decode-vbi

#!/usr/bin/python3
# Given a .tbc.json file, extract data from the VBI information in useful
# formats.

import argparse
import json
import re
import sys

from vbi import FieldInfo

def show_vbi(tbcjson):
    """Show the decoded VBI."""

    for index, fieldjson in enumerate(tbcjson["fields"]):
        fi = FieldInfo(fieldjson)
        print(index, fi)

def ffmpeg_escape(s):
    """Escape strings for ffmpeg metadata files."""
    return re.sub(r'[=;#\\\n]', lambda m: "\\" + m.group(0), s)

def make_metadata(tbcjson, metadata_fn):
    """Generate an ffmpeg metadata file with the chapter locations.

    Format description: <https://ffmpeg.org/ffmpeg-formats.html#Metadata-1>"""

    # We'll give positions using field indexes directly, rather than using the
    # times encoded in the VBI, because we might be working with a capture
    # of only part of a disc. Select the appropriate timebase to make this work.
    if tbcjson["videoParameters"]["isSourcePal"]:
        timebase = "1/50"
    else:
        timebase = "1001/60000"

    # Scan through the VBI...
    chapter_changes = []
    stopcodes = set()
    chapter = None
    first_field_index = 0
    for index, fieldjson in enumerate(tbcjson["fields"]):
        # Codes may be in either field; we want the index of the first
        if fieldjson["isFirstField"]:
            first_field_index = index

        fi = FieldInfo(fieldjson)

        if fi.chapter is not None and fi.chapter != chapter:
            # Chapter change
            chapter = fi.chapter
            chapter_changes.append((first_field_index, chapter))

        if fi.stopcode:
            # Stop code
            stopcodes.add(first_field_index)

    # Add a dummy change at the end of the input, so we can get the length of the last chapter
    chapter_changes.append((len(tbcjson["fields"]), None))

    # Because chapter markers have no error detection, a corrupt marker will
    # result in a spurious chapter change. Remove suspiciously short chapters.
    # XXX Do better when an error occurs near the start of a chapter
    clean_changes = []
    for i in range(len(chapter_changes) - 1):
        change = chapter_changes[i]
        next_change = chapter_changes[i + 1]

        if (next_change[0] - change[0]) < 10:
            # Too short - drop
            pass
        elif clean_changes != [] and change[1] == clean_changes[-1][1]:
            # Change to the same chapter - drop
            pass
        else:
            # Keep
            clean_changes.append(change)
    clean_changes.append(chapter_changes[-1])
    chapter_changes = clean_changes

    # Begin the file
    if metadata_fn == "-":
        f = sys.stdout
    else:
        f = open(metadata_fn, "w")
    f.write(";FFMETADATA1\n")

    # Write out the chapters
    for i, change in enumerate(chapter_changes[:-1]):
        f.write("\n")
        f.write("[CHAPTER]\n")
        f.write("TIMEBASE=%s\n" % timebase)
        f.write("START=%d\n" % change[0])
        f.write("END=%d\n" % (chapter_changes[i + 1][0] - 1))
        f.write("title=%s\n" % ffmpeg_escape("Chapter %d" % change[1]))

    # Write out the stop codes
    # XXX Is there a way to represent these properly?
    f.write("\n")
    for index in sorted(stopcodes):
        f.write("; Stop code at %d\n" % index)

    if metadata_fn != "-":
        f.close()

def main():
    # Parse command-line options
    parser = argparse.ArgumentParser()
    parser.add_argument('jsonfile', type=str, help='JSON file containing VBI data')
    parser.add_argument('-m', '--metadata', metavar='FILE', help='Generate ffmpeg metadata file')
    args = parser.parse_args()

    with open(args.jsonfile) as f:
        tbcjson = json.load(f)

    if args.metadata:
        make_metadata(tbcjson, args.metadata)
    else:
        show_vbi(tbcjson)

if __name__ == "__main__":
    main()