add script to check for duplicates in .prm files

from collections import Counter

from pathlib import Path

import sys

import pprint

HERE = Path(__file__).parent

CONFIGS_DIR = HERE.joinpath("config")

PRM_FILES = [

    CONFIGS_DIR.joinpath("self_test.prm"),

    CONFIGS_DIR.joinpath("self_test_ltv.prm"),

    CONFIGS_DIR.joinpath("self_test_basop_encoder.prm"),

    CONFIGS_DIR.joinpath("self_test_ltv_basop_encoder.prm"),

]

def main() -> bool:

    duplicates_found = False

    for prm_file in PRM_FILES:

        with open(prm_file) as f:

            contents = f.read()

        # first check if there are complete duplicates (i.e. comment line, encoder cmd and decoder cmd are the same)

        blocks = contents.split("\n\n")

        blocks = [b for b in blocks if len(b.strip()) > 0]

        blocks_counter = Counter(blocks)

        blocks_duplicates = [

            b.split("\n")[0] for b in blocks_counter if blocks_counter[b] > 1

        ]

        # also check if there are duplicates in the command names only

        lines = contents.split("\n")

        commands_lines = [l for l in lines if l.strip().startswith("//")]

        commands_counter = Counter(commands_lines)

        commands_duplicates = [l for l in commands_counter if commands_counter[l] > 1]

        if len(blocks_duplicates) > 0:

            duplicates_found = True

            print(f"Duplicate testcases in {prm_file.name} found:")

            pprint.pp(blocks_duplicates)

        if len(commands_duplicates) > 0:

            duplicates_found = True

            print(f"Duplicate command names in {prm_file.name} found:")

            pprint.pp(commands_duplicates)

        print()

    return duplicates_found

if __name__ == "__main__":

    sys.exit(int(main()))