internal/labels: generalize to multiple projects

IssueCategory now supports multiple projects. Each project's
categories are stored in a separate file.

For #64.

Change-Id: If941db708cc13ce229730d5f93a9d2548aab7ba7
Reviewed-on: https://go-review.googlesource.com/c/oscar/+/637856
Reviewed-by: Hyang-Ah Hana Kim <[email protected]>
LUCI-TryBot-Result: Go LUCI <[email protected]>

internal/devtools/cmd/labeleval/main.go

@@ -107,6 +107,7 @@ func run(ctx context.Context, categoryconfigFile, issueConfigFile string) error
 	}
 
 	var categoryConfig struct {
+		Project    string
 		Categories []labels.Category
 	}
 

internal/gaby/labels.go

@@ -100,7 +100,7 @@ func (g *Gaby) populateLabelsPage(r *http.Request) *labelsPage {
 		} else if isBot(i.User.Login) {
 			lr.Problem = "skipping: author is a bot"
 		} else {
-			cat, exp, err := labels.IssueCategory(r.Context(), g.llm, i)
+			cat, exp, err := labels.IssueCategory(r.Context(), g.llm, project, i)
 			if err != nil {
 				p.Error = err
 				return p

internal/labels/labeler.go

@@ -7,7 +7,6 @@ package labels
 import (
 	"context"
 	"encoding/json"
-	"errors"
 	"fmt"
 	"log/slog"
 	"maps"
@@ -126,7 +125,11 @@ func (l *Labeler) Run(ctx context.Context) error {
 
 	// Ensure that labels in GH match our config.
 	for p := range l.projects {
-		if err := l.syncLabels(ctx, p, config.Categories); err != nil {
+		cats, ok := config.Categories[p]
+		if !ok {
+			return fmt.Errorf("Labeler.Run: unknown project %q", p)
+		}
+		if err := l.syncLabels(ctx, p, cats); err != nil {
 			return err
 		}
 	}
@@ -183,7 +186,7 @@ func (l *Labeler) logLabelIssue(ctx context.Context, e *github.Event) (advance b
 	issue := e.Typed.(*github.Issue)
 	l.slog.Debug("labels.Labeler consider", "url", issue.HTMLURL)
 
-	cat, explanation, err := IssueCategory(ctx, l.cgen, issue)
+	cat, explanation, err := IssueCategory(ctx, l.cgen, e.Project, issue)
 	if err != nil {
 		return false, fmt.Errorf("IssueCategory(%s): %w", issue.HTMLURL, err)
 	}
@@ -230,10 +233,6 @@ func (l *Labeler) skip(e *github.Event) (bool, string) {
 // Otherwise, if the descriptions don't agree, a warning is logged and nothing is done on the issue tracker.
 // This function makes no other changes. In particular, it never deletes labels.
 func (l *Labeler) syncLabels(ctx context.Context, project string, cats []Category) error {
-	// TODO(jba): generalize to other projects.
-	if project != "golang/go" {
-		return errors.New("labeling only supported for golang/go")
-	}
 	l.slog.Info("syncing labels", "name", l.name, "project", project)
 	tlabList, err := l.github.ListLabels(ctx, project)
 	if err != nil {

internal/labels/labels.go

@@ -3,6 +3,9 @@
 // license that can be found in the LICENSE file.
 
 // Package labels classifies issues.
+//
+// The categories it uses are stored in static/*-categories.yaml
+// files, one file per project.
 package labels
 
 import (
@@ -13,6 +16,7 @@ import (
 	"errors"
 	"fmt"
 	"html/template"
+	"io/fs"
 	"iter"
 	"log"
 	"os"
@@ -36,8 +40,12 @@ type Category struct {
 
 // IssueCategory returns the category chosen by the LLM for the issue, along with an explanation
 // of why it was chosen. It uses the built-in list of categories.
-func IssueCategory(ctx context.Context, cgen llm.ContentGenerator, iss *github.Issue) (_ Category, explanation string, err error) {
-	return IssueCategoryFromList(ctx, cgen, iss, config.Categories)
+func IssueCategory(ctx context.Context, cgen llm.ContentGenerator, project string, iss *github.Issue) (_ Category, explanation string, err error) {
+	cats, ok := config.Categories[project]
+	if !ok {
+		return Category{}, "", fmt.Errorf("IssueCategory: unknown project %q", project)
+	}
+	return IssueCategoryFromList(ctx, cgen, iss, cats)
 }
 
 // IssueCategoryFromList is like [IssueCategory], but uses the given list of Categories.
@@ -226,22 +234,43 @@ type bodyArgs struct {
 }
 
 var config struct {
-	Categories []Category
+	// Key is project, e.g. "golang/go".
+	Categories map[string][]Category
 }
 
 //go:embed static/*
 var staticFS embed.FS
 
+// Read all category files into config.
 func init() {
-	f, err := staticFS.Open("static/categories.yaml")
+	catFiles, err := fs.Glob(staticFS, "static/*-categories.yaml")
 	if err != nil {
 		log.Fatal(err)
 	}
-	defer f.Close()
+	config.Categories = map[string][]Category{}
+	for _, file := range catFiles {
+		f, err := staticFS.Open(file)
+		if err != nil {
+			log.Fatalf("%s: %v", file, err)
+		}
 
-	dec := yaml.NewDecoder(f)
-	dec.KnownFields(true)
-	if err := dec.Decode(&config); err != nil {
-		log.Fatal(err)
+		var contents struct {
+			Project    string
+			Categories []Category
+		}
+
+		dec := yaml.NewDecoder(f)
+		dec.KnownFields(true)
+		if err := dec.Decode(&contents); err != nil {
+			log.Fatalf("%s: %v", file, err)
+		}
+		if contents.Project == "" {
+			log.Fatalf("%s: empty or missing project", file)
+		}
+		if _, ok := config.Categories[contents.Project]; ok {
+			log.Fatalf("%s: duplicate project %s", file, contents.Project)
+		}
+		config.Categories[contents.Project] = contents.Categories
+		f.Close()
 	}
 }

internal/labels/labels_test.go

@@ -21,7 +21,7 @@ func TestIssueLabels(t *testing.T) {
 		Body:  "body",
 	}
 
-	cat, exp, err := IssueCategory(ctx, llm, iss)
+	cat, exp, err := IssueCategory(ctx, llm, "golang/go", iss)
 	if err != nil {
 		t.Fatal(err)
 	}

internal/labels/static/categories.yaml → internal/labels/static/go-categories.yaml

@@ -5,6 +5,7 @@
 #  description: the label description on the issue tracker
 #  extra: additional information about the label, fed to the LLM along
 #         with the description
+project: golang/go
 categories:
   - name: bug
     label: Bug