def vis_doc(sents, lang):
from sagas.nlu.ruleset_procs import cached_chunks
from sagas.nlu.uni_remote_viz import list_contrast, display_doc_deps
from sagas.conf.conf import cf
chunks = cached_chunks(sents, lang, cf.engine(lang))
return display_doc_deps(chunks['doc'], None)
def get_domains(sents, lang, engine='corenlp', options=None):
"""
>>> from sagas.nlu.legacy.aiobj_kit import get_domains
>>> get_domains('你有几台笔记本电脑？', 'zh', 'ltp')
>>> get_domains('列出上周编辑的文件。', 'zh', 'ltp', DomainGetOptions(enable_predicts=True))
:param sents:
:param lang:
:param engine:
:param options:
:return:
"""
# from IPython.display import display
if options is None:
options=DomainGetOptions()
pipelines=['predicts'] if options.enable_predicts else []
doc_jsonify, resp = dep_parse(sents, lang, engine, pipelines)
result_set=[]
if doc_jsonify is not None:
tc.emp('cyan', resp)
if resp is not None and 'predicts' in resp and len(resp['predicts'])>0:
rs=resp['predicts']
# print(rs)
else:
# print(doc_jsonify.words_string())
rs = get_chunks(doc_jsonify)
if len(rs)>0:
if options.list_chunks:
list_rs(rs, lang)
if options.deps_graph:
# display(display_doc_deps(doc_jsonify, resp))
tc.gv(display_doc_deps(doc_jsonify, resp,
translit_lang=lang if lang in ('ja', 'ko', 'zh', 'fa', 'ar', 'he') else None))
# rs_represent(rs, data = {'lang': lang, "sents": sents, 'engine': engine,
# 'pipelines':pipelines})
data = {'lang': lang, "sents": sents, 'engine': engine,
'pipelines':pipelines}
for r in rs:
# fixture.print_table(r, False)
# print(f"lemma: {r['lemma']}")
# df = sagas.to_df(r['domains'], ['rel', 'index', 'text', 'lemma', 'children', 'features'])
# display(df)
domains = r['domains']
common = {'lemma': r['lemma'], 'word': r['word'],
'stems': r['stems']}
meta = {'rel': r['rel'], **common, **data}
result_set.append((domains, meta))
else:
tc.emp('red', '.. no found predefined chunk-patterns.')
tc.info(doc_jsonify.words_string())
tc.info(doc_jsonify.dependencies_string())
return result_set
def parse_subj():
lang = 'ru'
sents = 'Яблоко - это здоровый фрукт.'
st.write(sents)
data = {'lang': lang, "sents": sents, 'engine': cf.engine(lang)}
doc_jsonify, resp = parse_sents(data)
domains = get_subj_domain(doc_jsonify)
testing(domains, 'ru')
gv = display_doc_deps(doc_jsonify, resp)
st.graphviz_chart(gv)
st.write(domains)
def parse_aux():
lang = 'en'
sents = 'what will be the weather in three days?'
st.write(sents)
data = {'lang': lang, "sents": sents, 'engine': cf.engine(lang)}
doc_jsonify, resp = parse_sents(data)
domains = get_aux_domain(doc_jsonify)
testing(domains, 'en')
# show analyse graph
gv = display_doc_deps(doc_jsonify, resp)
st.graphviz_chart(gv)
st.write(domains)
def parse_deps(text, lang, translit=None):
text = fix_sents(text, lang)
engine = cf.engine(lang)
# g = sentence_view(lang, text, engine=engine, translit_lang=lang, enable_contrast=True)
doc_jsonify, resp = dep_parse(text, lang, engine, ['predicts'])
if doc_jsonify is not None:
list_chunks(doc_jsonify, resp, lang, enable_contrast=True)
g = display_doc_deps(doc_jsonify, resp, translit_lang=lang)
st.graphviz_chart(g)
if translit is not None:
st.text(f"♤ {translit}")
words = [word.text for word in doc_jsonify.words]
tools.contrast(text, lang, word_map=words)
def print_sents(self, sents, lang, engine=None):
"""
$ python -m sagas.nlu.ruleset_procs print_sents 'I want to play music.' en
$ python -m sagas.nlu.ruleset_procs print_sents "クモは4つの右の目をしています。" ja corenlp
:param sents:
:param lang:
:return:
"""
# lang = 'en'
if engine is None:
engine = cf.engine(lang)
data = {'lang': lang, "sents": sents, 'engine': engine}
doc_jsonify, resp = parse_sents(data)
rs = get_chunks(doc_jsonify)
if lang in non_spaces:
delim = ''
else:
delim = ' '
for serial, r in enumerate(rs):
meta = build_meta(r, data)
domains = r['domains']
# print([(x[0], x[2]) for x in domains])
#
keys = {x[0] for x in domains}
grp = lambda p, idx: [x[idx] for x in domains if x[0] == p]
tokens = {x: grp(x, 2) for x in keys}
words = {x: delim.join(grp(x, 2)) for x in keys}
lemmas = {x: delim.join(grp(x, 3)) for x in keys}
print('meta keys', meta.keys())
print('tokens', tokens)
print('words', meta['word'], words)
print('lemmas', lemmas)
#
ctx = Context(meta, domains)
# print(ctx.lemmas)
print('chunks', ctx._chunks)
g = display_doc_deps(doc_jsonify, resp, translit_lang=lang)
print(*[(w.index, w.text, w.governor,
doc_jsonify.words[w.governor - 1].text)
for w in doc_jsonify.words],
sep='\n')
tc.gv(g)
def row_view(row):
text = row[1]
if display_translit and len(row) > 2:
label = row[2]
else:
label = text
if st.button(f"{label} ✁ {row[0]}"):
text = fix_sents(text, lang)
engine = get_engine(lang)
# g = sentence_view(lang, text, engine=engine, translit_lang=lang, enable_contrast=True)
doc_jsonify, resp = dep_parse(text, lang, engine, ['predicts'])
if doc_jsonify is not None:
list_chunks(doc_jsonify, resp, lang, enable_contrast=True)
g=display_doc_deps(doc_jsonify, resp, translit_lang=lang)
st.graphviz_chart(g)
if len(row) > 2:
st.text(f"♤ {row[2]}")
words = [word.text for word in doc_jsonify.words]
tools.contrast(text, lang, word_map=words)
def verbs(self, sents, lang='en', do_action=False):
"""
单词的wordnet匹配使用专门的ruleset来评估, 匹配成功的rule会写入状态, 比如spec_xcomp_obj='music',
此处的music对应knowledgebase的object_type.
有效成分的单词wordnet引用依次放入ruleset评估, 这样会得到一个状态集, 这个状态集会放入句子结构,
供sents_ruleset评估.
sents_ruleset会收集到多个intents保存到状态中, 遍历intents, 如果intent有action可触发,
则触发这个action.
$ python -m sagas.nlu.ruleset_procs verbs 'I want to play music.' en
```
[{'ref': 'want/want', 'upos': 'verb'},
{'ref_xcomp': 'play/play', 'upos': 'verb'},
{'ref_xcomp_obj': 'music/music', 'upos': 'noun'}]
```
$ python -m sagas.nlu.ruleset_procs verbs 'I want to play video.' en
$ python -m sagas.nlu.ruleset_procs verbs 'I would like to play video.' en
$ python -m sagas.nlu.ruleset_procs verbs "i'd like to play sound." en
$ verbs 'I want to play music.' en True
:param sents:
:param lang:
:return:
"""
import sagas.nlu.ruleset_fixtures as rf
data = {'lang': lang, "sents": sents, 'engine': cf.engine(lang)}
doc_jsonify, resp = parse_sents(data)
v_domains = get_verb_domain(doc_jsonify)
if self.verbose:
tc.gv(display_doc_deps(doc_jsonify, resp))
pprint(v_domains)
json_utils.write_json_to_file('./out/v_domain.json', v_domains[0])
# list words
tc.emp('cyan', f"✁ list words. {'-' * 25}")
intents = []
host = create_host()
for d in v_domains:
tokens = list_words(d, lang, with_chains=True)
if self.verbose:
pprint(tokens)
# specs evaluate
tc.emp('cyan', f"✁ specs evaluate. {'-' * 25}")
r3 = {}
for token in tokens:
r3 = host.assert_fact('chains', token)
pprint(r3) # the last result is all state
[r3.pop(key) for key in ['$s', 'id', 'sid']]
tc.emp('red', f"specs state - {r3}")
# sents evaluate
tc.emp('cyan', f"✁ sents evaluate. {'-' * 25}")
sents_data = {**d, **r3}
tc.emp('cyan', f" keys: {', '.join(sents_data.keys())}")
result = host.assert_fact('sents', sents_data)
tc.emp('red', f"sents state - {result}")
if 'intents' in result:
intents.extend(result['intents'])
self.process_intents(sents, lang, intents, do_action)