parse_py / parser.py
broadfield-dev's picture
Update parser.py
4039137 verified
raw
history blame
5.09 kB
# parser.py
import ast
def get_category(node):
"""Determine the category of an AST node."""
if isinstance(node, (ast.Import, ast.ImportFrom)):
return 'import'
elif isinstance(node, (ast.Assign, ast.AnnAssign, ast.AugAssign)):
return 'assignment'
elif isinstance(node, ast.FunctionDef):
return 'function'
elif isinstance(node, ast.AsyncFunctionDef):
return 'async_function'
elif isinstance(node, ast.ClassDef):
return 'class'
elif isinstance(node, ast.If):
return 'if'
elif isinstance(node, ast.While):
return 'while'
elif isinstance(node, ast.For):
return 'for'
elif isinstance(node, ast.Expr):
return 'expression'
else:
return 'other'
def create_vector(category, level, location, total_lines, parent_path):
"""Create a vector representation including hierarchy info."""
category_map = {
'import': 1, 'assignment': 2, 'function': 3, 'async_function': 4, 'class': 5,
'if': 6, 'while': 7, 'for': 8, 'expression': 9, 'spacer': 10, 'other': 11
}
category_id = category_map.get(category, 0)
start_line, end_line = location
# Extend vector with parent path length and a hash of the path for uniqueness
parent_depth = len(parent_path)
parent_hash = hash(' -> '.join(parent_path)) % 1000 # Simple hash for vector
return [
category_id,
level,
start_line / total_lines,
end_line / total_lines,
parent_depth,
parent_hash / 1000 # Normalized hash (0.0 to 1.0)
]
def parse_node(node, lines, prev_end, level=0, total_lines=None, parent_path=None, counters=None):
"""Recursively parse an AST node with full hierarchy tracking."""
if total_lines is None:
total_lines = len(lines)
if parent_path is None:
parent_path = []
if counters is None:
counters = {'if': 0, 'while': 0, 'for': 0, 'function': 0, 'class': 0} # Track counts per type
parts = []
start_line = getattr(node, 'lineno', prev_end + 1)
end_line = getattr(node, 'end_lineno', start_line)
category = get_category(node)
# Assign a node ID for relevant categories
node_id = ''
if category in counters:
counters[category] += 1
node_id = f"{category.capitalize()}[{counters[category]}]"
# Handle spacers before the node
if start_line > prev_end + 1:
spacer_lines = lines[prev_end:start_line - 1]
spacer_vector = create_vector('spacer', level, (prev_end + 1, start_line - 1), total_lines, parent_path)
parts.append({
'category': 'spacer',
'source': ''.join(spacer_lines),
'location': (prev_end + 1, start_line - 1),
'level': level,
'vector': spacer_vector,
'parent_path': ' -> '.join(parent_path) if parent_path else 'Top-Level',
'node_id': ''
})
# Capture the node's source
stmt_lines = lines[start_line - 1:end_line]
current_path = parent_path + ([node_id] if node_id else [])
node_vector = create_vector(category, level, (start_line, end_line), total_lines, current_path)
parts.append({
'category': category,
'source': ''.join(stmt_lines),
'location': (start_line, end_line),
'level': level,
'vector': node_vector,
'parent_path': ' -> '.join(parent_path) if parent_path else 'Top-Level',
'node_id': node_id
})
# Process nested nodes (body and orelse)
nested_prev_end = end_line - 1
for attr in ('body', 'orelse'):
if hasattr(node, attr) and getattr(node, attr):
for child in getattr(node, attr):
child_parts = parse_node(child, lines, nested_prev_end, level + 1, total_lines, current_path, counters)
parts.extend(child_parts)
nested_prev_end = child_parts[-1]['location'][1]
return parts
def parse_python_code(code):
"""Parse Python code string and return parts with hierarchy and vectors."""
lines = code.splitlines(keepends=True)
total_lines = len(lines)
try:
tree = ast.parse(code)
except SyntaxError:
return [{'category': 'error', 'source': 'Invalid Python code', 'location': (1, 1), 'level': 0, 'vector': [0, 0, 1.0, 1.0, 0, 0], 'parent_path': 'Top-Level', 'node_id': ''}]
parts = []
prev_end = 0
for stmt in tree.body:
stmt_parts = parse_node(stmt, lines, prev_end, total_lines=total_lines)
parts.extend(stmt_parts)
prev_end = stmt_parts[-1]['location'][1]
if prev_end < total_lines:
remaining_lines = lines[prev_end:]
spacer_vector = create_vector('spacer', 0, (prev_end + 1, total_lines + 1), total_lines, [])
parts.append({
'category': 'spacer',
'source': ''.join(remaining_lines),
'location': (prev_end + 1, total_lines + 1),
'level': 0,
'vector': spacer_vector,
'parent_path': 'Top-Level',
'node_id': ''
})
return parts