extract c functions

extract_c_func.py

import re
from pathlib import Path
import clang.cindex


def extract_functions(filename):
    index = clang.cindex.Index.create()
    tu = index.parse(filename)
    functions = []
    for node in tu.cursor.walk_preorder():
        if node.kind == clang.cindex.CursorKind.FUNCTION_DECL:
            functions.append(node.spelling)
    return functions


def extract_functions_containing(filename, target):
    index = clang.cindex.Index.create()
    tu = index.parse(filename)
    functions_containing_target = []

    for node in tu.cursor.get_children():
        if node.kind == clang.cindex.CursorKind.FUNCTION_DECL:
            function_code = node.extent.start.file.name

            with open(function_code, 'r') as f:
                lines = f.readlines()

            start_line, end_line = node.extent.start.line, node.extent.end.line
            function_text = ''.join(lines[start_line - 1:end_line])

            if target in function_text:
                functions_containing_target.append(function_text)

    return functions_containing_target


def extract_functions_containing(code, cond):
    functions = []
    stack = []
    func = ""
    capture = False

    # Split the code into lines
    lines = code.split("\n")

    # Regular expression to identify function signatures
    func_signature_re = re.compile(r'\w+\s+\w+\s*\(.*\)\s*{')

    for line in lines:
        # Check for function signature
        if func_signature_re.match(line):
            capture = True

        # Capture lines if inside a function
        if capture:
            func += line + "\n"

        # Update stack based on braces
        stack += [c for c in line if c == '{']
        for c in line:
            if c == '}':
                if stack:
                    stack.pop()

        # Check if we have captured a complete function
        if capture and not stack:
            if cond(func):
                functions.append(func)
            func = ""
            capture = False

    return functions


# Sample C code as a string
sample_code = '''
int foo(int a, int b) {
    if (a > b) {
        // Nested block
    }
    for (int i = 0; i < 100; i++) {
        if (i % 2 == 0) {
            printf("even");
        }
    }
    // This is function foo
    return a + b;
}

void bar() {
    // This is function bar
    foo(1, 2);
}

int baz(int x) {
    // This is function baz
    return x * x;
}
'''

# Extract functions that contain the string "foo"
result = extract_functions_containing(sample_code, lambda x: "foo" in x)
print("Functions containing 'foo':")
for i, func in enumerate(result):
    print(f"----------------{i}-----------------\n{func}")