generic/conversion_utils
16
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Renamed internal fields

Browse Source
This commit is contained in:
Cyprian Laskowski 2021-09-10 15:21:55 +02:00
parent 2f7f90d073
commit c9cfb46314
1 changed files with 35 additions and 35 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

70
conversion_utils/jos_msds_and_properties.py
View File

@ -57,11 +57,11 @@ class Specifications:
def add_category(self, category): def add_category(self, category):
self.categories.append(category) self.categories.append(category)
def find_category_by_char(self, char, language): def find_category_by_code(self, char, language):
return next((category for category in self.categories if category.char_pair.get(language) == char), None) return next((category for category in self.categories if category.codes.get(language) == char), None)
def find_category_by_name(self, name, language): def find_category_by_name(self, name, language):
return next((category for category in self.categories if category.string_pair.get(language) == name), None) return next((category for category in self.categories if category.names.get(language) == name), None)
def __str__(self): def __str__(self):
return 'categories:{categories}'.format(categories=self.categories) return 'categories:{categories}'.format(categories=self.categories)
@ -82,9 +82,9 @@ class Pair:
class Category: class Category:
def __init__(self, string_pair, char_pair, *features): def __init__(self, names, codes, *features):
self.string_pair = string_pair self.names = names
self.char_pair = char_pair self.codes = codes
self.features = list(features) self.features = list(features)
def add_feature(self, feature): def add_feature(self, feature):
@ -94,17 +94,17 @@ class Category:
return next((feature for feature in self.features if feature.position == position), None) return next((feature for feature in self.features if feature.position == position), None)
def find_feature_by_name(self, name, language): def find_feature_by_name(self, name, language):
return next((feature for feature in self.features if feature.string_pair.get(language) == name), None) return next((feature for feature in self.features if feature.names.get(language) == name), None)
def __str__(self): def __str__(self):
return 'strings:{strings}, chars:{chars}, features:{features}'.\ return 'names:{names}, codes:{codes}, features:{features}'.\
format(strings=self.string_pair, chars=self.char_pair, features=self.features) format(strings=self.names, chars=self.codes, features=self.features)
class Feature: class Feature:
def __init__(self, string_pair, position, lexeme_level_flag, *values): def __init__(self, names, position, lexeme_level_flag, *values):
self.string_pair = string_pair self.names = names
self.position = position self.position = position
self.lexeme_level_flag = lexeme_level_flag self.lexeme_level_flag = lexeme_level_flag
self.values = list(values) self.values = list(values)
@ -113,25 +113,25 @@ class Feature:
self.values.append(value) self.values.append(value)
def find_value_by_char(self, char, language): def find_value_by_char(self, char, language):
return next((value for value in self.values if value.char_pair.get(language) == char), None) return next((value for value in self.values if value.codes.get(language) == char), None)
def find_value_by_name(self, name, language): def find_value_by_name(self, name, language):
return next((value for value in self.values if value.string_pair.get(language) == name), None) return next((value for value in self.values if value.names.get(language) == name), None)
def __str__(self): def __str__(self):
return 'strings:{strings}, position:{position}, level:{level}, values:{values}'.\ return 'names:{names}, position:{position}, level:{level}, values:{values}'.\
format(strings=self.string_pair, position=self.position, level='level' if self.lexeme_level_flag else 'form', values=self.values) format(strings=self.names, position=self.position, level='level' if self.lexeme_level_flag else 'form', values=self.values)
class Value: class Value:
def __init__(self, string_pair, char_pair): def __init__(self, names, codes):
self.char_pair = char_pair self.codes = codes
self.string_pair = string_pair self.names = names
def __str__(self): def __str__(self):
return 'chars:{chars}, strings:{strings}'.\ return 'codes:{codes}, names:{names}'.\
format(chars=self.char_pair, strings=self.strings_pair) format(codes=self.codes, names=self.names)
class SpecificationsBuilder: class SpecificationsBuilder:
@ -143,22 +143,22 @@ class SpecificationsBuilder:
for div_element in div_elements: for div_element in div_elements:
if (re.match(r'^msd\..-sl', get_xml_id(div_element))): if (re.match(r'^msd\..-sl', get_xml_id(div_element))):
category_element = xpath_find(div_element, 'tei:table/tei:row[@role="type"]')[0] category_element = xpath_find(div_element, 'tei:table/tei:row[@role="type"]')[0]
category_string_pair = self.get_cell_pair(category_element, 'value') category_names = self.get_cell_pair(category_element, 'value')
category_char_pair = self.get_cell_pair(category_element, 'code') category_codes = self.get_cell_pair(category_element, 'code')
category = Category(category_string_pair, category_char_pair) category = Category(category_names, category_codes)
specifications.add_category(category) specifications.add_category(category)
feature_elements = xpath_find(div_element, 'tei:table/tei:row[@role="attribute"]') feature_elements = xpath_find(div_element, 'tei:table/tei:row[@role="attribute"]')
for feature_element in feature_elements: for feature_element in feature_elements:
feature_string_pair = self.get_cell_pair(feature_element, 'name') feature_names = self.get_cell_pair(feature_element, 'name')
feature_position = int(self.get_cell(feature_element, 'position')) feature_position = int(self.get_cell(feature_element, 'position'))
lexeme_level_flag = feature_position in LEXEME_FEATURE_MAP[category_string_pair.get('en')] lexeme_level_flag = feature_position in LEXEME_FEATURE_MAP[category_names.get('en')]
feature = Feature(feature_string_pair, feature_position, lexeme_level_flag) feature = Feature(feature_names, feature_position, lexeme_level_flag)
category.add_feature(feature) category.add_feature(feature)
value_elements = xpath_find(feature_element, 'tei:cell[@role="values"]/tei:table/tei:row[@role="value"]') value_elements = xpath_find(feature_element, 'tei:cell[@role="values"]/tei:table/tei:row[@role="value"]')
for value_element in value_elements: for value_element in value_elements:
value_char_pair = self.get_cell_pair(value_element, 'name') value_codes = self.get_cell_pair(value_element, 'name')
value_string_pair = self.get_cell_pair(value_element, 'code') value_names = self.get_cell_pair(value_element, 'code')
value = Value(value_char_pair, value_string_pair) value = Value(value_codes, value_names)
feature.add_value(value) feature.add_value(value)
return specifications return specifications
@ -204,8 +204,8 @@ class Converter:
def msd_to_properties(self, msd, lemma, properties_language): def msd_to_properties(self, msd, lemma, properties_language):
category_char = msd.code[0].lower() category_char = msd.code[0].lower()
value_chars = msd.code[1:] value_chars = msd.code[1:]
category = self.specifications.find_category_by_char(category_char, msd.language) category = self.specifications.find_category_by_code(category_char, msd.language)
category_name = category.string_pair.get(properties_language) category_name = category.names.get(properties_language)
feature_value_list = [] feature_value_list = []
lexeme_feature_map = {} lexeme_feature_map = {}
form_feature_map = {} form_feature_map = {}
@ -213,8 +213,8 @@ class Converter:
if (value_char != '-'): if (value_char != '-'):
feature = category.find_feature_by_position(index) feature = category.find_feature_by_position(index)
value = feature.find_value_by_char(value_char, msd.language) value = feature.find_value_by_char(value_char, msd.language)
feature_name = feature.string_pair.get(properties_language) feature_name = feature.names.get(properties_language)
feature_value = value.string_pair.get(properties_language) feature_value = value.names.get(properties_language)
level_exception_flag = (category_name, feature.position, lemma) in LEVEL_EXCEPTIONS level_exception_flag = (category_name, feature.position, lemma) in LEVEL_EXCEPTIONS
lexeme_level_flag = feature.lexeme_level_flag if not level_exception_flag else not feature.lexeme_level_flag lexeme_level_flag = feature.lexeme_level_flag if not level_exception_flag else not feature.lexeme_level_flag
feature_value_list.append((feature, value)) feature_value_list.append((feature, value))
@ -226,14 +226,14 @@ class Converter:
def properties_to_msd(self, properties, msd_language): def properties_to_msd(self, properties, msd_language):
category = self.specifications.find_category_by_name(properties.category, properties.language) category = self.specifications.find_category_by_name(properties.category, properties.language)
category_char = category.char_pair.get(msd_language).upper() category_char = category.codes.get(msd_language).upper()
feature_map = properties.lexeme_feature_map.copy() feature_map = properties.lexeme_feature_map.copy()
feature_map.update(properties.form_feature_map.copy()) feature_map.update(properties.form_feature_map.copy())
position_map = {} position_map = {}
for (name, value) in feature_map.items(): for (name, value) in feature_map.items():
feature = category.find_feature_by_name(name, properties.language) feature = category.find_feature_by_name(name, properties.language)
value = feature.find_value_by_name(value, properties.language) value = feature.find_value_by_name(value, properties.language)
position_map[feature.position] = value.char_pair.get(msd_language) position_map[feature.position] = value.codes.get(msd_language)
msd_code = category_char msd_code = category_char
i = 0 i = 0
for position in sorted(position_map.keys()): for position in sorted(position_map.keys()):