contracts-app/site/app.py

"""
app.py — Точка входа и сборка слоёв.

Бизнес-логики НОЛЬ. Только:
  1. Инициализация схемы БД (schema.ensure_schema)
  2. Регистрация Blueprint-ов (test, api — будут добавляться)

Правило: если в app.py появляется if/for или бизнес-слово — это лишнее.
"""

from dotenv import load_dotenv
from flask import Flask, render_template, request

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:
    """
    Главный класс приложения.
    Собирает все слои вместе, не содержит логики.
    """

    def __init__(self):
        self.app = Flask(__name__)

        # 1. Создать таблицы при старте (если их ещё нет)
        schema.ensure_schema()

        # 2. Собрать маршруты
        self._add_routes()

    def _add_routes(self):
        """Регистрация всех Blueprint-ов и системных маршрутов."""

        # -- Системные маршруты --
        self.app.add_url_rule("/", "index", self._upload_page)
        self.app.add_url_rule("/health", "health", self._health)

        # -- Blueprint-ы (каждый слой — отдельный Blueprint) --
        self.app.register_blueprint(test_bp)
        self.app.register_blueprint(upload_bp)
        self.app.register_blueprint(api_bp)

    def _health(self):
        """Health check для платформы."""
        return "OK", 200, {"Content-Type": "text/plain"}

    def _upload_page(self):
        """Страница загрузки: GET — форма, POST — обработка."""
        if request.method == "POST":
            return self._process_upload()
        return render_template("upload.html")

    def _process_upload(self):
        """Обработка загруженных файлов."""
        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")

        # Создать договор
        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()

        results = []
        total_rows = 0

        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()

            # Сохранить документ
            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 not doc_id:
                results.append({"name": filename, "ok": False, "error": "doc save failed"})
                continue

            # Парсинг
            pr = parser_mod.parse(file_bytes, mime)
            if pr.get("errors"):
                results.append({"name": filename, "ok": False, "error": "; ".join(pr["errors"])})
                continue

            text = textify.to_text(pr.get("elements", []))
            db.execute("UPDATE documents SET parsed_text=%s, status='parsed' WHERE id=%s", (text, str(doc_id)))

            # Допник
            db.execute(
                "INSERT INTO supplements (contract_id, document_id, type) VALUES (%s, %s, 'initial')",
                (str(contract_id), str(doc_id)),
            )
            supp, _ = db.query_one(
                "SELECT id FROM supplements WHERE document_id=%s ORDER BY created_at DESC LIMIT 1",
                (str(doc_id),),
            )
            supp_id = supp["id"] if supp else None

            if not supp_id:
                results.append({"name": filename, "ok": False, "error": "supplement save failed"})
                continue

            # LLM-извлечение
            ext = extractor.extract(text)
            if "error" in ext:
                results.append({"name": filename, "ok": False, "error": ext["error"]})
                continue

            rows = ext.get("rows", [])
            total_rows += len(rows)

            # Сохранить строки
            for row in rows:
                db.execute(
                    "INSERT INTO spec_rows (supplement_id, row_num, name, price, qty, sum, date_start) "
                    "VALUES (%s,%s,%s,%s,%s,%s,%s)",
                    (str(supp_id), row.get("row_num"), row.get("name"),
                     row.get("price"), row.get("qty"), row.get("sum"), row.get("date_start")),
                )

            # Diff с предыдущим
            prev, _ = 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 AND s.id!=%s ORDER BY sr.row_num",
                (str(contract_id), str(supp_id)),
            )
            prev_rows = [dict(zip(prev["columns"], r)) for r in prev["rows"]] if prev else []
            diff_r = differ.diff(prev_rows, rows)

            for ch in diff_r.get("changes", []):
                db.execute(
                    "INSERT INTO spec_history (contract_id, supplement_id, row_num, change_type, old_values, new_values) "
                    "VALUES (%s,%s,%s,%s,%s,%s)",
                    (str(contract_id), str(supp_id), ch["row_num"], ch["change_type"],
                     __import__("json").dumps(ch.get("old_values"), ensure_ascii=False, default=str) if ch.get("old_values") else None,
                     __import__("json").dumps(ch.get("new_values"), ensure_ascii=False, default=str) if ch.get("new_values") else None),
                )

            db.execute("UPDATE documents SET status='extracted' WHERE id=%s", (str(doc_id),))
            results.append({"name": filename, "ok": True, "rows": len(rows), "summary": diff_r.get("summary", {})})

        # Получить финальные строки для отображения
        all_rows = []
        rows_result, _ = 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",
            (str(contract_id),),
        )
        if rows_result:
            all_rows = [dict(zip(rows_result["columns"], r)) for r in rows_result["rows"]]

        return render_template("upload.html", results=results, all_rows=all_rows, total_rows=total_rows)

    def run(self):
        """Запуск Flask (только для разработки, в production — gunicorn)."""
        self.app.run(host="0.0.0.0", port=5000)


# ── Точка входа ─────────────────────────────────────────────────

if __name__ == "__main__":
    app_instance = ContractsApp()
    app_instance.run()
    app_instance = ContractsApp()
    app_instance.run()