3

backend/app/processors/parsers/docx_parser.py

@@ -0,0 +1,54 @@
+"""DOCX 解析器（python-docx fallback）。"""
+
+from app.core.logging import get_logger
+
+logger = get_logger(__name__)
+
+
+def parse_docx_with_python_docx(file_content: bytes) -> str | None:
+    """用 python-docx 解析 DOCX，返回 Markdown 或 None（失败时）。"""
+    try:
+        import io
+
+        from docx import Document
+
+        doc = Document(io.BytesIO(file_content))
+        md_parts = []
+
+        for para in doc.paragraphs:
+            text = para.text.strip()
+            if not text:
+                continue
+
+            style = para.style.name.lower()
+            if style.startswith("heading 1"):
+                md_parts.append(f"# {text}")
+            elif style.startswith("heading 2"):
+                md_parts.append(f"## {text}")
+            elif style.startswith("heading 3"):
+                md_parts.append(f"### {text}")
+            elif style.startswith("heading"):
+                md_parts.append(f"#### {text}")
+            elif style.startswith("list"):
+                md_parts.append(f"- {text}")
+            else:
+                md_parts.append(text)
+
+        # 处理表格
+        for table in doc.tables:
+            rows = []
+            for row in table.rows:
+                cells = [cell.text.strip() for cell in row.cells]
+                rows.append("| " + " | ".join(cells) + " |")
+            if rows:
+                # 添加表头分隔行
+                if len(rows) > 1:
+                    sep = "| " + " | ".join(["---"] * len(table.columns)) + " |"
+                    rows.insert(1, sep)
+                md_parts.append("\n".join(rows))
+
+        markdown = "\n\n".join(md_parts)
+        return markdown if markdown.strip() else None
+    except Exception as e:
+        logger.warning("python-docx 解析失败: %s", e)
+        return None

backend/app/processors/parsers/markitdown_parser.py

@@ -0,0 +1,30 @@
+"""MarkItDown 解析器（优先）。"""
+
+from app.core.logging import get_logger
+
+logger = get_logger(__name__)
+
+
+def parse_with_markitdown(file_content: bytes, filename: str) -> str | None:
+    """用 MarkItDown 解析文档，返回 Markdown 或 None（失败时）。"""
+    try:
+        from markitdown import MarkItDown
+
+        md = MarkItDown()
+        # MarkItDown 需要文件路径或文件对象，用临时文件
+        import tempfile
+        from pathlib import Path
+
+        suffix = Path(filename).suffix.lower()
+        with tempfile.NamedTemporaryFile(suffix=suffix, delete=False) as tmp:
+            tmp.write(file_content)
+            tmp_path = tmp.name
+
+        try:
+            result = md.convert(tmp_path)
+            return result.text_content if result and result.text_content else None
+        finally:
+            Path(tmp_path).unlink(missing_ok=True)
+    except Exception as e:
+        logger.warning("MarkItDown 解析失败 (%s): %s", filename, e)
+        return None

backend/app/processors/parsers/pdf_parser.py

@@ -0,0 +1,47 @@
+"""PDF 解析器（PyMuPDF fallback）。"""
+
+from app.core.logging import get_logger
+
+logger = get_logger(__name__)
+
+
+def parse_pdf_with_pymupdf(file_content: bytes) -> str | None:
+    """用 PyMuPDF 解析 PDF，返回 Markdown 或 None（失败时）。"""
+    try:
+        import pymupdf  # PyMuPDF >= 1.24
+
+        doc = pymupdf.open(stream=file_content, filetype="pdf")
+        pages = []
+        for page_num in range(len(doc)):
+            page = doc.load_page(page_num)
+            text = page.get_text()
+            if text.strip():
+                pages.append(text)
+        doc.close()
+
+        if not pages:
+            return None
+
+        # 组合成 Markdown（每页一个段落）
+        markdown = "\n\n".join(pages)
+        return markdown
+    except Exception as e:
+        logger.warning("PyMuPDF 解析失败: %s", e)
+        return None
+
+
+def has_text_layer(file_content: bytes) -> bool:
+    """检查 PDF 是否有文本层。"""
+    try:
+        import pymupdf
+
+        doc = pymupdf.open(stream=file_content, filetype="pdf")
+        for page_num in range(min(3, len(doc))):  # 检查前 3 页
+            page = doc.load_page(page_num)
+            if page.get_text().strip():
+                doc.close()
+                return True
+        doc.close()
+        return False
+    except Exception:
+        return False