feat(ingestion): clips land in web-clip by rule; shared concepts/entities; approve parked notes; remove the classifier (#208, #209) #211

Merged
mathias merged 16 commits from wt/web-clip into main 2026-10-20 19:36:51 +00:00
3 changed files with 158 additions and 0 deletions
Showing only changes of commit 8342de5c49 - Show all commits
+1
View File
@@ -20,6 +20,7 @@ type Source string
const (
Explicit Source = "explicit" // subfolder or frontmatter named the wing
Default Source = "default" // no signal; the configured default wing
Classified Source = "classified" // classifier + scorer agreed, confident
Exception Source = "exception" // needs a human
)
+70
View File
@@ -0,0 +1,70 @@
package wing
import (
"context"
"fmt"
"regexp"
"slices"
"strings"
)
// Placer decides where a raw note goes. Router is the production
// implementation; the interface lets the watcher stay ignorant of how.
type Placer interface {
Resolve(ctx context.Context, n Note) Decision
}
// Router places a note with rules only, no model (hyperguild#208):
//
// 1. A subfolder or frontmatter `wing:` is the owner's deliberate choice. It
// must name a plain, EXISTING wing (a typo never creates one) and then it
// places, whatever the wing is: extraction runs on the local model, and it
// is the owner who dropped the note there. Two signals that disagree park.
// 2. With no signal, a note whose text matches ClientBlock is parked; any
// other note goes to Default, the one wing that is created on first use.
//
// Wings are read from disk on every decision, so a wing created while the
// server runs is usable at once.
type Router struct {
BrainDir string
Default string
ClientBlock *regexp.Regexp // nil = no content check
}
// Resolve implements Placer. It never returns an error: anything it cannot
// decide safely is an Exception with the reason spelled out.
func (r Router) Resolve(_ context.Context, n Note) Decision {
dir, fm := norm(n.Dir), norm(n.FrontmatterWing)
switch {
case dir != "" && fm != "" && dir != fm:
return exception("", 0, fmt.Sprintf("subfolder %q and frontmatter wing %q disagree", dir, fm))
case dir != "":
return r.explicit(dir, "subfolder")
case fm != "":
return r.explicit(fm, "frontmatter")
}
if r.ClientBlock != nil && r.ClientBlock.MatchString(n.Text) {
return exception("", 0, "text matches the client-content marker; not defaulted into a shared wing")
}
if r.Default == "" || !plainWingName(r.Default) {
return exception("", 0, "no valid default wing configured")
}
return Decision{Wing: r.Default, Source: Default, Confidence: 1, Reason: "no wing named; default wing"}
}
func (r Router) explicit(w, via string) Decision {
if !plainWingName(w) {
return exception("", 0, fmt.Sprintf("%s names %q, which is not a plain wing name", via, w))
}
if w == r.Default && plainWingName(w) {
return Decision{Wing: w, Source: Explicit, Confidence: 1, Reason: "named by " + via}
}
known, err := KnownWings(r.BrainDir)
if err != nil {
return exception(w, 0, fmt.Sprintf("cannot list wings: %v", err))
}
if !slices.Contains(known, w) {
return exception(w, 0, fmt.Sprintf("%s names %q: not an existing wing (new wings are never created from a folder name)", via, strings.TrimSpace(w)))
}
return Decision{Wing: w, Source: Explicit, Confidence: 1, Reason: "named by " + via}
}
+87
View File
@@ -0,0 +1,87 @@
package wing_test
import (
"context"
"regexp"
"testing"
"github.com/stretchr/testify/assert"
"github.com/mathiasbq/hyperguild/ingestion/internal/wing"
)
func routerBrain(t *testing.T) string {
dir := t.TempDir()
writePage(t, dir, "wiki/homelab/facts/a.md", "x")
writePage(t, dir, "wiki/rikatillsammans/facts/b.md", "x")
return dir
}
func TestRouter_Decisions(t *testing.T) {
tests := []struct {
name string
note wing.Note
want wing.Source
wantWing string
}{
{"no signal goes to the default wing", wing.Note{Text: "a talk about agents"}, wing.Default, "web-clip"},
{"a subfolder naming an existing wing places there", wing.Note{Dir: "homelab", Text: "x"}, wing.Explicit, "homelab"},
{"a deliberate subfolder into a client wing is the owner's call and is placed", wing.Note{Dir: "rikatillsammans", Text: "x"}, wing.Explicit, "rikatillsammans"},
{"frontmatter wing places like a subfolder", wing.Note{FrontmatterWing: "homelab", Text: "x"}, wing.Explicit, "homelab"},
{"subfolder is case-insensitive", wing.Note{Dir: "HomeLab", Text: "x"}, wing.Explicit, "homelab"},
{"a typo never creates a wing", wing.Note{Dir: "hoemlab", Text: "x"}, wing.Exception, "hoemlab"},
{"conflicting signals are parked", wing.Note{Dir: "homelab", FrontmatterWing: "rikatillsammans", Text: "x"}, wing.Exception, ""},
{"the default wing may be named explicitly even before it exists", wing.Note{Dir: "web-clip", Text: "x"}, wing.Explicit, "web-clip"},
{"client marker with no signal is parked, not defaulted", wing.Note{Text: "notes from the ACME-CLIENT kickoff"}, wing.Exception, ""},
{"client marker does not block a deliberate subfolder", wing.Note{Dir: "homelab", Text: "ACME-CLIENT"}, wing.Explicit, "homelab"},
}
for _, tt := range tests {
t.Run(tt.name, func(t *testing.T) {
r := wing.Router{BrainDir: routerBrain(t), Default: "web-clip", ClientBlock: regexp.MustCompile(`(?i)acme-client`)}
d := r.Resolve(context.Background(), tt.note)
assert.Equal(t, tt.want, d.Source)
assert.Equal(t, tt.wantWing, d.Wing)
assert.NotEmpty(t, d.Reason)
})
}
}
func TestRouter_ReadsTheWingsFreshOnEveryDecision(t *testing.T) {
dir := routerBrain(t)
r := wing.Router{BrainDir: dir, Default: "web-clip"}
assert.Equal(t, wing.Exception, r.Resolve(context.Background(), wing.Note{Dir: "newwing", Text: "x"}).Source)
writePage(t, dir, "wiki/newwing/facts/c.md", "x") // a wing created after startup
assert.Equal(t, wing.Explicit, r.Resolve(context.Background(), wing.Note{Dir: "newwing", Text: "x"}).Source,
"a wing made while the server runs is usable without a restart")
}
func TestRouter_AnUnreadableBrainParksInsteadOfGuessing(t *testing.T) {
r := wing.Router{BrainDir: t.TempDir() + "/missing", Default: "web-clip"}
d := r.Resolve(context.Background(), wing.Note{Dir: "homelab", Text: "x"})
assert.Equal(t, wing.Exception, d.Source)
}
func TestRouter_WithoutADefaultAnUnsignalledNoteIsParked(t *testing.T) {
r := wing.Router{BrainDir: routerBrain(t)}
assert.Equal(t, wing.Exception, r.Resolve(context.Background(), wing.Note{Text: "x"}).Source)
}
func TestRouter_ADefaultThatIsNotAPlainNameIsRefused(t *testing.T) {
r := wing.Router{BrainDir: routerBrain(t), Default: "../escape"}
assert.Equal(t, wing.Exception, r.Resolve(context.Background(), wing.Note{Text: "x"}).Source)
}
func TestRouter_ANameThatIsNotPlainIsRefusedAsSuch(t *testing.T) {
for _, bad := range []string{"../x", "a/b", "has space", "dot.ted"} {
t.Run(bad, func(t *testing.T) {
// Default equals the bad name: the "named explicitly" shortcut for the
// default wing must not wave a malformed name through.
r := wing.Router{BrainDir: routerBrain(t), Default: bad}
d := r.Resolve(context.Background(), wing.Note{Dir: bad, Text: "x"})
assert.Equal(t, wing.Exception, d.Source)
assert.Empty(t, d.Wing, "a malformed name is never echoed as a proposal")
assert.Contains(t, d.Reason, "not a plain wing name")
})
}
}