feat: 百度通用文字识别提取供应商/日期（config api_key 持久化）

app/services/order_service.py

@@ -5,6 +5,11 @@
 """
 
 import os
+import json
+import re
+import shutil
+from datetime import datetime
+from pathlib import Path
 from typing import Dict, List, Optional, Tuple, Union, Any, Callable
 
 from ..config.settings import ConfigManager
@@ -12,6 +17,8 @@ from ..core.utils.log_utils import get_logger
 from ..core.excel.processor import ExcelProcessor
 from ..core.excel.merger import PurchaseOrderMerger
 from ..core.db.product_db import ProductDatabase
+from ..core.db.order_metadata_db import OrderMetadataDB
+from ..core.ocr.metadata_extractor import OrderMetadataExtractor, sanitize_for_filename
 
 logger = get_logger(__name__)
 
@@ -40,7 +47,11 @@ class OrderService:
         # 创建Excel处理器和采购单合并器
         self.excel_processor = ExcelProcessor(self.config, product_db=self.product_db)
         self.order_merger = PurchaseOrderMerger(self.config)
-        
+
+        # 元信息识别器 + 单据元信息库
+        self.extractor = OrderMetadataExtractor()
+        self.metadata_db = OrderMetadataDB(db_path)
+
         logger.info("OrderService初始化完成")
     
     def get_latest_excel(self) -> Optional[str]:
@@ -89,7 +100,21 @@ class OrderService:
         except Exception as e:
             logger.error(f"检查特殊预处理时出错: {e}")
 
-        return self.excel_processor.process_specific_file(file_path, progress_cb=progress_cb)
+        result_path = self.excel_processor.process_specific_file(file_path, progress_cb=progress_cb)
+        if not result_path:
+            return None
+
+        # 应用单据元信息识别 + 重命名 result 与原图
+        try:
+            meta = self._extract_and_save_metadata(file_path)
+            if meta:
+                result_path = self._apply_metadata_to_filenames(
+                    result_path, file_path, meta
+                )
+        except Exception as e:
+            logger.error(f"应用单据元信息失败（不影响 result 文件）: {e}")
+
+        return result_path
 
     def _check_special_preprocess(self, file_path: str) -> Optional[str]:
         """检查并执行特殊的预处理（支持杨碧月、烟草公司、蓉城易购）"""
@@ -243,3 +268,139 @@ class OrderService:
         except Exception as e:
             logger.error(f"单价校验过程中发生错误: {e}")
             return []
+
+    # ══════════════════════════════════════════════════════════════
+    # 单据元信息识别 + 文件重命名
+    # ══════════════════════════════════════════════════════════════
+
+    def _extract_and_save_metadata(self, ocr_excel_path: str) -> Optional[Any]:
+        """从 OCR 输出的 xlsx 同目录的 .meta.json 提取元信息，写入 SQLite。
+
+        Returns:
+            OrderMetadata 或 None（失败时）
+        """
+        try:
+            base = Path(ocr_excel_path)
+            file_hash = base.stem
+            meta_path = base.with_suffix('.meta.json')
+
+            ocr_text = ''
+            ocr_rows: List[List[str]] = []
+            source_image = ''
+
+            if meta_path.exists():
+                try:
+                    payload = json.loads(meta_path.read_text(encoding='utf-8'))
+                    # 优先用通用识别文本（含手写抬头/日期）
+                    ocr_text = payload.get('general_text') or payload.get('ocr_text', '') or ''
+                    ocr_rows = payload.get('ocr_rows', []) or []
+                    source_image = payload.get('image_path', '') or ''
+                except Exception as e:
+                    logger.warning(f"读 meta.json 失败: {e}")
+
+            # 兜底：从 xlsx 拼文本（与 OCRService._write_meta_json 的兜底一致）
+            if not ocr_text:
+                try:
+                    import xlrd
+                    rb = xlrd.open_workbook(str(ocr_excel_path))
+                    ws = rb.sheet_by_index(0)
+                    lines = []
+                    for r in range(ws.nrows):
+                        row_vals = [str(ws.cell_value(r, c)) for c in range(ws.ncols)]
+                        ocr_rows.append(row_vals)
+                        lines.append(' '.join(row_vals))
+                    ocr_text = '\n'.join(lines)
+                except Exception as e:
+                    logger.debug(f"从 xlsx 拼 OCR 文本失败: {e}")
+
+            meta = self.extractor.extract(ocr_text, ocr_rows)
+            self.metadata_db.save(
+                file_hash=file_hash,
+                supplier=meta.supplier,
+                bill_date=meta.bill_date,
+                total_amount=meta.total_amount,
+                raw_supplier_text=meta.raw_supplier_text,
+                source_image=source_image,
+            )
+            logger.info(
+                f"元信息识别: hash={file_hash} supplier={meta.supplier!r} "
+                f"bill_date={meta.bill_date!r} total_amount={meta.total_amount:.2f}"
+            )
+            return meta
+        except Exception as e:
+            logger.error(f"_extract_and_save_metadata 失败: {e}")
+            return None
+
+    def _apply_metadata_to_filenames(self, result_path: str,
+                                      ocr_excel_path: str,
+                                      meta) -> str:
+        """应用新文件名规则：
+        - result: 采购单_{YYYYMMDD}_{供应商}_{hash}.xls
+        - 原图: {原stem}_{YYYYMMDD}_{供应商}_{hash}.{ext}
+
+        任一步骤失败不影响 result 文件本身。
+        Returns:
+            新 result 路径（无论重命名是否成功都返回；失败时返回原路径）
+        """
+        try:
+            file_hash = Path(ocr_excel_path).stem
+            supplier_clean = sanitize_for_filename(meta.supplier) or '未知供应商'
+            date_part = meta.bill_date or '未知日期'
+
+            # ── 1. result 重命名 ──
+            new_result_name = f"采购单_{date_part}_{supplier_clean}_{file_hash}.xls"
+            result_dir = Path(result_path).parent
+            new_result_path = result_dir / new_result_name
+            try:
+                # 冲突时加 _N
+                if new_result_path.exists() and str(new_result_path) != str(result_path):
+                    new_result_path = self._dedup_path(new_result_path)
+                if str(new_result_path) != str(result_path):
+                    os.rename(result_path, str(new_result_path))
+                    logger.info(f"result 重命名: {result_path} -> {new_result_path}")
+                    result_path = str(new_result_path)
+            except Exception as e:
+                logger.warning(f"result 重命名失败: {e}")
+
+            # ── 2. 原图重命名 ──
+            try:
+                meta_row = self.metadata_db.get(file_hash)
+                src_image = (meta_row or {}).get('source_image', '')
+                if src_image and os.path.exists(src_image):
+                    src_p = Path(src_image)
+                    stem = src_p.stem
+                    ext = src_p.suffix
+                    new_image_name = f"{stem}_{date_part}_{supplier_clean}_{file_hash}{ext}"
+                    new_image_path = src_p.parent / new_image_name
+                    # 不覆盖已重命名的图片
+                    if str(new_image_path) != str(src_p) and not new_image_path.exists():
+                        os.rename(src_p, new_image_path)
+                        logger.info(f"原图重命名: {src_p.name} -> {new_image_path.name}")
+                        # 更新 source_image 路径
+                        self.metadata_db.save(
+                            file_hash=file_hash,
+                            supplier=meta.supplier,
+                            bill_date=meta.bill_date,
+                            total_amount=meta.total_amount,
+                            raw_supplier_text=meta.raw_supplier_text,
+                            source_image=str(new_image_path),
+                        )
+            except Exception as e:
+                logger.warning(f"原图重命名失败: {e}")
+
+            return result_path
+        except Exception as e:
+            logger.error(f"_apply_metadata_to_filenames 失败: {e}")
+            return result_path
+
+    @staticmethod
+    def _dedup_path(p: Path) -> Path:
+        """路径冲突时加 _N 后缀。"""
+        stem, suffix = p.stem, p.suffix
+        parent = p.parent
+        n = 1
+        while True:
+            cand = parent / f"{stem}_{n}{suffix}"
+            if not cand.exists():
+                return cand
+            n += 1