Merge pull request #90 from wshlavacek/modelapi-b4-collapse-double-commas

Collapse repeated commas in action normalization

Author: jrfaeder

bionetgen/modelapi/bngparser.py

@@ -28,9 +28,49 @@ def _normalize_action_text(action: str) -> str:
     text = _strip_comment_outside_quotes(action)
     text = _collapse_unquoted_whitespace(text)
     text = _strip_unquoted_backslashes(text)
+    text = _collapse_unquoted_double_commas(text)
     return text.strip()
 
 
+def _collapse_unquoted_double_commas(text: str) -> str:
+    """Collapse repeated commas that appear outside string literals."""
+    out = []
+    in_single = False
+    in_double = False
+    escaped = False
+    prev_was_comma = False
+    for ch in text:
+        if escaped:
+            out.append(ch)
+            escaped = False
+            prev_was_comma = False
+            continue
+        if ch == "\\" and (in_single or in_double):
+            out.append(ch)
+            escaped = True
+            prev_was_comma = False
+            continue
+        if ch == '"' and not in_single:
+            in_double = not in_double
+            out.append(ch)
+            prev_was_comma = False
+            continue
+        if ch == "'" and not in_double:
+            in_single = not in_single
+            out.append(ch)
+            prev_was_comma = False
+            continue
+        if ch == "," and not in_single and not in_double:
+            if prev_was_comma:
+                continue
+            prev_was_comma = True
+            out.append(ch)
+            continue
+        prev_was_comma = False
+        out.append(ch)
+    return "".join(out)
+
+
 def _strip_unquoted_backslashes(text: str) -> str:
     """Drop ``\\`` characters that appear outside string literals."""
     return _filter_outside_quotes(text, lambda ch: ch != "\\")

tests/test_bng_parsing.py

@@ -90,3 +90,20 @@ def test_action_normalization_preserves_backslashes_inside_quotes():
 
     out = _normalize_action_text('action({arg=>"a\\b"})')
     assert '"a\\b"' in out
+
+
+def test_action_normalization_collapses_unquoted_double_commas():
+    from bionetgen.modelapi.bngparser import _normalize_action_text
+
+    out = _normalize_action_text(
+        'simulate({method=>"ode",t_end=>3000,n_steps=>20,,print_functions=>1})'
+    )
+    assert ",," not in out
+    assert ",n_steps=>20,print_functions=>1" in out
+
+
+def test_action_normalization_preserves_double_commas_inside_quotes():
+    from bionetgen.modelapi.bngparser import _normalize_action_text
+
+    out = _normalize_action_text('something({xs=>"0,,1,,2"})')
+    assert '"0,,1,,2"' in out