199 lines
7.8 KiB
Python
199 lines
7.8 KiB
Python
"""app.py — Точка входа и сборка слоёв."""
|
||
|
||
from dotenv import load_dotenv
|
||
from flask import Flask, render_template, request, redirect, url_for, Response, jsonify
|
||
import json
|
||
|
||
import schema
|
||
import db
|
||
import parser as parser_mod
|
||
import textify
|
||
import extractor
|
||
import differ
|
||
import mimeutil
|
||
from test_routes import test_bp
|
||
from upload import upload_bp
|
||
from api import api_bp
|
||
|
||
load_dotenv()
|
||
|
||
|
||
class ContractsApp:
|
||
def __init__(self):
|
||
self.app = Flask(__name__)
|
||
schema.ensure_schema()
|
||
self._add_routes()
|
||
|
||
def _add_routes(self):
|
||
self.app.add_url_rule("/", "index", self._index, methods=["GET", "POST"])
|
||
self.app.add_url_rule("/parse/<cid>", "parse", self._parse, methods=["GET"])
|
||
self.app.add_url_rule("/health", "health", self._health)
|
||
self.app.register_blueprint(test_bp)
|
||
self.app.register_blueprint(upload_bp)
|
||
self.app.register_blueprint(api_bp)
|
||
|
||
def _health(self):
|
||
return "OK", 200, {"Content-Type": "text/plain"}
|
||
|
||
# ── Шаг 1: загрузка файлов ──────────────────────────────────
|
||
|
||
def _index(self):
|
||
if request.method == "POST":
|
||
return self._upload_files()
|
||
cid = request.args.get("id")
|
||
if cid:
|
||
return self._show_contract(cid)
|
||
return render_template("upload.html")
|
||
|
||
def _upload_files(self):
|
||
"""Только сохранить файлы, без LLM."""
|
||
files = request.files.getlist("files")
|
||
if not files or not any(f.filename for f in files):
|
||
return render_template("upload.html", error="Нет файлов")
|
||
|
||
conn, err = db.connect()
|
||
if err:
|
||
return render_template("upload.html", error=f"БД: {err}")
|
||
cur = conn.cursor()
|
||
cur.execute(
|
||
"INSERT INTO contracts (number, client) VALUES (%s, %s) RETURNING id",
|
||
("б/н " + __import__("datetime").datetime.now().strftime("%Y%m%d-%H%M"), ""),
|
||
)
|
||
contract_id = cur.fetchone()[0]
|
||
conn.commit()
|
||
cur.close()
|
||
conn.close()
|
||
|
||
for f in files:
|
||
if not f.filename:
|
||
continue
|
||
filename = f.filename
|
||
mime = mimeutil.guess_mime(filename) or f.content_type or "application/octet-stream"
|
||
file_bytes = f.read()
|
||
|
||
db.execute(
|
||
"INSERT INTO documents (filename, mime_type, original_bytes, status) VALUES (%s,%s,%s,'uploaded')",
|
||
(filename, mime, file_bytes),
|
||
)
|
||
doc, _ = db.query_one(
|
||
"SELECT id FROM documents WHERE filename=%s AND status='uploaded' ORDER BY created_at DESC LIMIT 1",
|
||
(filename,),
|
||
)
|
||
doc_id = doc["id"] if doc else None
|
||
if doc_id:
|
||
db.execute(
|
||
"INSERT INTO supplements (contract_id, document_id, type) VALUES (%s,%s,'initial')",
|
||
(str(contract_id), str(doc_id)),
|
||
)
|
||
|
||
return jsonify({"contract_id": str(contract_id)})
|
||
|
||
def _show_contract(self, cid):
|
||
"""Показать договор с кнопкой «Обработать»."""
|
||
contract, _ = db.query_one("SELECT id,number,created_at FROM contracts WHERE id=%s", (cid,))
|
||
if not contract:
|
||
return render_template("upload.html", error="Договор не найден")
|
||
|
||
supps, _ = db.query(
|
||
"SELECT s.id, s.created_at, d.filename, d.status, d.parsed_text IS NOT NULL as has_text "
|
||
"FROM supplements s JOIN documents d ON s.document_id=d.id "
|
||
"WHERE s.contract_id=%s ORDER BY s.created_at",
|
||
(cid,),
|
||
)
|
||
supp_list = [dict(zip(supps["columns"], r)) for r in supps["rows"]] if supps else []
|
||
|
||
# Есть ли уже результаты?
|
||
rows_res, _ = db.query(
|
||
"SELECT sr.row_num,sr.name,sr.price,sr.qty,sr.sum,sr.date_start "
|
||
"FROM spec_rows sr JOIN supplements s ON sr.supplement_id=s.id "
|
||
"WHERE s.contract_id=%s ORDER BY sr.row_num",
|
||
(cid,),
|
||
)
|
||
all_rows = [dict(zip(rows_res["columns"], r)) for r in rows_res["rows"]] if rows_res else []
|
||
|
||
# Есть необработанные допники?
|
||
unprocessed = [s for s in supp_list if s["status"] in ("uploaded", "parsed")]
|
||
|
||
return render_template("upload.html",
|
||
contract=contract, supplements=supp_list,
|
||
all_rows=all_rows, unprocessed=unprocessed)
|
||
|
||
# ── Шаг 2: Парсинг (SSE) ────────────────────────────────────
|
||
|
||
def _parse(self, cid):
|
||
"""SSE-поток: парсинг файлов договора (без LLM)."""
|
||
import time as time_mod
|
||
|
||
def generate():
|
||
start_time = time_mod.time()
|
||
total_bytes = 0
|
||
files_processed = 0
|
||
|
||
supps, _ = db.query(
|
||
"SELECT s.id, d.id as doc_id, d.filename, d.mime_type, "
|
||
"LENGTH(d.original_bytes) as file_size "
|
||
"FROM supplements s JOIN documents d ON s.document_id=d.id "
|
||
"WHERE s.contract_id=%s",
|
||
(cid,),
|
||
)
|
||
if not supps:
|
||
yield f"data: {json.dumps({'type': 'error', 'message': 'Нет файлов'})}\n\n"
|
||
return
|
||
|
||
supp_rows = [dict(zip(supps["columns"], r)) for r in supps["rows"]]
|
||
|
||
for s in supp_rows:
|
||
# Пропустить уже распарсенные
|
||
existing, _ = db.query(
|
||
"SELECT 1 FROM documents WHERE id=%s AND status='parsed'", (s["doc_id"],)
|
||
)
|
||
if existing and existing["rows"]:
|
||
continue
|
||
|
||
file_start = time_mod.time()
|
||
file_bytes = s.get("file_size", 0) or 0
|
||
total_bytes += file_bytes
|
||
|
||
yield f"data: {json.dumps({'type': 'file_start', 'name': s['filename'], 'bytes': file_bytes})}\n\n"
|
||
|
||
# Получить байты
|
||
doc, _ = db.query_one("SELECT original_bytes FROM documents WHERE id=%s", (s["doc_id"],))
|
||
if not doc or not doc.get("original_bytes"):
|
||
yield f"data: {json.dumps({'type': 'file_error', 'name': s['filename'], 'error': 'Нет данных'})}\n\n"
|
||
continue
|
||
|
||
raw = doc["original_bytes"]
|
||
file_data = bytes(raw) if isinstance(raw, memoryview) else raw
|
||
|
||
# Парсинг
|
||
pr = parser_mod.parse(file_data, s["mime_type"])
|
||
elements = []
|
||
if s["mime_type"] == "application/zip" and "files" in pr:
|
||
for zf in pr["files"]:
|
||
elements.extend(zf.get("elements", []))
|
||
else:
|
||
elements = pr.get("elements", [])
|
||
|
||
text = textify.to_text(elements) if elements else ""
|
||
db.execute(
|
||
"UPDATE documents SET parsed_text=%s, status='parsed' WHERE id=%s",
|
||
(text, str(s["doc_id"])),
|
||
)
|
||
|
||
elapsed = round(time_mod.time() - file_start, 1)
|
||
files_processed += 1
|
||
|
||
yield f"data: {json.dumps({'type': 'file_done', 'name': s['filename'], 'time_s': elapsed, 'elements': len(elements)})}\n\n"
|
||
|
||
total_time = round(time_mod.time() - start_time, 1)
|
||
yield f"data: {json.dumps({'type': 'summary', 'total_time_s': total_time, 'total_bytes': total_bytes, 'files_processed': files_processed})}\n\n"
|
||
|
||
return Response(generate(), mimetype="text/event-stream")
|
||
|
||
def run(self):
|
||
self.app.run(host="0.0.0.0", port=5000)
|
||
|
||
|
||
if __name__ == "__main__":
|
||
ContractsApp().run()
|