def run(self, key, ctx:Context):
import fnmatch, re
if '/' in key:
lemma=key.split('/')[-1] # the key is formatted like 'word/lemma'
else:
lemma=ctx.lemmas[key]
if self.match_method=='equals':
return lemma==self.target
elif self.match_method=='in':
return lemma in self.target
elif self.match_method=='chunk':
if isinstance(self.target, list):
for t in self.target:
if t in ctx.chunk_pieces(key, lowercase=True):
return True
return False
else:
return self.target in ctx.chunk_pieces(key, lowercase=True)
elif self.match_method=='glob':
regex = fnmatch.translate(self.target)
reobj = re.compile(regex)
return reobj.match(lemma) is not None
elif self.match_method=='regex':
reobj = re.compile(self.target)
return reobj.match(lemma) is not None
else:
raise ValueError(f"Cannot support match method {self.match_method}")
def expand(dispathcer: DispatcherIntf, data, keys, specific_domains):
fixt = InspectorFixture()
domains, meta = fixt.request_domains(data)
ctx = Context(meta, domains)
for key in keys:
for chunk in ctx.chunk_pieces(key):
dispathcer.execute(chunk)
def run(self, key, ctx:Context):
checkers = []
lang = ctx.meta['lang']
# cnt = ' '.join(ctx.chunks['obl'])
# cnt = ' '.join(ctx.chunks[key])
if self.entire:
checkers.append(self.providers[self.provider](key, lang, ctx, 'sents'))
else:
for cnt in ctx.chunk_pieces(key):
checkers.append(self.providers[self.provider](cnt, lang, ctx, key))
# print('... put %s'%self.cache_key(key))
# print(ctx.meta['intermedia'])
return any(checkers)
def run(self, key, ctx:Context):
result = False
lang = ctx.meta['lang']
# cnt = ' '.join(ctx.chunks[key])
# cnt=ctx.get_single_chunk_text(key)
requestors={'ru':lambda rc: query_entities_by_url(cf.ensure('ner_ru'), rc),
}
for cnt in ctx.chunk_pieces(key):
data={'lang': lang, 'sents': cnt}
if lang in requestors:
resp=requestors[lang](data)
else:
resp = query_entities(data)
if resp['result'] == 'success':
dims = [d['entity'] for d in resp['data']]
# print('entities ->', ', '.join(dims))
logger.info('entities -> %s, self.dim -> %s', ', '.join(dims), self.dim)
if self.dim in dims:
print('\t%s ∈' % cnt, self.dim)
result = True
return result
def run(self, key, ctx: Context):
# 当pickup为'_'时, key就是value
comp_val = key if self.pickup == '_' else ''
key = self.pickup or key
ex_map = {
'date_search':
lambda cnt, comp: ex_date_search(key, cnt, comp, ctx),
# .. extract_for('plain+date_search+date_parse', '時間'),
'date_parse':
lambda cnt, comp: ex_date_parse(key, cnt, comp, ctx),
'plain':
lambda cnt, comp: ex_plain(key, cnt, comp, ctx),
'word':
lambda cnt, comp: ex_word(key, cnt, comp, ctx),
# .. extract_for('plain+translit', 'obj'),
'translit':
lambda cnt, comp: ex_translit(key, cnt, comp, ctx),
'email':
lambda cnt, comp: ex_dims(key, cnt, comp, ctx, 'email'),
# .. extract_for('number', 'obl'),
'number':
lambda cnt, comp: ex_dims(key, cnt, comp, ctx, 'number'),
# .. extract_for('time', 'advmod'),
'time':
lambda cnt, comp: ex_dims(key, cnt, comp, ctx, 'time'),
# .. extract_for('plain+temperature', 'ニ'),
'temperature':
lambda cnt, comp: ex_dims(key, cnt, comp, ctx, 'temperature'),
# example: extract_for('rasa', '_')
'rasa':
lambda cnt, comp: ex_rasa(key, cnt, comp, ctx),
# example: extract_for('chunk', 'verb:xcomp/obj')
'chunk':
lambda cnt, comp: ex_chunk(key, cnt, comp, ctx, lambda w:
(w.text, w.upos.lower())),
# example: extract_for('chunk_text', 'verb:xcomp/obj')
'chunk_text':
lambda cnt, comp: ex_chunk(key, cnt, comp, ctx, lambda w: w.text),
'chunk_feats':
lambda cnt, comp: ex_chunk(key, cnt, comp, ctx, lambda w: w.feats),
# .. extract_for('feats', 'verb:_'),
# extract_for('feats', 'verb:obj')
'feats':
lambda cnt, comp: ex_feats(key, cnt, comp, ctx),
# example: extract_for('ner', '_'), extract_for('ner', 'xcomp')
'ner':
lambda cnt, comp: ex_ner(key, cnt, comp, ctx),
}
if self.pickup == '_' or is_full_domain_path(self.pickup):
self.results['_'] = []
for comp in self.comp_as:
op = ex_map[comp](comp_val, comp)
self.results['_'].append((comp, op))
else:
for cnt in ctx.chunk_pieces(key):
self.results[key] = []
for comp in self.comp_as:
ex = ex_map[comp]
op = ex(cnt, comp)
# self.results[comp] = op
self.results[key].append((comp, op))
return True # 只负责提取, 并不参与判定, 所以始终返回True