refactor(store,tools): migrate IDs from UUID to bigserial int64

All internal entity lookups now use bigserial primary keys (int64) while GUIDs are retained for external/public identification. Updated store functions (TouchProject, UpdateThoughtMetadata, AddThoughtAttachment) to query by id instead of guid, added GetThoughtByID, changed semanticSearch and all tool helpers to use *int64 project IDs, and updated retry/backfill workers to use int64 thought IDs throughout.
2026-05-03 11:43:34 +02:00
parent 9e6d05e055
commit 91239bcf4b
58 changed files with 1208 additions and 2774 deletions
--- a/internal/store/thoughts.go
+++ b/internal/store/thoughts.go
@@ -33,12 +33,12 @@ func (db *DB) InsertThought(ctx context.Context, thought thoughttypes.Thought, e
 	row := tx.QueryRow(ctx, `
 		insert into thoughts (content, metadata, project_id)
 		values ($1, $2::jsonb, $3)
-		returning guid, created_at, updated_at
+		returning id, guid, created_at, updated_at
 	`, thought.Content, metadata, thought.ProjectID)

 	created := thought
 	created.Embedding = nil
-	if err := row.Scan(&created.ID, &created.CreatedAt, &created.UpdatedAt); err != nil {
+	if err := row.Scan(&created.ID, &created.GUID, &created.CreatedAt, &created.UpdatedAt); err != nil {
 		return thoughttypes.Thought{}, fmt.Errorf("insert thought: %w", err)
 	}

@@ -132,7 +132,7 @@ func (db *DB) ListThoughts(ctx context.Context, filter thoughttypes.ListFilter)
 	}

 	query := `
-		select guid, content, metadata, project_id, archived_at, created_at, updated_at
+		select id, guid, content, metadata, project_id, archived_at, created_at, updated_at
 		from thoughts
 	`
 	if len(conditions) > 0 {
@@ -151,7 +151,7 @@ func (db *DB) ListThoughts(ctx context.Context, filter thoughttypes.ListFilter)
 	thoughts := make([]thoughttypes.Thought, 0, filter.Limit)
 	for rows.Next() {
 		var model generatedmodels.ModelPublicThoughts
-		if err := rows.Scan(&model.GUID, &model.Content, &model.Metadata, &model.ProjectID, &model.ArchivedAt, &model.CreatedAt, &model.UpdatedAt); err != nil {
+		if err := rows.Scan(&model.ID, &model.GUID, &model.Content, &model.Metadata, &model.ProjectID, &model.ArchivedAt, &model.CreatedAt, &model.UpdatedAt); err != nil {
 			return nil, fmt.Errorf("scan listed thought: %w", err)
 		}
 		thought, err := thoughtFromModel(model)
@@ -218,13 +218,13 @@ func (db *DB) Stats(ctx context.Context) (thoughttypes.ThoughtStats, error) {

 func (db *DB) GetThought(ctx context.Context, id uuid.UUID) (thoughttypes.Thought, error) {
 	row := db.pool.QueryRow(ctx, `
-		select guid, content, metadata, project_id, archived_at, created_at, updated_at
+		select id, guid, content, metadata, project_id, archived_at, created_at, updated_at
 		from thoughts
 		where guid = $1
 	`, id)

 	var model generatedmodels.ModelPublicThoughts
-	if err := row.Scan(&model.GUID, &model.Content, &model.Metadata, &model.ProjectID, &model.ArchivedAt, &model.CreatedAt, &model.UpdatedAt); err != nil {
+	if err := row.Scan(&model.ID, &model.GUID, &model.Content, &model.Metadata, &model.ProjectID, &model.ArchivedAt, &model.CreatedAt, &model.UpdatedAt); err != nil {
 		if err == pgx.ErrNoRows {
 			return thoughttypes.Thought{}, err
 		}
@@ -239,7 +239,30 @@ func (db *DB) GetThought(ctx context.Context, id uuid.UUID) (thoughttypes.Though
 	return thought, nil
 }

-func (db *DB) UpdateThought(ctx context.Context, id uuid.UUID, content string, embedding []float32, embeddingModel string, metadata thoughttypes.ThoughtMetadata, projectID *uuid.UUID) (thoughttypes.Thought, error) {
+func (db *DB) GetThoughtByID(ctx context.Context, id int64) (thoughttypes.Thought, error) {
+	row := db.pool.QueryRow(ctx, `
+		select id, guid, content, metadata, project_id, archived_at, created_at, updated_at
+		from thoughts
+		where id = $1
+	`, id)
+
+	var model generatedmodels.ModelPublicThoughts
+	if err := row.Scan(&model.ID, &model.GUID, &model.Content, &model.Metadata, &model.ProjectID, &model.ArchivedAt, &model.CreatedAt, &model.UpdatedAt); err != nil {
+		if err == pgx.ErrNoRows {
+			return thoughttypes.Thought{}, err
+		}
+		return thoughttypes.Thought{}, fmt.Errorf("get thought by id: %w", err)
+	}
+
+	thought, err := thoughtFromModel(model)
+	if err != nil {
+		return thoughttypes.Thought{}, fmt.Errorf("map thought: %w", err)
+	}
+
+	return thought, nil
+}
+
+func (db *DB) UpdateThought(ctx context.Context, id uuid.UUID, content string, embedding []float32, embeddingModel string, metadata thoughttypes.ThoughtMetadata, projectID *int64) (thoughttypes.Thought, error) {
 	metadataBytes, err := json.Marshal(metadata)
 	if err != nil {
 		return thoughttypes.Thought{}, fmt.Errorf("marshal updated metadata: %w", err)
@@ -271,7 +294,7 @@ func (db *DB) UpdateThought(ctx context.Context, id uuid.UUID, content string, e
 	if len(embedding) > 0 && embeddingModel != "" {
 		if _, err := tx.Exec(ctx, `
 			insert into embeddings (thought_id, model, dim, embedding)
-			values ($1, $2, $3, $4)
+			select id, $2, $3, $4 from thoughts where guid = $1
 			on conflict (thought_id, model) do update
 				set embedding = excluded.embedding,
 				    dim = excluded.dim,
@@ -288,7 +311,7 @@ func (db *DB) UpdateThought(ctx context.Context, id uuid.UUID, content string, e
 	return db.GetThought(ctx, id)
 }

-func (db *DB) UpdateThoughtMetadata(ctx context.Context, id uuid.UUID, metadata thoughttypes.ThoughtMetadata) (thoughttypes.Thought, error) {
+func (db *DB) UpdateThoughtMetadata(ctx context.Context, id int64, metadata thoughttypes.ThoughtMetadata) (thoughttypes.Thought, error) {
 	metadataBytes, err := json.Marshal(metadata)
 	if err != nil {
 		return thoughttypes.Thought{}, fmt.Errorf("marshal updated metadata: %w", err)
@@ -298,7 +321,7 @@ func (db *DB) UpdateThoughtMetadata(ctx context.Context, id uuid.UUID, metadata
 		update thoughts
 		set metadata = $2::jsonb,
 		    updated_at = now()
-		where guid = $1
+		where id = $1
 	`, id, metadataBytes)
 	if err != nil {
 		return thoughttypes.Thought{}, fmt.Errorf("update thought metadata: %w", err)
@@ -307,7 +330,7 @@ func (db *DB) UpdateThoughtMetadata(ctx context.Context, id uuid.UUID, metadata
 		return thoughttypes.Thought{}, pgx.ErrNoRows
 	}

-	return db.GetThought(ctx, id)
+	return db.GetThoughtByID(ctx, id)
 }

 func (db *DB) DeleteThought(ctx context.Context, id uuid.UUID) error {
@@ -332,7 +355,7 @@ func (db *DB) ArchiveThought(ctx context.Context, id uuid.UUID) error {
 	return nil
 }

-func (db *DB) RecentThoughts(ctx context.Context, projectID *uuid.UUID, limit int, days int) ([]thoughttypes.Thought, error) {
+func (db *DB) RecentThoughts(ctx context.Context, projectID *int64, limit int, days int) ([]thoughttypes.Thought, error) {
 	filter := thoughttypes.ListFilter{
 		Limit:           limit,
 		ProjectID:       projectID,
@@ -342,7 +365,7 @@ func (db *DB) RecentThoughts(ctx context.Context, projectID *uuid.UUID, limit in
 	return db.ListThoughts(ctx, filter)
 }

-func (db *DB) ListThoughtsPendingMetadataRetry(ctx context.Context, limit int, projectID *uuid.UUID, includeArchived bool, olderThanDays int) ([]thoughttypes.Thought, error) {
+func (db *DB) ListThoughtsPendingMetadataRetry(ctx context.Context, limit int, projectID *int64, includeArchived bool, olderThanDays int) ([]thoughttypes.Thought, error) {
 	args := make([]any, 0, 4)
 	conditions := []string{
 		"(metadata->>'metadata_status' = 'pending' or metadata->>'metadata_status' = 'failed')",
@@ -361,7 +384,7 @@ func (db *DB) ListThoughtsPendingMetadataRetry(ctx context.Context, limit int, p
 	}

 	query := `
-		select guid, content, metadata, project_id, archived_at, created_at, updated_at
+		select id, guid, content, metadata, project_id, archived_at, created_at, updated_at
 		from thoughts
 		where ` + strings.Join(conditions, " and ")

@@ -376,12 +399,12 @@ func (db *DB) ListThoughtsPendingMetadataRetry(ctx context.Context, limit int, p

 	thoughts := make([]thoughttypes.Thought, 0, limit)
 	for rows.Next() {
-		var thought thoughttypes.Thought
-		var metadataBytes []byte
-		if err := rows.Scan(&thought.ID, &thought.Content, &metadataBytes, &thought.ProjectID, &thought.ArchivedAt, &thought.CreatedAt, &thought.UpdatedAt); err != nil {
+		var model generatedmodels.ModelPublicThoughts
+		if err := rows.Scan(&model.ID, &model.GUID, &model.Content, &model.Metadata, &model.ProjectID, &model.ArchivedAt, &model.CreatedAt, &model.UpdatedAt); err != nil {
 			return nil, fmt.Errorf("scan pending metadata retry thought: %w", err)
 		}
-		if err := json.Unmarshal(metadataBytes, &thought.Metadata); err != nil {
+		thought, err := thoughtFromModel(model)
+		if err != nil {
 			return nil, fmt.Errorf("decode pending metadata retry thought: %w", err)
 		}
 		thoughts = append(thoughts, thought)
@@ -394,7 +417,7 @@ func (db *DB) ListThoughtsPendingMetadataRetry(ctx context.Context, limit int, p
 	return thoughts, nil
 }

-func (db *DB) SearchSimilarThoughts(ctx context.Context, embedding []float32, embeddingModel string, threshold float64, limit int, projectID *uuid.UUID, excludeID *uuid.UUID) ([]thoughttypes.SearchResult, error) {
+func (db *DB) SearchSimilarThoughts(ctx context.Context, embedding []float32, embeddingModel string, threshold float64, limit int, projectID *int64, excludeID *uuid.UUID) ([]thoughttypes.SearchResult, error) {
 	args := []any{pgvector.NewVector(embedding), threshold, embeddingModel}
 	conditions := []string{
 		"t.archived_at is null",
@@ -412,9 +435,9 @@ func (db *DB) SearchSimilarThoughts(ctx context.Context, embedding []float32, em
 	args = append(args, limit)

 	query := `
-		select t.guid, t.content, t.metadata, 1 - (e.embedding <=> $1) as similarity, t.created_at
+		select t.id, t.content, t.metadata, 1 - (e.embedding <=> $1) as similarity, t.created_at
 		from thoughts t
-		join embeddings e on e.thought_id = t.guid
+		join embeddings e on e.thought_id = t.id
 		where ` + strings.Join(conditions, " and ") + fmt.Sprintf(`
 		order by e.embedding <=> $1
 		limit $%d`, len(args))
@@ -443,7 +466,7 @@ func (db *DB) SearchSimilarThoughts(ctx context.Context, embedding []float32, em
 	return results, nil
 }

-func (db *DB) HasEmbeddingsForModel(ctx context.Context, model string, projectID *uuid.UUID) (bool, error) {
+func (db *DB) HasEmbeddingsForModel(ctx context.Context, model string, projectID *int64) (bool, error) {
 	args := []any{model}
 	conditions := []string{
 		"e.model = $1",
@@ -454,7 +477,7 @@ func (db *DB) HasEmbeddingsForModel(ctx context.Context, model string, projectID
 		conditions = append(conditions, fmt.Sprintf("t.project_id = $%d", len(args)))
 	}

-	query := `select exists(select 1 from embeddings e join thoughts t on t.guid = e.thought_id where ` +
+	query := `select exists(select 1 from embeddings e join thoughts t on t.id = e.thought_id where ` +
 		strings.Join(conditions, " and ") + `)`

 	var exists bool
@@ -464,7 +487,7 @@ func (db *DB) HasEmbeddingsForModel(ctx context.Context, model string, projectID
 	return exists, nil
 }

-func (db *DB) ListThoughtsMissingEmbedding(ctx context.Context, model string, limit int, projectID *uuid.UUID, includeArchived bool, olderThanDays int) ([]thoughttypes.Thought, error) {
+func (db *DB) ListThoughtsMissingEmbedding(ctx context.Context, model string, limit int, projectID *int64, includeArchived bool, olderThanDays int) ([]thoughttypes.Thought, error) {
 	args := []any{model}
 	conditions := []string{"e.id is null"}

@@ -482,9 +505,9 @@ func (db *DB) ListThoughtsMissingEmbedding(ctx context.Context, model string, li
 	args = append(args, limit)

 	query := `
-		select t.guid, t.content, t.metadata, t.project_id, t.archived_at, t.created_at, t.updated_at
+		select t.id, t.guid, t.content, t.metadata, t.project_id, t.archived_at, t.created_at, t.updated_at
 		from thoughts t
-		left join embeddings e on e.thought_id = t.guid and e.model = $1
+		left join embeddings e on e.thought_id = t.id and e.model = $1
 		where ` + strings.Join(conditions, " and ") + `
 		order by t.created_at asc
 		limit $` + fmt.Sprintf("%d", len(args))
@@ -497,12 +520,12 @@ func (db *DB) ListThoughtsMissingEmbedding(ctx context.Context, model string, li

 	thoughts := make([]thoughttypes.Thought, 0, limit)
 	for rows.Next() {
-		var thought thoughttypes.Thought
-		var metadataBytes []byte
-		if err := rows.Scan(&thought.ID, &thought.Content, &metadataBytes, &thought.ProjectID, &thought.ArchivedAt, &thought.CreatedAt, &thought.UpdatedAt); err != nil {
+		var model generatedmodels.ModelPublicThoughts
+		if err := rows.Scan(&model.ID, &model.GUID, &model.Content, &model.Metadata, &model.ProjectID, &model.ArchivedAt, &model.CreatedAt, &model.UpdatedAt); err != nil {
 			return nil, fmt.Errorf("scan missing-embedding thought: %w", err)
 		}
-		if err := json.Unmarshal(metadataBytes, &thought.Metadata); err != nil {
+		thought, err := thoughtFromModel(model)
+		if err != nil {
 			return nil, fmt.Errorf("decode missing-embedding metadata: %w", err)
 		}
 		thoughts = append(thoughts, thought)
@@ -513,7 +536,7 @@ func (db *DB) ListThoughtsMissingEmbedding(ctx context.Context, model string, li
 	return thoughts, nil
 }

-func (db *DB) ListThoughtsForMetadataReparse(ctx context.Context, limit int, projectID *uuid.UUID, includeArchived bool, olderThanDays int) ([]thoughttypes.Thought, error) {
+func (db *DB) ListThoughtsForMetadataReparse(ctx context.Context, limit int, projectID *int64, includeArchived bool, olderThanDays int) ([]thoughttypes.Thought, error) {
 	args := make([]any, 0, 3)
 	conditions := make([]string, 0, 4)

@@ -531,7 +554,7 @@ func (db *DB) ListThoughtsForMetadataReparse(ctx context.Context, limit int, pro
 	args = append(args, limit)

 	query := `
-		select guid, content, metadata, project_id, archived_at, created_at, updated_at
+		select id, guid, content, metadata, project_id, archived_at, created_at, updated_at
 		from thoughts
 	`
 	if len(conditions) > 0 {
@@ -547,12 +570,12 @@ func (db *DB) ListThoughtsForMetadataReparse(ctx context.Context, limit int, pro

 	thoughts := make([]thoughttypes.Thought, 0, limit)
 	for rows.Next() {
-		var thought thoughttypes.Thought
-		var metadataBytes []byte
-		if err := rows.Scan(&thought.ID, &thought.Content, &metadataBytes, &thought.ProjectID, &thought.ArchivedAt, &thought.CreatedAt, &thought.UpdatedAt); err != nil {
+		var model generatedmodels.ModelPublicThoughts
+		if err := rows.Scan(&model.ID, &model.GUID, &model.Content, &model.Metadata, &model.ProjectID, &model.ArchivedAt, &model.CreatedAt, &model.UpdatedAt); err != nil {
 			return nil, fmt.Errorf("scan metadata-reparse thought: %w", err)
 		}
-		if err := json.Unmarshal(metadataBytes, &thought.Metadata); err != nil {
+		thought, err := thoughtFromModel(model)
+		if err != nil {
 			return nil, fmt.Errorf("decode metadata-reparse thought metadata: %w", err)
 		}
 		thoughts = append(thoughts, thought)
@@ -564,7 +587,7 @@ func (db *DB) ListThoughtsForMetadataReparse(ctx context.Context, limit int, pro
 	return thoughts, nil
 }

-func (db *DB) UpsertEmbedding(ctx context.Context, thoughtID uuid.UUID, model string, embedding []float32) error {
+func (db *DB) UpsertEmbedding(ctx context.Context, thoughtID int64, model string, embedding []float32) error {
 	_, err := db.pool.Exec(ctx, `
 		insert into embeddings (thought_id, model, dim, embedding)
 		values ($1, $2, $3, $4)
@@ -579,7 +602,7 @@ func (db *DB) UpsertEmbedding(ctx context.Context, thoughtID uuid.UUID, model st
 	return nil
 }

-func (db *DB) SearchThoughtsText(ctx context.Context, query string, limit int, projectID *uuid.UUID, excludeID *uuid.UUID) ([]thoughttypes.SearchResult, error) {
+func (db *DB) SearchThoughtsText(ctx context.Context, query string, limit int, projectID *int64, excludeID *uuid.UUID) ([]thoughttypes.SearchResult, error) {
 	args := []any{query}
 	conditions := []string{
 		"t.archived_at is null",
@@ -596,11 +619,11 @@ func (db *DB) SearchThoughtsText(ctx context.Context, query string, limit int, p
 	args = append(args, limit)

 	q := `
-		select t.guid, t.content, t.metadata,
+		select t.id, t.content, t.metadata,
 		       ts_rank_cd(to_tsvector('simple', t.content) || to_tsvector('simple', coalesce(p.name, '')), websearch_to_tsquery('simple', $1)) as similarity,
 		       t.created_at
 		from thoughts t
-		left join projects p on t.project_id = p.guid
+		left join projects p on t.project_id = p.id
 		where ` + strings.Join(conditions, " and ") + `
 		order by similarity desc
 		limit $` + fmt.Sprintf("%d", len(args))