fix(file-summary): 补强 Office 页数统计

tests/test_file_summary_page_count.py

@@ -1,4 +1,6 @@
 import pytest
+import shutil
+from zipfile import ZipFile
 from docx import Document
 from openpyxl import Workbook
 from pptx import Presentation
@@ -31,6 +33,89 @@ def test_count_document_pages_for_office_formats(tmp_path):
     assert count_document_pages(pptx_path).page_count == 1
 
 
+def test_count_docx_pages_from_extended_properties(tmp_path):
+    docx_path = tmp_path / "with-pages.docx"
+    Document().save(docx_path)
+    app_xml = (
+        '<?xml version="1.0" encoding="UTF-8"?>'
+        '<Properties xmlns="http://schemas.openxmlformats.org/officeDocument/2006/extended-properties">'
+        "<Pages>7</Pages>"
+        "</Properties>"
+    )
+    rewritten = tmp_path / "rewritten.docx"
+    with ZipFile(docx_path) as source, ZipFile(rewritten, "w") as target:
+        for entry in source.infolist():
+            if entry.filename != "docProps/app.xml":
+                target.writestr(entry, source.read(entry.filename))
+        target.writestr("docProps/app.xml", app_xml)
+    shutil.move(rewritten, docx_path)
+
+    result = count_document_pages(docx_path)
+
+    assert result.status == "success"
+    assert result.page_count == 7
+
+
+def test_count_docx_pages_uses_word_com_fallback(monkeypatch, tmp_path):
+    docx_path = tmp_path / "without-pages.docx"
+    Document().save(docx_path)
+    monkeypatch.setattr(
+        "review_agent.file_summary.services.page_count._count_docx_pages_from_extended_properties",
+        lambda path: None,
+    )
+    monkeypatch.setattr(
+        "review_agent.file_summary.services.page_count._count_word_pages_with_com",
+        lambda path: 22,
+    )
+
+    result = count_document_pages(docx_path)
+
+    assert result.status == "success"
+    assert result.page_count == 22
+
+
+def test_count_doc_pages_uses_word_com_fallback(monkeypatch, tmp_path):
+    doc_path = tmp_path / "legacy.doc"
+    doc_path.write_bytes(b"legacy-doc-placeholder")
+    monkeypatch.setattr(
+        "review_agent.file_summary.services.page_count._count_word_pages_with_com",
+        lambda path: 5,
+    )
+
+    result = count_document_pages(doc_path)
+
+    assert result.status == "success"
+    assert result.page_count == 5
+
+
+def test_count_ppt_pages_uses_powerpoint_com_fallback(monkeypatch, tmp_path):
+    ppt_path = tmp_path / "legacy.ppt"
+    ppt_path.write_bytes(b"legacy-ppt-placeholder")
+    monkeypatch.setattr(
+        "review_agent.file_summary.services.page_count._count_powerpoint_slides_with_com",
+        lambda path: 9,
+    )
+
+    result = count_document_pages(ppt_path)
+
+    assert result.status == "success"
+    assert result.page_count == 9
+
+
+def test_count_excel_pages_uses_excel_com_fallback(monkeypatch, tmp_path):
+    xls_path = tmp_path / "legacy.xls"
+    xls_path.write_bytes(b"legacy-xls-placeholder")
+    monkeypatch.setattr(
+        "review_agent.file_summary.services.page_count._count_excel_sheets_with_com",
+        lambda path: 3,
+    )
+
+    result = count_document_pages(xls_path)
+
+    assert result.status == "success"
+    assert result.page_count == 3
+
+
 def test_document_page_count_skill_marks_unsupported_and_success(tmp_path, django_user_model):
     xlsx_path = tmp_path / "a.xlsx"
     workbook = Workbook()