Reduce noisy Gemini batch failure warnings

TODO.md

@@ -233,6 +233,7 @@
 - The result modal should now stay within viewport height, but this still needs real browser confirmation on multiple short-height displays because CSS-only constraints were the source of the latest user-visible regression.
 - Artgrid preview playback now has a server-side ffmpeg transcode path for `.m3u8` style preview URLs, but this trades storage savings for runtime CPU cost.
 - The reverted `5ca7aef` experiment showed that simply widening collector caps and Gemini candidate count can backfire when the added candidates are weak: final visible count fell sharply even though backend raw candidate count increased.
+- Gemini Vision batch execution can still see mixed failure modes: true model/output-format failures and ignorable low-value/no-visual candidate skips. These should not be conflated in user-facing warnings.
 - The local self-test script is better than before, but it is still a smoke test, not full integration coverage.
 
 ## Current Risks Around Search Quality
@@ -554,6 +555,7 @@
 ## Highest-Value Next Steps
 - [ ] Reduce `/api/search` latency further without collapsing result count
 - [ ] Rebuild the reverted search-expansion work from the previous stable baseline, but only after measuring where candidate quality collapses between ranked pool and final merge
+- [ ] Continue hardening Gemini response parsing so truncated JSON or mixed-quality candidate batches degrade more gracefully without surfacing alarming warnings
 - [ ] Build a repeatable repo-local bootstrap script or documented setup command set for non-root machines so fresh PC setup does not depend on shell history
 - [ ] Improve Envato / Artgrid preview acquisition reliability so Gemini Vision sees real frames more often
 - [ ] Browser-verify the new result modal at multiple viewport heights and confirm translated Source Summary readability on real long descriptions
@@ -624,6 +626,25 @@
 - If behavior in the browser does not match the latest backend/frontend code, the first assumption should be stale frontend assets until proven otherwise
 
 ## Recent Change Log
+- Date: `2026-03-17`
+- What changed:
+  - Reduced Gemini Vision batch size from `8` to `6` candidates to lower the chance of oversized/truncated JSON responses.
+  - Added an explicit Gemini output token cap for the vision JSON response path.
+  - Changed Gemini batch failure accounting so ignorable per-candidate skips such as low-value thumbnails or missing visuals no longer count as a user-facing partial batch failure when useful recommendations were still recovered.
+  - Added unit coverage for the hard-error filtering behavior.
+- Why it changed:
+  - The user reported the warning `gemini vision partially failed on 2 of 2 batches`.
+  - The provided log `ai-media-hub-2026-03-17T04-35-33-028Z.log` showed two distinct failure classes during the second repeated search:
+    - a real batch-level JSON extraction failure caused by truncated Gemini output
+    - multiple ignorable sequential-retry skips caused by low-value thumbnails or missing visuals
+  - Those were being combined into an alarming partial-failure warning even though useful recommendations were still recovered.
+- How it was verified:
+  - `go test ./...`
+  - `bash scripts/selftest.sh`
+- What is still risky or incomplete:
+  - This reduces false-positive partial-failure warnings, but true Gemini output truncation can still happen occasionally if model behavior changes again.
+  - Further hardening may still be needed if Gemini starts returning different malformed JSON shapes.
+
 - Date: `2026-03-17`
 - What changed:
   - Reverted commit `5ca7aef` (`Strengthen search breadth and modal fitting`) to restore the previous stable search/modal baseline.

backend/services/gemini.go

@@ -306,6 +306,7 @@ User query: ` + query,
 		},
 		"generationConfig": map[string]any{
 			"responseMimeType": "application/json",
+			"maxOutputTokens":  1400,
 		},
 	}
 

backend/services/gemini_test.go

@@ -213,3 +213,18 @@ func TestMergeRecommendationsExcludesIrrelevantAndPendingFiller(t *testing.T) {
 		t.Fatalf("unexpected merged result: %#v", merged)
 	}
 }
+
+func TestFilterHardGeminiErrorsIgnoresLowValueVisualFailures(t *testing.T) {
+	errs := []string{
+		"candidate thumbnail is low value",
+		"no candidate thumbnails or preview frames could be fetched for gemini vision",
+		"gemini vision JSON extraction failed: no complete JSON object found",
+	}
+	filtered := filterHardGeminiErrors(errs)
+	if len(filtered) != 1 {
+		t.Fatalf("expected only hard errors to remain, got %#v", filtered)
+	}
+	if !strings.Contains(filtered[0], "JSON extraction failed") {
+		t.Fatalf("unexpected filtered errors: %#v", filtered)
+	}
+}

backend/services/ranker.go

@@ -105,7 +105,7 @@ func EvaluateAllCandidatesWithGemini(service *GeminiService, query string, ranke
 }
 
 func EvaluateAllCandidatesWithGeminiWithDeadline(service *GeminiService, query string, ranked []SearchResult, deadline time.Time) ([]AIRecommendation, GeminiBatchStats, error) {
-	const chunkSize = 8
+	const chunkSize = 6
 	const maxConcurrentBatches = 2
 	if service == nil {
 		return nil, GeminiBatchStats{}, fmt.Errorf("gemini service is not configured")
@@ -187,6 +187,7 @@ func EvaluateAllCandidatesWithGeminiWithDeadline(service *GeminiService, query s
 				})
 			}
 			recovered, recoveredErrs := recoverGeminiBatchSequentially(service, query, ranked, batch.index*chunkSize)
+			hardErrs := filterHardGeminiErrors(recoveredErrs)
 			if len(recovered) > 0 {
 				stats.SequentialRetried++
 				stats.Succeeded++
@@ -197,9 +198,9 @@ func EvaluateAllCandidatesWithGeminiWithDeadline(service *GeminiService, query s
 					seen[item.Link] = true
 					merged = append(merged, item)
 				}
-				if len(recoveredErrs) > 0 {
+				if len(hardErrs) > 0 {
 					stats.Failed++
-					for _, recoveredErr := range recoveredErrs {
+					for _, recoveredErr := range hardErrs {
 						if len(stats.Errors) < 5 {
 							stats.Errors = append(stats.Errors, recoveredErr)
 						}
@@ -207,6 +208,9 @@ func EvaluateAllCandidatesWithGeminiWithDeadline(service *GeminiService, query s
 				}
 				continue
 			}
+			if len(hardErrs) == 0 {
+				continue
+			}
 			stats.Failed++
 			if len(stats.Errors) < 5 {
 				stats.Errors = append(stats.Errors, batch.err.Error())
@@ -329,6 +333,35 @@ func MergeUniqueRecommendations(base, extra []AIRecommendation) []AIRecommendati
 	return merged
 }
 
+func filterHardGeminiErrors(errs []string) []string {
+	filtered := make([]string, 0, len(errs))
+	for _, item := range errs {
+		if isIgnorableGeminiError(item) {
+			continue
+		}
+		filtered = append(filtered, item)
+	}
+	return filtered
+}
+
+func isIgnorableGeminiError(message string) bool {
+	lower := strings.ToLower(strings.TrimSpace(message))
+	if lower == "" {
+		return false
+	}
+	for _, token := range []string{
+		"no candidate thumbnails or preview frames could be fetched for gemini vision",
+		"candidate thumbnail is low value",
+		"candidate has no thumbnail or preview video",
+		"image url is empty",
+	} {
+		if strings.Contains(lower, token) {
+			return true
+		}
+	}
+	return false
+}
+
 func MergeGeminiBatchStats(base, extra GeminiBatchStats) GeminiBatchStats {
 	merged := base
 	merged.CandidateCap += extra.CandidateCap