k1nq/TREK
1
0
Fork 0
mirror of https://github.com/mauriceboe/TREK.git synced 2026-07-09 15:05:59 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(extract): make AI imports reliable and fast on local models

Browse Source 
client: the import call inherited the global 8s axios timeout and aborted long LLM extractions even though the server finished it; remove the timeout. server: raise the OpenAI-compatible LLM timeout 60s->180s (a cold Ollama model can take ~45s to first token). server: cap extracted text to 8000 chars before the LLM - multi-page T&C tails (30k+ chars) overflowed the context window, truncating the relevant head and making CPU inference crawl; booking details sit at the top.
This commit is contained in:
Maurice
2026-06-24 21:20:20 +02:00
parent 22813f8d81
commit ac03b7ca13
3 changed files with 12 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
client/src/api/client.ts
+3 -1
View File
@@ -664,7 +664,9 @@ export const reservationsApi = {
const fd = new FormData()
for (const f of files) fd.append('files', f)
fd.append('mode', mode)
return apiClient.post(`/trips/${tripId}/reservations/import/booking`, fd, { headers: { 'Content-Type': 'multipart/form-data' } }).then(r => r.data)
// No client-side timeout: kitinerary + LLM extraction routinely exceeds the
// global 8s default (a cold local model alone can take ~45s).
return apiClient.post(`/trips/${tripId}/reservations/import/booking`, fd, { headers: { 'Content-Type': 'multipart/form-data' }, timeout: 0 }).then(r => r.data)
},
importBookingConfirm: (tripId: number | string, items: BookingImportPreviewItem[]): Promise<BookingImportConfirmResponse> =>
apiClient.post(`/trips/${tripId}/reservations/import/booking/confirm`, { items }).then(r => r.data),