refactor: excel parse

2026-04-16 10:01:11 +08:00
parent 680ecc320f
commit f62f95ec02
7941 changed files with 2899112 additions and 0 deletions
@@ -0,0 +1,93 @@
+"""Word (.docx) parser — full extraction of paragraphs, tables, lists, styles, images."""
+
+import os
+import tempfile
+from docx import Document
+from docx.table import Table
+from docx.text.paragraph import Paragraph
+from docx.oxml.ns import qn
+
+
+def parse_word(file_path: str, extract_images: bool = False) -> dict:
+    doc = Document(file_path)
+    parts = []
+    image_paths = []
+
+    if extract_images:
+        img_dir = tempfile.mkdtemp(prefix="engimind_word_img_")
+        for i, rel in enumerate(doc.part.rels.values()):
+            if "image" in rel.reltype:
+                try:
+                    blob = rel.target_part.blob
+                    ext = _ct_to_ext(rel.target_part.content_type)
+                    path = os.path.join(img_dir, f"image_{i + 1}.{ext}")
+                    with open(path, "wb") as f:
+                        f.write(blob)
+                    image_paths.append(path)
+                except Exception:
+                    pass
+
+    for element in doc.element.body:
+        tag = element.tag.split("}")[-1] if "}" in element.tag else element.tag
+        if tag == "p":
+            md = _para_to_md(Paragraph(element, doc))
+            if md:
+                parts.append(md)
+        elif tag == "tbl":
+            md = _table_to_md(Table(element, doc))
+            if md:
+                parts.append(md)
+
+    return {"markdown": "\n\n".join(parts), "images": image_paths}
+
+
+def _para_to_md(para: Paragraph) -> str:
+    text = para.text.strip()
+    if not text:
+        return ""
+    style = (para.style.name or "").lower()
+    if style.startswith("heading"):
+        try:
+            lvl = min(int(style.replace("heading", "").strip()), 6)
+        except ValueError:
+            lvl = 1
+        return f"{'#' * lvl} {text}"
+
+    numPr = para._element.find(qn("w:pPr"))
+    if numPr is not None and numPr.find(qn("w:numPr")) is not None:
+        ilvl_elem = numPr.find(qn("w:numPr")).find(qn("w:ilvl"))
+        indent = int(ilvl_elem.get(qn("w:val"), "0")) if ilvl_elem is not None else 0
+        return "  " * indent + "- " + text
+
+    formatted = []
+    for run in para.runs:
+        t = run.text
+        if not t:
+            continue
+        if run.bold and run.italic:
+            formatted.append(f"***{t}***")
+        elif run.bold:
+            formatted.append(f"**{t}**")
+        elif run.italic:
+            formatted.append(f"*{t}*")
+        else:
+            formatted.append(t)
+    return "".join(formatted) if formatted else text
+
+
+def _table_to_md(table: Table) -> str:
+    rows = table.rows
+    if not rows:
+        return ""
+    md = []
+    for i, row in enumerate(rows):
+        cells = [c.text.strip().replace("\n", " ") for c in row.cells]
+        md.append("| " + " | ".join(cells) + " |")
+        if i == 0:
+            md.append("| " + " | ".join("---" for _ in cells) + " |")
+    return "\n".join(md)
+
+
+def _ct_to_ext(ct: str) -> str:
+    return {"image/png": "png", "image/jpeg": "jpg", "image/gif": "gif",
+            "image/bmp": "bmp", "image/tiff": "tiff"}.get(ct, "png")