parsep4

#!/usr/bin/python

import sys
import re

header = '''// This file was extracted from the TCG Published
// Trusted Platform Module Library
// Part 4: Supporting Routines
// Family "2.0"
// Level 00 Revision 01.16
// October 30, 2014

'''

head_spaces = re.compile('^\s*[0-9]+\s{0,4}')
sec_num_in_comments = re.compile('^//\s+([A-D0-9]\.([0-9]\.?)+)')
source_lines = open(sys.argv[1], 'r').read().splitlines()

def postprocess_lines(buffer):
    # get rid of heading line numbers and spaces.
    postproc_buffer = []
    for big_line in buffer:
        big_line = head_spaces.sub('', big_line)
        for line in big_line.splitlines():
            m = sec_num_in_comments.match(line)
            if m:
                line = line.replace(m.groups()[0], '')
            postproc_buffer.append(line)

    return header + '\n'.join(postproc_buffer) + '\n'

text = []
for line in source_lines:
    text.append(line)
    if line == '' and text[-2].startswith('') and text[-5] == '':
        text = text[:-5]

func_file = None
file_name = ''
prev_num = 0
line_buffer = []
output_buffer = []
for line in text:
    f = re.match('^\s*[A-D0-9]+\.([0-9]+|([0-9]+\.)+)\s+(\S+\.[ch])$', line)
    if not f:
        f = re.match('^\s*[A-D0-9]+\.([0-9]+|([0-9]+\.)+)\s+[^\(]+\((\S+\.[ch])\)$', line)
    if f:
        file_name = f.groups(0)[2]
        continue

    num = re.match('^\s{0,3}([0-9]+)\s', line + ' ')
    if num:
        line_num = int(num.groups(0)[0])
        if line_num == 1:
            # this is the first line of a file
            if func_file:
                func_file.write(postprocess_lines(output_buffer))
                func_file.close()
            if file_name:
                func_file = open('%s' % file_name, 'w')
            output_buffer = [line,]
            prev_num = 1
            line_buffer = []
            continue
        if line_num == prev_num + 1:
            if line_buffer:
                output_buffer.append('\n'.join(line_buffer))
                line_buffer = []
            output_buffer.append(line)
            prev_num = line_num
        continue
    if not '//' in line:
        line = '//' + line
    line_buffer.append(line)

if func_file:
    func_file.write(postprocess_lines(output_buffer))
    func_file.close()