merge-y4m.py

#!/usr/bin/env python

import sys
import re

Y4M_HEADER_RE = re.compile("YUV4MPEG2 W([0-9]+) H([0-9]+) F([:0-9]+) .*")

def parse_header(header):
    print(header)
    m = Y4M_HEADER_RE.match(header)
    return (int(m.group(1)), int(m.group(2)), m.group(3))

left_video = open(sys.argv[1], "rb")
right_video = open(sys.argv[2], "rb")

left_sections = left_video.read().split("FRAME\n")
right_sections = right_video.read().split("FRAME\n")

(left_w, left_h, left_f) = parse_header(left_sections.pop(0))
(right_w, right_h, right_f) = parse_header(right_sections.pop(0))

assert left_w == right_w
assert left_h == right_h
assert left_f == right_f
assert len(left_sections) == len(right_sections)

output_video = open(sys.argv[3], "wb")
output_video.write("YUV4MPEG2 W%d H%d F%s Ip A0:0 C420jpeg\n" % (left_w * 2, left_h, left_f))

for frame in range(len(left_sections)):
    output_video.write("FRAME\n")

    # Y
    width = left_w
    for j in range(left_h):
        start = j * width
        stop = (j + 1) * width
        output_video.write(left_sections[frame][start:stop])
        output_video.write(right_sections[frame][start:stop])
    # Cb
    Y_len = left_w * left_h
    width = width / 2
    for j in range(left_h / 2):
        start = j * width + Y_len
        stop = (j + 1) * width + Y_len
        output_video.write(left_sections[frame][start:stop])
        output_video.write(right_sections[frame][start:stop])
    # Cr
    for j in range(left_h / 2):
        start = j * width + Y_len
        stop = (j + 1) * width + Y_len
        output_video.write(left_sections[frame][start:stop])
        output_video.write(right_sections[frame][start:stop])