def test_http():
ts = TextSearch("ignore", "norm")
ts.add_http_handler(keep_result=True)
assert ts.findall("http://google.com") == ["http://google.com"]
def test_add_dict():
ts = TextSearch("smart", "norm")
ts.add({"hi": "greeting", "bye": "bye", "goodbye": "bye"})
assert ts.findall("hi bye goodbye") == ["greeting", "bye", "bye"]
def test_serializable():
ts = TextSearch("sensitive", dict)
ts.add("hi")
result = ts.findall("hi")
assert result
assert json.dumps(result)
def test_ignore_match():
ts = TextSearch("ignore", "match")
ts.add("hi")
assert ts.findall("hi") == ["hi"]
assert ts.findall("HI") == ["hi"]
assert ts.findall("asdf") == []
def test_add_list():
ts = TextSearch("smart", "match")
ts.add(["hi", "bye", "hello"])
assert ts.findall("hi bye hello") == ["hi", "bye", "hello"]
def test_fast_no_bounds():
ts = TextSearch("sensitive", "match", set(), set())
ts.add("hi")
assert ts.findall("asdfhiadsfs")
def test_sensitive_match():
ts = TextSearch("sensitive", "object")
ts.add("hi")
assert ts.findall("hi")
assert not ts.findall("HI")
def test_insensitive_object():
ts = TextSearch("insensitive", "object")
ts.add("hi")
assert ts.findall("HI")[0].end == 2
def test_regex_norm():
ts = TextSearch("insensitive", "norm")
ts.add_regex_handler(["last "], r"\d", keep_result=True)
assert ts.findall("last 5") == ["last 5"]
def test_not_overlap_3():
ts = TextSearch("ignore", "norm")
ts.add("a")
ts.add("a a")
assert ts.findall("a a a") == ["a a", "a"]
def test_postfix_regex():
ts = TextSearch("ignore", "norm")
ts.add_regex_handler(["products"], r"\d+ ", keep_result=True, prefix=False)
assert ts.findall("90 products") == ["90 products"]
def test_not_overlap():
ts = TextSearch("ignore", "norm")
ts.add("http://")
ts.add_http_handler(True)
assert len(ts.findall("https://vks.ai")) == 1
def test_insensitive_match():
ts = TextSearch("insensitive", "match")
ts.add("hi")
assert ts.findall("HI") == ["HI"]
def test_right_bounds():
ts = TextSearch("sensitive", "match")
ts.add("hi")
assert not ts.findall("hiasf")
def test_http_no_keep():
ts = TextSearch("ignore", "norm")
ts.add_http_handler(keep_result=False)
ts.add("google")
assert ts.findall("http://google.com") == []
def test_regex_overlap():
ts = TextSearch("insensitive", "object")
ts.add_regex_handler(["last "], r"\d", keep_result=True)
ts.add("last")
assert ts.findall("last 5")[0].norm == "last 5"
def test_twitter():
ts = TextSearch("ignore", "norm")
ts.add_twitter_handler(keep_result=True)
assert ts.findall("@hello") == ["@hello"]
assert ts.findall("#hello") == ["#hello"]
def test_ignore_norm():
ts = TextSearch("ignore", "norm")
ts.add("hi", "HI")
assert ts.findall("hi") == ["HI"]
assert ts.findall("asdf") == []