Fix bug in _removeFormattingFromMathjax

anki/template/template.py

@@ -232,34 +232,45 @@ class Template:
         '{{c123' with a '{{C123'. The clozeText method interprets the upper-case
         C as "don't wrap this Cloze in a <span>".
         """
-        # TODO: There is a bug in this method.
-        # Say txt = r'\(a\) {{c1::b}} \[ {{c1::c}} \]', ord = 1.
-        #
-        # This method should return: '\(a\) {{c1::b}} \[ {{C1::c}} \]'.
-        # Since the {{c1::c}} occurs within a MathJax display formula.
-        # However, it returns '\(a\) {{c1::b}} \[ {{c1::c}} \]'.
-        # This causes the Cloze within the MathJax display formula
-        # to be erroneously formatted with a <span>.
-        opening = ["\\(", "\\["]
-        closing = ["\\)", "\\]"]
-        # flags in middle of expression deprecated
         creg = clozeReg.replace("(?si)", "")
-        regex = r"(?si)(\\[([])(.*?)"+(creg%ord)+r"(.*?)(\\[\])])"
 
-        def repl(m):
+        # Scan the string left to right.
-            enclosed = True
+        # After a MathJax opening - \( or \[ - flip in_mathjax to True.
-            for s in closing:
+        # After a MathJax closing - \) or \] - flip in_mathjax to False.
-                if s in m.group(1):
+        # When a Cloze pattern number `ord` is found and we are in MathJax,
-                    enclosed = False
+        # replace its '{{c' with '{{C'.
-            for s in opening:
+        #
-                if s in m.group(7):
+        # TODO: Report mismatching opens/closes - e.g. '\(\]'
-                    enclosed = False
+        # TODO: Report errors in this method better than printing to stdout.
-            if not enclosed:
+        # flags in middle of expression deprecated
-                return m.group(0)
+        in_mathjax = False
-            # remove formatting
+        def replace(match):
-            return m.group(0).replace("{{c", "{{C")
+            nonlocal in_mathjax
-        txt = re.sub(regex, repl, txt)
+            if match.group('mathjax_open'):
-        return txt
+                if in_mathjax:
+                    print("MathJax opening found while already in MathJax")
+                in_mathjax = True
+            elif match.group('mathjax_close'):
+                if not in_mathjax:
+                    print("MathJax close found while not in MathJax")
+                in_mathjax = False
+            elif match.group('cloze'):
+                if in_mathjax:
+                    return match.group(0).replace(
+                        '{{c{}::'.format(ord),
+                        '{{C{}::'.format(ord))
+            else:
+                print("Unexpected: no expected capture group is present")
+            return match.group(0)
+        # The following regex matches one of:
+        #  -  MathJax opening
+        #  -  MathJax close
+        #  -  Cloze deletion number `ord`
+        return re.sub(
+            r"(?si)"
+            r"(?P<mathjax_open>\\[([])|"
+            r"(?P<mathjax_close>\\[\])])|"
+            r"(?P<cloze>" + (creg%ord) + ")", replace, txt)
 
     @modifier('=')
     def render_delimiter(self, tag_name=None, context=None) -> str:

tests/test_models.py

@@ -199,18 +199,12 @@ def test_cloze_mathjax():
     assert "class=cloze" in f.cards()[3].q()
     assert "class=cloze" in f.cards()[4].q()
 
-def test_cloze_mathjax_bug():
-    d = getEmptyCol()
-    d.models.setCurrent(d.models.byName("Cloze"))
-
     f = d.newNote()
     f['Text'] = r'\(a\) {{c1::b}} \[ {{c1::c}} \]'
     assert d.addNote(f)
     assert len(f.cards()) == 1
+    assert f.cards()[0].q().endswith('\(a\) <span class=cloze>[...]</span> \[ [...] \]')
 
-    # TODO: The following assertion should work, but currently fails due
-    # to a bug in _removeFormatingFromMathjax.
-    #   assert f.cards()[0].q() == '\(a\) <span class=cloze>[...]</span> \[ [...] \]'
 
 def test_chained_mods():
     d = getEmptyCol()

tests/test_template.py

@@ -11,6 +11,6 @@ def test_remove_formatting_from_mathjax():
     # formatting.
     assert t._removeFormattingFromMathjax(txt, 2) == txt
 
-    # TODO: r'\(a\) {{c1::b}} \[ {{c1::c}} \]', ord=1 should return
+    txt = r'\(a\) {{c1::b}} \[ {{c1::c}} \]'
-    # r'\(a\) {{c1::b}} \[ {{C1::c}} \]', but actually fails to mark the cloze
+    assert t._removeFormattingFromMathjax(txt, 1) == (
-    # as not-to-be-formatted.
+        r'\(a\) {{c1::b}} \[ {{C1::c}} \]')