appview: parse reference links from markdown body

Defined `refResolver` which will parse useful data from markdown body
like @-mentions or issue/pr/comment mentions

Signed-off-by: Seongmin Lee <[e-mail]>

appview/db/reference.go

···
+package db
+
+import (
+	"database/sql"
+	"fmt"
+	"strings"
+
+	"github.com/bluesky-social/indigo/atproto/syntax"
+	"tangled.org/core/api/tangled"
+	"tangled.org/core/appview/models"
+)
+
+// FindReferences resolves refLinks to Issue/PR/IssueComment/PullComment ATURIs.
+// It will ignore missing refLinks.
+func FindReferences(e Execer, refLinks []models.ReferenceLink) ([]syntax.ATURI, error) {
+	var (
+		issueRefs []models.ReferenceLink
+		pullRefs  []models.ReferenceLink
+	)
+	for _, ref := range refLinks {
+		switch ref.Kind {
+		case models.RefKindIssue:
+			issueRefs = append(issueRefs, ref)
+		case models.RefKindPull:
+			pullRefs = append(pullRefs, ref)
+		}
+	}
+	issueUris, err := findIssueReferences(e, issueRefs)
+	if err != nil {
+		return nil, err
+	}
+	pullUris, err := findPullReferences(e, pullRefs)
+	if err != nil {
+		return nil, err
+	}
+
+	return append(issueUris, pullUris...), nil
+}
+
+func findIssueReferences(e Execer, refLinks []models.ReferenceLink) ([]syntax.ATURI, error) {
+	if len(refLinks) == 0 {
+		return nil, nil
+	}
+	vals := make([]string, len(refLinks))
+	args := make([]any, 0, len(refLinks)*4)
+	for i, ref := range refLinks {
+		vals[i] = "(?, ?, ?, ?)"
+		args = append(args, ref.Handle, ref.Repo, ref.SubjectId, ref.CommentId)
+	}
+	query := fmt.Sprintf(
+		`with input(owner_did, name, issue_id, comment_id) as (
+			values %s
+		)
+		select
+			i.did, i.rkey,
+			c.did, c.rkey
+		from input inp
+		join repos r
+			on r.did = inp.owner_did
+				and r.name = inp.name
+		join issues i
+			on i.repo_at = r.at_uri
+				and i.issue_id = inp.issue_id
+		left join issue_comments c
+			on inp.comment_id is not null
+				and c.issue_at = i.at_uri
+				and c.id = inp.comment_id
+		`,
+		strings.Join(vals, ","),
+	)
+	rows, err := e.Query(query, args...)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	var uris []syntax.ATURI
+
+	for rows.Next() {
+		// Scan rows
+		var issueOwner, issueRkey string
+		var commentOwner, commentRkey sql.NullString
+		var uri syntax.ATURI
+		if err := rows.Scan(&issueOwner, &issueRkey, &commentOwner, &commentRkey); err != nil {
+			return nil, err
+		}
+		if commentOwner.Valid && commentRkey.Valid {
+			uri = syntax.ATURI(fmt.Sprintf(
+				"at://%s/%s/%s",
+				commentOwner.String,
+				tangled.RepoIssueCommentNSID,
+				commentRkey.String,
+			))
+		} else {
+			uri = syntax.ATURI(fmt.Sprintf(
+				"at://%s/%s/%s",
+				issueOwner,
+				tangled.RepoIssueNSID,
+				issueRkey,
+			))
+		}
+		uris = append(uris, uri)
+	}
+	return uris, nil
+}
+
+func findPullReferences(e Execer, refLinks []models.ReferenceLink) ([]syntax.ATURI, error) {
+	if len(refLinks) == 0 {
+		return nil, nil
+	}
+	vals := make([]string, len(refLinks))
+	args := make([]any, 0, len(refLinks)*4)
+	for i, ref := range refLinks {
+		vals[i] = "(?, ?, ?, ?)"
+		args = append(args, ref.Handle, ref.Repo, ref.SubjectId, ref.CommentId)
+	}
+	query := fmt.Sprintf(
+		`with input(owner_did, name, pull_id, comment_id) as (
+			values %s
+		)
+		select
+			p.owner_did, p.rkey,
+			c.owner_did, c.rkey
+		from input inp
+		join repos r
+			on r.did = inp.owner_did
+				and r.name = inp.name
+		join pulls p
+			on p.repo_at = r.at_uri
+				and p.pull_id = inp.pull_id
+		left join pull_comments c
+			on inp.comment_id is not null
+				and c.repo_at = r.at_uri and c.pull_id = p.pull_id
+				and c.id = inp.comment_id
+		`,
+		strings.Join(vals, ","),
+	)
+	rows, err := e.Query(query, args...)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	var uris []syntax.ATURI
+
+	for rows.Next() {
+		// Scan rows
+		var pullOwner, pullRkey string
+		var commentOwner, commentRkey sql.NullString
+		var uri syntax.ATURI
+		if err := rows.Scan(&pullOwner, &pullRkey, &commentOwner, &commentRkey); err != nil {
+			return nil, err
+		}
+		if commentOwner.Valid && commentRkey.Valid {
+			uri = syntax.ATURI(fmt.Sprintf(
+				"at://%s/%s/%s",
+				commentOwner.String,
+				tangled.RepoPullCommentNSID,
+				commentRkey.String,
+			))
+		} else {
+			uri = syntax.ATURI(fmt.Sprintf(
+				"at://%s/%s/%s",
+				pullOwner,
+				tangled.RepoPullNSID,
+				pullRkey,
+			))
+		}
+		uris = append(uris, uri)
+	}
+	return uris, nil
+}

appview/issues/issues.go

···
 	"tangled.org/core/appview/notify"
 	"tangled.org/core/appview/oauth"
 	"tangled.org/core/appview/pages"
-	"tangled.org/core/appview/pages/markup"
 	"tangled.org/core/appview/pages/repoinfo"
 	"tangled.org/core/appview/pagination"
+	"tangled.org/core/appview/refresolver"
 	"tangled.org/core/appview/reporesolver"
 	"tangled.org/core/appview/validator"
 	"tangled.org/core/idresolver"
···
 	enforcer     *rbac.Enforcer
 	pages        *pages.Pages
 	idResolver   *idresolver.Resolver
+	refResolver  *refresolver.Resolver
 	db           *db.DB
 	config       *config.Config
 	notifier     notify.Notifier
···
 	enforcer *rbac.Enforcer,
 	pages *pages.Pages,
 	idResolver *idresolver.Resolver,
+	refResolver *refresolver.Resolver,
 	db *db.DB,
 	config *config.Config,
 	notifier notify.Notifier,
···
 		enforcer:     enforcer,
 		pages:        pages,
 		idResolver:   idResolver,
+		refResolver:  refResolver,
 		db:           db,
 		config:       config,
 		notifier:     notifier,
···
 		replyTo = &replyToUri
 	}
 
+	mentions, _ := rp.refResolver.Resolve(r.Context(), body)
+
 	comment := models.IssueComment{
 		Did:     user.Did,
 		Rkey:    tid.TID(),
···
 	// notify about the new comment
 	comment.Id = commentId
 
-	rawMentions := markup.FindUserMentions(comment.Body)
-	idents := rp.idResolver.ResolveIdents(r.Context(), rawMentions)
-	l.Debug("parsed mentions", "raw", rawMentions, "idents", idents)
-	var mentions []syntax.DID
-	for _, ident := range idents {
-		if ident != nil && !ident.Handle.IsInvalidHandle() {
-			mentions = append(mentions, ident.DID)
-		}
-	}
 	rp.notifier.NewIssueComment(r.Context(), &comment, mentions)
 
 	ownerSlashRepo := reporesolver.GetBaseRepoPath(r, f)
···
 			RepoInfo:     rp.repoResolver.GetRepoInfo(r, user),
 		})
 	case http.MethodPost:
+		body := r.FormValue("body")
+		mentions, _ := rp.refResolver.Resolve(r.Context(), body)
+
 		issue := &models.Issue{
 			RepoAt:  f.RepoAt(),
 			Rkey:    tid.TID(),
 			Title:   r.FormValue("title"),
-			Body:    r.FormValue("body"),
+			Body:    body,
 			Open:    true,
 			Did:     user.Did,
 			Created: time.Now(),
···
 		// everything is successful, do not rollback the atproto record
 		atUri = ""
 
-		rawMentions := markup.FindUserMentions(issue.Body)
-		idents := rp.idResolver.ResolveIdents(r.Context(), rawMentions)
-		l.Debug("parsed mentions", "raw", rawMentions, "idents", idents)
-		var mentions []syntax.DID
-		for _, ident := range idents {
-			if ident != nil && !ident.Handle.IsInvalidHandle() {
-				mentions = append(mentions, ident.DID)
-			}
-		}
 		rp.notifier.NewIssue(r.Context(), issue, mentions)
 
 		ownerSlashRepo := reporesolver.GetBaseRepoPath(r, f)

appview/models/reference.go

···
+package models
+
+type RefKind int
+
+const (
+	RefKindIssue RefKind = iota
+	RefKindPull
+)
+
+// /@alice.com/cool-proj/issues/123
+// /@alice.com/cool-proj/issues/123#comment-321
+type ReferenceLink struct {
+	Handle    string
+	Repo      string
+	Kind      RefKind
+	SubjectId int
+	CommentId *int
+}

appview/pages/markup/markdown.go

···
 	return path.Join(rctx.CurrentDir, dst)
 }
 
-// FindUserMentions returns Set of user handles from given markup soruce.
-// It doesn't guarntee unique DIDs
-func FindUserMentions(source string) []string {
-	var (
-		mentions    []string
-		mentionsSet = make(map[string]struct{})
-		md          = NewMarkdown()
-		sourceBytes = []byte(source)
-		root        = md.Parser().Parse(text.NewReader(sourceBytes))
-	)
-	ast.Walk(root, func(n ast.Node, entering bool) (ast.WalkStatus, error) {
-		if entering && n.Kind() == textension.KindAt {
-			handle := n.(*textension.AtNode).Handle
-			mentionsSet[handle] = struct{}{}
-			return ast.WalkSkipChildren, nil
-		}
-		return ast.WalkContinue, nil
-	})
-	for handle := range mentionsSet {
-		mentions = append(mentions, handle)
-	}
-	return mentions
-}
-
 func isAbsoluteUrl(link string) bool {
 	parsed, err := url.Parse(link)
 	if err != nil {

appview/pages/markup/reference_link.go

···
+package markup
+
+import (
+	"maps"
+	"net/url"
+	"path"
+	"slices"
+	"strconv"
+	"strings"
+
+	"github.com/yuin/goldmark/ast"
+	"github.com/yuin/goldmark/text"
+	"tangled.org/core/appview/models"
+	textension "tangled.org/core/appview/pages/markup/extension"
+)
+
+// FindReferences collects all links referencing tangled-related objects
+// like issues, PRs, comments or even @-mentions
+// This funciton doesn't actually check for the existence of records in the DB
+// or the PDS; it merely returns a list of what are presumed to be references.
+func FindReferences(baseUrl string, source string) ([]string, []models.ReferenceLink) {
+	var (
+		refLinkSet  = make(map[models.ReferenceLink]struct{})
+		mentionsSet = make(map[string]struct{})
+		md          = NewMarkdown()
+		sourceBytes = []byte(source)
+		root        = md.Parser().Parse(text.NewReader(sourceBytes))
+	)
+	// trim url scheme. the SSL shouldn't matter
+	baseUrl = strings.TrimPrefix(baseUrl, "https://")
+	baseUrl = strings.TrimPrefix(baseUrl, "http://")
+
+	ast.Walk(root, func(n ast.Node, entering bool) (ast.WalkStatus, error) {
+		if !entering {
+			return ast.WalkContinue, nil
+		}
+		switch n.Kind() {
+		case textension.KindAt:
+			handle := n.(*textension.AtNode).Handle
+			mentionsSet[handle] = struct{}{}
+			return ast.WalkSkipChildren, nil
+		case ast.KindLink:
+			dest := string(n.(*ast.Link).Destination)
+			ref := parseTangledLink(baseUrl, dest)
+			if ref != nil {
+				refLinkSet[*ref] = struct{}{}
+			}
+			return ast.WalkSkipChildren, nil
+		case ast.KindAutoLink:
+			an := n.(*ast.AutoLink)
+			if an.AutoLinkType == ast.AutoLinkURL {
+				dest := string(an.URL(sourceBytes))
+				ref := parseTangledLink(baseUrl, dest)
+				if ref != nil {
+					refLinkSet[*ref] = struct{}{}
+				}
+			}
+			return ast.WalkSkipChildren, nil
+		}
+		return ast.WalkContinue, nil
+	})
+	mentions := slices.Collect(maps.Keys(mentionsSet))
+	references := slices.Collect(maps.Keys(refLinkSet))
+	return mentions, references
+}
+
+func parseTangledLink(baseHost string, urlStr string) *models.ReferenceLink {
+	u, err := url.Parse(urlStr)
+	if err != nil {
+		return nil
+	}
+
+	if u.Host != "" && !strings.EqualFold(u.Host, baseHost) {
+		return nil
+	}
+
+	p := path.Clean(u.Path)
+	parts := strings.FieldsFunc(p, func(r rune) bool { return r == '/' })
+	if len(parts) < 4 {
+		// need at least: handle / repo / kind / id
+		return nil
+	}
+
+	var (
+		handle     = parts[0]
+		repo       = parts[1]
+		kindSeg    = parts[2]
+		subjectSeg = parts[3]
+	)
+
+	handle = strings.TrimPrefix(handle, "@")
+
+	var kind models.RefKind
+	switch kindSeg {
+	case "issues":
+		kind = models.RefKindIssue
+	case "pulls":
+		kind = models.RefKindPull
+	default:
+		return nil
+	}
+
+	subjectId, err := strconv.Atoi(subjectSeg)
+	if err != nil {
+		return nil
+	}
+	var commentId *int
+	if u.Fragment != "" {
+		if strings.HasPrefix(u.Fragment, "comment-") {
+			commentIdStr := u.Fragment[len("comment-"):]
+			if id, err := strconv.Atoi(commentIdStr); err == nil {
+				commentId = &id
+			}
+		}
+	}
+
+	return &models.ReferenceLink{
+		Handle:    handle,
+		Repo:      repo,
+		Kind:      kind,
+		SubjectId: subjectId,
+		CommentId: commentId,
+	}
+}

appview/pages/markup/reference_link_test.go

···
+package markup_test
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"tangled.org/core/appview/models"
+	"tangled.org/core/appview/pages/markup"
+)
+
+func TestMarkupParsing(t *testing.T) {
+	tests := []struct {
+		name         string
+		source       string
+		wantHandles  []string
+		wantRefLinks []models.ReferenceLink
+	}{
+		{
+			name:        "normal link",
+			source:      `[link](http://127.0.0.1:3000/alice.pds.tngl.boltless.dev/coolproj/issues/1)`,
+			wantHandles: make([]string, 0),
+			wantRefLinks: []models.ReferenceLink{
+				{Handle: "alice.pds.tngl.boltless.dev", Repo: "coolproj", Kind: models.RefKindIssue, SubjectId: 1, CommentId: nil},
+			},
+		},
+		{
+			name:        "commonmark style autolink",
+			source:      `<http://127.0.0.1:3000/alice.pds.tngl.boltless.dev/coolproj/issues/1>`,
+			wantHandles: make([]string, 0),
+			wantRefLinks: []models.ReferenceLink{
+				{Handle: "alice.pds.tngl.boltless.dev", Repo: "coolproj", Kind: models.RefKindIssue, SubjectId: 1, CommentId: nil},
+			},
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			handles, refLinks := markup.FindReferences("http://127.0.0.1:3000", tt.source)
+			assert.ElementsMatch(t, tt.wantHandles, handles)
+			assert.ElementsMatch(t, tt.wantRefLinks, refLinks)
+		})
+	}
+}

appview/pulls/pulls.go

···
 	"tangled.org/core/appview/pages"
 	"tangled.org/core/appview/pages/markup"
 	"tangled.org/core/appview/pages/repoinfo"
+	"tangled.org/core/appview/refresolver"
 	"tangled.org/core/appview/reporesolver"
 	"tangled.org/core/appview/validator"
 	"tangled.org/core/appview/xrpcclient"
···
 	repoResolver *reporesolver.RepoResolver
 	pages        *pages.Pages
 	idResolver   *idresolver.Resolver
+	refResolver  *refresolver.Resolver
 	db           *db.DB
 	config       *config.Config
 	notifier     notify.Notifier
···
 	repoResolver *reporesolver.RepoResolver,
 	pages *pages.Pages,
 	resolver *idresolver.Resolver,
+	refResolver *refresolver.Resolver,
 	db *db.DB,
 	config *config.Config,
 	notifier notify.Notifier,
···
 		repoResolver: repoResolver,
 		pages:        pages,
 		idResolver:   resolver,
+		refResolver:  refResolver,
 		db:           db,
 		config:       config,
 		notifier:     notifier,
···
 }
 
 func (s *Pulls) PullComment(w http.ResponseWriter, r *http.Request) {
-	l := s.logger.With("handler", "PullComment")
 	user := s.oauth.GetUser(r)
 	f, err := s.repoResolver.Resolve(r)
 	if err != nil {
···
 			s.pages.Notice(w, "pull", "Comment body is required")
 			return
 		}
+
+		mentions, _ := s.refResolver.Resolve(r.Context(), body)
 
 		// Start a transaction
 		tx, err := s.db.BeginTx(r.Context(), nil)
···
 			return
 		}
 
-		rawMentions := markup.FindUserMentions(comment.Body)
-		idents := s.idResolver.ResolveIdents(r.Context(), rawMentions)
-		l.Debug("parsed mentions", "raw", rawMentions, "idents", idents)
-		var mentions []syntax.DID
-		for _, ident := range idents {
-			if ident != nil && !ident.Handle.IsInvalidHandle() {
-				mentions = append(mentions, ident.DID)
-			}
-		}
 		s.notifier.NewPullComment(r.Context(), comment, mentions)
 
 		ownerSlashRepo := reporesolver.GetBaseRepoPath(r, f)

appview/refresolver/resolver.go

···
+package refresolver
+
+import (
+	"context"
+	"log/slog"
+
+	"github.com/bluesky-social/indigo/atproto/syntax"
+	"tangled.org/core/appview/config"
+	"tangled.org/core/appview/db"
+	"tangled.org/core/appview/models"
+	"tangled.org/core/appview/pages/markup"
+	"tangled.org/core/idresolver"
+)
+
+type Resolver struct {
+	config     *config.Config
+	idResolver *idresolver.Resolver
+	execer     db.Execer
+	logger     *slog.Logger
+}
+
+func New(
+	config *config.Config,
+	idResolver *idresolver.Resolver,
+	execer db.Execer,
+	logger *slog.Logger,
+) *Resolver {
+	return &Resolver{
+		config,
+		idResolver,
+		execer,
+		logger,
+	}
+}
+
+func (r *Resolver) Resolve(ctx context.Context, source string) ([]syntax.DID, []syntax.ATURI) {
+	l := r.logger.With("method", "find_references")
+	rawMentions, rawRefs := markup.FindReferences(r.config.Core.AppviewHost, source)
+	l.Debug("found possible references", "mentions", rawMentions, "refs", rawRefs)
+	idents := r.idResolver.ResolveIdents(ctx, rawMentions)
+	var mentions []syntax.DID
+	for _, ident := range idents {
+		if ident != nil && !ident.Handle.IsInvalidHandle() {
+			mentions = append(mentions, ident.DID)
+		}
+	}
+	l.Debug("found mentions", "mentions", mentions)
+
+	var resolvedRefs []models.ReferenceLink
+	for _, rawRef := range rawRefs {
+		ident, err := r.idResolver.ResolveIdent(ctx, rawRef.Handle)
+		if err != nil || ident == nil || ident.Handle.IsInvalidHandle() {
+			continue
+		}
+		rawRef.Handle = string(ident.DID)
+		resolvedRefs = append(resolvedRefs, rawRef)
+	}
+	aturiRefs, err := db.FindReferences(r.execer, resolvedRefs)
+	if err != nil {
+		l.Error("failed running query", "err", err)
+	}
+	l.Debug("found references", "refs", aturiRefs)
+
+	return mentions, aturiRefs
+}

appview/state/router.go

···
 		s.enforcer,
 		s.pages,
 		s.idResolver,
+		s.refResolver,
 		s.db,
 		s.config,
 		s.notifier,
···
 		s.repoResolver,
 		s.pages,
 		s.idResolver,
+		s.refResolver,
 		s.db,
 		s.config,
 		s.notifier,

appview/state/state.go

···
 	phnotify "tangled.org/core/appview/notify/posthog"
 	"tangled.org/core/appview/oauth"
 	"tangled.org/core/appview/pages"
+	"tangled.org/core/appview/refresolver"
 	"tangled.org/core/appview/reporesolver"
 	"tangled.org/core/appview/validator"
 	xrpcclient "tangled.org/core/appview/xrpcclient"
···
 	enforcer      *rbac.Enforcer
 	pages         *pages.Pages
 	idResolver    *idresolver.Resolver
+	refResolver   *refresolver.Resolver
 	posthog       posthog.Client
 	jc            *jetstream.JetstreamClient
 	config        *config.Config
···
 	validator := validator.New(d, res, enforcer)
 
 	repoResolver := reporesolver.New(config, enforcer, d)
+
+	refResolver := refresolver.New(config, res, d, log.SubLogger(logger, "refResolver"))
 
 	wrapper := db.DbWrapper{Execer: d}
 	jc, err := jetstream.NewJetstreamClient(
···
 		enforcer,
 		pages,
 		res,
+		refResolver,
 		posthog,
 		jc,
 		config,