Spaces:
Sleeping
Sleeping
Commit
·
f91bb25
1
Parent(s):
e5d6418
Upd upload stream
Browse files- app/routers/import_doc.py +62 -1
app/routers/import_doc.py
CHANGED
|
@@ -20,7 +20,8 @@ class ImportRequest(BaseModel):
|
|
| 20 |
source: str
|
| 21 |
ref: dict
|
| 22 |
|
| 23 |
-
|
|
|
|
| 24 |
@router.post("")
|
| 25 |
async def import_book(req: ImportRequest):
|
| 26 |
logger.info(f"📥 Received import request: {req.dict()}")
|
|
@@ -105,6 +106,66 @@ async def import_book(req: ImportRequest):
|
|
| 105 |
"uri": uri
|
| 106 |
}
|
| 107 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 108 |
|
| 109 |
# Fetch textbook on id
|
| 110 |
@router.get("/textbook/{doc_id}")
|
|
|
|
| 20 |
source: str
|
| 21 |
ref: dict
|
| 22 |
|
| 23 |
+
|
| 24 |
+
# Online stream: Embedding, query and PDF saver to buckets
|
| 25 |
@router.post("")
|
| 26 |
async def import_book(req: ImportRequest):
|
| 27 |
logger.info(f"📥 Received import request: {req.dict()}")
|
|
|
|
| 106 |
"uri": uri
|
| 107 |
}
|
| 108 |
|
| 109 |
+
|
| 110 |
+
# File upload stream: Embedding, query and PDF saver to buckets
|
| 111 |
+
from fastapi import UploadFile, File, Form
|
| 112 |
+
@router.post("/upload")
|
| 113 |
+
async def upload_book(
|
| 114 |
+
file: UploadFile = File(...),
|
| 115 |
+
title: str = Form(...),
|
| 116 |
+
candidate_id: str = Form(...),
|
| 117 |
+
source: str = Form("manual") # Optional default source
|
| 118 |
+
):
|
| 119 |
+
logger.info(f"📤 Received direct upload for: {candidate_id} ({title})")
|
| 120 |
+
db = get_db()
|
| 121 |
+
# Insert placeholder document first
|
| 122 |
+
placeholder_doc = {
|
| 123 |
+
"_id": candidate_id,
|
| 124 |
+
"title": title,
|
| 125 |
+
"status": "PENDING",
|
| 126 |
+
"metadata": {
|
| 127 |
+
"source": source,
|
| 128 |
+
"ref": {"upload": True}
|
| 129 |
+
}
|
| 130 |
+
}
|
| 131 |
+
await db.documents.replace_one({"_id": candidate_id}, placeholder_doc, upsert=True)
|
| 132 |
+
# Save uploaded file to temp
|
| 133 |
+
file_path = f"/tmp/{candidate_id}.pdf"
|
| 134 |
+
try:
|
| 135 |
+
async with aiofiles.open(file_path, 'wb') as out_file:
|
| 136 |
+
content = await file.read()
|
| 137 |
+
await out_file.write(content)
|
| 138 |
+
logger.info(f"✅ Uploaded PDF temporarily saved to {file_path}")
|
| 139 |
+
except Exception as e:
|
| 140 |
+
logger.error(f"🚨 Failed to save uploaded file: {e}")
|
| 141 |
+
raise HTTPException(500, "Failed to save uploaded PDF")
|
| 142 |
+
# Save to buckets
|
| 143 |
+
try:
|
| 144 |
+
grid_fs_bucket = get_gridfs()
|
| 145 |
+
with open(file_path, "rb") as f:
|
| 146 |
+
await grid_fs_bucket.upload_from_stream(f"{candidate_id}.pdf", f)
|
| 147 |
+
await save_to_textbook_fs(candidate_id, file_path)
|
| 148 |
+
os.remove(file_path)
|
| 149 |
+
except Exception as e:
|
| 150 |
+
logger.error(f"💥 Failed to upload to GridFS: {e}")
|
| 151 |
+
raise HTTPException(500, "Storage failed")
|
| 152 |
+
# Update metadata and trigger ingestion
|
| 153 |
+
await db.documents.update_one(
|
| 154 |
+
{"_id": candidate_id},
|
| 155 |
+
{"$set": {"status": "DOWNLOADING"}}
|
| 156 |
+
)
|
| 157 |
+
asyncio.create_task(parse_and_index(candidate_id))
|
| 158 |
+
logger.info(f"📚 Direct upload {candidate_id} queued for indexing")
|
| 159 |
+
# Final block
|
| 160 |
+
return {
|
| 161 |
+
"status": "QUEUED",
|
| 162 |
+
"id": candidate_id,
|
| 163 |
+
"title": title,
|
| 164 |
+
"source": source,
|
| 165 |
+
"documentId": candidate_id,
|
| 166 |
+
"uri": f"/import/textbook/{candidate_id}"
|
| 167 |
+
}
|
| 168 |
+
|
| 169 |
|
| 170 |
# Fetch textbook on id
|
| 171 |
@router.get("/textbook/{doc_id}")
|