Change regex pattern to include digraphs

.gitignore

@@ -1,2 +1,4 @@
 .pytest*
 *__pycache__*
+*.egg*
+venv

soundchanger/change.py

@@ -7,23 +7,6 @@ def apply(change, string, categories=None, apply=True):
     if not apply:
         return string
 
-    # Set up default categories
-    if not categories:
-        categories = {
-            'V': 'aeiou'
-        }
-
-    # Prepare change for regex
-    for k, v in {'{': '[', '}': ']', ',': '', ' ': ''}.items():
-        change = change.replace(k, v)
-    
-    # Replacements for categories
-    for k, v in categories.items():
-        # Replacement of categories inside brackets (before or after a comma)
-        change = re.sub(f'((?<=,){k}|{k}(?=,))', v, change)
-        # Replacements of categories anywhere else
-        change = change.replace(k, f"[{v}]")
-
     # Check validity of change
     if change.count('>') != 1:
         raise ValueError(f"Change {change} is not a valid sound change. (Missing character '>')")
@@ -31,17 +14,31 @@ def apply(change, string, categories=None, apply=True):
     if change.count('/') > 1:
         raise ValueError(f"Change {change} is not a valid sound change. (More than one '/' character)")
 
+    # Prepare change for regex
+    for k, v in {'{': '(', '}': ')', ',': '|'}.items():
+        change = change.replace(k, v)
+    
+    # Replacements for categories
+    for k, v in categories.items():
+        # Replacement of categories inside brackets (before or after a comma)
+        change = re.sub(f'((?<=,){k}|{k}(?=,))', '|'.join(v), change)
+        # Replacements of categories anywhere else
+        change = change.replace(k, '('+'|'.join(v)+')')
+
     if '/' in change:
         change, environment = change.split('/')
     else:
-        environment = '._.'
+        environment = '_'
 
     if environment.count('_') != 1:
         raise ValueError(f"Environment {environment} is not a valid environment. (Character '_' should exist exactly once)")
 
-    original, repl = change.split('>')
+    original, change_to = change.split('>')
     before, after = environment.split('_')
 
-    pattern = f"(?<={before}){original}(?={after})"
-    print(pattern)
+    n = before.count('(') + original.count('(') + 3
+    pattern = f"({before})({original})({after})"
+    repl = f"\\1{change_to}\\{n}"
+
     return re.sub(pattern, repl, f"#{string}#").strip('#')
+

tests/test_change.py

@@ -23,3 +23,14 @@ def test_apply_with_complex_group_and_categories():
         'P': 'ptk',
         'N': 'mn',
         }) == 'pani'
+
+def test_apply_with_complex_group_and_categories_and_digraphs():
+    assert apply('u>o/#V{ts,pf,t}_k{V,e,o}#', 'atsuka', categories={
+        'V': ['a', 'i', 'u']
+    }) == 'atsoka'
+    assert apply('u>o/#V{ts,pf,t}_k{V,e,o}#', 'atuka', categories={
+        'V': ['a', 'i', 'u']
+    }) == 'atoka'
+    assert apply('u>o/#V{ts,pf,t}_k{V,e,o}#', 'matsuka', categories={
+        'V': ['a', 'i', 'u']
+    }) == 'matsuka'