Script to remove specific module(s) results from antiSMASH results JSON

remove_results.py

#!/usr/bin/env python

"""
Removes specific module(s) results from antiSMASH results JSON 

Modules are expected to be named as per a python `__name__` result,
e.g. "antismash.modules.clusterblast"
"""


from typing import Set

import json
import sys


def remove_module_results(filename: str, targets: Set[str]):
    """ Removes each target module from the given results JSON"""
    with open(filename) as handle:
        data = json.load(handle)
    removed = set()
    for record in data["records"]:
        record_modules = record.get("modules", [])
        for module in targets:
            if module in record_modules:
                removed.add(module)
                del record_modules[module]
    new_content = json.dumps(data)
    with open(filename, "w") as handle:
        handle.write(new_content)
    for module in targets.difference(removed):
        print("no results existed for:", module)

if __name__ == "__main__":
    if len(sys.argv) < 3:
        print("Usage: %s results_file target [target ...]" % sys.argv[0])
        sys.exit(1)
    remove_module_results(sys.argv[1], set(sys.argv[2:]))