claude/kochwas
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(ai): reichhaltigeres Logging fuer AI_FAILED-Diagnose
All checks were successful
Build & Publish Docker Image / build-and-push (push) Successful in 2m15s
Details

Browse Source 
Der bisherige Log "[extract-from-photo] AI_FAILED after 43165ms,
385807 bytes" verriet nicht, ob es JSON-Parse, Schema-Validierung
oder ein SDK-Fehler war. Endpoint haengt jetzt e.message an;
gemini-client loggt den First-Attempt-Fehler vor dem Retry und
packt bei AI_FAILED beide Messages in den finalen Error.

Keine Prompt-/Response-Inhalte werden geloggt -- nur unsere eigenen
GeminiError-Messages (Zod-Pfade, "non-JSON output", SDK-toString).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
hsiegeln
2026-04-21 14:08:10 +02:00
parent fb7c2f0e9b
commit efdcace892
2 changed files with 22 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/routes/api/recipes/extract-from-photo/+server.ts
View File

@@ -121,9 +121,11 @@ export const POST: RequestHandler = async ({ request, getClientAddress }) => {
: e.code === 'AI_NOT_CONFIGURED'
? 503
: 503;
// Nur Code + Meta loggen, niemals Prompt/Response-Inhalt.
// Nur Code + Meta + Error-Message loggen, niemals Prompt/Response-Inhalt.
// e.message enthaelt z.B. Zod-Validierungspfade oder "non-JSON output" --
// kein AI-Content, aber die Diagnose-Info, warum AI_FAILED kam.
console.warn(
`[extract-from-photo] ${e.code} after ${Date.now() - startedAt}ms, ${preprocessed.buffer.byteLength} bytes`
`[extract-from-photo] ${e.code} after ${Date.now() - startedAt}ms, ${preprocessed.buffer.byteLength} bytes: ${e.message}`
);
return errJson(status, e.code, 'Die Bild-Analyse ist fehlgeschlagen.');
}