Files
accounted/extensions/general/invoice-inbox/lib/mirror-extraction.ts
T
c7a75d069d feat(ai): job-shaped AI service with OpenAI-compatible backend, extraction-first; stop extracting every inbox document twice (#1740)
* feat(ai): job-shaped AI service with OpenAI-compatible backend, extraction-first; stop extracting every inbox document twice

Sovereign plan WS1 PR1 (#1406 Tier 2, extraction-first, aligned with the
AI surface audit).

lib/ai grows a job-shaped service (generateText / generateStructured /
extractFromDocument; no streaming members yet, see plan rule R3):
- services/anthropic-family delegates to the existing createAiClient()
  and sends the exact request literals the inbox extractor sent before
  (request-shape tests deep-equal them), so hosted Bedrock stays
  byte-identical.
- services/openai-compatible talks to any chat-completions endpoint
  (BYO Swedish provider) via Vercel AI SDK 6.x, exact-pinned and
  guarded: images as parts, PDFs rasterized with poppler (AI_PDF_MODE)
  or sent natively, AI_VISION / AI_STRICT_JSON declared, honest skips
  (ai_no_vision, pdf_rasterizer_missing) instead of fake failures.
- config.ts: AI_PROVIDER/AI_BASE_URL/AI_API_KEY/AI_MODEL and per-tier
  AI_*_MODEL with the legacy BEDROCK_* names kept as the same overrides;
  getAiStatus() is the single source of truth for "is AI wired up".
- provider.ts: openai-compatible in the auto-detect chain (after Bedrock
  and the direct API); createAiClient() refuses it loudly.

Document extraction moves onto the service and gets the audit's fixes:
- Inbox documents were extracted TWICE (pipeline A ran inside
  uploadDocument() before the inbox row existed, so its dedupe branch
  never fired; 3 707 + 1 666 calls / 30 d). The inbox now declares
  extractionOwner on the upload, the extension yields, and the inbox
  mirrors its single outcome onto document_attachments from every
  writer (sync, deferred, attach, retry, MCP).
- Every "no extraction will ever happen" outcome is stamped
  (skipped:no_ai_entitlement / ai_unconfigured / system_generated /
  ...); the status route maps the quiet ones to 'disabled' on the first
  poll instead of a 30 s client timeout. Prod showed 309 of the 327
  never-extracted uploads were the paywall working silently.
- Self-generated documents (our own invoice PDFs, payout files) are no
  longer OCR'd.
- Agent invoke answers 503 ai_unconfigured when the deployment has no
  assistant backend, distinct from the paywall.

Guard: new direct-ai-client antipattern check (shrink-only allowlist of
the pre-abstraction SDK callers) plus exact pins for @anthropic-ai/sdk,
ai and @ai-sdk/openai-compatible.

Verified: 15 958 unit tests green, guards, lint ratchet, typecheck, and a
live smoke against hosted Bedrock through the new service (ping, streamed
tool turn, thinking+cache, PDF extraction).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* feat(ai): make AI_API_KEY optional for OpenAI-compatible endpoints (keyless local model servers)

A local model server (llama.cpp's server, Ollama /v1, LM Studio, vLLM)
usually has no auth. Before, the OpenAI-compatible backend required both
AI_BASE_URL and AI_API_KEY to count as configured, so running Accounted on a
local model meant setting a meaningless placeholder key.

- resolveAiProvider / hasAiCredentials: a base URL alone is now enough.
- services/openai-compatible: only send Authorization: Bearer when AI_API_KEY
  is set, so a keyless server is never handed an empty bearer; a hosted
  provider that needs a key still sets it.
- Docs (SELF-HOSTING Option 3: local-model example, key marked optional),
  DECISIONS.

Verified: with no AI_API_KEY, just AI_BASE_URL + AI_MODEL, getAiStatus()
reports configured=true / provider=openai-compatible (live). lib/ai suite
71 green; tsc, guards, lint clean. Bedrock/Anthropic logic unchanged.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

---------

Co-authored-by: Jakob Wennberg <311770904+jakobwennberg-oss@users.noreply.github.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-08-20 19:39:08 +02:00

65 lines
2.5 KiB
TypeScript

import { createServiceClientNoCookies } from '@/lib/auth/api-keys'
import { createLogger } from '@/lib/logger'
import type { InvoiceExtractionResult } from '@/types'
const log = createLogger('invoice-inbox-mirror')
export interface ExtractionOutcome {
/** The parsed result (may be the empty skeleton). Ignored unless `rawText` is set and nothing was skipped. */
data: InvoiceExtractionResult | null
/** Raw model output; null when the call failed or was skipped. */
rawText: string | null
/** Provider-form model id that answered, when a call was made. */
model?: string | null
/** Why no model call was made: inbox skip reasons and lib/ai skip reasons alike. */
skipped?: string | null
}
/**
* Mirror an inbox extraction outcome onto document_attachments
* (extracted_data / extracted_at / extraction_model).
*
* The invoice inbox owns extraction for the documents it ingests, and the
* document-extraction extension yields to it (see extractionOwner on the
* document.uploaded event). Everything that read the document row before
* (the extraction-status poll behind the upload UI, the agent intents'
* "what do we already know" reads) keeps working because the inbox writes
* the same columns the extension would have, with the one model call it
* actually made.
*
* Service-role client, same as the extension: this runs from routes, the
* deferred worker and the MCP server alike, and the write is a system
* side-effect rather than a user action. Never throws: a failed mirror
* leaves the document row unstamped, which the UI already tolerates.
*/
export async function mirrorExtractionToDocument(
documentId: string,
outcome: ExtractionOutcome
): Promise<void> {
try {
const succeeded = !outcome.skipped && outcome.rawText != null && outcome.data != null
const extractionModel = outcome.skipped
? `skipped:${outcome.skipped}`
: succeeded
? outcome.model || 'invoice-inbox'
: 'failed:no_raw_text'
const supabase = createServiceClientNoCookies()
const { error } = await supabase
.from('document_attachments')
.update({
extracted_data: succeeded ? (outcome.data as unknown as Record<string, unknown>) : null,
extracted_at: new Date().toISOString(),
extraction_model: extractionModel,
})
.eq('id', documentId)
if (error) {
log.warn('mirror failed', { doc: documentId, extractionModel, err: error.message })
}
} catch (err) {
log.warn('mirror threw', {
doc: documentId,
err: err instanceof Error ? err.message : String(err),
})
}
}