pdf2txt.py

#!/usr/bin/python

import sys, zlib, re

def print_streams(bytes, start=0):
	r = re.compile("BT.*?\((.*?)\)[\s]*?Tj.*?ET", re.DOTALL)
	while True:
		start = bytes.find("stream", bytes.find("FlateDecode", start)) + len("stream")
		if start < len("stream"): return
		end = bytes.find("endstream", start)
		print "\n".join(r.findall(zlib.decompress(bytes[start:end].lstrip())))

if len(sys.argv) > 1: [print_streams(open(arg, "rb").read()) for arg in sys.argv[1:]]
else: print_streams(sys.stdin.read())