|
| 1 | +package routes |
| 2 | + |
| 3 | +import ( |
| 4 | + "context" |
| 5 | + "log/slog" |
| 6 | + "net/http" |
| 7 | + "time" |
| 8 | + |
| 9 | + "github.com/appsprout-dev/mnemonic/internal/llm" |
| 10 | + "github.com/appsprout-dev/mnemonic/internal/store" |
| 11 | +) |
| 12 | + |
| 13 | +// BackfillResponse reports what the backfill operation did. |
| 14 | +type BackfillResponse struct { |
| 15 | + Total int `json:"total"` |
| 16 | + Embedded int `json:"embedded"` |
| 17 | + Failed int `json:"failed"` |
| 18 | + Skipped int `json:"skipped"` |
| 19 | + Errors []string `json:"errors,omitempty"` |
| 20 | +} |
| 21 | + |
| 22 | +// HandleBackfillEmbeddings finds memories with empty embeddings and generates them. |
| 23 | +func HandleBackfillEmbeddings(s store.Store, provider llm.Provider, log *slog.Logger) http.HandlerFunc { |
| 24 | + return func(w http.ResponseWriter, r *http.Request) { |
| 25 | + ctx, cancel := context.WithTimeout(r.Context(), 5*time.Minute) |
| 26 | + defer cancel() |
| 27 | + |
| 28 | + // Find all active memories missing embeddings |
| 29 | + memories, err := s.ListMemories(ctx, "", 500, 0) |
| 30 | + if err != nil { |
| 31 | + log.Error("backfill: failed to list memories", "error", err) |
| 32 | + writeError(w, http.StatusInternalServerError, "failed to list memories", "STORE_ERROR") |
| 33 | + return |
| 34 | + } |
| 35 | + |
| 36 | + var missing []store.Memory |
| 37 | + for _, m := range memories { |
| 38 | + if len(m.Embedding) == 0 { |
| 39 | + missing = append(missing, m) |
| 40 | + } |
| 41 | + } |
| 42 | + |
| 43 | + if len(missing) == 0 { |
| 44 | + writeJSON(w, http.StatusOK, BackfillResponse{Total: 0}) |
| 45 | + return |
| 46 | + } |
| 47 | + |
| 48 | + log.Info("backfill: starting embedding backfill", "missing", len(missing)) |
| 49 | + |
| 50 | + // Quick sanity check: can we embed at all? |
| 51 | + testEmb, testErr := provider.Embed(ctx, "test embedding sanity check") |
| 52 | + if testErr != nil { |
| 53 | + log.Error("backfill: embedding sanity check failed", "error", testErr) |
| 54 | + writeJSON(w, http.StatusOK, BackfillResponse{Total: len(missing), Errors: []string{"sanity check failed: " + testErr.Error()}}) |
| 55 | + return |
| 56 | + } |
| 57 | + log.Info("backfill: sanity check passed", "dims", len(testEmb)) |
| 58 | + |
| 59 | + resp := BackfillResponse{Total: len(missing)} |
| 60 | + |
| 61 | + for _, mem := range missing { |
| 62 | + select { |
| 63 | + case <-ctx.Done(): |
| 64 | + log.Warn("backfill: context cancelled", "embedded", resp.Embedded, "remaining", resp.Total-resp.Embedded-resp.Failed) |
| 65 | + writeJSON(w, http.StatusOK, resp) |
| 66 | + return |
| 67 | + default: |
| 68 | + } |
| 69 | + |
| 70 | + // Build embedding text from summary + content (same as encoding agent) |
| 71 | + text := mem.Summary + " " + mem.Content |
| 72 | + if len(text) > 4000 { |
| 73 | + text = text[:4000] |
| 74 | + } |
| 75 | + |
| 76 | + embedding, err := provider.Embed(ctx, text) |
| 77 | + if err != nil { |
| 78 | + resp.Errors = append(resp.Errors, "embed:"+mem.ID[:8]+":"+err.Error()) |
| 79 | + resp.Failed++ |
| 80 | + continue |
| 81 | + } |
| 82 | + |
| 83 | + if len(embedding) == 0 { |
| 84 | + resp.Skipped++ |
| 85 | + continue |
| 86 | + } |
| 87 | + |
| 88 | + // Use targeted update to avoid FK issues with raw_id |
| 89 | + if err := s.UpdateEmbedding(ctx, mem.ID, embedding); err != nil { |
| 90 | + resp.Errors = append(resp.Errors, "update:"+mem.ID[:8]+":"+err.Error()) |
| 91 | + resp.Failed++ |
| 92 | + continue |
| 93 | + } |
| 94 | + |
| 95 | + resp.Embedded++ |
| 96 | + log.Debug("backfill: embedded memory", "id", mem.ID, "dims", len(embedding)) |
| 97 | + } |
| 98 | + |
| 99 | + log.Info("backfill: completed", "total", resp.Total, "embedded", resp.Embedded, "failed", resp.Failed) |
| 100 | + writeJSON(w, http.StatusOK, resp) |
| 101 | + } |
| 102 | +} |
0 commit comments