Python script for fixing duplicates in minipro's device database.

fix_dupes.py

#!/usr/bin/env python

# This script parses devices.h and prints a copy to stdout in which
# all duplicate entries are either removed, or renamed with a number
# appended if they have different contents.

import re
import sys

def log(msg):
    sys.stderr.write("{}\n".format(msg))

def parse_structs():
    with open("devices.h") as f:
        return f.read().split("},")[:-1]

name_pattern = re.compile(r'(name\s*=\s*)"([^"]+)"')

def main():
    name_structs = {}
    all_structs = parse_structs()
    log("found {} structs".format(len(all_structs)))
    for s in all_structs:
        m = name_pattern.search(s)
        name = m.group(2)
        structs = name_structs.setdefault(name, [])
        occurence = len(structs) + 1
        if s in structs:
            log("redundant: {}".format(name))
        elif occurence > 1:
            log("duplicate name: {} (occurence {})".format(name, occurence))
            s_ = name_pattern.sub(r'\1"\2 #{}"'.format(occurence), s)
            sys.stdout.write(s_ + "},")
            structs.append(s)
        else:
            sys.stdout.write(s + "},")
            structs.append(s)

if __name__ == "__main__":
    main()