def if_no_such_template_redirect(template_name: str) -> Optional[RRV]:
if template_name not in templates:
return flask.render_template(
'no-such-template.html',
template_name=template_name,
)
elif isinstance(templates[template_name], str):
return flask.redirect(flask.url_for(
cast(str, flask.request.endpoint),
**(dict(cast(dict[str, Any], flask.request.view_args), template_name=templates[template_name])),
**flask.request.args.to_dict(flat=False),
), code=307)
elif isinstance(templates[template_name], list):
replacement_templates = [
templates_without_redirects[replacement_name]
for replacement_name in templates[template_name]
]
flask.g.interface_language_code = lang_lex2int(replacement_templates[0]['language_code'])
return flask.render_template(
'ambiguous-template.html',
template_name=template_name,
replacement_templates=replacement_templates,
)
else:
return None
def term_span(term: Term) -> flask.Markup:
interface_language_code = lang_lex2int(term['language'])
return (flask.Markup(r'<span lang="') +
flask.Markup.escape(lang_int2html(interface_language_code)) +
flask.Markup(r'" dir="') +
flask.Markup.escape(text_direction(interface_language_code)) +
flask.Markup(r'">') +
flask.Markup.escape(term['value']) +
flask.Markup(r'</span>'))
def process_template_edit(template_name: str, lexeme_id: str) -> RRV:
response = if_no_such_template_redirect(template_name)
if response:
return response
template = templates_without_redirects[template_name]
template_language_code = template['language_code']
flask.g.interface_language_code = lang_lex2int(template_language_code)
representation_language_code = flask.request.args.get('language_code', template_language_code)
wiki = 'test' if 'test' in template else 'www'
if flask.request.method == 'POST':
lexeme_revision = flask.request.form['_lexeme_revision']
lexeme_data = get_lexeme_data(lexeme_id, wiki, lexeme_revision)
else:
lexeme_data = get_lexeme_data(lexeme_id, wiki)
lexeme_revision = str(lexeme_data['lastrevid'])
lexeme_match = match_template_to_lexeme_data(template, lexeme_data)
lexeme_matches_template = (
lexeme_match['language'] and
lexeme_match['lexical_category'] and
not lexeme_match['conflicting_statements']
)
template = match_lexeme_forms_to_template(lexeme_data['forms'], template)
template = cast(BoundTemplate, template)
template['lexeme_id'] = lexeme_id
template['lexeme_revision'] = lexeme_revision
readonly = 'OAUTH' in app.config and 'oauth_access_token' not in flask.session
if (flask.request.method == 'POST' and
'_edit_mode' in flask.request.form and
csrf_token_matches(flask.request.form) and
not readonly):
form_data = flask.request.form
lexeme_data = update_lexeme(lexeme_data, template, form_data, representation_language_code, missing_statements=lexeme_match['missing_statements'])
summary = build_summary(template, form_data)
if 'OAUTH' in app.config:
lexeme_id, lexeme_uri = submit_lexeme(template, lexeme_data, summary)
target = add_hash_to_uri(lexeme_uri, form_data.get('target_hash'))
return flask.redirect(target, code=303)
else:
print(summary)
return flask.jsonify(lexeme_data)
for template_form in template['forms']:
template_form = cast(MatchedTemplateForm, template_form)
if lexeme_forms := template_form.get('lexeme_forms'):
template_form = cast(EditedTemplateForm, template_form)
template_form['value'] = '/'.join(lexeme_form['representations'][representation_language_code]['value']
for lexeme_form in lexeme_forms
if representation_language_code in lexeme_form['representations'])
def test_translations_available():
missing_language_codes = set()
for template in templates.templates_without_redirects.values():
language_code = lang_lex2int(template['language_code'])
if language_code not in translations.translations:
missing_language_codes.add(language_code)
# language code disabled on translatewiki.net, needs more consideration?
missing_language_codes.remove('zh')
assert not missing_language_codes
def get_duplicates_api(wiki: str, language_code: str, lemma: str) -> RRV:
flask.g.interface_language_code = lang_lex2int(language_code)
matches = get_duplicates(wiki, language_code, lemma)
if not matches:
return flask.Response(status=204)
if flask.request.accept_mimetypes.accept_html:
return render_duplicates(
matches,
in_bulk_mode=False,
template_name=flask.request.args.get('template_name'),
)
else:
return flask.jsonify(matches)
def process_template_advanced(template_name: str, advanced: bool = True) -> RRV:
response = if_no_such_template_redirect(template_name)
if response:
return response
template = templates_without_redirects[template_name]
flask.g.interface_language_code = lang_lex2int(template['language_code'])
form_data = flask.request.form # type: werkzeug.datastructures.MultiDict
readonly = 'OAUTH' in app.config and 'oauth_access_token' not in flask.session
if (flask.request.method == 'POST' and
form_data.get('_advanced_mode', 'None') == str(advanced) and
not readonly):
response = if_has_duplicates_redirect(template, advanced, form_data)
if response:
return response
response = if_needs_csrf_redirect(template, advanced, form_data)
if response:
return response
lexeme_data = build_lexeme(template, form_data)
summary = build_summary(template, form_data)
if 'OAUTH' in app.config:
lexeme_id, lexeme_uri = submit_lexeme(template, lexeme_data, summary)
target = add_hash_to_uri(lexeme_uri, form_data.get('target_hash'))
return flask.redirect(target, code=303)
else:
print(summary)
return flask.jsonify(lexeme_data)
else:
if not form_data:
form_data = flask.request.args
return flask.render_template(
'template.html',
template=add_form_data_to_template(form_data, template),
lemmas=build_lemmas(template, form_data),
lexeme_id=form_data.get('lexeme_id'),
advanced=advanced,
can_use_bulk_mode=can_use_bulk_mode(),
readonly=readonly,
)
def language_name_with_code(language_code: str) -> flask.Markup:
code_zxx = (flask.Markup(r'<span lang=zxx>') +
flask.Markup.escape(language_code) +
flask.Markup(r'</span>'))
language_name = autonym(language_code)
if language_name is None:
language_name = label(language_code)
if language_name is None:
return code_zxx
interface_language_code = lang_lex2int(language_code)
return (flask.Markup(r'<span lang="') +
flask.Markup.escape(lang_int2html(interface_language_code)) +
flask.Markup(r'" dir="') +
flask.Markup.escape(text_direction(interface_language_code)) +
flask.Markup(r'">') +
flask.Markup.escape(language_name) +
flask.Markup(r' (') +
code_zxx +
flask.Markup(r')</span>'))
def add_labels_to_lexeme_forms_grammatical_features(session, language, lexeme_forms):
grammatical_features_item_ids = set()
for lexeme_form in lexeme_forms:
grammatical_features_item_ids.update(lexeme_form['grammaticalFeatures'])
grammatical_features_item_ids = list(grammatical_features_item_ids)
labels_map = {} # item ID to label
while grammatical_features_item_ids:
chunk, grammatical_features_item_ids = grammatical_features_item_ids[:50], grammatical_features_item_ids[50:]
response = session.get(action='wbgetentities',
ids=chunk,
props=['labels'],
languages=[lang_lex2int(language)],
languagefallback=1, # TODO use True once mediawiki-utilities/python-mwapi#38 is in a released version
formatversion=2)
for item_id, item in response['entities'].items():
labels_map[item_id] = item['labels'].get(language, {'language': 'zxx', 'value': item_id})
for lexeme_form in lexeme_forms:
lexeme_form['grammaticalFeatures_labels'] = [labels_map[grammatical_feature_item_id]
for grammatical_feature_item_id in lexeme_form['grammaticalFeatures']]
def get_duplicates(wiki: str, language_code: str, lemma: str) -> list[Duplicate]:
session = anonymous_session(f'https://{wiki}.wikidata.org')
api_language_code = lang_lex2int(language_code)
response = session.get(
action='wbsearchentities',
search=lemma,
language=api_language_code,
uselang=api_language_code, # for the result descriptions
type='lexeme',
limit=50,
)
matches: dict[str, Duplicate] = {}
for result in response['search']:
if (result.get('label') == lemma and
(result['match']['language'] == language_code or
(len(language_code) > 2 and result['match']['language'] == 'und'))): # T230833
match = {
'id': result['id'],
'uri': result['concepturi'],
'label': result['label'],
'description': result['description'],
}
matches[result['id']] = cast(Duplicate, match) # missing forms_count, senses_count added below
if matches:
response = session.get( # no, this can’t be combined with the previous call by using generator=wbsearch – then we don’t get the match language
action='query',
titles=['Lexeme:' + id for id in matches],
prop=['pageprops'],
ppprop=['wbl-forms', 'wbl-senses'],
)
for page in response['query']['pages'].values():
id = page['title'][len('Lexeme:'):]
pageprops = page.get('pageprops', {})
matches[id]['forms_count'] = pageprops.get('wbl-forms')
matches[id]['senses_count'] = pageprops.get('wbl-senses')
return list(matches.values()) # list() to turn odict_values (not JSON serializable) into plain list
def render_advanced_partial_forms_hint(language_code: str) -> RRV:
flask.g.interface_language_code = lang_lex2int(language_code)
return flask.render_template(
'advanced_partial_forms_hint.html',
)
def render_no_duplicate(language_code: str) -> RRV:
flask.g.interface_language_code = lang_lex2int(language_code)
return flask.render_template(
'no_duplicate.html',
)
def process_template_bulk(template_name: str) -> RRV:
response = if_no_such_template_redirect(template_name)
if response:
return response
template = templates_without_redirects[template_name]
flask.g.interface_language_code = lang_lex2int(template['language_code'])
readonly = 'OAUTH' in app.config and 'oauth_access_token' not in flask.session
if not can_use_bulk_mode() and not readonly:
return flask.render_template(
'bulk-not-allowed.html',
)
if (flask.request.method == 'POST' and
'_bulk_mode' in flask.request.form and
csrf_token_matches(flask.request.form) and
not readonly):
form_data = flask.request.form
parse_error = None
show_optional_forms_hint = False
try:
lexemes = parse_lexemes(form_data['lexemes'], template)
except FirstFieldNotLexemeIdError as error:
parse_error = message_with_kwargs(
'bulk_first_field_not_lexeme_id',
num_forms=error.num_forms,
num_fields=error.num_fields,
first_field=error.first_field,
line_number=error.line_number,
)
except FirstFieldLexemeIdError as error:
parse_error = message_with_kwargs(
'bulk_first_field_lexeme_id',
num_forms=error.num_forms,
num_fields=error.num_fields,
first_field=error.first_field,
line_number=error.line_number,
)
except WrongNumberOfFieldsError as error:
show_optional_forms_hint = error.num_fields < error.num_forms
parse_error = message_with_kwargs(
'bulk_wrong_number_of_fields',
num_forms=error.num_forms,
num_fields=error.num_fields,
line_number=error.line_number,
)
except ValueError as error:
parse_error = str(error)
if parse_error is not None:
return flask.render_template(
'bulk.html',
template=template,
value=form_data['lexemes'],
parse_error=parse_error,
show_optional_forms_hint=show_optional_forms_hint,
)
results = [] # type: list[dict]
for lexeme in lexemes:
if not lexeme.get('lexeme_id'):
duplicates = find_duplicates(template, lexeme)
if duplicates:
results.append({
'duplicates': duplicates,
'form_representations': lexeme.getlist('form_representation'),
})
continue
lexeme_data = build_lexeme(template, lexeme)
summary = build_summary(template, form_data)
if 'OAUTH' in app.config:
lexeme_id, lexeme_uri = submit_lexeme(template, lexeme_data, summary)
results.append({
'lexeme_data': lexeme_data,
'lexeme_id': lexeme_id,
'lexeme_uri': lexeme_uri,
})
else:
print(summary)
results.append({
'lexeme_data': lexeme_data,
})
if 'OAUTH' in app.config:
return flask.render_template(
'bulk-result.html',
template=template,
results=results,
)
else:
return flask.jsonify(results)
else:
placeholder = ''
for form in template['forms']:
if placeholder:
placeholder += '|'
(prefix, form_placeholder, suffix) = split_example(form)
placeholder += form_placeholder
placeholder += '\n...'
csrf_error = False
if flask.request.method == 'POST':
form_data = flask.request.form
if 'form_representation' in form_data:
# user came from non-bulk mode
representations = form_data.getlist('form_representation')
value = '|'.join(representations)
if value == '|' * (len(representations) - 1):
# ...but had not typed anything into non-bulk mode yet,
# clear the value so that the placeholder is shown
value = ''
else:
value += '\n' # for convenience when adding more
else:
# user came from bulk mode with CSRF error
value = form_data['lexemes']
csrf_error = True
else:
value = None
return flask.render_template(
'bulk.html',
template=template,
placeholder=placeholder,
value=value,
csrf_error=csrf_error,
show_optional_forms_hint=False,
readonly=readonly,
)