refactor: recursively extract text

openedx · Jan 9, 2025 · 48ad916 · 48ad916
1 parent a3e2087
commit 48ad916
Show file tree

Hide file tree

Showing 2 changed files with 20 additions and 9 deletions.
diff --git a/openedx/core/djangoapps/content/search/plain_text_math.py b/openedx/core/djangoapps/content/search/plain_text_math.py
@@ -9,7 +9,6 @@
 
 class InvalidMathEquation(Exception):
     """Raised when converting mathjax equations to plain text fails"""
-    pass
 
 
 class PlainTextMath:
@@ -36,7 +35,6 @@ class PlainTextMath:
     )
     regex_replacements = (
         # Makes text bold, so not required in plain text.
-        (re.compile(r'\\mathbf{(.*?)}'), r"\1"),
         (re.compile(r'{\\bf (.*?)}'), r"\1"),
     )
     extract_inner_texts = (
@@ -52,7 +50,7 @@ def _nested_bracket_matcher(equation: str, opening_pattern: str) -> str:
 
         Args:
             equation: string
-            opening_pattern: for example, \mathbf{
+            opening_pattern: for example, `\\mathbf{`
 
         Returns:
             String inside the eqn brackets
@@ -108,18 +106,27 @@ def _fraction_handler(self, equation: str) -> str:
         equation = equation[:n_start] + f"({numerator}/{denominator})" + equation[n_end + d_end:]
         return equation
 
+    def _nested_text_extractor(self, equation: str, pattern: str) -> str:
+        """
+        Recursively extracts text from equation for given pattern
+        """
+        try:
+            start, inner_start, inner_end, end = self._nested_bracket_matcher(equation, pattern)
+            inner_text = equation[inner_start:inner_end]
+            inner_text = self._nested_text_extractor(inner_text, pattern)
+            equation = equation[:start] + inner_text + equation[end:]
+        except InvalidMathEquation:
+            pass
+        return equation
+
     def _handle_replacements(self, equation: str) -> str:
         """
         Makes a bunch of replacements in equation string.
         """
         for q, replacement in self.eqn_replacements:
             equation = equation.replace(q, replacement)
         for pattern in self.extract_inner_texts:
-            try:
-                start, inner_start, inner_end, end = self._nested_bracket_matcher(equation, pattern)
-                equation = equation[:start] + equation[inner_start:inner_end] + equation[end:]
-            except InvalidMathEquation:
-                continue
+            equation = self._nested_text_extractor(equation, pattern)
         for pattern, replacement in self.regex_replacements:
             equation = re.sub(pattern, replacement, equation)
         return equation

diff --git a/openedx/core/djangoapps/content/search/tests/test_documents.py b/openedx/core/djangoapps/content/search/tests/test_documents.py
@@ -553,7 +553,11 @@ def test_mathjax_plain_text_conversion_for_search(self):
                 'Bold text:  a ⋅ b = |a| |b| cos(θ)',
             ),
             ('Bold text: \\( \\frac{\\sqrt{\\mathbf{2}+3}}{\\sqrt{4}} \\)', 'Bold text:  (√{2+3}/√{4})'),
-            ('Nested Bold text: \\( \\mathbf{ \\frac{1}{2} } \\)', 'Nested Bold text:   (1/2)'),
+            ('Nested Bold text 1: \\( \\mathbf{ \\frac{1}{2} } \\)', 'Nested Bold text 1:   (1/2)'),
+            (
+                'Nested Bold text 2: \\( \\mathbf{a \\cdot (a \\mathbf{\\times} b)} \\)',
+                'Nested Bold text 2:  a ⋅ (a × b)'
+            ),
             ('Sqrt test 1: \\(\\sqrt\\)', 'Sqrt test 1: √'),
             ('Sqrt test 2: \\(x^2 + \\sqrt(y)\\)', 'Sqrt test 2: x² + √(y)'),
             ('Sqrt test 3: [mathjaxinline]x^2 + \\sqrt(y)[/mathjaxinline]', 'Sqrt test 3: x² + √(y)'),