def check_if_tag_is_valid(self, original_tag, formatted_tag, previous_original_tag='', previous_formatted_tag=''):
"""Reports a validation error if the tag provided is not a valid tag or doesn't take a value.
Parameters
----------
original_tag: string
The original tag that is used to report the error.
formatted_tag: string
The tag that is used to do the validation.
previous_original_tag: string
The previous original tag that is used to report the error.
previous_formatted_tag: string
The previous tag that is used to do the validation.
Returns
-------
string
A validation error string. If no errors are found then an empty string is returned.
"""
validation_error = '';
is_extension_tag = self.is_extension_allowed_tag(formatted_tag);
if self._hed_dictionary_dictionaries[TagValidator.TAG_DICTIONARY_KEY].get(formatted_tag) or \
self.tag_takes_value(formatted_tag) or formatted_tag == TagValidator.TILDE:
pass;
elif is_extension_tag:
pass;
elif not is_extension_tag and self.tag_takes_value(previous_formatted_tag) and TagValidator.COMMA in previous_formatted_tag:
validation_error = error_reporter.report_error_type(TagValidator.COMMA_VALID_ERROR_TYPE,
tag=original_tag,
previous_tag=previous_original_tag);
self._increment_issue_count();
elif not is_extension_tag:
validation_error = error_reporter.report_error_type(TagValidator.VALID_ERROR_TYPE, tag=original_tag);
self._increment_issue_count();
return validation_error;
def check_if_duplicate_tags_exist(self, original_tag_list, formatted_tag_list):
"""Reports a validation error if two or more tags are the same.
Parameters
----------
original_tag_list: list
A list containing tags that are used to report the error.
formatted_tag_list: list
A list containing tags that are used to do the validation.
Returns
-------
string
A validation error string. If no errors are found then an empty string is returned.
"""
validation_error = '';
duplicate_indices = set([]);
for tag_index, tag in enumerate(formatted_tag_list):
all_indices = range(len(formatted_tag_list));
for duplicate_index in all_indices:
if tag_index == duplicate_index:
continue;
if formatted_tag_list[tag_index] != TagValidator.TILDE and \
formatted_tag_list[tag_index] == formatted_tag_list[duplicate_index] and \
tag_index not in duplicate_indices and duplicate_index not in duplicate_indices:
duplicate_indices.add(tag_index);
duplicate_indices.add(duplicate_index);
validation_error += error_reporter.report_error_type(TagValidator.DUPLICATE_ERROR_TYPE,
tag=original_tag_list[tag_index]);
self._increment_issue_count();
return validation_error;
def check_if_multiple_unique_tags_exist(self, original_tag_list, formatted_tag_list):
"""Reports a validation error if two or more tags start with a tag prefix that has the 'unique' attribute.
Parameters
----------
original_tag_list: list
A list containing tags that are used to report the error.
formatted_tag_list: list
A list containing tags that are used to do the validation.
Returns
-------
string
A validation error string. If no errors are found then an empty string is returned.
"""
validation_error = '';
unique_tag_prefixes = self._hed_dictionary_dictionaries[TagValidator.UNIQUE_ERROR_TYPE];
for unique_tag_prefix in unique_tag_prefixes:
unique_tag_prefix_boolean_mask = [x.startswith(unique_tag_prefix) for x in formatted_tag_list];
if sum(unique_tag_prefix_boolean_mask) > 1:
validation_error += error_reporter.report_error_type(
TagValidator.UNIQUE_ERROR_TYPE,
tag_prefix=self._hed_dictionary_dictionaries[TagValidator.UNIQUE_ERROR_TYPE][unique_tag_prefix]);
self._increment_issue_count();
return validation_error;
def check_if_tag_unit_class_units_are_valid(self, original_tag, formatted_tag):
"""Reports a validation error if the tag provided has a unit class and the units are incorrect.
Parameters
----------
original_tag: string
The original tag that is used to report the error.
formatted_tag: string
The tag that is used to do the validation.
Returns
-------
string
A validation error string. If no errors are found then an empty string is returned.
"""
validation_error = '';
if not self.tag_is_valid(formatted_tag) and self.is_unit_class_tag(formatted_tag):
tag_unit_classes = self.get_tag_unit_classes(formatted_tag);
tag_unit_values = self.get_tag_name(formatted_tag);
tag_unit_class_units = tuple(self.get_tag_unit_class_units(formatted_tag));
if TagValidator.TIME_UNIT_CLASS in tag_unit_classes and TagValidator.is_hh_mm_time(tag_unit_values):
pass;
elif re.search(TagValidator.DIGIT_EXPRESSION,
TagValidator.strip_off_units_if_valid(tag_unit_values, tag_unit_class_units)):
pass
else:
validation_error = error_reporter.report_error_type('unitClass', tag=original_tag,
unit_class_units=','.join(tag_unit_class_units));
self._increment_issue_count();
return validation_error;
def report_invalid_character_error(error_tag):
"""Reports a error that is related with an invalid character
Parameters
----------
error_tag: string
The tag that caused the error.
Returns
-------
string
The error message associated with a missing comma.
"""
error_tag = error_tag.strip();
return error_reporter.report_error_type(TagValidator.CHARACTER_ERROR_TYPE, tag=error_tag);
def report_missing_comma_error(error_tag):
"""Reports a error that is related with a missing comma.
Parameters
----------
error_tag: string
The tag that caused the error.
Returns
-------
string
The error message associated with a missing comma.
"""
error_tag = error_tag[:-1].strip();
return error_reporter.report_error_type(TagValidator.COMMA_ERROR_TYPE, tag=error_tag);
def generate_row_issue_message(row_number, has_headers=True):
"""Generates a row issue message that is associated with a particular row in a spreadsheet.
Parameters
----------
row_number: integer
The row number that the issue is associated with.
Returns
-------
string
The row issue message.
"""
if has_headers:
row_number += 1;
return error_reporter.report_error_type('row', error_row=row_number);
def check_number_of_group_tildes(self, tag_group):
"""Reports a validation error if the tag group has too many tildes.
Parameters
----------
tag_group: list
A list containing the tags in a group.
Returns
-------
string
A validation error string. If no errors are found then an empty string is returned.
"""
validation_error = '';
if tag_group.count('~') > 2:
validation_error = error_reporter.report_error_type(TagValidator.TILDE_ERROR_TYPE, tag_group);
self._increment_issue_count();
return validation_error;
def find_comma_issues_in_hed_string(self, hed_string):
"""Reports a validation error if there are missing commas or commas in tags that take values.
Parameters
----------
hed_string: string
A hed string.
Returns
-------
string
A validation error string. If no errors are found then an empty string is returned.
"""
validation_error = '';
current_tag = '';
last_non_empty_character = '';
character_indices_iterator = iter(range(len(hed_string)));
for character_index in character_indices_iterator:
character = hed_string[character_index];
current_tag += character;
if not character.isspace():
if TagValidator.character_is_delimiter(character):
current_tag = '';
if character == TagValidator.OPENING_GROUP_BRACKET:
# If we have an opening group bracket by itself without a tag, it's actually starting a new group.
if current_tag.strip() == TagValidator.OPENING_GROUP_BRACKET:
current_tag = ''
else:
validation_error = error_reporter.report_error_type(TagValidator.VALID_ERROR_TYPE,
tag=current_tag)
self._increment_issue_count()
break
elif TagValidator.comma_is_missing_before_opening_bracket(last_non_empty_character, character):
validation_error = TagValidator.report_missing_comma_error(current_tag);
self._increment_issue_count();
break;
elif TagValidator.comma_is_missing_after_closing_bracket(last_non_empty_character, character):
validation_error = TagValidator.report_missing_comma_error(current_tag);
self._increment_issue_count();
break;
last_non_empty_character = character;
return validation_error;
def check_if_tag_requires_child(self, original_tag, formatted_tag):
"""Reports a validation error if the tag provided has the 'requireChild' attribute.
Parameters
----------
original_tag: string
The original tag that is used to report the error.
formatted_tag: string
The tag that is used to do the validation.
Returns
-------
string
A validation error string. If no errors are found then an empty string is returned.
"""
validation_error = '';
if self._hed_dictionary_dictionaries[TagValidator.REQUIRE_CHILD_ERROR_TYPE].get(formatted_tag):
validation_error = error_reporter.report_error_type(TagValidator.REQUIRE_CHILD_ERROR_TYPE,
tag=original_tag);
self._increment_issue_count();
return validation_error;
def count_tag_group_brackets(self, hed_string):
"""Reports a validation error if there are an unequal number of opening or closing parentheses. This is the
first check before the tags are parsed.
Parameters
----------
hed_string: string
A hed string.
Returns
-------
string
A validation error string. If no errors are found then an empty string is returned.
"""
validation_error = '';
number_of_opening_brackets = hed_string.count('(');
number_of_closing_brackets = hed_string.count(')');
if number_of_opening_brackets != number_of_closing_brackets:
validation_error = error_reporter.report_error_type(TagValidator.BRACKET_ERROR_TYPE,
opening_bracket_count=number_of_opening_brackets,
closing_bracket_count=number_of_closing_brackets);
self._increment_issue_count();
return validation_error;
def test_report_error_type(self):
for error_type in self.error_types:
error_report = error_reporter.report_error_type(error_type)
self.assertIsInstance(error_report, str)
