fix(query): handle semicolons inside string literals (#131)

The query parser was using query.split(';') to separate statements,
which incorrectly split on semicolons inside string literals, causing
QueryParseException for valid queries like 'RETURN = "hello;world"'.

Replace with _split_query_statements() that tracks quote state and
only splits on semicolons that are actual statement terminators.

Fixes ActivityWatch/aw-server#145

Author: TimeToBuildBob

aw_query/query2.py

@@ -401,6 +401,33 @@ def get_return(namespace):
     return namespace["RETURN"]
 
 
+def _split_query_statements(query: str) -> List[str]:
+    """Split query into statements on semicolons, ignoring semicolons inside string literals."""
+    statements = []
+    current = ""
+    in_single_quote = False
+    in_double_quote = False
+    prev_char = None
+
+    for char in query:
+        if char == "'" and prev_char != "\\" and not in_double_quote:
+            in_single_quote = not in_single_quote
+            current += char
+        elif char == '"' and prev_char != "\\" and not in_single_quote:
+            in_double_quote = not in_double_quote
+            current += char
+        elif char == ";" and not in_single_quote and not in_double_quote:
+            statements.append(current)
+            current = ""
+        else:
+            current += char
+        prev_char = char
+
+    if current:
+        statements.append(current)
+    return statements
+
+
 def query(
     name: str, query: str, starttime: datetime, endtime: datetime, datastore: Datastore
 ) -> Any:
@@ -409,7 +436,7 @@ def query(
     namespace["STARTTIME"] = starttime.isoformat()
     namespace["ENDTIME"] = endtime.isoformat()
 
-    query_stmts = query.split(";")
+    query_stmts = _split_query_statements(query)
     for statement in query_stmts:
         statement = statement.strip()
         if statement:

tests/test_query2.py

@@ -19,6 +19,7 @@
     QString,
     QVariable,
     _parse_token,
+    _split_query_statements,
 )
 
 from .utils import param_datastore_objects
@@ -229,6 +230,41 @@ def test_query2_return_value():
         result = query(qname, example_query, starttime, endtime, ds)
 
 
+def test_query2_semicolon_in_string():
+    """Test that semicolons inside string literals don't break query parsing (issue #145)."""
+    ds = mock_ds
+    qname = "asd"
+    starttime = iso8601.parse_date("1970-01-01")
+    endtime = iso8601.parse_date("1970-01-02")
+
+    # Semicolon in double-quoted string
+    example_query = 'RETURN = "hello;world"'
+    result = query(qname, example_query, starttime, endtime, ds)
+    assert result == "hello;world"
+
+    # Semicolon in single-quoted string
+    example_query = "RETURN = 'hello;world'"
+    result = query(qname, example_query, starttime, endtime, ds)
+    assert result == "hello;world"
+
+    # Multiple statements where one value contains a semicolon in a string
+    example_query = 'a = "foo;bar"; RETURN = a'
+    result = query(qname, example_query, starttime, endtime, ds)
+    assert result == "foo;bar"
+
+
+def test_split_query_statements():
+    """Unit test for _split_query_statements."""
+    # Basic split
+    assert _split_query_statements("a=1;b=2") == ["a=1", "b=2"]
+    # Semicolon inside double quotes should NOT split
+    assert _split_query_statements('a="x;y";b=2') == ['a="x;y"', "b=2"]
+    # Semicolon inside single quotes should NOT split
+    assert _split_query_statements("a='x;y';b=2") == ["a='x;y'", "b=2"]
+    # Trailing semicolon — empty trailing part is omitted
+    assert _split_query_statements("a=1;") == ["a=1"]
+
+
 def test_query2_multiline():
     ds = mock_ds
     qname = "asd"