def process_kwargs(self, mapping):
''' Apply kwargs in the order they are provided. kwargs are ordered as of python 3.6
'''
# Add defaults
if 'as_is' not in self.kwargs:
self.kwargs['as_is'] = False
if 'case_sensitive' not in self.kwargs:
self.kwargs['case_sensitive'] = True
if 'escape_special' not in self.kwargs:
self.kwargs['escape_special'] = False
if 'norm_form' not in self.kwargs:
self.kwargs['norm_form'] = 'NFD'
if 'reverse' not in self.kwargs:
self.kwargs['reverse'] = False
# Process kwargs in order received
for kwarg, val in self.kwargs.items():
if kwarg == 'as_is' and not val:
# sort by reverse len
mapping = sorted(mapping,
key=lambda x: len(x["in"]),
reverse=True)
elif kwarg == 'escape_special' and val:
mapping = [escape_special_characters(x) for x in mapping]
elif kwarg == 'case_sensitive' and not val:
mapping = self.lower_mappings(mapping)
elif kwarg == 'norm_form' and val:
for io in mapping:
for k, v in io.items():
if isinstance(v, str):
io[k] = normalize(v, self.kwargs['norm_form'])
elif kwarg == 'reverse' and val:
mapping = self.reverse_mappings(mapping)
# After all processing is done, turn into regex
for io in mapping:
io['match_pattern'] = self.rule_to_regex(io)
self.processed = True
return mapping
def process_kwargs(self, mapping):
''' Apply kwargs in the order they are provided. kwargs are ordered as of python 3.6
'''
if 'as_is' in self.kwargs:
as_is = self.kwargs['as_is']
if as_is:
appropriate_setting = "as-written"
else:
appropriate_setting = "apply-longest-first"
self.kwargs["rule_ordering"] = appropriate_setting
LOGGER.warning(
f"mapping from {self.kwargs.get('in_lang')} to {self.kwargs.get('out_lang')} "
'is using the deprecated parameter "as_is"; '
f"replace `as_is: {as_is}` with `rule_ordering: {appropriate_setting}`"
)
# Add defaults
if 'rule_ordering' in self.kwargs:
# right now, "rule-ordering" is a more explict alias of the "as-is" option.
ordering = self.kwargs["rule_ordering"]
if ordering not in ("as-written", "apply-longest-first"):
LOGGER.error(
f"mapping from {self.kwargs.get('in_lang')} to {self.kwargs.get('out_lang')} "
f"has invalid value '{ordering}' for rule_ordering parameter; "
"rule_ordering must be one of "
'"as-written" or "apply-longest-first"')
else:
self.kwargs["rule_ordering"] = "as-written"
if 'case_sensitive' not in self.kwargs:
self.kwargs['case_sensitive'] = True
if 'escape_special' not in self.kwargs:
self.kwargs['escape_special'] = False
if 'norm_form' not in self.kwargs:
self.kwargs['norm_form'] = 'NFD'
if 'reverse' not in self.kwargs:
self.kwargs['reverse'] = False
if 'prevent_feeding' not in self.kwargs:
self.kwargs['prevent_feeding'] = False
if 'in_lang' not in self.kwargs:
self.kwargs['in_lang'] = 'und'
if 'out_lang' not in self.kwargs:
self.kwargs['out_lang'] = 'und'
# Process kwargs in order received
for kwarg, val in self.kwargs.items():
if kwarg == 'rule_ordering' and self.wants_rules_sorted():
# sort by reverse len
mapping = sorted(mapping,
key=lambda x: len(x["in"]),
reverse=True)
elif kwarg == 'escape_special' and val:
mapping = [escape_special_characters(x) for x in mapping]
elif kwarg == 'norm_form' and val:
for io in mapping:
for k, v in io.items():
if isinstance(v, str):
io[k] = normalize(v, self.kwargs['norm_form'])
elif kwarg == 'reverse' and val:
mapping = self.reverse_mappings(mapping)
# After all processing is done, turn into regex
for io in mapping:
if self.kwargs['prevent_feeding'] or ('prevent_feeding' in io
and io['prevent_feeding']):
io['intermediate_form'] = self._string_to_pua(
io['out'], mapping.index(io))
io['match_pattern'] = self.rule_to_regex(io)
if not io['match_pattern']:
mapping.remove(io)
self.processed = True
return mapping
def test_escape_special(self):
self.assertEqual(utils.escape_special_characters({'in': '?'}), {'in': '\?'})
def process_kwargs(self, mapping):
""" Apply kwargs in the order they are provided. kwargs are ordered as of python 3.6
"""
if "as_is" in self.kwargs:
as_is = self.kwargs["as_is"]
if as_is:
appropriate_setting = "as-written"
else:
appropriate_setting = "apply-longest-first"
self.kwargs["rule_ordering"] = appropriate_setting
del self.kwargs["as_is"]
LOGGER.warning(
f"mapping from {self.kwargs.get('in_lang')} to {self.kwargs.get('out_lang')} "
'is using the deprecated parameter "as_is"; '
f"replace `as_is: {as_is}` with `rule_ordering: {appropriate_setting}`"
)
# Add defaults
if "rule_ordering" in self.kwargs:
# right now, "rule-ordering" is a more explict alias of the "as-is" option.
ordering = self.kwargs["rule_ordering"]
if ordering not in ("as-written", "apply-longest-first"):
LOGGER.error(
f"mapping from {self.kwargs.get('in_lang')} to {self.kwargs.get('out_lang')} "
f"has invalid value '{ordering}' for rule_ordering parameter; "
"rule_ordering must be one of "
'"as-written" or "apply-longest-first"')
else:
self.kwargs["rule_ordering"] = "as-written"
if "case_sensitive" not in self.kwargs:
self.kwargs["case_sensitive"] = True
if "escape_special" not in self.kwargs:
self.kwargs["escape_special"] = False
if "norm_form" not in self.kwargs:
self.kwargs["norm_form"] = "NFD"
if "reverse" not in self.kwargs:
self.kwargs["reverse"] = False
if "prevent_feeding" not in self.kwargs:
self.kwargs["prevent_feeding"] = False
if "in_lang" not in self.kwargs:
self.kwargs["in_lang"] = "und"
if "out_lang" not in self.kwargs:
self.kwargs["out_lang"] = "und"
# Process kwargs in order received
for kwarg, val in self.kwargs.items():
if kwarg == "rule_ordering" and self.wants_rules_sorted():
# sort by reverse len
mapping = sorted(mapping,
key=lambda x: len(x["in"]),
reverse=True)
elif kwarg == "escape_special" and val:
mapping = [escape_special_characters(x) for x in mapping]
elif kwarg == "norm_form" and val:
for io in mapping:
for k, v in io.items():
if isinstance(v, str):
io[k] = normalize(v, self.kwargs["norm_form"])
elif kwarg == "reverse" and val:
mapping = self.reverse_mappings(mapping)
# After all processing is done, turn into regex
for i, io in enumerate(mapping):
if self.kwargs["prevent_feeding"] or ("prevent_feeding" in io
and io["prevent_feeding"]):
io["intermediate_form"] = self._string_to_pua(io["out"], i)
io["match_pattern"] = self.rule_to_regex(io)
# Finally, remove rules with an empty match pattern, typically empty rules
mapping = [io for io in mapping if io["match_pattern"]]
self.processed = True
return mapping