Function to rewrite index files generated by PE Harmony V6 to be compatible with https://github.com/broadinstitute/pe2loaddata

rewrite_harmonyV6_index_file.py

import os
import xml.etree.ElementTree as ET

def rewrite_harmonyV6_index_file(index_path):
    """Fix Harmony V6 index file to be compatible with pe2loaddata
    Warning: This will rewrite the index file in place!
    Dependencies: os, xml.etree.ElementTree as ET

    Parameters
        ----------
    index_path: str
        Path to Harmony V6 index file

    Returns
    -------
    """

    namespaces = dict([node for _, node in ET.iterparse(index_path, events=['start-ns'])])

    assert "HarmonyV6" in namespaces[''], "Index file does not have a HarmonyV6 namespace!"

    for key, value in namespaces.items():
        ET.register_namespace(key, value)

    tree = ET.parse(index_path)
    xmlroot = tree.getroot()
    maps = xmlroot.find("Maps", namespaces=namespaces)[-1] # Last map seems to have Channel information
    images = xmlroot.find("Images", namespaces=namespaces)

    for image in images:
        if len(image.findall("ChannelName", namespaces=namespaces)) == 0: # Check that no channel name exists
            for child in maps: # Find map with matching ChannelID
                if (image.find("ChannelID", namespaces=namespaces).text == child.get('ChannelID')): # Stop if the Channel ID in the image matches the map entry
                    for sub in child: # Loop through every entry in the map and copy it to the image
                        temp = ET.SubElement(image, sub.tag, attrib=sub.attrib)
                        temp.text = sub.text
                    break # No need to continue looping through the map entries

    # Save the updated xml
    ET.indent(tree, '  ')
    tree.write(index_path, encoding="utf-8", xml_declaration=False)
    
    
if __name__ == "__main__":
    rewrite_harmonyV6_index_file(index_path="path/to/Index.idx.xml")