Triage of .inc files on GitHub by languages

triage_inc.py

#!/usr/bin/env python
import os
import re

re_php = re.compile(r"<\?(php)?\s", re.MULTILINE)
re_clarion = re.compile(r"^([A-Z][a-z]+\s+PROCEDURE\(|![\s\w-]+)", re.MULTILINE)
re_pascal = re.compile(r"^\s*(end;(\s*{[^}]+})?\s*$|{\$\w+\s)", re.MULTILINE | re.IGNORECASE)
re_sourcepawn = re.compile(r"^\s*(native|forward|stock)\s+[\w:]+\s*\(", re.MULTILINE | re.IGNORECASE)
re_cpp = re.compile(r"^\s*#(include|define|if|endif|pragma)\s*", re.MULTILINE)
re_html = re.compile(r"<\/\w+>", re.MULTILINE)
re_mysql = re.compile(r"^\s*--[\s\w-]+", re.MULTILINE)
re_asm = re.compile(r"(equ\s+[\dA-F]{8}h|^\s*(\.(const|code|data)|(add|mov)\s*\(?[&@%\w]+,\s*[&@%\w]+\)?))", re.MULTILINE | re.IGNORECASE)
re_js = re.compile(r"^\s*var\s+\w+\s+=", re.MULTILINE)
re_tex = re.compile(r"\\(section|begin){[\s\w-]+}")

php_files = []
clarion_files = []
pascal_files = []
html_files = []
sourcepawn_files = []
cpp_files = []
mysql_files = []
js_files = []
asm_files = []
tex_files = []
nb_files = 0
nb_unrecognized_files = 0
for root, dirs, filenames in os.walk('samples-inc-sourcepawn'):
	for filename in filenames:
		if root.startswith("samples-inc-sourcepawn/.git"):
			continue

		nb_files += 1
		filepath = os.path.join(root, filename)
		content = open(filepath, 'r').read()
		if re_php.search(content):
			php_files.append(filepath)
		elif re_clarion.search(content):
			clarion_files.append(filepath)
		elif re_pascal.search(content):
			pascal_files.append(filepath)
		elif re_mysql.search(content):
			mysql_files.append(filepath)
		elif re_sourcepawn.search(content):
			sourcepawn_files.append(filepath)
		elif re_cpp.search(content):
			cpp_files.append(filepath)
		elif re_html.search(content):
			html_files.append(filepath)
		elif re_asm.search(content):
			asm_files.append(filepath)
		elif re_js.search(content):
			js_files.append(filepath)
		elif re_tex.search(content):
			tex_files.append(filepath)
		else:
			nb_unrecognized_files += 1
			print(filepath)

print("%d files." % nb_files)
print("%d unrecognized files." % nb_unrecognized_files)
print("%d PHP files." % len(php_files))
print("%d Clarion files." % len(clarion_files))
print("%d Pascal files." % len(pascal_files))
print("%d SourcePawn files." % len(sourcepawn_files))
print("%d C++ files." % len(cpp_files))
print("%d HTML files." % len(html_files))
print("%d MySQL files." % len(mysql_files))
print("%d Assembly files." % len(asm_files))
print("%d JavaScript files." % len(js_files))
print("%d TeX files." % len(tex_files))