Function commandline

pattern/text/__init__.py:1046–1088

(parse=Parser().parse)

# python -m pattern.en.parser xml -s "Hello, my name is Dr. Sbaitso. Nice to meet you." -OTCLI

def commandline(parse=Parser().parse):
    import optparse
    import codecs
    p = optparse.OptionParser()
    p.add_option("-f", "--file",      dest="file",      action="store",      help="text file to parse",   metavar="FILE")
    p.add_option("-s", "--string",    dest="string",    action="store",      help="text string to parse", metavar="STRING")
    p.add_option("-O", "--tokenize",  dest="tokenize",  action="store_true", help="tokenize the input")
    p.add_option("-T", "--tags",      dest="tags",      action="store_true", help="parse part-of-speech tags")
    p.add_option("-C", "--chunks",    dest="chunks",    action="store_true", help="parse chunk tags")
    p.add_option("-R", "--relations", dest="relations", action="store_true", help="find verb/predicate relations")
    p.add_option("-L", "--lemmata",   dest="lemmata",   action="store_true", help="find word lemmata")
    p.add_option("-e", "--encoding",  dest="encoding",  action="store_true", help="character encoding", default="utf-8")
    p.add_option("-v", "--version",   dest="version",   action="store_true", help="version info")
    o, arguments = p.parse_args()
    # Version info.
    if o.version:
        sys.path.insert(0, os.path.join(MODULE, "..", ".."))
        from pattern import __version__
        print __version__
        sys.path.pop(0)
    # Either a text file (-f) or a text string (-s) must be supplied.
    s = o.file and codecs.open(o.file, "r", o.encoding).read() or o.string
    # The given text can be parsed in two modes: 
    # - implicit: parse everything (tokenize, tag/chunk, find relations, lemmatize).
    # - explicit: define what to parse manually.
    if s:
        explicit = False
        for option in [o.tokenize, o.tags, o.chunks, o.relations, o.lemmata]:
            if option is not None: explicit=True; break
        if not explicit:
            a = {"encoding": o.encoding }
        else:
            a = {"tokenize": o.tokenize  or False,
                     "tags": o.tags      or False,
                   "chunks": o.chunks    or False,
                "relations": o.relations or False,
                  "lemmata": o.lemmata   or False,
                 "encoding": o.encoding }
        s = parse(s, **a)
        # The output can be either slash-formatted string or XML.
        if "xml" in arguments:
            s = Tree(s, s.tags).xml
        print s

#### VERBS #########################################################################################