Transcode Found Video Files to HEVC (H.265) and optimally remove the older versions

undupe-transode.py

#!/usr/bin/env python3
import os
import sys
import glob
import subprocess
import shlex
import shutil
import pathlib
import itertools

def multiple_file_types(*patterns):
    return itertools.chain.from_iterable(glob.iglob(pattern, recursive=True) for pattern in patterns)

def convert_bytes(num):
    """
    this function will convert bytes to MB.... GB... etc
    """
    for x in ['bytes', 'KB', 'MB', 'GB', 'TB']:
        if num < 1024.0:
            return "%3.1f %s" % (num, x)
        num /= 1024.0

def get_first_line_from_command(command):
    command = shlex.split(command)
    output = subprocess.run(command, capture_output=True, text=True, universal_newlines=True)
    for line in output.stdout.split('\n'):
        return line.strip()

if __name__ == '__main__':
    FFPROBE = '/usr/local/bin/ffprobe'
    TVPATH  = '/usr/local/bin/transcode-video'
    MAXCOMPRESSRATIO = 0.95  # size of H.265 file over size of original file, max ratio allowed
    MAXSECONDSSHIFT = 2.5  # max seconds the new file is alowed to be shorter or longer than the old file
    REPLACEIFMP4ORMKV = True

    assert os.path.isfile(FFPROBE), f"ERROR: please install ffmpeg which includes a seperate ffprobe binary and fix the listed path: {FFPROBE}"
    assert os.path.isfile(TVPATH), f"ERROR: please install transcode-video and fix the listed path: {TVPATH}"
    assert 0.25 <= MAXCOMPRESSRATIO <= 1.1, f"ERROR: value given for MAXCOMPRESSRATIO is unreasonable: {MAXCOMPRESSRATIO} (it should probably be somewhere between .95 & .75)"

    for vidFile in multiple_file_types('**/*.mov', '**/*.mp4', '**/*.mkv', '**/*.m4v', '**/*.avi', '**/*.wmv', '**/*.webm', '**/*.flv'):
        already265 = False
        foundVideoType = False
        if vidFile[-3:] == 'mp4' and REPLACEIFMP4ORMKV:
            vidFile265 = ".".join(vidFile.split('.')[:-1]) + ".h265.mp4"
        else:
            vidFile265 = ".".join(vidFile.split('.')[:-1]) + ".h265.mkv"
        poorTranscodeLogPath = os.path.join(os.path.split(vidFile)[0],'.poorTranscodeLog')
        badTranscodeLogPath  = os.path.join(os.path.split(vidFile)[0],'.badTranscodeLog')

        if os.path.isfile(poorTranscodeLogPath):
            lines = [line.strip('\n') for line in open(poorTranscodeLogPath)]
            if vidFile265 in lines:
                # print(f'Skipping {vidFile} because it had a poor H.265 encode (A)')
                continue
            elif vidFile in lines:
                print(f'Skipping {vidFile} because this is the poor H.265 encode, perhaps delete this file? (B)')
                continue

        if os.path.isfile(badTranscodeLogPath):
            lines = [line.strip('\n') for line in open(badTranscodeLogPath)]
            if vidFile265 in lines:
                # print(f'Skipping {vidFile} because it had a bad H.265 encode (C)')
                continue
            elif vidFile in lines:
                print(f'Skipping {vidFile} because this is the bad H.265 encode, perhaps delete this file? (D)')
                continue

        if "hevc" in vidFile or "HEVC" in vidFile or "265" in vidFile:
            already265 = True
        else:
            command = shlex.split(f'{FFPROBE} -hide_banner "{vidFile}"')
            output = subprocess.run(command, capture_output=True, text=True, universal_newlines=True)
            for line in output.stderr.split('\n'):
                line = line.strip()
                if line.startswith('Stream') and "Video" in line:
                    foundVideoType = True
                    if 'hevc' in line:
                        already265 = True
            if not foundVideoType:
                print("Command failed:")
                for i in command:
                    print(i + " ", end = '')
                print('')
                sys.exit()

        if already265:
            # print(f'--------------------\nSkipping {vidFile} which is alredy H.265 (this is probably good)')
            continue
        else:
            print(f'--------------------\nProcessing {vidFile} to {vidFile265}')
            command = shlex.split(f'{TVPATH} -q --encoder vt_h265 --target small -o "{vidFile265}" "{vidFile}"')
            print(command)
            subprocess.run(command)

        #---------------------------------------------------------------------
        # Compare resulting file sizes
        # technically this could be part of the previous statement after we run the encode but this way it's resumable
        if os.path.isfile(vidFile) and os.path.isfile(vidFile265):
            vidFileSize = os.path.getsize(vidFile)
            vidFile265Size = os.path.getsize(vidFile265)
            vidFilesSizeDiff = vidFileSize - vidFile265Size

            if vidFile265Size >=vidFileSize * MAXCOMPRESSRATIO:
                print(f"Poor encode, not sufficiently smaller: {vidFile265} (E)")
                with open(poorTranscodeLogPath, "a") as pTLO:
                    pTLO.write(vidFile265 + '\n')
                os.remove(vidFile265)
            else:
                try:
                    # first check their length with ffprobe, sexagesimal give an HOURS:MM:SS.MICROSECONDS string vs a raw second count with decimals
                    vidFile265LengthString = get_first_line_from_command(f'{FFPROBE} -v error -show_entries format=duration -of default=noprint_wrappers=1:nokey=1 -sexagesimal "{vidFile265}"')
                    vidFile265Length = float(get_first_line_from_command(f'{FFPROBE} -v error -show_entries format=duration -of default=noprint_wrappers=1:nokey=1 "{vidFile265}"'))
                except ValueError:
                    # we probably have a bad conversion here somwhere, we should be able to read this info about our new conversion
                    print(f"Bad encode, error getting video length from {vidFile265} (F)")
                    with open(poorTranscodeLogPath, "a") as pTLO:
                        pTLO.write(vidFile265 + '\n')
                    os.remove(vidFile265)
                    continue

                try:
                    vidFileLengthString = get_first_line_from_command(f'{FFPROBE} -v error -show_entries format=duration -of default=noprint_wrappers=1:nokey=1 -sexagesimal "{vidFile}"')
                    vidFileLength = float(get_first_line_from_command(f'{FFPROBE} -v error -show_entries format=duration -of default=noprint_wrappers=1:nokey=1 "{vidFile}"'))
                except ValueError:
                    # we probably have a bad conversion here somwhere
                    print(f"Bad source, error getting video length from {vidFile}, check file in player (G)")
                    with open(poorTranscodeLogPath, "a") as pTLO:
                        pTLO.write(vidFile265 + '\n')
                    continue

                if abs(vidFile265Length - vidFileLength) > MAXSECONDSSHIFT:
                    print(f"Bad encode, orginal video length: {vidFileLengthString.split('.')[0]}, {vidFile265} vidFile265Length: {vidFile265LengthString.split('.')[0]}, more than max diff {MAXSECONDSSHIFT} (H)")
                    with open(poorTranscodeLogPath, "a") as pTLO:
                        pTLO.write(vidFile265 + '\n')
                    os.remove(vidFile265)
                    continue
                else:
                    print(f"Saved {convert_bytes(vidFilesSizeDiff)} ({int(100*(vidFilesSizeDiff)/float(vidFileSize))}%) by converting to {vidFile265}")
                    # os.remove(vidFile)
                    if REPLACEIFMP4ORMKV:
                        shutil.move(vidFile265, vidFile)
                        print(f"Also moved {vidFile265} to {vidFile}")

This script uses Don Melton's scripts to do the transcoding, but will try to transcode any video files beneath the current working directory when invoked.

Initially it's set to use hardware acceleration from Apple's VideoToolbox but you should configure this for whatever acceleration your system supports.

Someday I'd like to add more CLI argument parsing for more options but for right most adjustable values are coded into the script (in ALLCAPS).