feat(新的文档结构化规则)

core/construction_review/component/minimal_pipeline/pdf_extractor2.py

@@ -0,0 +1,960 @@
+"""
+PDF 结构提取器 - 同步并发 OCR 版本
+
+基于 splitter_pdf 逻辑，直接提取章节结构并记录页码。
+支持 OCR 增强：检测表格区域并使用 ThreadPoolExecutor 5并发 OCR，其他文本保持 PyMuPDF 提取。
+输出格式兼容后续分类与组装流程。
+"""
+
+import base64
+import io
+import re
+from concurrent.futures import ThreadPoolExecutor, as_completed
+from dataclasses import dataclass
+from typing import Dict, Any, List, Optional, Tuple
+
+import fitz
+import numpy as np
+import requests
+
+from foundation.observability.logger.loggering import review_logger as logger
+
+# 尝试导入 RapidLayout
+try:
+    from rapid_layout import RapidLayout
+    RAPID_LAYOUT_AVAILABLE = True
+except ImportError:
+    RAPID_LAYOUT_AVAILABLE = False
+    RapidLayout = None
+
+
+@dataclass
+class TableRegion:
+    """表格区域信息"""
+    page_num: int
+    page: fitz.Page
+    bbox: Tuple[float, float, float, float]
+    score: float
+
+
+@dataclass
+class OcrResult:
+    """OCR 结果"""
+    page_num: int
+    bbox: Tuple[float, float, float, float]
+    score: float
+    text: str
+    success: bool
+
+
+class PdfStructureExtractor:
+    """PDF 章节结构提取器（支持 OCR 异步并发）"""
+
+    RULE_LIB = {
+        "Rule_1_纯数字派": {
+            "l1": re.compile(r"^\d{1,2}(?:[\.．。、])?\s*(?!\d)[\u4e00-\u9fa5A-Za-z].*"),
+            "l2": re.compile(r"^(\d+)\.(\d+)(?!\.\d)\.?\s*([\u4e00-\u9fa5A-Za-z].*)"),
+        },
+        "Rule_2_混合章派": {
+            "l1": re.compile(r"^第\s*(\d+)\s*[章部分篇][\s、]*(.*)"),
+            "l2": re.compile(r"^(\d+)\.(\d+)(?!\.\d)\.?\s*([\u4e00-\u9fa5A-Za-z].*)"),
+        },
+        "Rule_3_中英混血派": {
+            "l1": re.compile(r"^第\s*[一二三四五六七八九十百零两]+\s*[章部分篇][\s、]*(.*)"),
+            "l2": re.compile(r"^(\d+)\.(\d+)(?!\.\d)\.?\s*([\u4e00-\u9fa5A-Za-z].*)"),
+        },
+        "Rule_4_传统公文派": {
+            "l1": re.compile(r"^第\s*[一二三四五六七八九十百零两]+\s*[章部分篇][\s、]*(.*)"),
+            "l2": re.compile(r"^([一二三四五六七八九十百零两]+)[、\s]+([\u4e00-\u9fa5A-Za-z].*)"),
+        },
+        "Rule_5_单边括号派": {
+            "l1": re.compile(r"^第\s*[一二三四五六七八九十百零两]+\s*[章部分篇][\s、]*(.*)"),
+            "l2": re.compile(r"^([一二三四五六七八九十百零两]+)[）\)\]][\s]*([\u4e00-\u9fa5A-Za-z].*)"),
+        },
+        "Rule_6_小节派": {
+            "l1": re.compile(r"^第\s*[一二三四五六七八九十百零两]+\s*[章部分篇][\s、]*(.*)"),
+            "l2": re.compile(r"^第\s*([一二三四五六七八九十百零两]+)\s*节[\s、]*([\u4e00-\u9fa5A-Za-z].*)"),
+        },
+        "Rule_7_粗体括号派": {
+            "l1": re.compile(r"^第\s*[一二三四五六七八九十百零两]+\s*[章部分篇][\s、]*(.*)"),
+            "l2": re.compile(r"^[【\[]\s*(\d+)\s*[\]】][\s]*([\u4e00-\u9fa5A-Za-z].*)"),
+        },
+    }
+    TOC_PATTERN = re.compile(r"\.{3,}|…{2,}")
+
+    # OCR 配置
+    MAX_SHORT_EDGE = 1024
+    JPEG_QUALITY = 90
+    OCR_DPI = 200
+    OCR_CONFIDENCE_THRESHOLD = 0.5
+    OCR_CONCURRENT_WORKERS = 5
+
+    def __init__(
+        self,
+        clip_top: float = 60,
+        clip_bottom: float = 60,
+        use_ocr: bool = False,
+        ocr_api_url: str = "http://183.220.37.46:25429/v1/chat/completions",
+        ocr_timeout: int = 600,
+        ocr_api_key: str = "",
+        detect_toc: bool = True,
+        toc_model_path: str = "config/yolo/best.pt",
+    ):
+        self.clip_top = clip_top
+        self.clip_bottom = clip_bottom
+        self.use_ocr = use_ocr and RAPID_LAYOUT_AVAILABLE
+
+        # OCR 配置
+        self.ocr_api_url = ocr_api_url
+        self.ocr_timeout = ocr_timeout
+        self.ocr_api_key = ocr_api_key
+        self._layout_engine: Optional[Any] = None
+
+        # 目录检测配置
+        self.detect_toc = detect_toc
+        self.toc_model_path = toc_model_path
+        self._toc_extractor = None
+
+        if use_ocr and not RAPID_LAYOUT_AVAILABLE:
+            logger.warning("RapidLayout 未安装，OCR 功能不可用")
+
+    def _get_layout_engine(self) -> Optional[Any]:
+        """延迟初始化 RapidLayout"""
+        if self._layout_engine is None and RAPID_LAYOUT_AVAILABLE:
+            self._layout_engine = RapidLayout()
+        return self._layout_engine
+
+    def extract(self, file_content: bytes, progress_callback=None) -> Dict[str, Any]:
+        """
+        从 PDF 字节流提取章节结构。
+
+        Args:
+            file_content: PDF 文件字节流
+            progress_callback: 进度回调函数，接收 (stage, current, message) 参数
+
+        Returns:
+            {
+                "chapters": {
+                    "第一章 xxx": {
+                        "章节标题": {"content": "...", "page_start": 1, "page_end": 1},
+                        "一、xxx": {"content": "...", "page_start": 2, "page_end": 3},
+                    }
+                },
+                "total_pages": N,
+                "catalog": {  # 目录结构（YOLO检测+OCR提取）
+                    "chapters": [...],
+                    "total_chapters": N
+                }
+            }
+        """
+        result = {"chapters": {}, "total_pages": 0, "catalog": None}
+
+        # === 阶段0: 目录页检测与提取（如果启用）===
+        if self.detect_toc:
+            try:
+                catalog = self._extract_catalog(file_content, progress_callback)
+                if catalog:
+                    catalog = self._normalize_catalog(catalog)
+                    result["catalog"] = catalog
+                    logger.info(f"[PDF提取] 目录提取完成: {catalog.get('total_chapters', 0)} 章")
+            except Exception as e:
+                logger.warning(f"[PDF提取] 目录提取失败: {e}")
+
+        # === 阶段1-3: 文档结构提取 ===
+        doc = fitz.open(stream=file_content)
+        try:
+            structure = self._extract_from_doc(doc, progress_callback)
+            if result.get("catalog"):
+                structure["chapters"] = self._reconcile_structure_with_catalog(
+                    structure.get("chapters", {}),
+                    result["catalog"],
+                )
+            result["chapters"] = structure.get("chapters", {})
+            result["total_pages"] = len(doc)
+            return result
+        finally:
+            doc.close()
+
+    def _extract_catalog(self, file_content: bytes, progress_callback=None) -> Optional[Dict[str, Any]]:
+        """
+        提取目录结构（YOLO检测 + OCR识别）
+
+        Returns:
+            {"chapters": [...], "total_chapters": N} 或 None
+        """
+        # 延迟导入避免循环依赖（YOLO依赖必须存在，否则报错）
+        from .toc_detector import TOCCatalogExtractor
+
+        if self._toc_extractor is None:
+            self._toc_extractor = TOCCatalogExtractor(
+                model_path=self.toc_model_path,
+                ocr_api_url=self.ocr_api_url,
+                ocr_api_key=self.ocr_api_key,
+                ocr_timeout=self.ocr_timeout,
+            )
+
+        return self._toc_extractor.detect_and_extract(file_content, progress_callback)
+
+    def _extract_from_doc(self, doc: fitz.Document, progress_callback=None) -> Dict[str, Any]:
+        """提取文档结构（支持 OCR 异步并发）"""
+
+        def _emit_progress(stage: str, current: int, message: str):
+            """发送进度回调"""
+            if progress_callback:
+                try:
+                    progress_callback(stage, current, message)
+                except Exception:
+                    pass
+
+        # === 阶段1: 收集所有需要 OCR 的表格区域 ===
+        table_regions: List[TableRegion] = []
+
+        if self.use_ocr:
+            logger.info("[OCR预处理] 扫描所有页面的表格区域...")
+            total_pages = len(doc)
+            for page_num in range(total_pages):
+                page = doc.load_page(page_num)
+                rect = page.rect
+                clip_box = fitz.Rect(0, self.clip_top, rect.width, rect.height - self.clip_bottom)
+                regions = self._detect_table_regions(page, page_num + 1, clip_box)
+                for bbox, score in regions:
+                    table_regions.append(TableRegion(
+                        page_num=page_num + 1,
+                        page=page,
+                        bbox=bbox,
+                        score=score
+                    ))
+                # 每5页或最后一页推送一次进度
+                if (page_num + 1) % 5 == 0 or page_num == total_pages - 1:
+                    progress = int((page_num + 1) / total_pages * 30)  # OCR预处理占30%进度
+                    _emit_progress("版面分析", progress, f"扫描页面 {page_num + 1}/{total_pages}")
+            logger.info(f"[OCR预处理] 共发现 {len(table_regions)} 个表格区域需要 OCR")
+
+        # === 阶段2: 异步并发执行 OCR (5并发) ===
+        ocr_results: List[OcrResult] = []
+
+        if table_regions:
+            logger.info(f"[OCR执行] 使用 {self.OCR_CONCURRENT_WORKERS} 并发执行 OCR...")
+            _emit_progress("版面分析", 35, f"发现 {len(table_regions)} 个表格，开始OCR识别...")
+            ocr_results = self._process_ocr_concurrent(table_regions, progress_callback=_emit_progress)
+            success_count = sum(1 for r in ocr_results if r.success)
+            logger.info(f"[OCR执行] 完成 {success_count}/{len(table_regions)} 个表格 OCR")
+            _emit_progress("版面分析", 50, f"OCR识别完成 {success_count}/{len(table_regions)}")
+
+        # 按页码分组 OCR 结果
+        ocr_by_page: Dict[int, List[OcrResult]] = {}
+        for result in ocr_results:
+            if result.success:
+                if result.page_num not in ocr_by_page:
+                    ocr_by_page[result.page_num] = []
+                ocr_by_page[result.page_num].append(result)
+
+        # === 阶段3: 提取页面文本（应用 OCR 结果）并切分章节 ===
+        structured_data: Dict[str, Dict[str, Dict[str, Any]]] = {}
+        current_chapter = "未分类前言"
+        current_section = "默认部分"
+        in_body = False
+        candidate_rule_names: Optional[List[str]] = None
+        active_rule_name: Optional[str] = None
+
+        logger.info("[文本提取] 提取页面内容并切分章节...")
+
+        for page_num in range(len(doc)):
+            page = doc.load_page(page_num)
+            rect = page.rect
+            clip_box = fitz.Rect(0, self.clip_top, rect.width, rect.height - self.clip_bottom)
+
+            # 获取页面文本（应用 OCR 结果）
+            if page_num + 1 in ocr_by_page:
+                original_text = page.get_text("text", clip=clip_box)
+                ocr_results_list = [
+                    {
+                        "region_index": i,
+                        "bbox": r.bbox,
+                        "score": r.score,
+                        "ocr_text": r.text,
+                    }
+                    for i, r in enumerate(ocr_by_page[page_num + 1])
+                ]
+                text = self._replace_table_regions(page, original_text, ocr_results_list, clip_box)
+            else:
+                text = page.get_text("text", clip=clip_box)
+
+            lines = text.split("\n")
+
+            for line in lines:
+                line = line.strip()
+                if not line:
+                    continue
+                if self._is_header_footer(line):
+                    continue
+
+                # 跳过目录阶段
+                if not in_body:
+                    matched_rules = self._matching_rule_names(line, "l1")
+                    if matched_rules and not self.TOC_PATTERN.search(line):
+                        in_body = True
+                        candidate_rule_names = matched_rules
+                    else:
+                        continue
+
+                # 跳过残余目录格式
+                if self.TOC_PATTERN.search(line):
+                    continue
+
+                active_scope = [active_rule_name] if active_rule_name else candidate_rule_names
+
+                # 匹配章标题
+                matched_chapter_rules = self._matching_rule_names(line, "l1", active_scope)
+                if matched_chapter_rules:
+                    if active_rule_name is None:
+                        candidate_rule_names = matched_chapter_rules
+                    current_chapter = self._clean_chapter_title(line)
+                    current_section = "章节标题"
+                    if current_chapter not in structured_data:
+                        structured_data[current_chapter] = {}
+                    if current_section not in structured_data[current_chapter]:
+                        structured_data[current_chapter][current_section] = {
+                            "lines": [],
+                            "page_start": page_num + 1,
+                            "page_end": page_num + 1,
+                        }
+                    continue
+
+                # 匹配节标题
+                matched_section_rules = self._matching_rule_names(line, "l2", active_scope)
+                if matched_section_rules:
+                    if active_rule_name is None:
+                        if candidate_rule_names:
+                            for rule_name in candidate_rule_names:
+                                if rule_name in matched_section_rules:
+                                    active_rule_name = rule_name
+                                    break
+                        if active_rule_name is None:
+                            active_rule_name = matched_section_rules[0]
+                    current_section = self._clean_section_title(line)
+                    if current_chapter not in structured_data:
+                        structured_data[current_chapter] = {}
+                    if current_section not in structured_data[current_chapter]:
+                        structured_data[current_chapter][current_section] = {
+                            "lines": [],
+                            "page_start": page_num + 1,
+                            "page_end": page_num + 1,
+                        }
+                    continue
+
+                # 确保结构存在
+                if current_chapter not in structured_data:
+                    structured_data[current_chapter] = {}
+                if current_section not in structured_data[current_chapter]:
+                    structured_data[current_chapter][current_section] = {
+                        "lines": [],
+                        "page_start": page_num + 1,
+                        "page_end": page_num + 1,
+                    }
+
+                # 添加内容
+                structured_data[current_chapter][current_section]["lines"].append(line)
+                structured_data[current_chapter][current_section]["page_end"] = page_num + 1
+
+        # 将行列表拼接为文本
+        result: Dict[str, Any] = {"chapters": {}}
+        for chap, sections in structured_data.items():
+            result["chapters"][chap] = {}
+            for sec, data in sections.items():
+                result["chapters"][chap][sec] = {
+                    "content": "\n".join(data["lines"]),
+                    "page_start": data["page_start"],
+                    "page_end": data["page_end"],
+                }
+
+        logger.info(f"[PdfExtractor] 提取完成，共 {len(result['chapters'])} 个章节")
+        return result
+
+    def _normalize_catalog(self, catalog: Dict[str, Any]) -> Dict[str, Any]:
+        if not catalog:
+            return {}
+
+        normalized = dict(catalog)
+        raw_text = catalog.get("raw_ocr_text", "")
+        parsed_chapters = self._parse_catalog_from_raw_text(raw_text) if isinstance(raw_text, str) else []
+        if parsed_chapters:
+            normalized["chapters"] = parsed_chapters
+            normalized["total_chapters"] = len(parsed_chapters)
+            normalized["formatted_text"] = self._format_catalog_chapters(parsed_chapters)
+        return normalized
+
+    def _parse_catalog_from_raw_text(self, text: str) -> List[Dict[str, Any]]:
+        if not text or not text.strip():
+            return []
+
+        chapters: List[Dict[str, Any]] = []
+        current_chapter: Optional[Dict[str, Any]] = None
+        active_l2_rule: Optional[str] = None
+
+        for raw_line in text.splitlines():
+            title_text, page = self._split_catalog_entry(raw_line)
+            if not title_text:
+                continue
+
+            compact = re.sub(r"\s+", "", title_text)
+            if compact in {"目录", "目錄"}:
+                continue
+
+            chapter_matches = self._matching_rule_names(title_text, "l1")
+            if chapter_matches:
+                current_chapter = {
+                    "index": len(chapters) + 1,
+                    "title": self._clean_chapter_title(title_text),
+                    "page": str(page or 1),
+                    "original": raw_line.strip(),
+                    "subsections": [],
+                }
+                chapters.append(current_chapter)
+                active_l2_rule = None
+                continue
+
+            if current_chapter is None:
+                continue
+
+            section_matches = self._matching_rule_names(title_text, "l2")
+            if not section_matches:
+                continue
+
+            if active_l2_rule is None:
+                active_l2_rule = section_matches[0]
+            if active_l2_rule not in section_matches:
+                continue
+
+            section_title = self._clean_section_title(title_text)
+            section_key = self._normalize_heading_key(section_title)
+            existing_keys = {
+                self._normalize_heading_key(sub.get("title", ""))
+                for sub in current_chapter.get("subsections", [])
+            }
+            if section_key in existing_keys:
+                continue
+
+            current_chapter["subsections"].append({
+                "title": section_title,
+                "page": str(page or current_chapter.get("page", 1)),
+                "level": 2,
+                "original": raw_line.strip(),
+            })
+
+        return chapters
+
+    @staticmethod
+    def _split_catalog_entry(line: str) -> Tuple[str, Optional[int]]:
+        cleaned = line.strip()
+        if not cleaned:
+            return "", None
+
+        cleaned = re.sub(r"\s+", " ", cleaned).strip()
+        page_match = re.search(r"(?:[.\u2026\u00b7\u2022 ]{2,})(\d+)\s*$", cleaned)
+        if page_match:
+            title_text = cleaned[:page_match.start()].strip()
+            title_text = re.sub(r"[.\u2026\u00b7\u2022 ]+$", "", title_text).strip()
+            return title_text, int(page_match.group(1))
+
+        return cleaned, None
+
+    @staticmethod
+    def _format_catalog_chapters(chapters: List[Dict[str, Any]]) -> str:
+        lines: List[str] = []
+        for chapter in chapters:
+            title = chapter.get("title", "").strip()
+            if not title:
+                continue
+            lines.append(title)
+            for sub in chapter.get("subsections", []):
+                sub_title = sub.get("title", "").strip()
+                if sub_title:
+                    lines.append(f"  {sub_title}")
+        return "\n".join(lines)
+
+    def _reconcile_structure_with_catalog(
+        self,
+        chapters: Dict[str, Dict[str, Dict[str, Any]]],
+        catalog: Dict[str, Any],
+    ) -> Dict[str, Dict[str, Dict[str, Any]]]:
+        catalog_chapters = catalog.get("chapters", []) if isinstance(catalog, dict) else []
+        if not chapters or not catalog_chapters:
+            return chapters
+
+        section_title_key = "章节标题"
+        chapter_title_payloads: Dict[str, List[Dict[str, Any]]] = {}
+        flat_sections: List[Tuple[str, Dict[str, Any]]] = []
+
+        for chapter_title, sections in chapters.items():
+            title_key = self._normalize_heading_key(chapter_title)
+            title_payload = sections.get(section_title_key)
+            if title_payload is not None:
+                chapter_title_payloads.setdefault(title_key, []).append({
+                    "content": title_payload.get("content", ""),
+                    "page_start": title_payload.get("page_start", 1),
+                    "page_end": title_payload.get("page_end", title_payload.get("page_start", 1)),
+                })
+
+            for section_title, payload in sections.items():
+                if section_title == section_title_key:
+                    continue
+                flat_sections.append((
+                    self._normalize_heading_key(section_title),
+                    {
+                        "content": payload.get("content", ""),
+                        "page_start": payload.get("page_start", 1),
+                        "page_end": payload.get("page_end", payload.get("page_start", 1)),
+                    },
+                ))
+
+        rebuilt: Dict[str, Dict[str, Dict[str, Any]]] = {}
+        search_start = 0
+        used_indices = set()
+
+        for chapter in catalog_chapters:
+            chapter_title = (chapter.get("title", "") or "").strip()
+            if not chapter_title:
+                continue
+
+            chapter_page = self._safe_page_number(chapter.get("page"))
+            chapter_key = self._normalize_heading_key(chapter_title)
+            title_candidates = chapter_title_payloads.get(chapter_key, [])
+            title_payload = title_candidates.pop(0) if title_candidates else self._empty_section_payload(chapter_page)
+
+            rebuilt[chapter_title] = {
+                section_title_key: title_payload,
+            }
+
+            for subsection in chapter.get("subsections", []):
+                section_title = (subsection.get("title", "") or "").strip()
+                if not section_title:
+                    continue
+
+                target_key = self._normalize_heading_key(section_title)
+                match_index = None
+                for idx in range(search_start, len(flat_sections)):
+                    if idx in used_indices:
+                        continue
+                    if flat_sections[idx][0] == target_key:
+                        match_index = idx
+                        break
+                if match_index is None:
+                    for idx, (section_key, _) in enumerate(flat_sections):
+                        if idx in used_indices:
+                            continue
+                        if section_key == target_key:
+                            match_index = idx
+                            break
+
+                if match_index is not None:
+                    used_indices.add(match_index)
+                    search_start = max(search_start, match_index + 1)
+                    rebuilt[chapter_title][section_title] = flat_sections[match_index][1]
+                else:
+                    rebuilt[chapter_title][section_title] = self._empty_section_payload(
+                        self._safe_page_number(subsection.get("page"), chapter_page)
+                    )
+
+        return rebuilt or chapters
+
+    @staticmethod
+    def _normalize_heading_key(text: str) -> str:
+        normalized = (text or "").strip()
+        normalized = normalized.replace("【", "[").replace("】", "]")
+        normalized = normalized.replace("（", "(").replace("）", ")")
+        normalized = normalized.replace("．", ".").replace("。", ".")
+        normalized = re.sub(r"\s+", "", normalized)
+        return normalized
+
+    @staticmethod
+    def _safe_page_number(value: Any, default: int = 1) -> int:
+        try:
+            return max(1, int(str(value).strip()))
+        except Exception:
+            return default
+
+    @staticmethod
+    def _empty_section_payload(page_num: int) -> Dict[str, Any]:
+        return {
+            "content": "",
+            "page_start": page_num,
+            "page_end": page_num,
+        }
+
+    def _process_ocr_concurrent(self, regions: List[TableRegion], progress_callback=None) -> List[OcrResult]:
+        """同步并发处理 OCR（使用 ThreadPoolExecutor）"""
+        results: List[OcrResult] = []
+        total = len(regions)
+        completed = 0
+
+        with ThreadPoolExecutor(max_workers=self.OCR_CONCURRENT_WORKERS) as executor:
+            # 提交所有任务
+            future_to_region = {
+                executor.submit(self._ocr_table_region, r.page, r.bbox): r
+                for r in regions
+            }
+
+            # 处理完成的结果
+            for future in as_completed(future_to_region):
+                region = future_to_region[future]
+                completed += 1
+                try:
+                    text = future.result()
+                    results.append(OcrResult(
+                        page_num=region.page_num,
+                        bbox=region.bbox,
+                        score=region.score,
+                        text=text,
+                        success=True,
+                    ))
+                except Exception as e:
+                    logger.error(f"  第 {region.page_num} 页表格 OCR 失败: {e}")
+                    results.append(OcrResult(
+                        page_num=region.page_num,
+                        bbox=region.bbox,
+                        score=region.score,
+                        text="",
+                        success=False,
+                    ))
+
+                # 每完成5个或最后一个时推送进度
+                if progress_callback and (completed % 5 == 0 or completed == total):
+                    progress = 35 + int(completed / total * 15)  # OCR执行占15%进度(35-50)
+                    progress_callback("版面分析", progress, f"OCR识别中 {completed}/{total}")
+
+        return results
+
+    def _detect_table_regions(
+        self,
+        page: fitz.Page,
+        page_num: int,
+        clip_box: fitz.Rect
+    ) -> List[Tuple[Tuple[float, float, float, float], float]]:
+        """检测页面中的表格区域，返回坐标列表"""
+        table_regions: List[Tuple[Tuple[float, float, float, float], float]] = []
+
+        if not RAPID_LAYOUT_AVAILABLE:
+            return table_regions
+
+        layout_engine = self._get_layout_engine()
+        if layout_engine is None:
+            return table_regions
+
+        # 渲染页面（裁剪区域）
+        pix = page.get_pixmap(dpi=self.OCR_DPI, clip=clip_box)
+        img = np.frombuffer(pix.samples, dtype=np.uint8).reshape(pix.height, pix.width, 3)
+
+        try:
+            layout_output = layout_engine(img)
+
+            # 解析版面结果
+            if hasattr(layout_output, 'boxes') and hasattr(layout_output, 'class_names'):
+                # 获取缩放比例
+                scale_x = clip_box.width / img.shape[1]
+                scale_y = clip_box.height / img.shape[0]
+
+                for box, label, score in zip(layout_output.boxes, layout_output.class_names, layout_output.scores):
+                    if label == "table" and score > self.OCR_CONFIDENCE_THRESHOLD:
+                        # 转换为 PDF 坐标
+                        pdf_x1 = clip_box.x0 + box[0] * scale_x
+                        pdf_y1 = clip_box.y0 + box[1] * scale_y
+                        pdf_x2 = clip_box.x0 + box[2] * scale_x
+                        pdf_y2 = clip_box.y0 + box[3] * scale_y
+
+                        table_regions.append(((pdf_x1, pdf_y1, pdf_x2, pdf_y2), score))
+
+        except Exception as e:
+            logger.warning(f"  第 {page_num} 页: 版面分析失败 ({e})")
+
+        return table_regions
+
+    def _ocr_table_region(self, page: fitz.Page, bbox: Tuple[float, float, float, float], max_retries: int = 3) -> str:
+        """对指定区域进行 OCR 识别（使用 GLM-OCR），支持指数退避重试"""
+        import time
+
+        # 渲染指定区域
+        rect = fitz.Rect(bbox)
+        pix = page.get_pixmap(dpi=self.OCR_DPI, clip=rect)
+        img_bytes = pix.tobytes("jpeg")
+
+        # 压缩图片
+        compressed = self._compress_image(img_bytes)
+        img_base64 = base64.b64encode(compressed).decode('utf-8')
+
+        # 请求 OCR
+        payload = {
+            "model": "GLM-OCR",
+            "messages": [
+                {
+                    "role": "user",
+                    "content": [
+                        {
+                            "type": "text",
+                            "text": "识别图片中的表格内容，按原文排版输出。"
+                                    "注意："
+                                    "1. 表格用 Markdown 表格格式"
+                                    "2. 保持换行和列对齐"
+                                    "3. 只输出表格内容，不要其他说明"
+                        },
+                        {
+                            "type": "image_url",
+                            "image_url": {"url": f"data:image/jpeg;base64,{img_base64}"}
+                        }
+                    ]
+                }
+            ],
+            "max_tokens": 2048,
+            "temperature": 0.1
+        }
+
+        headers = {"Content-Type": "application/json"}
+        if self.ocr_api_key:
+            headers["Authorization"] = f"Bearer {self.ocr_api_key}"
+
+        # 指数退避重试
+        last_error = None
+        for attempt in range(max_retries):
+            try:
+                response = requests.post(
+                    self.ocr_api_url,
+                    headers=headers,
+                    json=payload,
+                    timeout=self.ocr_timeout
+                )
+                response.raise_for_status()
+
+                result = response.json()
+                return self._extract_ocr_content(result)
+
+            except Exception as e:
+                last_error = e
+                if attempt < max_retries - 1:
+                    # 指数退避: 2, 4, 8 秒
+                    wait_time = 2 ** (attempt + 1)
+                    logger.warning(f"  第 {page.number + 1} 页表格 OCR 第 {attempt + 1} 次失败: {e}, {wait_time}秒后重试...")
+                    time.sleep(wait_time)
+                else:
+                    logger.error(f"  第 {page.number + 1} 页表格 OCR 最终失败（已重试{max_retries}次）: {e}")
+
+        # 所有重试都失败，抛出最后一个错误
+        raise last_error
+
+    def _replace_table_regions(
+        self,
+        page: fitz.Page,
+        original_text: str,
+        ocr_results: List[Dict],
+        clip_box: fitz.Rect
+    ) -> str:
+        """用 OCR 结果替换原始文本中的表格区域"""
+        if not ocr_results:
+            return original_text
+
+        # 获取页面上的文本块及其坐标
+        text_blocks = []
+        for block in page.get_text("blocks"):
+            x0, y0, x1, y1, text, _, _ = block
+            # 只考虑裁剪区域内的文本
+            if y0 >= clip_box.y0 and y1 <= clip_box.y1:
+                text_blocks.append({
+                    "bbox": (x0, y0, x1, y1),
+                    "text": text.strip(),
+                })
+
+        # 按 Y 坐标排序
+        text_blocks.sort(key=lambda b: (b["bbox"][1], b["bbox"][0]))
+
+        # 找出属于表格区域的文本块
+        replaced_indices: Set[int] = set()
+        for ocr_result in ocr_results:
+            bbox = ocr_result["bbox"]
+            rx0, ry0, rx1, ry1 = bbox
+
+            for idx, block in enumerate(text_blocks):
+                if idx in replaced_indices:
+                    continue
+                bx0, by0, bx1, by1 = block["bbox"]
+
+                # 检查重叠
+                overlap_x = max(0, min(bx1, rx1) - max(bx0, rx0))
+                overlap_y = max(0, min(by1, ry1) - max(by0, ry0))
+                overlap_area = overlap_x * overlap_y
+                block_area = (bx1 - bx0) * (by1 - by0)
+
+                if block_area > 0 and overlap_area / block_area > 0.5:
+                    replaced_indices.add(idx)
+
+        # 构建新文本
+        result_parts: List[str] = []
+        last_idx = 0
+
+        for ocr_result in sorted(ocr_results, key=lambda r: r["bbox"][1]):
+            bbox = ocr_result["bbox"]
+            rx0, ry0, rx1, ry1 = bbox
+
+            # 找到该表格区域之前的文本
+            region_start_idx = None
+            for idx, block in enumerate(text_blocks):
+                if idx in replaced_indices:
+                    bx0, by0, bx1, by1 = block["bbox"]
+                    if (bx0 >= rx0 - 5 and bx1 <= rx1 + 5 and
+                        by0 >= ry0 - 5 and by1 <= ry1 + 5):
+                        if region_start_idx is None:
+                            region_start_idx = idx
+                        last_idx = idx + 1
+
+            if region_start_idx is not None:
+                # 添加表格前的非表格文本
+                for idx in range(last_idx - (last_idx - region_start_idx), region_start_idx):
+                    if idx not in replaced_indices and idx < len(text_blocks):
+                        result_parts.append(text_blocks[idx]["text"])
+                        result_parts.append("\n")
+
+                # 添加 OCR 结果
+                result_parts.append(ocr_result["ocr_text"])
+                result_parts.append("\n")
+
+        # 添加剩余文本
+        for idx in range(last_idx, len(text_blocks)):
+            if idx not in replaced_indices:
+                result_parts.append(text_blocks[idx]["text"])
+                result_parts.append("\n")
+
+        return "".join(result_parts).strip() or original_text
+
+    def _compress_image(self, img_bytes: bytes) -> bytes:
+        """压缩图片"""
+        try:
+            from PIL import Image
+            img = Image.open(io.BytesIO(img_bytes))
+
+            if img.mode in ('RGBA', 'LA', 'P'):
+                background = Image.new('RGB', img.size, (255, 255, 255))
+                if img.mode == 'P':
+                    img = img.convert('RGBA')
+                if img.mode in ('RGBA', 'LA'):
+                    background.paste(img, mask=img.split()[-1])
+                img = background
+            elif img.mode != 'RGB':
+                img = img.convert('RGB')
+
+            min_edge = min(img.size)
+            if min_edge > self.MAX_SHORT_EDGE:
+                ratio = self.MAX_SHORT_EDGE / min_edge
+                new_size = (int(img.width * ratio), int(img.height * ratio))
+                img = img.resize(new_size, Image.Resampling.LANCZOS)
+
+            buffer = io.BytesIO()
+            img.save(buffer, format='JPEG', quality=self.JPEG_QUALITY, optimize=True)
+            return buffer.getvalue()
+
+        except Exception as e:
+            logger.warning(f"图片压缩失败，使用原图: {e}")
+            return img_bytes
+
+    def _extract_ocr_content(self, result: Dict) -> str:
+        """从 OCR 响应提取内容，并将 HTML 表格转换为 Markdown"""
+        content = ""
+        if "choices" in result and isinstance(result["choices"], list):
+            if len(result["choices"]) > 0:
+                message = result["choices"][0].get("message", {})
+                content = message.get("content", "")
+
+        # 如果内容包含 HTML 标签，转换为 Markdown
+        if content and "<" in content and ">" in content:
+            try:
+                from ..doc_worker.pdf_worker.html_to_markdown import convert_html_to_markdown
+                content = convert_html_to_markdown(content)
+            except Exception as e:
+                logger.debug(f"HTML 转 Markdown 失败，保留原始内容: {e}")
+
+        return content
+
+    @staticmethod
+    def _is_header_footer(line: str) -> bool:
+        return (
+            "四川路桥建设集团股份有限公司" in line
+            or "T梁运输及安装专项施工方案" in line
+            or line.isdigit()
+        )
+
+    @classmethod
+    def _matching_rule_names(
+        cls,
+        line: str,
+        level: str,
+        rule_names: Optional[List[str]] = None,
+    ) -> List[str]:
+        clean_line = line.strip()
+        names = rule_names or list(cls.RULE_LIB.keys())
+        return [
+            rule_name
+            for rule_name in names
+            if cls.RULE_LIB[rule_name][level].match(clean_line)
+        ]
+
+    @classmethod
+    def _matches_chapter_heading(cls, line: str, rule_names: Optional[List[str]] = None) -> bool:
+        return bool(cls._matching_rule_names(line, "l1", rule_names))
+
+    @classmethod
+    def _matches_section_heading(cls, line: str, rule_names: Optional[List[str]] = None) -> bool:
+        return bool(cls._matching_rule_names(line, "l2", rule_names))
+
+    @staticmethod
+    def _clean_chapter_title(line: str) -> str:
+        cleaned = line.strip()
+        cleaned = re.sub(r"\s+\d+\s*$", "", cleaned)
+        cleaned = re.sub(r"[\._\-]{3,}[^\u4e00-\u9fa5a-zA-Z0-9]*$", "", cleaned)
+        cleaned = re.sub(r"\s+", " ", cleaned).strip()
+
+        cn_chapter_match = re.match(
+            r"^(第\s*(?:\d+|[一二三四五六七八九十百零两]+)\s*[章部分篇])[\s、:：\.-]*(.*)$",
+            cleaned,
+        )
+        if cn_chapter_match:
+            prefix = re.sub(r"\s+", "", cn_chapter_match.group(1))
+            title = cn_chapter_match.group(2).strip()
+            return f"{prefix} {title}".strip()
+
+        num_chapter_match = re.match(r"^(\d{1,2})(?:[\.．。、])?\s*(.*)$", cleaned)
+        if num_chapter_match:
+            prefix = num_chapter_match.group(1)
+            title = num_chapter_match.group(2).strip()
+            return f"{prefix} {title}".strip()
+
+        return cleaned
+
+    @staticmethod
+    def _clean_section_title(line: str) -> str:
+        cleaned = line.strip()
+        cleaned = re.sub(r"\s+\d+\s*$", "", cleaned)
+        cleaned = re.sub(r"[\._\-]{3,}[^\u4e00-\u9fa5a-zA-Z0-9]*$", "", cleaned)
+        cleaned = re.sub(r"\s+", " ", cleaned).strip()
+
+        numeric_section_match = re.match(r"^(\d+\.\d+)(?!\.\d)\.?\s*(.*)$", cleaned)
+        if numeric_section_match:
+            prefix = numeric_section_match.group(1)
+            title = numeric_section_match.group(2).strip()
+            return f"{prefix} {title}".strip()
+
+        cn_section_match = re.match(r"^(第\s*[一二三四五六七八九十百零两]+\s*节)[\s、:：\.-]*(.*)$", cleaned)
+        if cn_section_match:
+            prefix = re.sub(r"\s+", "", cn_section_match.group(1))
+            title = cn_section_match.group(2).strip()
+            return f"{prefix} {title}".strip()
+
+        cn_list_match = re.match(r"^([一二三四五六七八九十百零两]+[、）\)\]])[\s]*(.*)$", cleaned)
+        if cn_list_match:
+            prefix = cn_list_match.group(1).strip()
+            title = cn_list_match.group(2).strip()
+            return f"{prefix} {title}".strip()
+
+        bracket_match = re.match(r"^([【\[]\s*\d+\s*[\]】])[\s]*(.*)$", cleaned)
+        if bracket_match:
+            prefix = re.sub(r"\s+", "", bracket_match.group(1))
+            title = bracket_match.group(2).strip()
+            return f"{prefix} {title}".strip()
+
+        return cleaned

core/construction_review/component/minimal_pipeline/simple_processor.py

@@ -9,13 +9,14 @@
 """
 
 import asyncio
+import json
 import uuid
 from collections import defaultdict
 from typing import Dict, Any, Optional, Tuple, List
 
 from foundation.observability.logger.loggering import review_logger as logger
 
-from .pdf_extractor import PdfStructureExtractor
+from .pdf_extractor2 import PdfStructureExtractor
 from .toc_builder import build_toc_items_from_structure
 from .chunk_assembler import assemble_chunks
 from ..doc_worker.classification.hierarchy_classifier import HierarchyClassifier
@@ -121,6 +122,9 @@ class SimpleDocumentProcessor:
                     pass
 
         structure = self.pdf_extractor.extract(file_content, progress_callback=_extraction_progress)
+        print("-"*50)
+        print(f'{json.dumps(structure, ensure_ascii=False, indent=2)}')
+        print("-"*50)
         catalog = structure.get("catalog")  # 获取YOLO检测+OCR提取的目录
 
         # 对 catalog 进行分类（如果存在）