Spaces:
Sleeping
Sleeping
File size: 11,313 Bytes
ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd ff86bc9 ace5bcd |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 |
# rfp_api_app.py
# -*- coding: utf-8 -*-
"""
API FastAPI pour RFP Parser & Exports (DeepInfra en backend)
===========================================================
Ce que fait ce fichier
----------------------
- Expose 4 endpoints principaux :
- GET /health : ping + modèle utilisé.
- POST /submit : lance un job asynchrone {text} -> job_id.
- GET /status : renvoie l'état du job, liens de résultats (raw.json, own.csv, xlsx),
et un aperçu court du JSON pour “live JSON”.
- GET /results/{job_id}/... : sert les artefacts (raw.json, own.csv, feuille_de_charge.xlsx).
Ce que propose chaque fonction
------------------------------
- new_job(text: str) -> str :
Crée un job en mémoire, status "queued", retourne job_id.
- set_job_status(job_id: str, **updates) -> None :
Met à jour les métadonnées d'un job (thread-safe).
- parse_with_deepinfra(text: str) -> dict :
Construit le payload LLM (via rfp_parser.prompting.build_chat_payload),
appelle DeepInfra et parse le JSON renvoyé (strippé si fenced).
- persist_doc(job_dir: Path, doc: dict) -> tuple[str, str] :
Sauvegarde doc dans raw.json (UTF-8), retourne (path, url).
- build_csv_if_available(doc: dict, job_dir: Path) -> tuple[str|None, str|None] :
(Optionnel) si ton repo expose un export CSV, le génère; sinon None.
- build_xlsx(doc: dict, job_dir: Path) -> str :
Construit la feuille Excel dynamique (via rfp_parser.exports_xls.build_xls_from_doc).
- run_job(job_id: str, text: str) -> None :
Orchestration : parse -> persist raw.json -> (own.csv si dispo) -> xlsx -> maj status.
Logs
----
- Logger [API] avec niveau INFO (ou DEBUG si RFP_DEBUG=1).
- Traces détaillées sur DeepInfra / persistences / erreurs.
Notes
-----
- Jobs en mémoire : si le process redémarre, l'état est perdu (simple mais suffisant en Space).
- CORS permissif (*): autorise les requêtes depuis le Space Gradio.
"""
from __future__ import annotations
from typing import Dict, Any, Tuple, Optional
import os, json, uuid, threading, time, traceback
from pathlib import Path
import logging
import requests
from fastapi import FastAPI, HTTPException, Query
from fastapi.responses import JSONResponse, FileResponse
from fastapi.middleware.cors import CORSMiddleware
# === Imports depuis ton repo RFPmaster ===
from rfp_parser.prompting import build_chat_payload
from rfp_parser.exports_xls import build_xls_from_doc
# Si tu as un export CSV (ex: rfp_parser.exports_csv), décommente et branche ici :
# from rfp_parser.exports_csv import build_csv_from_doc
# --------- Config ---------
DEEPINFRA_API_KEY = os.environ.get("DEEPINFRA_API_KEY", "")
MODEL_NAME = os.environ.get("RFP_MODEL", "meta-llama/Meta-Llama-3.1-70B-Instruct")
DEEPINFRA_URL = os.environ.get("DEEPINFRA_URL", "https://api.deepinfra.com/v1/openai/chat/completions")
RFP_DEBUG = str(os.environ.get("RFP_DEBUG", "0")).lower() in {"1", "true", "yes"}
BASE_TMP = Path("/tmp/rfp_jobs"); BASE_TMP.mkdir(parents=True, exist_ok=True)
logger = logging.getLogger("RFP_API")
if not logger.handlers:
h = logging.StreamHandler()
h.setFormatter(logging.Formatter("[API] %(levelname)s: %(message)s"))
logger.addHandler(h)
logger.setLevel(logging.DEBUG if RFP_DEBUG else logging.INFO)
# --------- Jobs en mémoire ---------
JOBS: Dict[str, Dict[str, Any]] = {}
JOBS_LOCK = threading.Lock()
def new_job(text: str) -> str:
job_id = uuid.uuid4().hex[:12]
with JOBS_LOCK:
JOBS[job_id] = {
"status": "queued",
"error": None,
"raw_json_path": None,
"raw_json_url": None,
"own_csv_path": None,
"own_csv_url": None,
"xlsx_path": None,
"xlsx_url": None,
"started_at": time.time(),
"done_at": None,
"meta": {"model": MODEL_NAME, "length": len(text or "")},
"json_preview": None, # court extrait pour le “live JSON”
}
return job_id
def set_job_status(job_id: str, **updates):
with JOBS_LOCK:
if job_id in JOBS:
JOBS[job_id].update(**updates)
# --------- Cœur pipeline ---------
def parse_with_deepinfra(text: str) -> Dict[str, Any]:
if not DEEPINFRA_API_KEY:
raise RuntimeError("DEEPINFRA_API_KEY manquant (Settings → Secrets du Space).")
payload = build_chat_payload(text, model=MODEL_NAME)
headers = {"Authorization": f"Bearer {DEEPINFRA_API_KEY}", "Content-Type": "application/json"}
logger.info("DeepInfra call model=%s max_tokens=%s", payload.get("model"), payload.get("max_tokens"))
r = requests.post(DEEPINFRA_URL, headers=headers, json=payload, timeout=120)
if r.status_code // 100 != 2:
raise RuntimeError(f"DeepInfra HTTP {r.status_code}: {r.text}")
data = r.json()
try:
content = data["choices"][0]["message"]["content"]
except Exception:
raise RuntimeError(f"Réponse inattendue DeepInfra: {json.dumps(data)[:400]}")
try:
doc = json.loads(content)
except Exception as e:
logger.warning("json.loads(content) a échoué; strip fallback. Err=%s", e)
doc = json.loads(content.strip().strip('`').strip())
if not isinstance(doc, dict):
raise RuntimeError("Le contenu renvoyé n'est pas un objet JSON.")
return doc
def persist_doc(job_dir: Path, doc: Dict[str, Any]) -> Tuple[str, str]:
job_dir.mkdir(parents=True, exist_ok=True)
raw_path = job_dir / "raw.json"
with open(raw_path, "w", encoding="utf-8") as f:
json.dump(doc, f, indent=2, ensure_ascii=False)
raw_url = f"/results/{job_dir.name}/raw.json"
return str(raw_path), raw_url
def build_csv_if_available(doc: Dict[str, Any], job_dir: Path) -> Tuple[Optional[str], Optional[str]]:
"""
Si tu as un export CSV dans ton repo, branche-le ici.
Sinon, on renvoie (None, None) sans erreur pour rester permissif.
"""
try:
# Exemple si tu as build_csv_from_doc(doc, out_path)
# out_path = job_dir / "own.csv"
# build_csv_from_doc(doc, str(out_path))
# return str(out_path), f"/results/{job_dir.name}/own.csv"
return None, None
except Exception as e:
logger.warning("CSV non généré: %s", e)
return None, None
def build_xlsx(doc: Dict[str, Any], job_dir: Path) -> str:
job_dir.mkdir(parents=True, exist_ok=True)
out_path = str(job_dir / "feuille_de_charge.xlsx")
baseline = (doc.get("assumptions") or {}).get("baseline_uop_kg") or 100.0
try:
baseline = float(baseline)
except Exception:
baseline = 100.0
build_xls_from_doc(doc, out_path, baseline_kg=baseline)
return out_path
def run_job(job_id: str, text: str) -> None:
set_job_status(job_id, status="running")
job_dir = BASE_TMP / job_id
try:
# 1) Parse LLM
doc = parse_with_deepinfra(text)
# 2) Persist raw.json immédiatement (pour JSON-first côté client)
raw_path, raw_url = persist_doc(job_dir, doc)
preview = json.dumps(doc, ensure_ascii=False)[:1500] # court extrait
set_job_status(job_id, raw_json_path=raw_path, raw_json_url=raw_url, json_preview=preview)
# 3) (Optionnel) Génère own.csv si dispo
csv_path, csv_url = build_csv_if_available(doc, job_dir)
if csv_path and csv_url:
set_job_status(job_id, own_csv_path=csv_path, own_csv_url=csv_url)
# 4) XLSX (peut être le plus long)
xlsx_path = build_xlsx(doc, job_dir)
xlsx_url = f"/results/{job_id}/feuille_de_charge.xlsx"
# 5) Terminé
set_job_status(
job_id,
status="done",
xlsx_path=xlsx_path,
xlsx_url=xlsx_url,
done_at=time.time(),
meta={**JOBS[job_id]["meta"], "assumptions": doc.get("assumptions")},
)
logger.info("Job %s terminé -> %s", job_id, xlsx_path)
except Exception as e:
logger.error("Job %s échoué: %s\n%s", job_id, e, traceback.format_exc())
set_job_status(job_id, status="error", error=str(e), done_at=time.time())
# --------- FastAPI ---------
app = FastAPI(title="RFP_MASTER API", version="1.1.0")
app.add_middleware(
CORSMiddleware,
allow_origins=["*"], # restreins ici si tu veux limiter au Space Gradio
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
@app.get("/health")
def health():
return {"ok": True, "ts": time.time(), "model": MODEL_NAME}
@app.post("/submit")
def submit(payload: Dict[str, Any]):
text = (payload or {}).get("text", "")
if not isinstance(text, str) or not text.strip():
raise HTTPException(400, "Champ 'text' manquant ou vide.")
job_id = new_job(text)
logger.info("Submit reçu job_id=%s len(text)=%d", job_id, len(text))
t = threading.Thread(target=run_job, args=(job_id, text), daemon=True)
t.start()
return JSONResponse({"job_id": job_id, "status": "queued"})
@app.get("/status")
def status(job_id: str = Query(..., description="Identifiant renvoyé par /submit")):
with JOBS_LOCK:
info = JOBS.get(job_id)
if not info:
raise HTTPException(404, f"job_id inconnu: {job_id}")
# On expose les URLs disponibles + un petit aperçu JSON (pour “live JSON”)
return JSONResponse({
"job_id": job_id,
"status": info.get("status"),
"error": info.get("error"),
"meta": info.get("meta"),
"raw_json_url": info.get("raw_json_url"),
"own_csv_url": info.get("own_csv_url"),
"xlsx_url": info.get("xlsx_url"),
"json_preview": info.get("json_preview"),
})
# ---- Résultats ----
@app.get("/results/{job_id}/raw.json")
def download_raw(job_id: str):
with JOBS_LOCK:
info = JOBS.get(job_id)
if not info:
raise HTTPException(404, f"job_id inconnu: {job_id}")
raw_path = info.get("raw_json_path")
if not raw_path or not Path(raw_path).exists():
raise HTTPException(404, "raw.json indisponible.")
return FileResponse(raw_path, media_type="application/json", filename="raw.json")
@app.get("/results/{job_id}/own.csv")
def download_csv(job_id: str):
with JOBS_LOCK:
info = JOBS.get(job_id)
if not info:
raise HTTPException(404, f"job_id inconnu: {job_id}")
csv_path = info.get("own_csv_path")
if not csv_path:
raise HTTPException(404, "own.csv non généré sur ce job.")
if not Path(csv_path).exists():
raise HTTPException(404, "own.csv indisponible.")
return FileResponse(csv_path, media_type="text/csv", filename="own.csv")
@app.get("/results/{job_id}/feuille_de_charge.xlsx")
def download_xlsx(job_id: str):
with JOBS_LOCK:
info = JOBS.get(job_id)
if not info:
raise HTTPException(404, f"job_id inconnu: {job_id}")
if info.get("status") != "done":
raise HTTPException(409, f"job {job_id} non prêt (status={info.get('status')})")
xlsx_path = info.get("xlsx_path")
if not xlsx_path or not Path(xlsx_path).exists():
raise HTTPException(404, "Fichier indisponible.")
return FileResponse(
xlsx_path,
media_type="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
filename="feuille_de_charge.xlsx",
)
|