示例#1
0
文件: conll.py 项目: hsensoy/uparse
    elif args.directory:
        if not args.section:
            sections = ["%02d" % s for s in range(0, 25)]
        elif "-" in args.section:
            begin_section, end_section = [int(s) for s in args.section.split("-")]
            sections = ["%02d" % s for s in range(begin_section, end_section + 1)]
        elif "," in args.section:
            sections = ["%02d" % int(s) for s in args.section]
        else:
            sections = ["%02d" % int(args.section)]

        import os

        if args.subsmode != "nochange":
            for subs, (conll, section) in izip(
                subssentiter(file=args.subsfile), ConLLiter(directory=args.directory, extension=args.extension)
            ):
                if section in sections:
                    if not any([c == s for c, s in zip(conll.sentence(), subs.orginal())]):
                        sys.stderr.write(repr(subs))
                        sys.stderr.write("\n")
                        sys.stderr.write(repr(conll))
                        sys.stderr.write("\n")

                        break
                    else:
                        for word, substitution in zip(conll, subs):
                            if args.subsmode == "best":
                                word._form = substitution.best()

                            if args.tagmode == "nochange":