finds cpp files from given dir that are never included in given dir

orphans.py

#!/usr/bin/env python

import sys
import re
import pprint
import os
import fnmatch

headers = { 'cpp': ['*.h', '*.hh', '*.hpp'] }
sources = { 'cpp': ['*.c', '*.cpp', '*.cc']}
main_returns = ['void', 'int', 'long', 'char', 'short']
mains = [ '%s main'%ret for ret in main_returns]

def cpp_line_check(line):
    pos = line.find('include')
    if pos > -1:
        imp = (line[pos+len('include'):]).strip(' "<>\n\r')
        last = imp.split('/')[-1]
        
        return last
    return None

def get_files(search_dir, lang):
    os.chdir(search_dir)
    ext = headers[lang] + sources[lang]
    patterns = re.compile('|'.join(fnmatch.translate(p) for p in ext))
    matches = []
    for root, dirs, files in os.walk(search_dir):
            matches.extend([os.path.join(root,f) for f in files if patterns.match(f)])
    return matches

def build_imports(fn, lang, line_check):
    with open(fn,'rb') as code:
        imps = set()
        for line in code:
            imp = line_check(line)
            if imp:
                imps.add(imp)
        return imps

def has_main(fn):
    with open(fn,'rb') as code:
        for line in code:
            if 'main' in line:
                if line.strip()[:line.find('main')-1] in main_returns:
                    return True
    return False
    
try:
    search_dir = sys.argv[1]
except:
    search_dir = '.'

lang = 'cpp'
all_files = get_files(search_dir, lang)
includes = set()
for fn in all_files:
    includes.update( build_imports(fn, lang, cpp_line_check) )

orphans = [fn for fn in all_files if (os.path.basename(fn) not in includes) and not has_main(fn) ]
    
pp = pprint.PrettyPrinter(indent=2)
pp.pprint(orphans)

print('total orphans %d of %d files checked'%(len(orphans), len(includes)))