def regex_parse(r, config: "SessionConfig", op: ExtendedOp) -> SpacyPattern:
if op.ignore_case(config):
d = {"LOWER": {"REGEX": r.lower()}}
else:
d = {"TEXT": {"REGEX": r}}
if not op.empty():
d["OP"] = op.value
yield d
def any_of_parse(lst, config: "SessionConfig", op: ExtendedOp) -> SpacyPattern:
if op.ignore_case(config):
normalized = sorted([item.lower() for item in lst])
base = {"LOWER": {"REGEX": r"^({0})$".format("|".join(normalized))}}
else:
base = {"TEXT": {"REGEX": r"^({0})$".format("|".join(sorted(lst)))}}
if not op.empty():
base["OP"] = op.value
yield base
def apply_operator(syntax, op: ExtendedOp) -> str:
if op.empty():
return syntax
elif str(op) == "!": # A bit complicated one
return (r"((?!{})\w+)".format(syntax
.rstrip(")")
.lstrip("(")))
else:
return syntax + str(op)
def generic_parse(tag, value, config: "SessionConfig",
op: ExtendedOp) -> SpacyPattern:
d = {}
if tag == "ORTH" and op.ignore_case(config):
d["LOWER"] = value.lower()
else:
d[tag] = value
if not op.empty():
d["OP"] = op.value
yield d
def tag_parse(values, config: "SessionConfig", op: ExtendedOp) -> SpacyPattern:
"""
For generating POS/TAG patterns based on a Regex
e.g. TAG("^NN|^JJ") for adjectives or nouns
also deals with TAG_WORD for tag and word or tag and list
"""
d = {"TAG": {"REGEX": values["tag"]}}
if "word" in values:
if op.ignore_case(config):
d["LOWER"] = values["word"].lower()
else:
d["TEXT"] = values["word"]
elif "list" in values:
lst = values["list"]
if op.ignore_case(config):
normalized = sorted([item.lower() for item in lst])
d["LOWER"] = {"REGEX": r"^({0})$".format("|".join(normalized))}
else:
d["TEXT"] = {"REGEX": r"^({0})$".format("|".join(sorted(lst)))}
if not op.empty():
d["OP"] = op.value
yield d
def punct_parse(_, config: "SessionConfig", op: ExtendedOp) -> SpacyPattern:
d = dict()
d["IS_PUNCT"] = True
if not op.empty():
d["OP"] = op.value
yield d
def fuzzy_parse(r, config: "SessionConfig", op: ExtendedOp) -> SpacyPattern:
# TODO: build premutations
d = {"LOWER": {"REGEX": "({0})[.,?;!]?".format("|".join(r))}}
if not op.empty():
d["OP"] = op.value
yield d
def orth_parse(value, config: "SessionConfig", op: ExtendedOp) -> SpacyPattern:
d = {}
d["ORTH"] = value
if not op.empty():
d["OP"] = op.value
yield d