edm_pset_tweak.py draft

#!/usr/bin/env python
# usage
# edm_pset_tweak.py --input_pkl RunPromptRecoCfg.pkl --output_pkl ouput.pkl --json tweaks.json
#
# tweaks.json can be
#  [{ "process.maxEvents.input" : 100}]
# or
# { "process.maxEvents.input" : 100}
# or
# [{ "maxEvents.input" : 100}]
# or
# { "maxEvents.input" : 100}
# 
# properly handles multiple pkls and mulitple jsons (all jsons are applied to all pkls)
#

import pickle
import argparse
import sys
import json

def apply_tweak(process, key, value):
    key_split = key.split('.')
    param = process
    if key_split[0] == "process":
        key_split = key_split[1:]
    for p in key_split:
        param = getattr(param, p)
        if param is None:
            return 1
    param = value
    return 0


def init_argparse():
    parser = argparse.ArgumentParser(
        usage="%(prog)s [OPTION] [FILE]...",
        description="Tweak pset(s) based on json list of tweaks to apply and write out resulting pset(s)"
    )
    parser.add_argument('--json', nargs='+', required=True)
    parser.add_argument('--input_pkl', nargs='+', required=True)
    parser.add_argument('--output_pkl', nargs='+', required=False)
    parser.add_argument('--allow_failed_tweaks', action='store_true')
    return parser


def main():
    parser = init_argparse()
    args = parser.parse_args()
    if len(args.output_pkl) != 0:
        output_file_names = args.output_pkl
        if len(args.input_pkl) != len(args.output_pkl):
            print(
                "Incorrect arguments. --input_pkl and --output_pkl must be the same length")
            sys.exit(1)
    else:
        output_file_names=args.input_pkl

    tweaks = []
    for json_file_name in args.json:
        try:
            with open(json_file_name) as json_file:
                json_data = json.load(json_file)
        except Exception as e:
            print("Error opening file "+json_file_name)
            sys.exit(1)

        if not isinstance(json_data,(dict,list)):
            print("Error loading json "+json_file_name)
            sys.exit(1)

        if isinstance(json_data, list):
            for d in json_data:
                for key in d:
                    tweaks.append((key, d[key]))
        elif isinstance(json_data, dict):
            for key in json_data:
                tweaks.append((key, json_data[key]))
        else:
            print("Error loading json "+json_file_name)
            sys.exit(1)

    for i in range(len(args.input_pkl)):
        with open(args.input_pkl[i], "rb") as pkl_file:
            try:
                process = pickle.load(pkl_file)
            except (IOError, OSError, pickle.PickleError, pickle.UnpicklingError):
                print("Not a valid pickle file " +
                      args.input_pkl[i]) + "- stopping"
                sys.exit(1)

        for tweak in tweaks:
            err_val = apply_tweak(process, tweak[0], tweak[1])
            if err_val != 0:
                print("Tweak not applied "+tweak[0]+" "+str(tweak[1]))
                if not args.allow_failed_tweaks:
                    sys.exit(1)
            else:
                print("Tweak applied "+tweak[0]+" "+str(tweak[1]))

        with open(output_file_names[i], "wb") as output_file:
            if output_file.closed:
                print("Error loading pickle input "+args.output_pkl[i])
                sys.exit(1)
            pickle.dump(process, output_file, protocol=0)


main()

both

[{"param1": value1}, {"param2": value2}, etc etc

and

{"param1": value1, "param2": value2, etc}

are supported.

'''
In central production, we are supposed to invoke this tweak once per cmsRun process. For StepChain workflows, we would invoke the pset tweak multiple times during runtime, but still once per cmsRun step (cmsRun1, cmsRun2, etc).
'''

is there some action needed by the script? It doesn't sound like it.

I revised the tool given comments above. I didn't yet consider the case where --output_pkl is not specified. I think that is easy to do.

updated for --output_pkl and isinstance
(still mostly untested)

I believe there is a little mistake here:

if isinstance(json_data,list) and isinstance(json_data,dict):
    print("Error loading json "+json_file_name)
    sys.exit(1)

It should probably be (the negation):

if not isinstance(json_data, (list, dict)):
    print("Error loading json " + json_file_name)
    sys.exit(1)

Or even if you bring back elif isinstance(json_data, dict), error can be moved to else:

if isinstance(json_data, list):
    for d in json_data:
        for key in d:
            tweaks.append((key, d[key]))
elif isinstance(json_data, dict):
    for key in json_data:
        tweaks.append((key, json_data[key]))
else:
    print("Error loading json "+json_file_name)
    sys.exit(1)

But this is not necessary, just might save one isinstance.

definitely a bug.. I'll take your version

now to write some tests for all the different cases..