fix some problems in libreoffice writer

desktop_env/evaluators/metrics/__init__.py

@@ -43,6 +43,7 @@ from .docs import (
     compare_highlighted_text,
     is_first_line_centered,
     check_file_exists,
+    check_tabstops,
     compare_contains_image
 )
 from .general import (

desktop_env/evaluators/metrics/docs.py

@@ -6,11 +6,13 @@ import zipfile
 from typing import List, Dict, Any
 
 from docx import Document
-from docx.enum.text import WD_PARAGRAPH_ALIGNMENT
+from docx.enum.text import WD_PARAGRAPH_ALIGNMENT, WD_TAB_ALIGNMENT
 from docx.shared import RGBColor
 from odf.opendocument import load
 from odf.text import P
 from odf.text import Span
+from skimage.color import deltaE_ciede2000
+from skimage.color import rgb2lab
 
 logger = logging.getLogger("desktopenv.metric.docs")
 
@@ -141,7 +143,7 @@ def compare_docx_tables(docx_file1, docx_file2):
         # Compare each cell
         for i in range(len(table1.rows)):
             for j in range(len(table1.columns)):
-                if table1.cell(i, j).text != table2.cell(i, j).text:
+                if table1.cell(i, j).text.strip() != table2.cell(i, j).text.strip():
                     return 0
 
     return 1
@@ -234,6 +236,40 @@ def check_file_exists(directory, filename):
     return 1 if os.path.isfile(file_path) else 0
 
 
+def check_tabstops(docx_file1, docx_file2, **kwargs) -> float:
+    doc1: Document = Document(docx_file1)
+    doc2: Document = Document(docx_file2)
+    para1 = [p for p in doc1.paragraphs if p.text.strip()]
+    para2 = [p for p in doc2.paragraphs if p.text.strip()]
+    if len(para1) != len(para2): return .0
+
+    if kwargs.get('word_number_split_by_tabstop', None) is not None:
+        number = kwargs['word_number_split_by_tabstop']
+        index = kwargs.get('index', 0)
+        for p1 in para1:
+            splits = p1.text.split('\t')
+            if len(splits) == 0: return .0
+            words = list(filter(lambda x: x.strip(), re.split(r'\s', splits[index])))
+            if len(words) != number: return .0        
+    
+    section = doc2.sections[0]
+    paragraph_width = section.page_width - section.left_margin - section.right_margin
+    ignore_tabs = lambda x: x.alignment == WD_TAB_ALIGNMENT.CLEAR or (x.alignment == WD_TAB_ALIGNMENT.LEFT and x.position == 0)
+    minus = .0
+    for p1, p2 in zip(para1, para2):
+        # filter CLEAR tabstop and default left-0 tabstop
+        tabs1 = [tst for tst in p1.paragraph_format.tab_stops if not ignore_tabs(tst)]
+        tabs2 = [tst for tst in p2.paragraph_format.tab_stops if not ignore_tabs(tst)]
+        if len(tabs1) != len(tabs2): return .0
+        difference = .0
+        for t1, t2 in zip(tabs1, tabs2):
+            if t1.alignment != t2.alignment: return .0
+            difference += abs(t1.position - t2.position)
+        minus += difference / paragraph_width
+    score = 1 - (minus / len(para1))
+    return score
+
+
 def compare_contains_image(docx_file1, docx_file2):
     doc1 = Document(docx_file1)
     doc2 = Document(docx_file2)
@@ -258,10 +294,18 @@ def compare_contains_image(docx_file1, docx_file2):
 # print(find_default_font("Ani", config_path))
 
 
-def evaluate_colored_words_in_tables(file_path1, file_path2):
+def evaluate_colored_words_in_tables(file_path1, file_path2, **kwargs):
     if not compare_docx_files(file_path1, file_path2):
         return 0
     document = Document(file_path1)
+    threshold = kwargs.get('threshold', 3.5)
+
+    def _calculate_color_difference(rgb1, rgb2):
+        srgb1 = [rgb1[0] / 255.0, rgb1[1] / 255.0, rgb1[2] / 255.0]
+        srgb2 = [rgb2[0] / 255.0, rgb2[1] / 255.0, rgb2[2] / 255.0]
+        lab1, lab2 = rgb2lab(srgb1), rgb2lab(srgb2)
+        delta_e = deltaE_ciede2000(lab1, lab2)
+        return delta_e
 
     for table in document.tables:
         # Iterate through rows and cells in the table
@@ -273,9 +317,9 @@ def evaluate_colored_words_in_tables(file_path1, file_path2):
                         if word:
                             first_letter = word[0].lower()
 
-                            if first_letter in 'aeiou' and run.font.color.rgb != RGBColor(255, 0, 0):
+                            if first_letter in 'aeiou' and _calculate_color_difference(run.font.color.rgb, RGBColor(255, 0, 0)) > threshold:
                                 return 0  # Vowel-colored words should be red
-                            elif first_letter not in 'aeiou' and run.font.color.rgb != RGBColor(0, 0, 255):
+                            elif first_letter not in 'aeiou' and _calculate_color_difference(run.font.color.rgb, RGBColor(0, 0, 255)) > threshold:
                                 return 0  # Non-vowel-colored words should be blue
 
     return 1  # All words in tables are correctly colored