From f62ce21f50b540315a708ebfbf36878ddec9d1c4 Mon Sep 17 00:00:00 2001 From: marc Date: Sat, 5 Nov 2016 20:51:38 -0600 Subject: [mod] fetch supported languages for several engines utils/fetch_languages.py gets languages supported by each engine and generates engines_languages.json with each engine's supported language. --- utils/fetch_languages.py | 164 +++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 164 insertions(+) create mode 100644 utils/fetch_languages.py (limited to 'utils/fetch_languages.py') diff --git a/utils/fetch_languages.py b/utils/fetch_languages.py new file mode 100644 index 000000000..ae4a2def9 --- /dev/null +++ b/utils/fetch_languages.py @@ -0,0 +1,164 @@ +# -*- coding: utf-8 -*- + +# This script generates languages.py from intersecting each engine's supported languages. +# +# The country names are obtained from http://api.geonames.org which requires registering as a user. +# +# Output files (engines_languages.json and languages.py) +# are written in current directory to avoid overwriting in case something goes wrong. + +from requests import get +from urllib import urlencode +from lxml.html import fromstring +from json import loads, dumps +import io +from sys import path +path.append('../searx') # noqa +from searx.engines import engines + +# Geonames API for country names. +geonames_user = '' # ADD USER NAME HERE +country_names_url = 'http://api.geonames.org/countryInfoJSON?{parameters}' + +# Output files. +engines_languages_file = 'engines_languages.json' +languages_file = 'languages.py' + +engines_languages = {} +languages = {} + + +# To filter out invalid codes and dialects. +def valid_code(lang_code): + # filter invalid codes + # sl-SL is technically not invalid, but still a mistake + if lang_code[:2] == 'xx'\ + or lang_code == 'sl-SL'\ + or lang_code == 'wt-WT'\ + or lang_code == 'jw'\ + or lang_code[-2:] == 'UK'\ + or lang_code[-2:] == 'XA'\ + or lang_code[-2:] == 'XL': + return False + + # filter dialects + lang_code = lang_code.split('-') + if len(lang_code) > 2 or len(lang_code[0]) > 3: + return False + if len(lang_code) == 2 and len(lang_code[1]) > 2: + return False + + return True + + +# Get country name in specified language. +def get_country_name(locale): + if geonames_user is '': + return '' + + locale = locale.split('-') + if len(locale) != 2: + return '' + + url = country_names_url.format(parameters=urlencode({'lang': locale[0], + 'country': locale[1], + 'username': geonames_user})) + response = get(url) + json = loads(response.text) + content = json.get('geonames', None) + if content is None or len(content) != 1: + print "No country name found for " + locale[0] + "-" + locale[1] + return '' + + return content[0].get('countryName', '') + + +# Fetchs supported languages for each engine and writes json file with those. +def fetch_supported_languages(): + for engine_name in engines: + if hasattr(engines[engine_name], 'fetch_supported_languages'): + try: + engines_languages[engine_name] = engines[engine_name].fetch_supported_languages() + except Exception as e: + print e + + # write json file + f = io.open(engines_languages_file, "w", encoding="utf-8") + f.write(unicode(dumps(engines_languages, indent=4, ensure_ascii=False, encoding="utf-8"))) + f.close() + + +# Join all language lists. +# Iterate all languages supported by each engine. +def join_language_lists(): + # include wikipedia first for more accurate language names + # exclude languages with too few articles + languages.update({code: lang for code, lang + in engines_languages['wikipedia'].iteritems() + if valid_code(code) and lang['articles'] >= 100000}) + + for engine_name in engines_languages: + for locale in engines_languages[engine_name]: + if not valid_code(locale): + continue + + # if language is not on list or if it has no name yet + if locale not in languages or not languages[locale].get('name'): + if isinstance(engines_languages[engine_name], dict) \ + and engines_languages[engine_name][locale].get('articles', float('inf')) >= 100000: + languages[locale] = engines_languages[engine_name][locale] + else: + languages[locale] = {} + + # get locales that have no name yet + for locale in languages.keys(): + if not languages[locale].get('name'): + # try to get language and country names + name = languages.get(locale.split('-')[0], {}).get('name', None) + if name: + languages[locale]['name'] = name + languages[locale]['country'] = get_country_name(locale) or '' + languages[locale]['english_name'] = languages.get(locale.split('-')[0], {}).get('english_name', '') + else: + # filter out locales with no name + del languages[locale] + + +# Remove countryless language if language is featured in only one country. +def filter_single_country_languages(): + prev_lang = None + for code in sorted(languages): + lang = code.split('-')[0] + if lang == prev_lang: + countries += 1 + else: + if prev_lang is not None and countries == 1: + del languages[prev_lang] + countries = 0 + prev_lang = lang + + +# Write languages.py. +def write_languages_file(): + new_file = open(languages_file, 'w') + file_content = '# -*- coding: utf-8 -*-\n' + file_content += '# list of language codes\n' + file_content += '# this file is generated automatically by utils/update_search_languages.py\n' + file_content += '\nlanguage_codes = (' + for code in sorted(languages): + file_content += '\n (u"' + code + '"'\ + + ', u"' + languages[code]['name'].split(' (')[0] + '"'\ + + ', u"' + languages[code].get('country', '') + '"'\ + + ', u"' + languages[code].get('english_name', '').split(' (')[0] + '"),' + # remove last comma + file_content = file_content[:-1] + file_content += '\n)\n' + new_file.write(file_content.encode('utf8')) + new_file.close() + + +if __name__ == "__main__": + fetch_supported_languages() + join_language_lists() + filter_single_country_languages() + write_languages_file() -- cgit v1.2.3 From af35eee10b98940c51c6e5e18629de514b4bd48d Mon Sep 17 00:00:00 2001 From: marc Date: Thu, 15 Dec 2016 00:34:43 -0600 Subject: tests for _fetch_supported_languages in engines and refactor method to make it testable without making requests --- utils/fetch_languages.py | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) (limited to 'utils/fetch_languages.py') diff --git a/utils/fetch_languages.py b/utils/fetch_languages.py index ae4a2def9..aed875daf 100644 --- a/utils/fetch_languages.py +++ b/utils/fetch_languages.py @@ -84,7 +84,7 @@ def fetch_supported_languages(): # write json file f = io.open(engines_languages_file, "w", encoding="utf-8") - f.write(unicode(dumps(engines_languages, indent=4, ensure_ascii=False, encoding="utf-8"))) + f.write(unicode(dumps(engines_languages, ensure_ascii=False, encoding="utf-8"))) f.close() @@ -110,18 +110,22 @@ def join_language_lists(): else: languages[locale] = {} - # get locales that have no name yet + # get locales that have no name or country yet for locale in languages.keys(): if not languages[locale].get('name'): - # try to get language and country names + # try to get language names name = languages.get(locale.split('-')[0], {}).get('name', None) if name: languages[locale]['name'] = name - languages[locale]['country'] = get_country_name(locale) or '' languages[locale]['english_name'] = languages.get(locale.split('-')[0], {}).get('english_name', '') else: # filter out locales with no name del languages[locale] + continue + + # try to get country name + if locale.find('-') > 0 and not languages[locale].get('country'): + languages[locale]['country'] = get_country_name(locale) or '' # Remove countryless language if language is featured in only one country. -- cgit v1.2.3 From 4a1ff56389d6ad560594ba82b448aef1d70bbbf4 Mon Sep 17 00:00:00 2001 From: marc Date: Fri, 16 Dec 2016 22:14:14 -0600 Subject: minor fixes in utils/fetch_languages.py --- utils/fetch_languages.py | 49 +++++++++++++++++++++++++----------------------- 1 file changed, 26 insertions(+), 23 deletions(-) (limited to 'utils/fetch_languages.py') diff --git a/utils/fetch_languages.py b/utils/fetch_languages.py index aed875daf..3510a3234 100644 --- a/utils/fetch_languages.py +++ b/utils/fetch_languages.py @@ -32,23 +32,26 @@ languages = {} def valid_code(lang_code): # filter invalid codes # sl-SL is technically not invalid, but still a mistake + invalid_codes = ['sl-SL', 'wt-WT', 'jw'] + invalid_countries = ['UK', 'XA', 'XL'] if lang_code[:2] == 'xx'\ - or lang_code == 'sl-SL'\ - or lang_code == 'wt-WT'\ - or lang_code == 'jw'\ - or lang_code[-2:] == 'UK'\ - or lang_code[-2:] == 'XA'\ - or lang_code[-2:] == 'XL': + or lang_code in invalid_codes\ + or lang_code[-2:] in invalid_countries\ + or is_dialect(lang_code): return False - # filter dialects + return True + + +# Language codes with any additional tags other than language and country. +def is_dialect(lang_code): lang_code = lang_code.split('-') if len(lang_code) > 2 or len(lang_code[0]) > 3: - return False + return True if len(lang_code) == 2 and len(lang_code[1]) > 2: - return False + return True - return True + return False # Get country name in specified language. @@ -83,19 +86,17 @@ def fetch_supported_languages(): print e # write json file - f = io.open(engines_languages_file, "w", encoding="utf-8") - f.write(unicode(dumps(engines_languages, ensure_ascii=False, encoding="utf-8"))) - f.close() + with io.open(engines_languages_file, "w", encoding="utf-8") as f: + f.write(unicode(dumps(engines_languages, ensure_ascii=False, encoding="utf-8"))) # Join all language lists. # Iterate all languages supported by each engine. def join_language_lists(): # include wikipedia first for more accurate language names - # exclude languages with too few articles languages.update({code: lang for code, lang in engines_languages['wikipedia'].iteritems() - if valid_code(code) and lang['articles'] >= 100000}) + if valid_code(code)}) for engine_name in engines_languages: for locale in engines_languages[engine_name]: @@ -104,25 +105,27 @@ def join_language_lists(): # if language is not on list or if it has no name yet if locale not in languages or not languages[locale].get('name'): - if isinstance(engines_languages[engine_name], dict) \ - and engines_languages[engine_name][locale].get('articles', float('inf')) >= 100000: + if isinstance(engines_languages[engine_name], dict): languages[locale] = engines_languages[engine_name][locale] else: languages[locale] = {} # get locales that have no name or country yet for locale in languages.keys(): + # try to get language names if not languages[locale].get('name'): - # try to get language names name = languages.get(locale.split('-')[0], {}).get('name', None) if name: languages[locale]['name'] = name - languages[locale]['english_name'] = languages.get(locale.split('-')[0], {}).get('english_name', '') else: # filter out locales with no name del languages[locale] continue + # try to get language name in english + if not languages[locale].get('english_name'): + languages[locale]['english_name'] = languages.get(locale.split('-')[0], {}).get('english_name', '') + # try to get country name if locale.find('-') > 0 and not languages[locale].get('country'): languages[locale]['country'] = get_country_name(locale) or '' @@ -145,10 +148,10 @@ def filter_single_country_languages(): # Write languages.py. def write_languages_file(): new_file = open(languages_file, 'w') - file_content = '# -*- coding: utf-8 -*-\n' - file_content += '# list of language codes\n' - file_content += '# this file is generated automatically by utils/update_search_languages.py\n' - file_content += '\nlanguage_codes = (' + file_content = '# -*- coding: utf-8 -*-\n'\ + + '# list of language codes\n'\ + + '# this file is generated automatically by utils/update_search_languages.py\n'\ + + '\nlanguage_codes = (' for code in sorted(languages): file_content += '\n (u"' + code + '"'\ + ', u"' + languages[code]['name'].split(' (')[0] + '"'\ -- cgit v1.2.3