hyungi/ai-server
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

chore: save WIP before importing Document-AI subtree

Browse Source
This commit is contained in:
hyungi
2025-08-13 08:38:30 +09:00
parent 4c81686657
commit 9c70d3e8a1
9 changed files with 154 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
server/main.py
View File

@@ -188,6 +188,7 @@ def paperless_sync(req: PaperlessSyncRequest, _: None = Depends(require_api_key)
client = PaperlessClient(settings.paperless_base_url, settings.paperless_token)
from .index_store import IndexRow
added_total = 0
skipped = 0
next_url: str | None = None
fetched = 0
@@ -205,13 +206,18 @@ def paperless_sync(req: PaperlessSyncRequest, _: None = Depends(require_api_key)
doc_id = doc.get("id")
if not doc_id:
continue
text = client.get_document_text(int(doc_id))
if not text:
try:
text = client.get_document_text(int(doc_id))
if not text:
skipped += 1
continue
parts = chunk_text(text)
for i, t in enumerate(parts):
vec = ollama.embeddings(settings.embedding_model, t)
to_append.append(IndexRow(id=f"paperless:{doc_id}:{i}", text=t, vector=vec, source="paperless"))
except Exception:
skipped += 1
continue
parts = chunk_text(text)
for i, t in enumerate(parts):
vec = ollama.embeddings(settings.embedding_model, t)
to_append.append(IndexRow(id=f"paperless:{doc_id}:{i}", text=t, vector=vec, source="paperless"))
if to_append:
added_total += index.append(to_append)
fetched += len(results)
@@ -221,7 +227,7 @@ def paperless_sync(req: PaperlessSyncRequest, _: None = Depends(require_api_key)
if not next_url:
break
return {"status": "synced", "added": added_total}
return {"status": "synced", "added": added_total, "skipped": skipped}
# OpenAI-compatible chat completions (minimal)