Added convert_unicode_dashes utility and its corresponding tests

Author: m-messer

app/tests/expression_utilities_test.py

@@ -0,0 +1,40 @@
+import pytest
+from ..utility.expression_utilities import convert_unicode_dashes
+
+
+class TestConvertUnicodeDashes:
+
+    @pytest.mark.parametrize(
+        "expr, expected",
+        [
+            # No dashes at all
+            ("x+y", []),
+            # ASCII hyphen-minus is not a unicode dash — no substitution
+            ("x-y", []),
+            # Empty string
+            ("", []),
+            # HYPHEN (U+2010)
+            ("x‐y", [("‐", "-")]),
+            # NON-BREAKING HYPHEN (U+2011)
+            ("x‑y", [("‑", "-")]),
+            # FIGURE DASH (U+2012)
+            ("x‒y", [("‒", "-")]),
+            # EN DASH (U+2013)
+            ("x–y", [("–", "-")]),
+            # EM DASH (U+2014)
+            ("x—y", [("—", "-")]),
+            # MINUS SIGN (U+2212)
+            ("x−y", [("−", "-")]),
+            # SMALL HYPHEN-MINUS (U+FE63)
+            ("x﹣y", [("﹣", "-")]),
+            # FULLWIDTH HYPHEN-MINUS (U+FF0D)
+            ("x－y", [("－", "-")]),
+            # Multiple different unicode dashes in one expression
+            ("x–y−z", [("–", "-"), ("−", "-")]),
+            # Repeated occurrences of the same dash — still one substitution pair
+            ("x−y−z", [("−", "-")]),
+        ]
+    )
+    def test_convert_unicode_dashes(self, expr, expected):
+        result = convert_unicode_dashes(expr)
+        assert result == expected

app/utility/expression_utilities.py

@@ -290,6 +290,20 @@ def transform_unicode_greek_symbols(expr):
                 alias_substitutions += [(alias, " "+name+" ")]
     return alias_substitutions
 
+def convert_unicode_dashes(expr):
+    unicode_dashes = [
+        "‐",  # HYPHEN
+        "‑",  # NON-BREAKING HYPHEN
+        "‒",  # FIGURE DASH
+        "–",  # EN DASH
+        "—",  # EM DASH
+        "−",  # MINUS SIGN
+        "﹣",  # SMALL HYPHEN-MINUS
+        "－",  # FULLWIDTH HYPHEN-MINUS
+    ]
+    return [(dash, "-") for dash in unicode_dashes if dash in expr]
+
+
 def protect_elementary_functions_substitutions(expr):
     alias_substitutions = []
     for (name, alias_list) in elementary_functions_names: