24798516591

Committed 22 Apr 2026 07:19PM UTC coverage: 90.606% (-0.2%) from 90.758%

Build # 24798516591

Build Type

push

github

Committed by

speedyk-005

Commit Message

refactor: remove redundant type hints from docstrings

- Strip (type) from Args/Returns where signature already has types
- Simplify Returns format to prose description
- Run clean_docstrings.py on src/chunklet (26 files)
- Add ExtractionState TypedDict for type safety (from earlier refactor)

Coverage Stats

1360 of 1501 relevant lines covered (90.61%)

3.62 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

79.17

/src/chunklet/code_chunker/utils.py

import ast
import re
from pathlib import Path

from pygments.lexers import guess_lexer
from pygments.util import ClassNotFound

from chunklet.common.path_utils import is_path_like
from chunklet.common.validation import validate_input


@validate_input
def is_python_code(source: str | Path) -> bool:
    """
    Check if a source is written in Python.

    This function uses multiple indicators, prioritizing syntactic validity
    via the Abstract Syntax Tree (AST) parser for maximum confidence.

    Indicators used:
      - File extension check for path inputs (e.g., .py, .pyi, .pyx, .pyw).
      - Shebang line detection (e.g., "#!/usr/bin/python").
      - Definitive syntax check using Python's `ast.parse()`.
      - Fallback heuristic via Pygments lexer guessing.

    Note:
        The function is definitive for complete, syntactically correct code blocks.
        It falls back to a Pygments heuristic only for short, incomplete, or
        ambiguous code snippets that fail AST parsing.

    Args:
        source: raw code string or Path to source file to check.

    Returns:
        True if the source is written in Python.
    """
    # Path-based check
    if isinstance(source, Path) or (isinstance(source, str) and is_path_like(source)):
        path = Path(source)
        return path.suffix.lower() in {".py", ".pyi", ".pyx", ".pyw"}

    # Shebang line check
    if re.match(r"#!/usr/bin/(env\s+)?python", source.strip()):
        return True

    # Definitive syntactic check
    # If parsing succeeds, it's definitely Python code
    try:
        ast.parse(source)
        return True
    except Exception:  # noqa: S110
        # If fails, it might still be Python code (e.g., incomplete snippet), so continue with heuristics
        pass

    # Pygments heuristic (Lowest confidence, last resort)
    try:
        lexer = guess_lexer(source)
        return lexer.name.lower() == "python"
    except ClassNotFound:
        return False

1	import ast	4✔
2	import re	4✔
3	from pathlib import Path	4✔
4
5	from pygments.lexers import guess_lexer	4✔
6	from pygments.util import ClassNotFound	4✔
7
8	from chunklet.common.path_utils import is_path_like	4✔
9	from chunklet.common.validation import validate_input	4✔
10
11
12	@validate_input	4✔
13	def is_python_code(source: str \| Path) -> bool:	4✔
14	"""
15	Check if a source is written in Python.
16
17	This function uses multiple indicators, prioritizing syntactic validity
18	via the Abstract Syntax Tree (AST) parser for maximum confidence.
19
20	Indicators used:
21	- File extension check for path inputs (e.g., .py, .pyi, .pyx, .pyw).
22	- Shebang line detection (e.g., "#!/usr/bin/python").
23	- Definitive syntax check using Python's `ast.parse()`.
24	- Fallback heuristic via Pygments lexer guessing.
25
26	Note:
27	The function is definitive for complete, syntactically correct code blocks.
28	It falls back to a Pygments heuristic only for short, incomplete, or
29	ambiguous code snippets that fail AST parsing.
30
31	Args:
32	source: raw code string or Path to source file to check.
33
34	Returns:
35	True if the source is written in Python.
36	"""
37	# Path-based check
38	if isinstance(source, Path) or (isinstance(source, str) and is_path_like(source)):	4✔
39	path = Path(source)	×
40	return path.suffix.lower() in {".py", ".pyi", ".pyx", ".pyw"}	×
41
42	# Shebang line check
43	if re.match(r"#!/usr/bin/(env\s+)?python", source.strip()):	4✔
44	return True	×
45
46	# Definitive syntactic check
47	# If parsing succeeds, it's definitely Python code
48	try:	4✔
49	ast.parse(source)	4✔
50	return True	4✔
51	except Exception: # noqa: S110	4✔
52	# If fails, it might still be Python code (e.g., incomplete snippet), so continue with heuristics
53	pass	4✔
54
55	# Pygments heuristic (Lowest confidence, last resort)
56	try:	4✔
57	lexer = guess_lexer(source)	4✔
58	return lexer.name.lower() == "python"	4✔
59	except ClassNotFound:	×
60	return False	×

speedyk-005 / chunklet-py / 24798516591

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous