示例#1
0
def main(args):
    goldF = args.goldfile
    sysFs = args.sysfile

    ss_mapper = lambda ss: coarsen_pss(ss, args.depth) if ss.startswith(
        'p.') else ss

    # Load gold data
    gold_sents = list(load_sents(goldF, ss_mapper=ss_mapper))
    for sent in gold_sents:
        sent['punits'] = {
            tuple(e['toknums']): (e['lexcat'], e['ss'], e['ss2'])
            for e in list(sent['swes'].values()) + list(sent['smwes'].values())
            if e['ss'] and (e['ss'].startswith('p.') or e['ss'] == '??')
        }

    all_sys_scores = {}
    for sysF in sysFs:
        sysscores = eval_sys(sysF, gold_sents, ss_mapper)
        syspath = sysF.name
        basename = syspath.rsplit('.', 2)[0]
        if basename not in all_sys_scores:
            all_sys_scores[basename] = [
                defaultdict(lambda: defaultdict(Counter)),
                defaultdict(lambda: defaultdict(Counter))
            ]
        if syspath.split('.')[-2] == 'goldid':
            all_sys_scores[basename][0] = sysscores
        else:
            all_sys_scores[basename][1] = sysscores

    # Print output
    args.output_format(all_sys_scores, depth=args.depth)
示例#2
0
def main(args):
    goldF = args.goldfile
    sysFs = args.sysfile

    ss_mapper = lambda ss: coarsen_pss(ss, args.depth) if ss.startswith(
        'p.') else ss

    # Load gold data
    gold_sents = list(load_sents(goldF, ss_mapper=ss_mapper))

    all_sys_scores = {}
    for sysF in sysFs:
        sysscores = eval_sys(sysF, gold_sents, ss_mapper)
        syspath = sysF.name
        basename = syspath.rsplit('.', 2)[0]
        if basename not in all_sys_scores:
            all_sys_scores[basename] = [
                defaultdict(lambda: defaultdict(Counter)),
                defaultdict(lambda: defaultdict(Counter))
            ]
        if syspath.split('.')[-2] == 'goldid':
            all_sys_scores[basename][0] = sysscores
        else:
            all_sys_scores[basename][1] = sysscores

    # Print output
    args.output_format(all_sys_scores, depth=args.depth, mode=args.output_mode)
示例#3
0
 def __call__(self, ss):
     return coarsen_pss(ss, self.depth) if ss.startswith('p.') else ss
示例#4
0
def ss_mapper(ss):
    return coarsen_pss(ss, DEPTH) if ss.startswith('p.') else ss
示例#5
0
def main(args):
    if args.colorless or not sys.stdin.isatty():
        for c in dir(Colors):
            if not c.startswith('_'):
                setattr(Colors, c, '')
        for s in dir(Styles):
            if not s.startswith('_'):
                setattr(Styles, s, '')


    goldF = args.goldfile
    sysFs = args.sysfile

    ss_mapper = lambda ss: coarsen_pss(ss, args.depth) if ss.startswith('p.') else ss

    # Load gold data
    gold_sents = list(load_sents(goldF, ss_mapper=ss_mapper))

    predFs = [load_sents(predFP, ss_mapper=ss_mapper) for predFP in sysFs]

    all_sys_scores = {}

    def filter_labels(ll):
        result = dict(ll)
        for k,l in ll.items():
            if l.startswith('n.') and args.no_noun: del result[k]
            elif l.startswith('v.') and args.no_verb: del result[k]
            elif l.startswith('p.') and args.no_snacs: del result[k]
        return result

    R = lambda ww,sg,wg,ll: render(ww, sg if not args.no_mwe else [], wg if not args.no_mwe else [], filter_labels(ll))

    for i,sent in enumerate(gold_sents):
        # gold analysis
        words = [t["word"] for t in sent["toks"]]
        rendered = []
        rendered.append(R(words,
                           [e["toknums"] for e in sent["smwes"].values()],
                           [e["toknums"] for e in sent["wmwes"].values()],
                           makelabelmap(sent, include_lexcat=args.lexcats, include_supersenses=True)))
        for predF in predFs:
            psent = next(predF)
            assert psent['sent_id']==sent['sent_id']
            rendered.append(R(words,
                               [e["toknums"] for e in psent["smwes"].values()],
                               [e["toknums"] for e in psent["wmwes"].values()],
                               makelabelmap(sent, include_lexcat=args.lexcats, include_supersenses=True)))

        diff_classes = set()
        if not args.no_diff:
            diff_classes.add('special')
            if not args.no_mwe_diff: diff_classes.add('mwe')
            if not args.no_noun_diff: diff_classes.add('n')
            if not args.no_snacs_diff: diff_classes.add('p')
            if not args.no_verb_diff: diff_classes.add('v')

        if args.sent_ids:
            print(sent['sent_id'], end='\t')
        print(color_rendered(words, rendered, diff_classes))
        #assert False,(color_rendered(words, rendered),words,rendered)

    # restore the terminal's default colors
    print(Colors.ENDC, end='')
def ss_mapper(supersense):
    return coarsen_pss(supersense,
                       DEPTH) if supersense.startswith('p.') else supersense