Opens an fcpxml (from Apple FCP X) and modifies the asset elements to contain enough audio information to be used properly by any audio elements

fixfcpxml.py

#! /usr/bin/env python

import logging

def get_assets_dict(asset_elements):
        #Create dictionary
        asset_elements_dict = {}
        #Add each asset to dictionary with id as key
        for asset_element in asset_elements:
                asset_id = asset_element.get("id")
                asset_elements_dict[asset_id] = asset_element
        return asset_elements_dict

def fix_asset(asset_element, audio_element):
        logging.debug("Processing audio element that references asset " + asset_element.get("id") + " (" + asset_element.get("name") + ")")
        
        # Add or update hasAudio attribute
        if not asset_element.get("hasAudio") or asset_element.get("hasAudio") == "0":
            logging.info("Changed asset " + asset_element.get("id") + " (" + asset_element.get("name") + ") to have audio")
            asset_element.set("hasAudio", "1")

        #Get highest audio channel used by audio element
        if audio_element.get("srcCh"):
                #Get the last channel number in the comma separated list
                used_channels = audio_element.get("srcCh").split(", ")
                highest_used_channel = int(used_channels[-1])
        else:
                #If no attribute, channel must be 1
                highest_used_channel = 1

        #Get number of channels currently in asset
        if asset_element.get("audioChannels"):
                asset_audio_channels = int(asset_element.get("audioChannels"))
        else:
                asset_audio_channels = 0

        if asset_audio_channels < highest_used_channel:
                #Update audioChannels attribute
                asset_element.set("audioChannels", str(highest_used_channel))
                logging.info("Changed asset " + asset_element.get("id") + " (" + asset_element.get("name") + ") to have " + asset_element.get("audioChannels") + " channels")

        #Get audio id used by audio element
        if audio_element.get("srcID"):
                used_id = int(audio_element.get("srcID"))
        else:
                #If no attribute, id must be 1
                used_id = 1

        #Get number of ids currently in asset
        if asset_element.get("audioSources"):
                asset_audio_ids = int(asset_element.get("audioSources"))
        else:
                asset_audio_ids = 0

        if asset_audio_ids < used_id:
                #Update audioSources attribute
                asset_element.set("audioSources", str(used_id))
                logging.info("Changed asset " + asset_element.get("id") + " (" + asset_element.get("name") + ") to have " + asset_element.get("audioSources") + " sources")


def main():
        import argparse, os, codecs, sys

        from xml.etree.ElementTree import parse        #Set up argument parser
        parser = argparse.ArgumentParser(description="Fix an invalid fcpxml",
                                         epilog="Latest version available at https://gist.github.com/Grezzo/ed62fe2292373f14d17d")
        parser.add_argument("input", metavar='SOURCE', help="an invalid fcpxml file")
        parser.add_argument("-o", "--output", help="the name of the fcpxml file to create")
        parser.add_argument("-v", "--verbose", help="print verbose logging information", action="store_true")
        parser.add_argument("-d", "--debug", help="print debug logging information", action="store_true")
        args = parser.parse_args()

        #If output omitted, set output to input so it overwrites source
        if not args.output:
                args.output = args.input

        #Turn on logging if specified
        if args.debug:
                logging.getLogger().setLevel("DEBUG")
        elif args.verbose:
                logging.getLogger().setLevel("INFO")

        #Check that input file exists
        if not os.path.isfile(args.input):
                sys.exit(os.path.basename(__file__) + ": error: " + args.input + " does not exist")
                
        #Set logging format
        logging.basicConfig(format="%(levelname)s:%(message)s")

        #Parse fcpxml and catch any errors
        try:
            tree = parse(args.input)
            root = tree.getroot()
            #Create a parent map for each element so audio elements can be removed if they have no ref
            parent_map = {c:p for p in tree.iter() for c in p}
        except:
            sys.exit(os.path.basename(__file__) + ": error: " + args.input + " cannot be parsed")
            
        #Check if it is really an fcpxml
        if root.tag != "fcpxml":
            sys.exit(os.path.basename(__file__) + ": error: " + args.input + " is not an fcpxml")

        #Check if destination folder exists
        if not os.path.exists(os.path.dirname(args.output)):
            sys.exit(os.path.basename(__file__) + ": error: " + os.path.dirname(args.output) + " does not exist")

        #Add assets to dictionary so they can be looked up by id
        assets_dict = get_assets_dict(root.findall("resources/asset"))

        #Loop through each audio element
        audios_list = root.findall(".//audio")
        for audio_element in audios_list:
                ref = audio_element.get("ref")
                #If audio element has no ref, it is invalid
                if not ref:
                    #Audio element is invalid so remove it
                    logging.info("Removed an audio element because it has no ref attribute")
                    parent = parent_map[audio_element]
                    parent.remove(audio_element)
                else:
                    #Update asset element to contain audio info referenced by audio element
                    fix_asset(assets_dict[ref], audio_element)

        #Write out new fcpxml file
        tree.write(args.output)

if __name__ == "__main__":
        main()