dotfiles/scripts/compressPictureMovies

#!/usr/bin/env python3

import os
import shutil
import subprocess
import sys

# Constants
PICTURES_FOLDER = os.path.join(os.path.expanduser("~"), "Images")
ORIGNAL_FOLDER = os.path.join(PICTURES_FOLDER, ".Originaux")
MOVIE_EXTENSIONS = ["mov", "avi", "mp4"]
OUTPUT_EXTENSION = "mp4"
OUTPUT_FFMPEG_PARAMETERS = ["-codec:v", "libx265", "-crf", "28", "-preset:v", "slower", "-codec:a", "libfdk_aac", "-movflags", "+faststart", "-vbr", "5"]
OUTPUT_METADATA_FIELD = ["episode_id"]


# Walk folders
for root, dirs, files in os.walk(PICTURES_FOLDER):
    # If folder is in ORIGINAL_FOLDER, skip it
    if root.startswith(ORIGNAL_FOLDER):
        continue
    # Iterate over files
    for inputName in files:
        # If the file is not a video, skip it
        inputNameBase, inputExt = os.path.splitext(inputName)
        inputExt = inputExt[1:].lower()
        if inputExt not in MOVIE_EXTENSIONS:
            continue

        # Generates all needed filepaths
        ## Found file
        inputFull = os.path.join(root, inputName)
        inputRel = os.path.relpath(inputFull, PICTURES_FOLDER)
        ## Original file
        originalFull = os.path.join(ORIGNAL_FOLDER, inputRel)
        originalRel = inputRel
        ## Compressed file
        outputFull = os.path.join(root, inputNameBase + "." + OUTPUT_EXTENSION)

        # If the extension is the same of the output one
        if inputExt == OUTPUT_EXTENSION:
            # Read the metadata of the video
            metadataRaw = subprocess.run(["ffmpeg", "-i", inputFull, "-f", "ffmetadata", "-"], stdout=subprocess.PIPE).stdout
            # If it has the field with the original file 
            originalRel = None
            wantedPattern = OUTPUT_METADATA_FIELD.encode() + b"="
            for metadataLine in metadataRaw.split('\n'):
                if metadataLine.startswith(wantedPattern):
                    originalRel = metadataLine[len(wantedPattern)+1:]
                    break
            if originalRel:
                # If the original file does not exists, warn about it
                originalFull = os.path.join(ORIGNAL_FOLDER, originalRel)
                if not os.path.isfile(originalFull):
                    print("WARN {inputRel} states to have {originalRel} as original but this file doesn't exist".format(inputRel=inputRel, originalRel=originalRel))
                # If the original is not aligned with the compressed, warn about it (TODO move it automatically)
                if inputRel != originalRel:
                    print("WARN {inputRel} is not aligned with original {originalRel}".format(inputRel=inputRel, originalRel=originalRel))
                # Skip file
                continue
            # Initiate a conversion in a temporary file
            # If the temporary file does not have the same caracteristics as the original
                # Warn about it
                # Delete it
                # Skip file
            # Move the original to the corresponding original folder
            # Move the converted file in place of the original

# TODO Iterate over the orignal folder to find non-matching compressed videos not found in the above pass

sys.exit(0)

# Constants
SOURCE_FOLDER = os.path.join(os.path.expanduser("~"), "Musique")
OUTPUT_FOLDER = os.path.join(os.path.expanduser("~"), ".MusiqueCompressed")
CONVERSIONS = {"flac": "m4a"}
FORBIDDEN_EXTENSIONS = ["jpg", "pdf", "ffs_db"]
FORGIVEN_FILENAMES = ["cover.jpg"]
IGNORED_EMPTY_FOLDER = [".stfolder"]



# Listing files
sourceFiles = dict()
for root, dirs, files in os.walk(SOURCE_FOLDER):
    for f in files:
        fullPath = os.path.join(root, f)
        path = os.path.relpath(fullPath, SOURCE_FOLDER)
        sourceFiles[path] = os.path.getctime(fullPath)

outputFiles = dict()
for root, dirs, files in os.walk(OUTPUT_FOLDER):
    for f in files:
        fullPath = os.path.join(root, f)
        path = os.path.relpath(fullPath, OUTPUT_FOLDER)
        outputFiles[path] = os.path.getctime(fullPath)

# Sorting files
remainingConversions = dict()
extraFiles = list(outputFiles.keys())

def convertPath(path):
    filename, extension = os.path.splitext(path)
    extension = extension[1:].lower()
    # If the extension isn't allowed
    if extension in FORBIDDEN_EXTENSIONS:
        basename = os.path.basename(path)
        # And the filename is not an exception
        if basename not in FORGIVEN_FILENAMES:
            # This file shouldn't be copied nor converted
            return False
    # If this needs a conversion
    elif extension in CONVERSIONS:
        extension = CONVERSIONS[extension]
        return filename + "." + extension
    # In all other case, this is a simple copy
    return path

for sourceFile in sourceFiles:
    outputFile = convertPath(sourceFile)
    # If the file should not be converted, do nothing
    if outputFile == False:
        continue
    # If the file already has something as an output
    elif outputFile in outputFiles:
        extraFiles.remove(outputFile)
        # If the output file is newer than the source file, do not initiate a conversion
        if outputFiles[outputFile] >= sourceFiles[sourceFile]:
            continue
    # If the file needs to be converted, do it
    remainingConversions[sourceFile] = outputFile

# Converting files
for sourceFile in remainingConversions:
    outputFile = remainingConversions[sourceFile]

    # Creating folder if it doesn't exists
    fullOutputFile = os.path.join(OUTPUT_FOLDER, outputFile)
    fullOutputDir = os.path.dirname(fullOutputFile)
    os.makedirs(fullOutputDir, exist_ok=True)

    # Converting
    fullSourceFile = os.path.join(SOURCE_FOLDER, sourceFile)
    print(fullSourceFile, "→", fullOutputFile)
    if sourceFile == outputFile:
        # shutil.copy(fullSourceFile, fullOutputFile)
        os.link(fullSourceFile, fullOutputFile)
    else:
        subprocess.run(["ffmpeg", "-y", "-i", fullSourceFile, "-codec:a", "libfdk_aac", "-cutoff", "18000", "-movflags", "+faststart", "-vbr", "5", fullOutputFile])

# Removing extra files
for extraFile in extraFiles:
    fullExtraFile = os.path.join(OUTPUT_FOLDER, extraFile)
    os.remove(fullExtraFile)

# Removing empty dirs
for root, dirs, files in os.walk(OUTPUT_FOLDER):
    if not dirs and not files:
        dirBasename = os.path.basename(root)
        if dirBasename not in IGNORED_EMPTY_FOLDER:
            os.rmdir(root)