQueryDoesNotHangUnderAFIFOSwapRace(t *testing.T) {
-	brain, facts, outside := searchBrain(t)
-	startSwap(t, filepath.Join(facts, "r.md"), outside, true)
-	done := make(chan struct{})
-	go func() {
-		for deadline := time.Now().Add(1500 * time.Millisecond); time.Now().Before(deadline); {
-			_, _ = Query(brain, QueryOptions{Query: "zqracenote", Limit: 10})
-		}
-		close(done)
-	}()
-	select {
-	case <-done:
-	case <-time.After(6 * time.Second):
-		if f, err := os.OpenFile(filepath.Join(facts, "r.md"), os.O_WRONLY|syscall.O_NONBLOCK, 0); err == nil {
-			_ = f.Close()
-		}
-		t.Fatal("search hung on a swapped-in FIFO")
-	}
-}
-
-// hydrate (vector-only hits) reads by relative path; it must refuse a symlink.
-func TestHydrateRefusesASymlink(t *testing.T) {
-	brain, facts, outside := searchBrain(t)
-	require.NoError(t, os.Symlink(outside, filepath.Join(facts, "leak.md")))
-	r, err := hydrate(brain, "wiki/w/facts/leak.md")
-	assert.Error(t, err)
-	assert.NotContains(t, r.Excerpt, sentinel157)
-}

ingestion/internal/search/search.go

@@ -5,7 +5,6 @@ import (
 	"bufio"
 	"context"
 	"fmt"
-	"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
 	sharedfm "github.com/mathiasbq/hyperguild/ingestion/internal/frontmatter"
 	"log/slog"
 	"os"
@@ -120,30 +119,20 @@ func QueryContext(ctx context.Context, brainDir string, opts QueryOptions) ([]Re
 		return nil, err
 	}
 
-	// Walk and read through an os.Root on the brain dir, regular files only
-	// (hyperguild#157): a symlink planted in the Syncthing-shared brain must
-	// not put an outside file in results, nor a FIFO hang the query.
-	bfs, err := brainfs.Open(brainDir)
-	if err != nil {
-		return nil, fmt.Errorf("search: %w", err)
-	}
-	defer func() { _ = bfs.Close() }()
-
 	var results []Result
 	for _, dir := range roots {
 		if _, statErr := os.Stat(dir); os.IsNotExist(statErr) {
 			continue
 		}
-		relDir, relErr := bfs.Rel(dir)
+		err := filepath.WalkDir(dir, func(path string, d os.DirEntry, err error) error {
-		if relErr != nil {
+			if err != nil {
-			return nil, fmt.Errorf("search: %w", relErr)
+				slog.Warn("search: skipping path", "path", path, "err", err)
-		}
-		err := bfs.WalkRegular(relDir, func(relPath string) error {
-			if !strings.HasSuffix(relPath, ".md") {
 				return nil
 			}
-			path := filepath.Join(brainDir, filepath.FromSlash(relPath))
+			if d.IsDir() || !strings.HasSuffix(path, ".md") {
-			content, err := bfs.ReadRegular(relPath)
+				return nil
+			}
+			content, err := os.ReadFile(path)
 			if err != nil {
 				slog.Warn("search: skipping unreadable file", "path", path, "err", err)
 				return nil
@@ -166,7 +155,7 @@ func QueryContext(ctx context.Context, brainDir string, opts QueryOptions) ([]Re
 			origin, writtenBy := extractProvenance(string(content))
 			results = append(results, Result{
 				Path:    rel,
-				Title:   extractTitle(string(content), filepath.Base(relPath)),
+				Title:   extractTitle(string(content), d.Name()),
 				Excerpt: excerpt(string(content), 300),
 				Score:   score,
 				Wing:    wing,
@@ -177,8 +166,6 @@ func QueryContext(ctx context.Context, brainDir string, opts QueryOptions) ([]Re
 				WrittenBy: writtenBy,
 			})
 			return nil
-		}, func(p string, err error) {
-			slog.Warn("search: skipping path", "path", p, "err", err)
 		})
 		if err != nil {
 			return nil, err
@@ -288,13 +275,8 @@ func pathInScope(relPath, wing, hall string) bool {
 // excerpt, wing, and hall populated. Used for paths that surface only
 // via vector search.
 func hydrate(brainDir, relPath string) (Result, error) {
-	// Through the brain root, regular files only (hyperguild#157).
+	full := filepath.Join(brainDir, filepath.FromSlash(relPath))
-	bfs, err := brainfs.Open(brainDir)
+	content, err := os.ReadFile(full)
-	if err != nil {
-		return Result{}, err
-	}
-	defer func() { _ = bfs.Close() }()
-	content, err := bfs.ReadRegular(filepath.ToSlash(relPath))
 	if err != nil {
 		return Result{}, err
 	}

ingestion/internal/symlinkguard/symlinkguard.go

@@ -1,153 +0,0 @@
-// Package symlinkguard is the hyperguild#157 tripwire. The brain directory is
-// a Syncthing sendreceive folder and Syncthing syncs symlinks; several brain
-// readers (search, vectorstore, context, graph) follow a symlink out of the
-// brain and would index or return whatever it points to. Until every reader
-// reads through an os.Root, a symlink inside the brain, or a directory the
-// scan cannot read (which could hide one), is an incident: the server refuses
-// to start on one (main), and Watch reports one that appears while it runs.
-package symlinkguard
-
-import (
-	"context"
-	"fmt"
-	"io/fs"
-	"log/slog"
-	"os"
-	"path/filepath"
-	"sort"
-	"strconv"
-	"strings"
-	"time"
-)
-
-// DefaultInterval is used when Watch is given a non-positive interval.
-const DefaultInterval = 5 * time.Minute
-
-// Notifier is the out-of-band alert sink (audit.Notifier satisfies it).
-type Notifier interface {
-	Notify(ctx context.Context, msg string) error
-}
-
-// Result is what a scan found: symlinks, and directories it could not read
-// (each of which could hide a symlink). Paths are brain-relative.
-type Result struct {
-	Symlinks   []string
-	Unreadable []string
-}
-
-// Clean reports whether the scan found nothing to act on.
-func (r Result) Clean() bool { return len(r.Symlinks) == 0 && len(r.Unreadable) == 0 }
-
-// String describes the findings for a log line or alert. Paths are quoted, so
-// a name containing a newline cannot blur two findings together.
-func (r Result) String() string {
-	q := func(xs []string) string {
-		out := make([]string, len(xs))
-		for i, x := range xs {
-			out[i] = strconv.Quote(x)
-		}
-		return strings.Join(out, ", ")
-	}
-	var parts []string
-	if len(r.Symlinks) > 0 {
-		parts = append(parts, fmt.Sprintf("%d symlink(s): %s", len(r.Symlinks), q(r.Symlinks)))
-	}
-	if len(r.Unreadable) > 0 {
-		parts = append(parts, fmt.Sprintf("%d unreadable dir(s): %s", len(r.Unreadable), q(r.Unreadable)))
-	}
-	return strings.Join(parts, "; ")
-}
-
-// Scan walks the brain dir and reports every symlink in it and every
-// directory it could not read, continuing past both. Only the ROOT .git is
-// skipped (it is brain-sync's, and no reader walks it); a nested .git is
-// scanned like any other directory. brainDir itself may be a symlink (a
-// volume mount); it is resolved first. The error is non-nil when the root
-// cannot be resolved (wrapping os.ErrNotExist when it is missing) or cannot be
-// listed.
-func Scan(brainDir string) (Result, error) {
-	root, err := filepath.EvalSymlinks(brainDir)
-	if err != nil {
-		return Result{}, fmt.Errorf("resolve brain dir: %w", err)
-	}
-	var r Result
-	rootGit := filepath.Join(root, ".git")
-	walkErr := filepath.WalkDir(root, func(p string, d fs.DirEntry, err error) error {
-		rel, _ := filepath.Rel(root, p)
-		rel = filepath.ToSlash(rel)
-		if err != nil {
-			if p == root {
-				return err
-			}
-			r.Unreadable = append(r.Unreadable, rel)
-			if d != nil && d.IsDir() {
-				return filepath.SkipDir
-			}
-			return nil
-		}
-		if d.IsDir() && p == rootGit {
-			return filepath.SkipDir
-		}
-		if d.Type()&os.ModeSymlink != 0 {
-			r.Symlinks = append(r.Symlinks, rel)
-		}
-		return nil
-	})
-	sort.Strings(r.Symlinks)
-	sort.Strings(r.Unreadable)
-	if walkErr != nil {
-		// Only the root's own error reaches here (deeper errors are recorded
-		// as Unreadable). A root that cannot be listed cannot be shown clean
-		// (PR #159 round 2: discarding this reported clean while search
-		// served a planted symlink).
-		return r, fmt.Errorf("scan brain dir: %w", walkErr)
-	}
-	return r, nil
-}
-
-// Watch rescans every interval (DefaultInterval if non-positive) until ctx
-// ends, and notifies once per distinct non-clean result. A result counts as
-// reported only after Notify succeeds, so an alert lost to a notifier outage
-// is retried on the next tick.
-func Watch(ctx context.Context, brainDir string, interval time.Duration, n Notifier) {
-	if interval <= 0 {
-		interval = DefaultInterval
-	}
-	t := time.NewTicker(interval)
-	defer t.Stop()
-	reported := ""
-	for {
-		select {
-		case <-ctx.Done():
-			return
-		case <-t.C:
-		}
-		r, err := Scan(brainDir)
-		var key, msg string
-		switch {
-		case err != nil:
-			key = "error: " + err.Error()
-			msg = "brain symlink tripwire (hyperguild#157): cannot scan the brain dir: " + err.Error()
-		case r.Clean():
-			reported = ""
-			continue
-		default:
-			key = r.String()
-			msg = "brain symlink tripwire (hyperguild#157): " + key +
-				". Readers may follow symlinks out of the brain; remove them."
-		}
-		if key == reported {
-			continue
-		}
-		slog.Error("brain symlink tripwire", "finding", key)
-		if n == nil {
-			reported = key
-			continue
-		}
-		if err := n.Notify(ctx, msg); err != nil {
-			slog.Warn("symlink tripwire notify failed; will retry", "err", err)
-			continue
-		}
-		reported = key
-	}
-}

ingestion/internal/symlinkguard/symlinkguard_test.go

@@ -1,189 +0,0 @@
-package symlinkguard
-
-import (
-	"context"
-	"errors"
-	"os"
-	"path/filepath"
-	"sync"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-)
-
-// hyperguild#157: the brain dir is a Syncthing sendreceive folder and
-// Syncthing syncs symlinks; several readers follow them out of the brain.
-// Until every reader goes through an os.Root, a symlink anywhere in the brain,
-// or a directory the scan cannot read (which could hide one), is an incident.
-
-func TestScanReportsSymlinksAtEveryDepthIncludingNestedGit(t *testing.T) {
-	brain := t.TempDir()
-	outside := t.TempDir()
-	require.NoError(t, os.MkdirAll(filepath.Join(brain, "wiki", "w", "facts", ".git"), 0o755))
-	require.NoError(t, os.MkdirAll(filepath.Join(brain, ".git", "objects"), 0o755))
-	require.NoError(t, os.Symlink(filepath.Join(outside, "f"), filepath.Join(brain, "wiki", "w", "facts", "leak.md")))
-	require.NoError(t, os.Symlink(outside, filepath.Join(brain, "wiki", "dirlink")))
-	// PR #159 review: only the ROOT .git is skipped; a nested .git is not
-	// special to the brain readers, which walk into it.
-	require.NoError(t, os.Symlink(filepath.Join(outside, "g"), filepath.Join(brain, "wiki", "w", "facts", ".git", "leak.md")))
-	require.NoError(t, os.Symlink(outside, filepath.Join(brain, ".git", "objects", "alt")))
-
-	r, err := Scan(brain)
-	require.NoError(t, err)
-	assert.ElementsMatch(t, []string{"wiki/w/facts/leak.md", "wiki/dirlink", "wiki/w/facts/.git/leak.md"}, r.Symlinks)
-	assert.Empty(t, r.Unreadable)
-}
-
-// PR #159 review: WalkDir stopped at the first unreadable directory and the
-// startup check passed, while search (which keeps walking) served a planted
-// symlink. Scan now records the unreadable dir and keeps going.
-func TestScanReportsUnreadableDirsAndKeepsWalkingPastThem(t *testing.T) {
-	if os.Geteuid() == 0 {
-		t.Skip("root reads mode-000 directories")
-	}
-	brain := t.TempDir()
-	locked := filepath.Join(brain, "aaa")
-	require.NoError(t, os.MkdirAll(locked, 0o755))
-	require.NoError(t, os.Chmod(locked, 0o000))
-	t.Cleanup(func() { _ = os.Chmod(locked, 0o755) })
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(brain, "zzz-planted")))
-
-	r, err := Scan(brain)
-	require.NoError(t, err)
-	assert.Equal(t, []string{"aaa"}, r.Unreadable)
-	assert.Equal(t, []string{"zzz-planted"}, r.Symlinks, "the walk continues past the unreadable dir")
-}
-
-func TestScanResolvesASymlinkedRootAndReportsSymlinkedEntriesInsideIt(t *testing.T) {
-	real := t.TempDir()
-	link := filepath.Join(t.TempDir(), "brain")
-	require.NoError(t, os.Symlink(real, link))
-	require.NoError(t, os.WriteFile(filepath.Join(real, "n.md"), []byte("x"), 0o644))
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(real, "inner")))
-	r, err := Scan(link)
-	require.NoError(t, err)
-	assert.Equal(t, []string{"inner"}, r.Symlinks)
-}
-
-func TestScanOfAMissingRootIsErrNotExist(t *testing.T) {
-	_, err := Scan(filepath.Join(t.TempDir(), "nope"))
-	assert.True(t, errors.Is(err, os.ErrNotExist), "got %v", err)
-}
-
-type recNotifier struct {
-	mu   sync.Mutex
-	msgs []string
-	fail int // fail the first N calls
-}
-
-func (r *recNotifier) Notify(_ context.Context, m string) error {
-	r.mu.Lock()
-	defer r.mu.Unlock()
-	r.msgs = append(r.msgs, m)
-	if r.fail > 0 {
-		r.fail--
-		return errors.New("ntfy down")
-	}
-	return nil
-}
-
-func (r *recNotifier) count() int { r.mu.Lock(); defer r.mu.Unlock(); return len(r.msgs) }
-
-func TestWatchNotifiesOncePerDistinctSetAndAgainForANewSet(t *testing.T) {
-	brain := t.TempDir()
-	n := &recNotifier{}
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go Watch(ctx, brain, 10*time.Millisecond, n)
-
-	time.Sleep(50 * time.Millisecond)
-	assert.Zero(t, n.count(), "a clean brain raises nothing")
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(brain, "first")))
-	require.Eventually(t, func() bool { return n.count() == 1 }, 2*time.Second, 10*time.Millisecond)
-	time.Sleep(60 * time.Millisecond)
-	assert.Equal(t, 1, n.count(), "the same set is reported once, not every tick")
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(brain, "second")))
-	require.Eventually(t, func() bool { return n.count() == 2 }, 2*time.Second, 10*time.Millisecond)
-}
-
-// PR #159 review: last was recorded before Notify, so an alert lost to an
-// ntfy outage was never retried while the symlink stayed.
-func TestWatchRetriesAnAlertThatFailedToSend(t *testing.T) {
-	brain := t.TempDir()
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(brain, "planted")))
-	n := &recNotifier{fail: 2}
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go Watch(ctx, brain, 10*time.Millisecond, n)
-	require.Eventually(t, func() bool { return n.count() == 3 }, 2*time.Second, 10*time.Millisecond)
-	time.Sleep(60 * time.Millisecond)
-	assert.Equal(t, 3, n.count(), "after one successful send, no more")
-}
-
-func TestWatchAlertsOnAnUnreadableDir(t *testing.T) {
-	if os.Geteuid() == 0 {
-		t.Skip("root reads mode-000 directories")
-	}
-	brain := t.TempDir()
-	locked := filepath.Join(brain, "locked")
-	require.NoError(t, os.MkdirAll(locked, 0o755))
-	require.NoError(t, os.Chmod(locked, 0o000))
-	t.Cleanup(func() { _ = os.Chmod(locked, 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)
-	assert.Contains(t, n.msgs[0], "locked")
-}
-
-// PR #159 review: a non-positive interval panicked time.NewTicker inside the
-// goroutine and crashlooped the server.
-func TestWatchToleratesANonPositiveInterval(t *testing.T) {
-	brain := t.TempDir()
-	ctx, cancel := context.WithCancel(context.Background())
-	done := make(chan struct{})
-	go func() { Watch(ctx, brain, 0, nil); close(done) }()
-	cancel()
-	select {
-	case <-done:
-	case <-time.After(2 * time.Second):
-		t.Fatal("Watch did not return on cancel")
-	}
-}
-
-// PR #159 round 2: a brain root that can be entered but not listed (mode
-// 0100) made WalkDir fail at the root; the error was discarded and Scan
-// reported clean, while search (which goes straight into wiki/) still served
-// a planted symlink. A root that cannot be listed is now an error.
-func TestScanOfAnUnlistableRootIsAnError(t *testing.T) {
-	if os.Geteuid() == 0 {
-		t.Skip("root lists mode-0100 directories")
-	}
-	brain := t.TempDir()
-	require.NoError(t, os.MkdirAll(filepath.Join(brain, "wiki"), 0o755))
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(brain, "wiki", "leak.md")))
-	require.NoError(t, os.Chmod(brain, 0o100))
-	t.Cleanup(func() { _ = os.Chmod(brain, 0o755) })
-	_, err := Scan(brain)
-	require.Error(t, err)
-	assert.False(t, errors.Is(err, os.ErrNotExist), "an unlistable root is not a missing root")
-}
-
-// The same set that disappears and comes back is reported again.
-func TestWatchReportsASetAgainAfterItWasCleared(t *testing.T) {
-	brain := t.TempDir()
-	link := filepath.Join(brain, "planted")
-	n := &recNotifier{}
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go Watch(ctx, brain, 10*time.Millisecond, n)
-	require.NoError(t, os.Symlink(t.TempDir(), link))
-	require.Eventually(t, func() bool { return n.count() == 1 }, 2*time.Second, 10*time.Millisecond)
-	require.NoError(t, os.Remove(link))
-	time.Sleep(60 * time.Millisecond)
-	require.NoError(t, os.Symlink(t.TempDir(), link))
-	require.Eventually(t, func() bool { return n.count() == 2 }, 2*time.Second, 10*time.Millisecond)
-}