Gitlog extraction

gistfile1.py

def extract_git_log_by_hashes(hash_file_path, raw_log_path, output_file_path):
	f = open(hash_file_path, 'r')
	hashes = []
	for h in f:
		hashes.append(str(h).strip())
	f.close()

	raw_log_file = open(raw_log_path, 'r')
	processed_data = {}
	is_in_block = False
	current_block_text = ''
	current_hash = ''
	for line in raw_log_file:
		if not is_in_block:
			for h in hashes:
				start_trigger = 'commit ' + h
				if start_trigger in str(line):
					current_block_text += line
					current_hash = h
					is_in_block = True
					break
		else:
			current_block_text += line
			if 'file' in line and 'changed' in line and ('insertion' in line or 'deletion' in line):
				processed_data[current_hash] = current_block_text
				current_block_text = ''
				current_hash = ''
				is_in_block = False
	raw_log_file.close()

	log_file = open(output_file_path, 'w')
	for h in hashes:
		log_file.write(processed_data.get(h, ''))
		log_file.write('\n======================================================================================\n\n')
	log_file.close()