count the code, comments, empty lines in the project

count.py

# Python 3 Script
# Count the "real" code lines (exclude empty lines, and comments) in a project.
# Languages supported: Python, JavaScript, C, C++, Java, C#, etc....

# Analyse:
# A source code file contains code, and comments.
# Single line comments are easy. Just // or #
# Multi-line comments are begin usually with /*, and ends with */, this is
# accomplished with a stack, maybe. /* // */ counts as a single block of comments.

# In Python is begin with ''' or """, then ends with three quotes too.

# Good news is: the total lines = code + comments + empty lines.
# So if we can count the empty lines AND comments,
# Then we can calculate the code lines.

import os
from typing import List, Tuple

C_FAMILY = ['js', 'ts', 'mjs', 'c', 'cpp', 'java', 'cs']
PYTHON_FAMILY = ['py']
EXTENSIONS = [] + C_FAMILY + PYTHON_FAMILY


def _lstrip(this_line: str) -> str:
    ''' Strip the leading whitespaces off the line '''
    return this_line.lstrip()


def _rstrip(this_line: str) -> str:
    ''' Strip the tailing whitespaces off the line '''
    return this_line.rstrip()


def _strip(this_line: str) -> str:
    ''' Strip on the head and tail off the line '''
    return _lstrip(_rstrip(this_line))


def _get_ext(path_name: str):
    path_name = _strip(path_name)
    if not ('.' in path_name):
        raise ValueError('path does not contain ., no extension?')
    ext = path_name.split('.')[-1]
    return ext


def _is_empty_line(ext: str, this_line: str) -> bool:
    ''' Judge if this line is empty or only with white spaces.'''
    this_line = _strip(this_line)
    if this_line == '':
        return True
    
    return False


def _is_begin_multi_comments(ext: str, this_line: str) -> bool:
    ''' Judge if this line is the begin of multi-line comments '''
    this_line = _strip(this_line)
    if ext in PYTHON_FAMILY:
        if this_line.startswith("'''") or this_line.startswith('"""'):
            return True
        return False
    
    if ext in C_FAMILY:
        if this_line.startswith('/*'):
            return True
        return False
    
    raise ValueError('file with extension: ' + ext + ' not supported!')


def _is_end_multi_comments(ext: str, this_line: str) -> bool:
    ''' Judge if this line is the end of multi-line comments'''
    this_line = _strip(this_line)
    if ext in PYTHON_FAMILY:
        if this_line.endswith("'''") or this_line.endswith('"""'):
            return True
        return False
    
    if ext in C_FAMILY:
        if this_line.endswith('*/'):
            return True
        return False
    
    raise ValueError('file with extension: ' + ext + ' not supported!')


def _is_single_comment_line(ext: str, this_line: str) -> bool:
    ''' Judge if this line is a pure single line of comment '''
    this_line = _strip(this_line)
    if ext in PYTHON_FAMILY:
        if this_line.startswith('#'):
            return True
        
        if _is_begin_multi_comments(ext, this_line) and _is_end_multi_comments(ext, this_line):
            return True

        return False
    
    if ext in C_FAMILY:
        if this_line.startswith('//'):
            return True
        
        if _is_begin_multi_comments(ext, this_line) and _is_end_multi_comments(ext, this_line):
            return True

        return False
    
    raise ValueError('file with extension: ' + ext + ' not supported!')


def _read_a_file(abs_path: str) -> List[str]:
    ''' Read a file, get its lines of string '''
    with open(abs_path, 'r') as f:
        lines = []
        lines = f.readlines()
        return lines


def detect_file(ext: str, abs_path: str) -> Tuple[int, int, int, int]:
    ''' Read a file, get its lines of code, comments and empty lines '''
    if not os.path.isfile(abs_path):
        raise ValueError('path must point to a file...')

    lines = _read_a_file(abs_path)
    total_lines = len(lines)
    code_lines = 0
    comment_lines = 0
    empty_lines = 0

    in_comments_block_flag = False
    for idx, line in enumerate(lines):
        if not in_comments_block_flag:
            if _is_empty_line(ext, line):
                empty_lines += 1
                continue

            if _is_single_comment_line(ext, line):
                comment_lines += 1
                continue
            
            if _is_begin_multi_comments(ext, line):
                in_comments_block_flag = True
                comment_lines += 1
                continue

            if _is_end_multi_comments(ext, line):
                raise ValueError('not in comments block but see a end of block sign. ' + abs_path + ' :' + idx )
        else:
            comment_lines += 1
            if _is_end_multi_comments(ext, line):
                in_comments_block_flag = False

    code_lines = total_lines - empty_lines - comment_lines
    return total_lines, code_lines, comment_lines, empty_lines


def summarize(abs_dir_path: str, exclude_abs_folders: List, include_extensions: List, debug: bool = False) -> Tuple[int, int, int, int]:
    '''
    Summarize the absolute_directory_path folder, 
    but exclude the chosen immediate child folders.
    Include the extensions.
    '''
    total_lines = 0
    code_lines = 0
    comment_lines = 0
    empty_lines = 0

    if not os.path.isdir(abs_dir_path):
        raise ValueError('abs_dir_path should be a dir.')
    
    items = os.listdir(abs_dir_path)
    for item in items:
        item_abs_path = os.path.join(abs_dir_path, item)
        if os.path.isdir(item_abs_path):
            if not item in exclude_abs_folders:
                _total, _code, _comment, _empty = summarize(item_abs_path, exclude_abs_folders, include_extensions, debug)
                if debug:
                    print(_total, _code, _comment, _empty, item_abs_path)
                total_lines += _total
                code_lines += _code
                comment_lines += _comment
                empty_lines += _empty
            else:
                if debug:
                    print(item_abs_path, 'ignored.')

        elif os.path.isfile(item_abs_path):
            ext = _get_ext(item_abs_path)
            ext = ext.lower()
            if ext in include_extensions:
                _total, _code, _comment, _empty = detect_file(ext, item_abs_path)
                if debug:
                    print(_total, _code, _comment, _empty, item_abs_path)
                total_lines += _total
                code_lines += _code
                comment_lines += _comment
                empty_lines += _empty
        else:
            raise ValueError(item_abs_path + ' is not a file or dir.')

    return total_lines, code_lines, comment_lines, empty_lines


def pretty_print(abs_dir_path: str, exclude_abs_folders: List, include_extensions: List, debug: bool = False):
    _total, _code, _comment, _empty = summarize(abs_dir_path, exclude_abs_folders, include_extensions, debug)

    print('Summary:', abs_dir_path)
    print('-' * (len('Summary: ') + len(abs_dir_path)))

    print('Total\t', _total)
    print('Code\t', _code)
    print('Comment\t', _comment)
    print('Empty\t', _empty)
    print('-' * 16)
    print('Code Percentage:\t', '{:.2f}%'.format(_code/_total * 100))
    print('Comment Percentage:\t', '{:.2f}%'.format(_comment/_total * 100))
    print()


if __name__ == "__main__":

    pretty_print('/Users/username/github/thor-devkit.js', ['node_modules', 'dist', '.git'], ['ts', 'js', 'mjs'], False)
    pretty_print('/Users/username/github/thor-devkit.netcore', ['.git'], ['cs'], False)
    pretty_print('/Users/username/github/thor-devkit.py', ['.pytest_cache', '__pycache__', '.vscode', 'build', 'dist', '.git'], ['py'], False)