add eval libreoffice write compare content

desktop_env/evaluators/metrics/__init__.py

@@ -1,3 +1,3 @@
 from .table import compare_table, compare_with_sparklines, compare_with_charts
 from .table import check_sheet_list, check_xlsx_freeze
-from .docs import find_default_font, contains_page_break
+from .docs import find_default_font, contains_page_break, compare_docx_files

desktop_env/evaluators/metrics/docs.py

@@ -2,8 +2,6 @@ import xml.etree.ElementTree as ET
     
 from docx import Document
 
-from lxml import etree
-
 def find_default_font(expected, config_file_path):
     """Find the default font in LibreOffice Writer."""
     default_font = None
@@ -36,6 +34,29 @@ def contains_page_break(docx_file):
                 if br is not None and '{http://schemas.openxmlformats.org/wordprocessingml/2006/main}type' in br.attrib and br.attrib['{http://schemas.openxmlformats.org/wordprocessingml/2006/main}type'] == 'page':
                     return 1
     return 0
+
+def compare_docx_files(file1, file2):
+
+    doc1 = Document(file1)
+    doc2 = Document(file2)
+
+    doc1_paragraphs = [p.text for p in doc1.paragraphs]
+    doc2_paragraphs = [p.text for p in doc2.paragraphs]
+
+    if len(doc1_paragraphs) != len(doc2_paragraphs):
+        return 0
+
+    # Compare each paragraph
+    for p1, p2 in zip(doc1_paragraphs, doc2_paragraphs):
+        if p1 != p2:
+            return 0
+
+    return 1
+
+# file1 = 'path/to/file1.docx'
+# file2 = 'path/to/file2.docx'
+
+# print(are_docx_files_same(file1, file2))
 # Replace 'your_document.docx' with the path to your document
 # result = contains_page_break('your_document.docx')
 # print(result)