def standalone_engine(rules, **kwargs):
parser = rita.compile_string(rules, use_engine="standalone", **kwargs)
print(parser.patterns)
def parse(text):
results = list(parser.execute(text, include_submatches=False))
return list([(r["text"], r["label"]) for r in results])
return parse
def test_save_and_load_rules_from_file(self):
rules = '''
{WORD("Hello"), WORD("world")}->MARK("HELLO")
'''
engine = rita.compile_string(rules, use_engine="standalone")
with tempfile.TemporaryDirectory() as tmpdir:
rules_path = os.path.join(tmpdir, "rules-example.json")
engine.save(rules_path)
engine.load(rules_path)
engine.execute("Hello world")
def spacy_engine(rules, **kwargs):
spacy = pytest.importorskip("spacy", minversion="2.1")
nlp = spacy.load("en_core_web_sm")
setup_spacy(nlp, rules_string=rules, override_ents=True, **kwargs)
patterns = rita.compile_string(rules, **kwargs)
print(patterns)
def parse(text):
doc = nlp(text)
return list([(e.text, e.label_) for e in doc.ents])
return parse
def rust_engine(rules, **kwargs):
from rita.engine.translate_rust import load_lib
lib = load_lib()
if lib is None:
pytest.skip("Missing rita-rust dynamic lib, skipping related tests")
print("Trying to run: {}".format(rules))
parser = rita.compile_string(rules, use_engine="rust", **kwargs)
print(parser.patterns)
def parse(text):
results = list(parser.execute(text, include_submatches=False))
return list([(r["text"], r["label"]) for r in results])
return parse
def _spacy_v3(model, patterns=None, rules_path=None, rules_string=None, override_ents=True):
ruler = model.add_pipe("entity_ruler", config={"overwrite_ents": override_ents, "validate": True})
if not patterns:
if rules_path:
patterns = rita.compile(rules_path, use_engine="spacy")
elif rules_string:
patterns = rita.compile_string(rules_string, use_engine="spacy")
else:
raise RuntimeError("Please provides rules. Either `patterns`, `rules_path` or `rules_string`")
ruler.add_patterns(patterns)
else:
ruler.from_disk(patterns)
return model
def spacy_engine(rules, **kwargs):
spacy = pytest.importorskip("spacy", minversion="2.1")
patterns = rita.compile_string(rules, **kwargs)
nlp = spacy.load("en")
ruler = spacy.pipeline.EntityRuler(nlp, overwrite_ents=True)
print(patterns)
ruler.add_patterns(patterns)
nlp.add_pipe(ruler)
def parse(text):
doc = nlp(text)
return list([(e.text, e.label_) for e in doc.ents])
return parse
def _spacy_v2(model, patterns=None, rules_path=None, rules_string=None, override_ents=True):
from spacy.pipeline import EntityRuler
ruler = EntityRuler(model, overwrite_ents=override_ents)
if not patterns:
if rules_path:
patterns = rita.compile(rules_path, use_engine="spacy")
elif rules_string:
patterns = rita.compile_string(rules_string, use_engine="spacy")
else:
raise RuntimeError("Please provides rules. Either `patterns`, `rules_path` or `rules_string`")
ruler.add_patterns(patterns)
else:
ruler.from_disk(patterns)
model.add_pipe(ruler)
return model
def test_compile_context(engine):
if engine == "rust":
from rita.engine.translate_rust import load_lib
lib = load_lib()
if lib is None:
pytest.skip(
"Missing rita-rust dynamic lib, skipping related tests")
rules = """
{WORD*, IN_LIST(companies), WORD*}->MARK("SUSPISCIOUS_COMPANY")
"""
parser = rita.compile_string(rules,
use_engine=engine,
companies=["CompanyA", "CompanyB"])
print(parser.patterns)
results = list(parser.execute("CompanyB is doing it's dirty work."))
assert results[0] == {
"start":
0,
"end":
33,
"label":
"SUSPISCIOUS_COMPANY",
"text":
"CompanyB is doing it's dirty work",
"submatches": [{
"start": 0,
"end": 33,
"key": "SUSPISCIOUS_COMPANY",
"text": "CompanyB is doing it's dirty work"
}, {
"start": 0,
"end": 9,
"key": "s2",
"text": "CompanyB"
}, {
"start": 9,
"end": 33,
"key": "s4",
"text": "is doing it's dirty work"
}],
}
def compiler(self, rules):
return rita.compile_string(rules, use_engine="standalone").patterns
def compiler(self, rules):
pytest.importorskip("spacy", minversion="2.1")
return rita.compile_string(rules, use_engine="spacy")