| Index: trunk/tools/editor_trends/config.py |
| — | — | @@ -1,90 +0,0 @@ |
| 2 | | -#!/usr/bin/python |
| 3 | | -# -*- coding: utf-8 -*- |
| 4 | | -''' |
| 5 | | -Copyright (C) 2010 by Diederik van Liere (dvanliere@gmail.com) |
| 6 | | -This program is free software; you can redistribute it and/or |
| 7 | | -modify it under the terms of the GNU General Public License version 2 |
| 8 | | -as published by the Free Software Foundation. |
| 9 | | -This program is distributed in the hope that it will be useful, |
| 10 | | -but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 11 | | -MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. |
| 12 | | -See the GNU General Public License for more details, at |
| 13 | | -http://www.fsf.org/licenses/gpl.html |
| 14 | | -''' |
| 15 | | - |
| 16 | | -__author__ = '''\n'''.join(['Diederik van Liere (dvanliere@gmail.com)', ]) |
| 17 | | -__email__ = 'dvanliere at gmail dot com' |
| 18 | | -__date__ = '2010-10-21' |
| 19 | | -__version__ = '0.1' |
| 20 | | - |
| 21 | | - |
| 22 | | -import os |
| 23 | | -import ConfigParser |
| 24 | | - |
| 25 | | -from utils import file_utils |
| 26 | | -from classes import wikiprojects |
| 27 | | - |
| 28 | | - |
| 29 | | -def show_choices(settings, attr): |
| 30 | | - choices = getattr(settings, attr).items() |
| 31 | | - choices.sort() |
| 32 | | - choices = ['%s\t%s' % (choice[0], choice[1]) for choice in choices] |
| 33 | | - return choices |
| 34 | | - |
| 35 | | - |
| 36 | | -def create_configuration(settings, args): |
| 37 | | - force = getattr(args, 'force', False) |
| 38 | | - |
| 39 | | - if not os.path.exists('wiki.cfg') or force: |
| 40 | | - config = ConfigParser.RawConfigParser() |
| 41 | | - project = None |
| 42 | | - language = None |
| 43 | | - dumpversion = None |
| 44 | | - #language_map = languages.language_map() |
| 45 | | - working_directory = raw_input('Please indicate where you installed Editor Trends Analytics.\nCurrent location is %s\nPress Enter to accept default.\n' % os.getcwd()) |
| 46 | | - input_location = raw_input('Please indicate where to store the Wikipedia dump files.\nDefault is: %s\nPress Enter to accept default.\n' % settings.input_location) |
| 47 | | - |
| 48 | | - while project not in settings.projects.keys(): |
| 49 | | - project = raw_input('Please indicate which project you would like to analyze.\nDefault is: %s\nPress Enter to accept default.\n' % settings.projects[args.project].capitalize()) |
| 50 | | - project = project if len(project) > 0 else args.project |
| 51 | | - if project not in settings.projects.keys(): |
| 52 | | - print 'Valid choices for a project are: %s' % ','.join(settings.projects.keys()) |
| 53 | | - |
| 54 | | - wiki = wikiprojects.Wiki(settings.encoding, project=project) |
| 55 | | - while language not in wiki.valid_languages: |
| 56 | | - language = raw_input('Please indicate which language of project %s you would like to analyze.\nDefault is: %s\nPress Enter to accept default.\n' % (settings.projects[project].capitalize(), language_map[args.language])) |
| 57 | | - if len(language) == 0: |
| 58 | | - language = language_map[args.language] |
| 59 | | - language = language if language in wiki.valid_languages else args.language |
| 60 | | - |
| 61 | | -# while dumpversion not in settings.dumpversions.keys(): |
| 62 | | -# choices = '\n'.join(show_choices(settings, 'dumpversions')) |
| 63 | | -# dumpversion = raw_input('Please indicate the version of the Wikipedia project you are analyzing.\nValid choices are:\n%s\nDefault is: 0 (%s)\nPress Enter to accept default.\n' % (choices, settings.dumpversions['0'])) |
| 64 | | -# if len(dumpversion) == 0: |
| 65 | | -# dumpversion = settings.dumpversions['0'] |
| 66 | | - |
| 67 | | - #dumpversion = settings.dumpversions[dumpversion] |
| 68 | | - input_location = input_location if len(input_location) > 0 else settings.input_location |
| 69 | | - working_directory = working_directory if len(working_directory) > 0 else os.getcwd() |
| 70 | | - |
| 71 | | - config = ConfigParser.RawConfigParser() |
| 72 | | - config.add_section('file_locations') |
| 73 | | - config.set('file_locations', 'working_directory', working_directory) |
| 74 | | - config.set('file_locations', 'input_location', input_location) |
| 75 | | - config.add_section('wiki') |
| 76 | | - config.set('wiki', 'project', project) |
| 77 | | - config.set('wiki', 'language', language) |
| 78 | | - #config.set('wiki', 'dumpversion', dumpversion) |
| 79 | | - |
| 80 | | - fh = file_utils.create_binary_filehandle(working_directory, 'wiki.cfg', 'wb') |
| 81 | | - config.write(fh) |
| 82 | | - fh.close() |
| 83 | | - |
| 84 | | - settings.working_directory = config.get('file_locations', 'working_directory') |
| 85 | | - settings.input_location = config.get('file_locations', 'input_location') |
| 86 | | - #settings.xml_namespace = config.get('wiki', 'dumpversion') |
| 87 | | - return settings |
| 88 | | - |
| 89 | | - |
| 90 | | -if __name__ == '__main__': |
| 91 | | - pass |
| Index: trunk/tools/editor_trends/languages.py |
| — | — | @@ -1,613 +0,0 @@ |
| 2 | | -#!/usr/bin/python |
| 3 | | -# coding=utf-8 |
| 4 | | -''' |
| 5 | | -Copyright (C) 2010 by Diederik van Liere (dvanliere@gmail.com) |
| 6 | | -This program is free software; you can redistribute it and/or |
| 7 | | -modify it under the terms of the GNU General Public License version 2 |
| 8 | | -as published by the Free Software Foundation. |
| 9 | | -This program is distributed in the hope that it will be useful, |
| 10 | | -but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 11 | | -MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. |
| 12 | | -See the GNU General Public License for more details, at |
| 13 | | -http,//www.fsf.org/licenses/gpl.html |
| 14 | | -''' |
| 15 | | - |
| 16 | | -__author__ = '''\n'''.join(['Diederik van Liere (dvanliere@gmail.com)', ]) |
| 17 | | -__author__email = 'dvanliere at gmail dot com' |
| 18 | | -__date__ = '2010-10-21' |
| 19 | | -__version__ = '0.1' |
| 20 | | - |
| 21 | | -''' |
| 22 | | -This file provides mapper between language name and locale language name and |
| 23 | | -Wikipedia acronym. |
| 24 | | -Gothic and Birmese are not yet supported, see rows 450 and 554. |
| 25 | | -''' |
| 26 | | - |
| 27 | | -from utils import ordered_dict as odict |
| 28 | | -from utils import utils |
| 29 | | - |
| 30 | | -MAPPING = odict.OrderedDict([ |
| 31 | | -(u'English', 'en'), |
| 32 | | -(u'German', 'de'), |
| 33 | | -(u'French', 'fr'), |
| 34 | | -(u'Italian', 'it'), |
| 35 | | -(u'Polish', 'pl'), |
| 36 | | -(u'Japanese', 'ja'), |
| 37 | | -(u'Spanish', 'es'), |
| 38 | | -(u'Dutch', 'nl'), |
| 39 | | -(u'Portuguese', 'pt'), |
| 40 | | -(u'Russian', 'ru'), |
| 41 | | -(u'Swedish', 'sv'), |
| 42 | | -(u'Chinese', 'zh'), |
| 43 | | -(u'Catalan', 'ca'), |
| 44 | | -(u'Norwegian', 'no'), |
| 45 | | -(u'Bokmål', 'no'), |
| 46 | | -(u'Finnish', 'fi'), |
| 47 | | -(u'Ukrainian', 'uk'), |
| 48 | | -(u'Hungarian', 'hu'), |
| 49 | | -(u'Czech', 'cs'), |
| 50 | | -(u'Romanian', 'ro'), |
| 51 | | -(u'Turkish', 'tr'), |
| 52 | | -(u'Korean', 'ko'), |
| 53 | | -(u'Vietnamese', 'vi'), |
| 54 | | -(u'Danish', 'da'), |
| 55 | | -(u'Arabic', 'ar'), |
| 56 | | -(u'Esperanto', 'eo'), |
| 57 | | -(u'Serbian', 'sr'), |
| 58 | | -(u'Indonesian', 'id'), |
| 59 | | -(u'Lithuanian', 'lt'), |
| 60 | | -(u'Volapük', 'vo'), |
| 61 | | -(u'Slovak', 'sk'), |
| 62 | | -(u'Hebrew', 'he'), |
| 63 | | -(u'Bulgarian', 'bg'), |
| 64 | | -(u'Persian', 'fa'), |
| 65 | | -(u'Slovenian', 'sl'), |
| 66 | | -(u'Waray-Waray', 'war'), |
| 67 | | -(u'Croatian', 'hr'), |
| 68 | | -(u'Estonian', 'et'), |
| 69 | | -(u'Malay', 'ms'), |
| 70 | | -(u'Newar', 'new'), |
| 71 | | -(u'Nepal Bhasa', 'new'), |
| 72 | | -(u'Simple English', 'simple'), |
| 73 | | -(u'Galician', 'gl'), |
| 74 | | -(u'Thai', 'th'), |
| 75 | | -(u'Aromanian', 'roa-rup'), |
| 76 | | -(u'Nynorsk', 'nn'), |
| 77 | | -(u'Basque', 'eu'), |
| 78 | | -(u'Hindi', 'hi'), |
| 79 | | -(u'Greek', 'el'), |
| 80 | | -(u'Haitian', 'ht'), |
| 81 | | -(u'Latin', 'la'), |
| 82 | | -(u'Telugu', 'te'), |
| 83 | | -(u'Georgian', 'ka'), |
| 84 | | -(u'Cebuano', 'ceb'), |
| 85 | | -(u'Macedonian', 'mk'), |
| 86 | | -(u'Azeri', 'az'), |
| 87 | | -(u'Tagalog', 'tl'), |
| 88 | | -(u'Breton', 'br'), |
| 89 | | -(u'Serbo-Croatian', 'sh'), |
| 90 | | -(u'Marathi', 'mr'), |
| 91 | | -(u'Luxembourgish', 'lb'), |
| 92 | | -(u'Javanese', 'jv'), |
| 93 | | -(u'Latvian', 'lv'), |
| 94 | | -(u'Bosnian', 'bs'), |
| 95 | | -(u'Icelandic', 'is'), |
| 96 | | -(u'Welsh', 'cy'), |
| 97 | | -(u'Belarusian', 'be-x-old'), |
| 98 | | -(u'Taraškievica', 'be-x-old'), |
| 99 | | -(u'Piedmontese', 'pms'), |
| 100 | | -(u'Albanian', 'sq'), |
| 101 | | -(u'Tamil', 'ta'), |
| 102 | | -(u'Bishnupriya Manipuri', 'bpy'), |
| 103 | | -(u'Belarusian', 'be'), |
| 104 | | -(u'Aragonese', 'an'), |
| 105 | | -(u'Occitan', 'oc'), |
| 106 | | -(u'Bengali', 'bn'), |
| 107 | | -(u'Swahili', 'sw'), |
| 108 | | -(u'Ido', 'io'), |
| 109 | | -(u'Ripuarian', 'ksh'), |
| 110 | | -(u'Lombard', 'lmo'), |
| 111 | | -(u'West Frisian', 'fy'), |
| 112 | | -(u'Gujarati', 'gu'), |
| 113 | | -(u'Low Saxon', 'nds'), |
| 114 | | -(u'Afrikaans', 'af'), |
| 115 | | -(u'Sicilian', 'scn'), |
| 116 | | -(u'Quechua', 'qu'), |
| 117 | | -(u'Kurdish', 'ku'), |
| 118 | | -(u'Urdu', 'ur'), |
| 119 | | -(u'Sundanese', 'su'), |
| 120 | | -(u'Malayalam', 'ml'), |
| 121 | | -(u'Cantonese', 'zh-yue'), |
| 122 | | -(u'Asturian', 'ast'), |
| 123 | | -(u'Neapolitan', 'nap'), |
| 124 | | -(u'Samogitian', 'bat-smg'), |
| 125 | | -(u'Walloon', 'wa'), |
| 126 | | -(u'Chuvash', 'cv'), |
| 127 | | -(u'Irish', 'ga'), |
| 128 | | -(u'Armenian', 'hy'), |
| 129 | | -(u'Yoruba', 'yo'), |
| 130 | | -(u'Kannada', 'kn'), |
| 131 | | -(u'Tajik', 'tg'), |
| 132 | | -(u'Tarantino', 'roa-tara'), |
| 133 | | -(u'Venetian', 'vec'), |
| 134 | | -(u'Western Panjabi', 'pnb'), |
| 135 | | -(u'Nepali', 'ne'), |
| 136 | | -(u'Scottish Gaelic', 'gd'), |
| 137 | | -(u'Yiddish', 'yi'), |
| 138 | | -(u'Min Nan', 'zh-min-nan'), |
| 139 | | -(u'Uzbek', 'uz'), |
| 140 | | -(u'Tatar', 'tt'), |
| 141 | | -(u'Kapampangan', 'pam'), |
| 142 | | -(u'Ossetian', 'os'), |
| 143 | | -(u'Sakha', 'sah'), |
| 144 | | -(u'Alemannic', 'als'), |
| 145 | | -(u'Maori', 'mi'), |
| 146 | | -(u'Egyptian Arabic', 'arz'), |
| 147 | | -(u'Kazakh', 'kk'), |
| 148 | | -(u'Nahuatl', 'nah'), |
| 149 | | -(u'Limburgian', 'li'), |
| 150 | | -(u'Upper Sorbian', 'hsb'), |
| 151 | | -(u'Gilaki', 'glk'), |
| 152 | | -(u'Corsican', 'co'), |
| 153 | | -(u'Gan', 'gan'), |
| 154 | | -(u'Amharic', 'am'), |
| 155 | | -(u'Mongolian', 'mn'), |
| 156 | | -(u'Interlingua', 'ia'), |
| 157 | | -(u'Central Bicolano', 'bcl'), |
| 158 | | -(u'Võro', 'fiu-vro'), |
| 159 | | -(u'Dutch Low Saxon', 'nds-nl'), |
| 160 | | -(u'Faroese', 'fo'), |
| 161 | | -(u'Turkmen', 'tk'), |
| 162 | | -(u'Scots', 'sco'), |
| 163 | | -(u'West Flemish', 'vls'), |
| 164 | | -(u'Sinhalese', 'si'), |
| 165 | | -(u'Sanskrit', 'sa'), |
| 166 | | -(u'Bavarian', 'bar'), |
| 167 | | -(u'Burmese', 'my'), |
| 168 | | -(u'Manx', 'gv'), |
| 169 | | -(u'Divehi', 'dv'), |
| 170 | | -(u'Norman', 'nrm'), |
| 171 | | -(u'Pangasinan', 'pag'), |
| 172 | | -(u'Romansh', 'rm'), |
| 173 | | -(u'Banyumasan', 'map-bms'), |
| 174 | | -(u'Zazaki', 'diq'), |
| 175 | | -(u'Sorani', 'ckb'), |
| 176 | | -(u'Northern Sami', 'se'), |
| 177 | | -(u'Mazandarani', 'mzn'), |
| 178 | | -(u'Wu', 'wuu'), |
| 179 | | -(u'Uyghur', 'ug'), |
| 180 | | -(u'Friulian', 'fur'), |
| 181 | | -(u'Ligurian', 'lij'), |
| 182 | | -(u'Maltese', 'mt'), |
| 183 | | -(u'Bihari', 'bh'), |
| 184 | | -(u'Novial', 'nov'), |
| 185 | | -(u'Malagasy', 'mg'), |
| 186 | | -(u'Kashubian', 'csb'), |
| 187 | | -(u'Ilokano', 'ilo'), |
| 188 | | -(u'Sardinian', 'sc'), |
| 189 | | -(u'Classical Chinese', 'zh-classical'), |
| 190 | | -(u'Khmer', 'km'), |
| 191 | | -(u'Ladino', 'lad'), |
| 192 | | -(u'Pali', 'pi'), |
| 193 | | -(u'Anglo-Saxon', 'ang'), |
| 194 | | -(u'Zamboanga Chavacano', 'cbk-zam'), |
| 195 | | -(u'Tibetan', 'bo'), |
| 196 | | -(u'Fiji Hindi', 'hif'), |
| 197 | | -(u'Franco-Provençal', 'frp'), |
| 198 | | -(u'Arpitan', 'frp'), |
| 199 | | -(u'Hakka', 'hak'), |
| 200 | | -(u'Cornish', 'kw'), |
| 201 | | -(u'Punjabi', 'pa'), |
| 202 | | -(u'Pashto', 'ps'), |
| 203 | | -(u'Kalmyk', 'xal'), |
| 204 | | -(u'Silesian', 'szl'), |
| 205 | | -(u'Pennsylvania German', 'pdc'), |
| 206 | | -(u'Hawaiian', 'haw'), |
| 207 | | -(u'Saterland Frisian', 'stq'), |
| 208 | | -(u'Interlingue', 'ie'), |
| 209 | | -(u'Navajo', 'nv'), |
| 210 | | -(u'Fijian', 'fj'), |
| 211 | | -(u'Crimean Tatar', 'crh'), |
| 212 | | -(u'Komi', 'kv'), |
| 213 | | -(u'Tongan', 'to'), |
| 214 | | -(u'Acehnese', 'ace'), |
| 215 | | -(u'Somali', 'so'), |
| 216 | | -(u'Erzya', 'myv'), |
| 217 | | -(u'Guarani', 'gn'), |
| 218 | | -(u'Karachay-Balkar', 'krc'), |
| 219 | | -(u'Extremaduran', 'ext'), |
| 220 | | -(u'Lingala', 'ln'), |
| 221 | | -(u'Kirghiz', 'ky'), |
| 222 | | -(u'Meadow Mari', 'mhr'), |
| 223 | | -(u'Assyrian Neo-Aramaic', 'arc'), |
| 224 | | -(u'Emilian-Romagnol', 'eml'), |
| 225 | | -(u'Lojban', 'jbo'), |
| 226 | | -(u'Picard', 'pcd'), |
| 227 | | -(u'Aymara', 'ay'), |
| 228 | | -(u'Wolof', 'wo'), |
| 229 | | -(u'Tumbuka', 'tum'), |
| 230 | | -(u'Kabyle', 'kab'), |
| 231 | | -(u'Bashkir', 'ba'), |
| 232 | | -(u'North Frisian', 'frr'), |
| 233 | | -(u'Tahitian', 'ty'), |
| 234 | | -(u'Tok Pisin', 'tpi'), |
| 235 | | -(u'Papiamentu', 'pap'), |
| 236 | | -(u'Zealandic', 'zea'), |
| 237 | | -(u'Sranan', 'srn'), |
| 238 | | -(u'Greenlandic', 'kl'), |
| 239 | | -(u'Udmurt', 'udm'), |
| 240 | | -(u'Chechen', 'ce'), |
| 241 | | -(u'Igbo', 'ig'), |
| 242 | | -(u'Komi-Permyak', 'koi'), |
| 243 | | -(u'Oriya', 'or'), |
| 244 | | -(u'Lower Sorbian', 'dsb'), |
| 245 | | -(u'Kongo', 'kg'), |
| 246 | | -(u'Lao', 'lo'), |
| 247 | | -(u'Abkhazian', 'ab'), |
| 248 | | -(u'Moksha', 'mdf'), |
| 249 | | -(u'Romani', 'rmy'), |
| 250 | | -(u'Hill Mari', 'mrj'), |
| 251 | | -(u'Banjar', 'bjn'), |
| 252 | | -(u'Old Church Slavonic', 'cu'), |
| 253 | | -(u'Mirandese', 'mwl'), |
| 254 | | -(u'Karakalpak', 'kaa'), |
| 255 | | -(u'Samoan', 'sm'), |
| 256 | | -(u'Moldovan', 'mo'), |
| 257 | | -(u'Tetum', 'tet'), |
| 258 | | -(u'Avar', 'av'), |
| 259 | | -(u'Kashmiri', 'ks'), |
| 260 | | -(u'Gothic', 'got'), |
| 261 | | -(u'Sindhi', 'sd'), |
| 262 | | -(u'Bambara', 'bm'), |
| 263 | | -(u'Nauruan', 'na'), |
| 264 | | -(u'Norfolk', 'pih'), |
| 265 | | -(u'Pontic', 'pnt'), |
| 266 | | -(u'Inuktitut', 'iu'), |
| 267 | | -(u'Inupiak', 'ik'), |
| 268 | | -(u'Bislama', 'bi'), |
| 269 | | -(u'Cherokee', 'chr'), |
| 270 | | -(u'Assamese', 'as'), |
| 271 | | -(u'Min Dong', 'cdo'), |
| 272 | | -(u'Ewe', 'ee'), |
| 273 | | -(u'Swati', 'ss'), |
| 274 | | -(u'Oromo', 'om'), |
| 275 | | -(u'Zhuang', 'za'), |
| 276 | | -(u'Zulu', 'zu'), |
| 277 | | -(u'Tigrinya', 'ti'), |
| 278 | | -(u'Venda', 've'), |
| 279 | | -(u'Tsonga', 'ts'), |
| 280 | | -(u'Hausa', 'ha'), |
| 281 | | -(u'Dzongkha', 'dz'), |
| 282 | | -(u'Sango', 'sg'), |
| 283 | | -(u'Chamorro', 'ch'), |
| 284 | | -(u'Cree', 'cr'), |
| 285 | | -(u'Xhosa', 'xh'), |
| 286 | | -(u'Akan', 'ak'), |
| 287 | | -(u'Sesotho', 'st'), |
| 288 | | -(u'Kinyarwanda', 'rw'), |
| 289 | | -(u'Tswana', 'tn'), |
| 290 | | -(u'Kikuyu', 'ki'), |
| 291 | | -(u'Buryat', 'bxr'), |
| 292 | | -(u'Buginese', 'bug'), |
| 293 | | -(u'Chichewa', 'ny'), |
| 294 | | -(u'Lak', 'lbe'), |
| 295 | | -(u'Twi', 'tw'), |
| 296 | | -(u'Shona', 'sn'), |
| 297 | | -(u'Kirundi', 'rn'), |
| 298 | | -(u'Fula', 'ff'), |
| 299 | | -(u'Cheyenne', 'chy'), |
| 300 | | -(u'Luganda', 'lg'), |
| 301 | | -(u'Ndonga', 'ng'), |
| 302 | | -(u'Sichuan Yi', 'ii'), |
| 303 | | -(u'Choctaw', 'cho'), |
| 304 | | -(u'Marshallese', 'mh'), |
| 305 | | -(u'Afar', 'aa'), |
| 306 | | -(u'Kuanyama', 'kj'), |
| 307 | | -(u'Hiri Motu', 'ho'), |
| 308 | | -(u'Muscogee', 'mus'), |
| 309 | | -(u'Kanuri', 'kr'), |
| 310 | | -(u'Herero', 'hz'), |
| 311 | | -(u'English', 'en'), |
| 312 | | -(u'Deutsch', 'de'), |
| 313 | | -(u'Français', 'fr'), |
| 314 | | -(u'Italiano', 'it'), |
| 315 | | -(u'Polski', 'pl'), |
| 316 | | -(u'日本語', 'ja'), |
| 317 | | -(u'Español', 'es'), |
| 318 | | -(u'Nederlands', 'nl'), |
| 319 | | -(u'Português', 'pt'), |
| 320 | | -(u'Русский', 'ru'), |
| 321 | | -(u'Svenska', 'sv'), |
| 322 | | -(u'中文', 'zh'), |
| 323 | | -(u'Català', 'ca'), |
| 324 | | -(u'Norsk', 'no'), |
| 325 | | -(u'Bokmål', 'no'), |
| 326 | | -(u'Suomi', 'fi'), |
| 327 | | -(u'Українська', 'uk'), |
| 328 | | -(u'Magyar', 'hu'), |
| 329 | | -(u'Čeština', 'cs'), |
| 330 | | -(u'Română', 'ro'), |
| 331 | | -(u'Türkçe', 'tr'), |
| 332 | | -(u'한국어', 'ko'), |
| 333 | | -(u'Tiếng Việt', 'vi'), |
| 334 | | -(u'Dansk', 'da'), |
| 335 | | -(u'العربية', 'ar'), |
| 336 | | -(u'Esperanto', 'eo'), |
| 337 | | -(u'Српски', 'sr'), |
| 338 | | -(u'Srpski', 'sr'), |
| 339 | | -(u'Bahasa Indonesia', 'id'), |
| 340 | | -(u'Lietuvių', 'lt'), |
| 341 | | -(u'Volapük', 'vo'), |
| 342 | | -(u'Slovenčina', 'sk'), |
| 343 | | -(u'עברית', 'he'), |
| 344 | | -(u'Български', 'bg'), |
| 345 | | -(u'فارسی', 'fa'), |
| 346 | | -(u'Slovenščina', 'sl'), |
| 347 | | -(u'Winaray', 'war'), |
| 348 | | -(u'Hrvatski', 'hr'), |
| 349 | | -(u'Eesti', 'et'), |
| 350 | | -(u'Bahasa Melayu', 'ms'), |
| 351 | | -(u'नेपाल भाषा', 'new'), |
| 352 | | -(u'Simple English', 'simple'), |
| 353 | | -(u'Galego', 'gl'), |
| 354 | | -(u'ไทย', 'th'), |
| 355 | | -(u'Armãneashce', 'roa-rup'), |
| 356 | | -(u'Nynorsk', 'nn'), |
| 357 | | -(u'Euskara', 'eu'), |
| 358 | | -(u'हिन्दी', 'hi'), |
| 359 | | -(u'Ελληνικά', 'el'), |
| 360 | | -(u'Krèyol ayisyen', 'ht'), |
| 361 | | -(u'Latina', 'la'), |
| 362 | | -(u'తెలుగు', 'te'), |
| 363 | | -(u'ქართული', 'ka'), |
| 364 | | -(u'Sinugboanong Binisaya', 'ceb'), |
| 365 | | -(u'Македонски', 'mk'), |
| 366 | | -(u'Azərbaycan', 'az'), |
| 367 | | -(u'Tagalog', 'tl'), |
| 368 | | -(u'Brezhoneg', 'br'), |
| 369 | | -(u'Srpskohrvatski', 'sh'), |
| 370 | | -(u'Српскохрватски', 'sh'), |
| 371 | | -(u'मराठी', 'mr'), |
| 372 | | -(u'Lëtzebuergesch', 'lb'), |
| 373 | | -(u'Basa Jawa', 'jv'), |
| 374 | | -(u'Latviešu', 'lv'), |
| 375 | | -(u'Bosanski', 'bs'), |
| 376 | | -(u'Íslenska', 'is'), |
| 377 | | -(u'Cymraeg', 'cy'), |
| 378 | | -(u'Беларуская', 'be-x-old'), |
| 379 | | -(u'тарашкевіца', 'be-x-old'), |
| 380 | | -(u'Piemontèis', 'pms'), |
| 381 | | -(u'Shqip', 'sq'), |
| 382 | | -(u'தமிழ்', 'ta'), |
| 383 | | -(u'ইমার ঠার', 'bpy'), |
| 384 | | -(u'বিষ্ণুপ্রিয়া মণিপুরী', 'bpy'), |
| 385 | | -(u'Беларуская', 'be'), |
| 386 | | -(u'Aragonés', 'an'), |
| 387 | | -(u'Occitan', 'oc'), |
| 388 | | -(u'বাংলা', 'bn'), |
| 389 | | -(u'Kiswahili', 'sw'), |
| 390 | | -(u'Ido', 'io'), |
| 391 | | -(u'Ripoarisch', 'ksh'), |
| 392 | | -(u'Lumbaart', 'lmo'), |
| 393 | | -(u'Frysk', 'fy'), |
| 394 | | -(u'ગુજરાતી', 'gu'), |
| 395 | | -(u'Plattdüütsch', 'nds'), |
| 396 | | -(u'Afrikaans', 'af'), |
| 397 | | -(u'Sicilianu', 'scn'), |
| 398 | | -(u'Runa Simi', 'qu'), |
| 399 | | -(u'Kurdî', 'ku'), |
| 400 | | -(u'كوردی', 'ku'), |
| 401 | | -(u'اردو', 'ur'), |
| 402 | | -(u'Basa Sunda', 'su'), |
| 403 | | -(u'മലയാളം', 'ml'), |
| 404 | | -(u'粵語', 'zh-yue'), |
| 405 | | -(u'Asturianu', 'ast'), |
| 406 | | -(u'Nnapulitano', 'nap'), |
| 407 | | -(u'Žemaitėška', 'bat-smg'), |
| 408 | | -(u'Walon', 'wa'), |
| 409 | | -(u'Чăваш', 'cv'), |
| 410 | | -(u'Gaeilge', 'ga'), |
| 411 | | -(u'Հայերեն', 'hy'), |
| 412 | | -(u'Yorùbá', 'yo'), |
| 413 | | -(u'ಕನ್ನಡ', 'kn'), |
| 414 | | -(u'Тоҷикӣ', 'tg'), |
| 415 | | -(u'Tarandíne', 'roa-tara'), |
| 416 | | -(u'Vèneto', 'vec'), |
| 417 | | -(u'شاہ مکھی پنجابی', 'pnb'), |
| 418 | | -(u'Shāhmukhī Pañjābī', 'pnb'), |
| 419 | | -(u'नेपाली', 'ne'), |
| 420 | | -(u'Gàidhlig', 'gd'), |
| 421 | | -(u'ייִדיש', 'yi'), |
| 422 | | -(u'Bân-lâm-gú', 'zh-min-nan'), |
| 423 | | -(u'O‘zbek', 'uz'), |
| 424 | | -(u'Tatarça', 'tt'), |
| 425 | | -(u'Татарча', 'tt'), |
| 426 | | -(u'Kapampangan', 'pam'), |
| 427 | | -(u'Иронау', 'os'), |
| 428 | | -(u'Саха тыла', 'sah'), |
| 429 | | -(u'Saxa Tyla', 'sah'), |
| 430 | | -(u'Alemannisch', 'als'), |
| 431 | | -(u'Māori', 'mi'), |
| 432 | | -(u'مصرى', 'arz'), |
| 433 | | -(u'Maṣrī', 'arz'), |
| 434 | | -(u'Қазақша', 'kk'), |
| 435 | | -(u'Nāhuatl', 'nah'), |
| 436 | | -(u'Limburgs', 'li'), |
| 437 | | -(u'Hornjoserbsce', 'hsb'), |
| 438 | | -(u'گیلکی', 'glk'), |
| 439 | | -(u'Corsu', 'co'), |
| 440 | | -(u'贛語', 'gan'), |
| 441 | | -(u'አማርኛ', 'am'), |
| 442 | | -(u'Монгол', 'mn'), |
| 443 | | -(u'Interlingua', 'ia'), |
| 444 | | -(u'Bikol', 'bcl'), |
| 445 | | -(u'Võro', 'fiu-vro'), |
| 446 | | -(u'Nedersaksisch', 'nds-nl'), |
| 447 | | -(u'Føroyskt', 'fo'), |
| 448 | | -(u'تركمن ', 'tk'), |
| 449 | | -(u'Туркмен', 'tk'), |
| 450 | | -(u'Scots', 'sco'), |
| 451 | | -(u'West-Vlams', 'vls'), |
| 452 | | -(u'සිංහල', 'si'), |
| 453 | | -(u'संस्कृतम्', 'sa'), |
| 454 | | -(u'Boarisch', 'bar'), |
| 455 | | -(u'မ္ရန္မာစာ', 'my'), #Needs fix |
| 456 | | -(u'Gaelg', 'gv'), |
| 457 | | -(u'ދިވެހިބަސް', 'dv'), |
| 458 | | -(u'Nouormand', 'nrm'), |
| 459 | | -(u'Normaund', 'nrm'), |
| 460 | | -(u'Pangasinan', 'pag'), |
| 461 | | -(u'Rumantsch', 'rm'), |
| 462 | | -(u'Basa Banyumasan', 'map-bms'), |
| 463 | | -(u'Zazaki', 'diq'), |
| 464 | | -(u'Soranî', 'ckb'), |
| 465 | | -(u'کوردی', 'ckb'), |
| 466 | | -(u'Sámegiella', 'se'), |
| 467 | | -(u'مَزِروني', 'mzn'), |
| 468 | | -(u'吴语', 'wuu'), |
| 469 | | -(u'Oyghurque', 'ug'), |
| 470 | | -(u'Furlan', 'fur'), |
| 471 | | -(u'Líguru', 'lij'), |
| 472 | | -(u'Malti', 'mt'), |
| 473 | | -(u'भोजपुरी', 'bh'), |
| 474 | | -(u'Novial', 'nov'), |
| 475 | | -(u'Malagasy', 'mg'), |
| 476 | | -(u'Kaszëbsczi', 'csb'), |
| 477 | | -(u'Ilokano', 'ilo'), |
| 478 | | -(u'Sardu', 'sc'), |
| 479 | | -(u'古文', 'zh-classical'), |
| 480 | | -(u'文言文', 'zh-classical'), |
| 481 | | -(u'ភាសាខ្មែរ', 'km'), |
| 482 | | -(u'Dzhudezmo', 'lad'), |
| 483 | | -(u'पाऴि', 'pi'), |
| 484 | | -(u'Englisc', 'ang'), |
| 485 | | -(u'Chavacano de Zamboanga', 'cbk-zam'), |
| 486 | | -(u'བོད་སྐད', 'bo'), |
| 487 | | -(u'Fiji Hindi', 'hif'), |
| 488 | | -(u'Arpitan', 'frp'), |
| 489 | | -(u'Hak-kâ-fa', 'hak'), |
| 490 | | -(u'客家話', 'hak'), |
| 491 | | -(u'Kernewek', 'kw'), |
| 492 | | -(u'Karnuack', 'kw'), |
| 493 | | -(u'ਪੰਜਾਬੀ', 'pa'), |
| 494 | | -(u'پښتو', 'ps'), |
| 495 | | -(u'Хальмг', 'xal'), |
| 496 | | -(u'Ślůnski', 'szl'), |
| 497 | | -(u'Deitsch', 'pdc'), |
| 498 | | -(u'Hawai`i', 'haw'), |
| 499 | | -(u'Seeltersk', 'stq'), |
| 500 | | -(u'Interlingue', 'ie'), |
| 501 | | -(u'Diné bizaad', 'nv'), |
| 502 | | -(u'Na Vosa Vakaviti', 'fj'), |
| 503 | | -(u'Qırımtatarca', 'crh'), |
| 504 | | -(u'Коми', 'kv'), |
| 505 | | -(u'faka Tonga', 'to'), |
| 506 | | -(u'Bahsa Acèh', 'ace'), |
| 507 | | -(u'Soomaaliga', 'so'), |
| 508 | | -(u'Эрзянь', 'myv'), |
| 509 | | -(u'Erzjanj Kelj', 'myv'), |
| 510 | | -(u"Avañe'ẽ", 'gn'), |
| 511 | | -(u'Къарачай-Малкъар', 'krc'), |
| 512 | | -(u'Qarachay-Malqar', 'krc'), |
| 513 | | -(u'Estremeñu', 'ext'), |
| 514 | | -(u'Lingala', 'ln'), |
| 515 | | -(u'Кыргызча', 'ky'), |
| 516 | | -(u'Олык Марий', 'mhr'), |
| 517 | | -(u'Olyk Marij', 'mhr'), |
| 518 | | -(u'ܐܪܡܝܐ', 'arc'), |
| 519 | | -(u'Emiliàn e rumagnòl', 'eml'), |
| 520 | | -(u'Lojban', 'jbo'), |
| 521 | | -(u'Picard', 'pcd'), |
| 522 | | -(u'Aymar', 'ay'), |
| 523 | | -(u'Wolof', 'wo'), |
| 524 | | -(u'chiTumbuka', 'tum'), |
| 525 | | -(u'Taqbaylit', 'kab'), |
| 526 | | -(u'Башҡорт', 'ba'), |
| 527 | | -(u'Frasch', 'frr'), |
| 528 | | -(u'Reo Mā`ohi', 'ty'), |
| 529 | | -(u'Tok Pisin', 'tpi'), |
| 530 | | -(u'Papiamentu', 'pap'), |
| 531 | | -(u'Zeêuws', 'zea'), |
| 532 | | -(u'Sranantongo', 'srn'), |
| 533 | | -(u'Kalaallisut', 'kl'), |
| 534 | | -(u'Удмурт кыл', 'udm'), |
| 535 | | -(u'Нохчийн', 'ce'), |
| 536 | | -(u'Igbo', 'ig'), |
| 537 | | -(u'Перем Коми', 'koi'), |
| 538 | | -(u'Perem Komi', 'koi'), |
| 539 | | -(u'ଓଡ଼ିଆ', 'or'), |
| 540 | | -(u'Dolnoserbski', 'dsb'), |
| 541 | | -(u'KiKongo', 'kg'), |
| 542 | | -(u'ລາວ', 'lo'), |
| 543 | | -(u'Аҧсуа', 'ab'), |
| 544 | | -(u'Мокшень', 'mdf'), |
| 545 | | -(u'Mokshanj Kälj', 'mdf'), |
| 546 | | -(u'romani - रोमानी', 'rmy'), |
| 547 | | -(u'Кырык Мары', 'mrj'), |
| 548 | | -(u'Kyryk Mary', 'mrj'), |
| 549 | | -(u'Bahasa Banjar', 'bjn'), |
| 550 | | -(u'Словѣньскъ', 'cu'), |
| 551 | | -(u'Páigina Percipal', 'mwl'), |
| 552 | | -(u'Qaraqalpaqsha', 'kaa'), |
| 553 | | -(u'Gagana Samoa', 'sm'), |
| 554 | | -(u'Молдовеняскэ', 'mo'), |
| 555 | | -(u'Tetun', 'tet'), |
| 556 | | -(u'Авар', 'av'), |
| 557 | | -(u'कश्मीरी', 'ks'), |
| 558 | | -(u'كشميري', 'ks'), |
| 559 | | -(u'𐌲𐌿𐍄𐌹𐍃𐌺', 'got'), #Needs fix |
| 560 | | -(u'سنڌي، سندھی ، सिन्ध', 'sd'), |
| 561 | | -(u'Bamanankan', 'bm'), |
| 562 | | -(u'dorerin Naoero', 'na'), |
| 563 | | -(u'Norfuk', 'pih'), |
| 564 | | -(u'Ποντιακά', 'pnt'), |
| 565 | | -(u'ᐃᓄᒃᑎᑐᑦ', 'iu'), |
| 566 | | -(u'Iñupiak', 'ik'), |
| 567 | | -(u'Bislama', 'bi'), |
| 568 | | -(u'ᏣᎳᎩ', 'chr'), |
| 569 | | -(u'অসমীয়া', 'as'), |
| 570 | | -(u'Mìng-dĕ̤ng-ngṳ̄', 'cdo'), |
| 571 | | -(u'Eʋegbe', 'ee'), |
| 572 | | -(u'SiSwati', 'ss'), |
| 573 | | -(u'Oromoo', 'om'), |
| 574 | | -(u'Cuengh', 'za'), |
| 575 | | -(u'isiZulu', 'zu'), |
| 576 | | -(u'ትግርኛ', 'ti'), |
| 577 | | -(u'Tshivenda', 've'), |
| 578 | | -(u'Xitsonga', 'ts'), |
| 579 | | -(u'هَوُسَ', 'ha'), |
| 580 | | -(u'ཇོང་ཁ', 'dz'), |
| 581 | | -(u'Sängö', 'sg'), |
| 582 | | -(u'Chamoru', 'ch'), |
| 583 | | -(u'Nehiyaw', 'cr'), |
| 584 | | -(u'isiXhosa', 'xh'), |
| 585 | | -(u'Akana', 'ak'), |
| 586 | | -(u'Sesotho', 'st'), |
| 587 | | -(u'Ikinyarwanda', 'rw'), |
| 588 | | -(u'Setswana', 'tn'), |
| 589 | | -(u'Gĩkũyũ', 'ki'), |
| 590 | | -(u'Буряад', 'bxr'), |
| 591 | | -(u'Basa Ugi', 'bug'), |
| 592 | | -(u'Chi-Chewa', 'ny'), |
| 593 | | -(u'Лакку', 'lbe'), |
| 594 | | -(u'Twi', 'tw'), |
| 595 | | -(u'chiShona', 'sn'), |
| 596 | | -(u'Kirundi', 'rn'), |
| 597 | | -(u'Fulfulde', 'ff'), |
| 598 | | -(u'Tsetsêhestâhese', 'chy'), |
| 599 | | -(u'Luganda', 'lg'), |
| 600 | | -(u'Oshiwambo', 'ng'), |
| 601 | | -(u'ꆇꉙ', 'ii'), |
| 602 | | -(u'Choctaw', 'cho'), |
| 603 | | -(u'Ebon', 'mh'), |
| 604 | | -(u'Afar', 'aa'), |
| 605 | | -(u'Kuanyama', 'kj'), |
| 606 | | -(u'Hiri Motu', 'ho'), |
| 607 | | -(u'Muskogee', 'mus'), |
| 608 | | -(u'Kanuri', 'kr'), |
| 609 | | -(u'Otsiherero', 'hz'), |
| 610 | | -]) |
| 611 | | - |
| 612 | | - |
| 613 | | -def language_map(): |
| 614 | | - return utils.invert_dict(MAPPING) |
| Index: trunk/tools/editor_trends/manage.py |
| — | — | @@ -24,6 +24,7 @@ |
| 25 | 25 | import datetime |
| 26 | 26 | from argparse import ArgumentParser |
| 27 | 27 | from argparse import RawTextHelpFormatter |
| | 28 | +import ConfigParser |
| 28 | 29 | |
| 29 | 30 | |
| 30 | 31 | import configuration |
| — | — | @@ -44,15 +45,61 @@ |
| 45 | 46 | from analyses import count_editors |
| 46 | 47 | |
| 47 | 48 | |
| | 49 | +def show_choices(settings, attr): |
| | 50 | + choices = getattr(settings, attr).items() |
| | 51 | + choices.sort() |
| | 52 | + choices = ['%s\t%s' % (choice[0], choice[1]) for choice in choices] |
| | 53 | + return choices |
| 48 | 54 | |
| 49 | 55 | |
| | 56 | + |
| 50 | 57 | def config_launcher(properties, settings, logger): |
| 51 | 58 | ''' |
| 52 | 59 | Config launcher is used to reconfigure editor trends toolkit. |
| 53 | 60 | ''' |
| 54 | | - settings.load_configuration() |
| | 61 | +# settings.load_configuration() |
| | 62 | +# |
| | 63 | + if not os.path.exists('wiki.cfg') or properties.force: |
| | 64 | + config = ConfigParser.RawConfigParser() |
| | 65 | + project = None |
| | 66 | + language = None |
| | 67 | + #language_map = languages.language_map() |
| | 68 | + working_directory = raw_input('Please indicate where you installed Editor Trends Analytics.\nCurrent location is %s\nPress Enter to accept default.\n' % os.getcwd()) |
| | 69 | + input_location = raw_input('Please indicate where to store the Wikipedia dump files.\nDefault is: %s\nPress Enter to accept default.\n' % settings.input_location) |
| 55 | 70 | |
| | 71 | + while project not in properties.projects.keys(): |
| | 72 | + project = raw_input('Please indicate which project you would like to analyze.\nDefault is: %s\nPress Enter to accept default.\n' % properties.projects[properties.short_project].capitalize()) |
| | 73 | + project = project if len(project) > 0 else properties.short_project |
| | 74 | + if project not in properties.projects.keys(): |
| | 75 | + print 'Valid choices for a project are: %s' % ','.join(properties.projects.keys()) |
| 56 | 76 | |
| | 77 | + while language not in properties.valid_languages: |
| | 78 | + language = raw_input('Please indicate which language of project %s you would like to analyze.\nDefault is: %s\nPress Enter to accept default.\n' % (properties.projects[project].capitalize(), properties.language)) |
| | 79 | + if len(language) == 0: |
| | 80 | + language = properties.language_code |
| | 81 | + language = language if language in properties.valid_languages else properties.language |
| | 82 | + |
| | 83 | + input_location = input_location if len(input_location) > 0 else settings.input_location |
| | 84 | + working_directory = working_directory if len(working_directory) > 0 else os.getcwd() |
| | 85 | + |
| | 86 | + config = ConfigParser.RawConfigParser() |
| | 87 | + config.add_section('file_locations') |
| | 88 | + config.set('file_locations', 'working_directory', working_directory) |
| | 89 | + config.set('file_locations', 'input_location', input_location) |
| | 90 | + config.add_section('wiki') |
| | 91 | + config.set('wiki', 'project', project) |
| | 92 | + config.set('wiki', 'language', language) |
| | 93 | + |
| | 94 | + fh = file_utils.create_binary_filehandle(working_directory, 'wiki.cfg', 'wb') |
| | 95 | + config.write(fh) |
| | 96 | + fh.close() |
| | 97 | + |
| | 98 | + settings.working_directory = config.get('file_locations', 'working_directory') |
| | 99 | + settings.input_location = config.get('file_locations', 'input_location') |
| | 100 | + |
| | 101 | + |
| | 102 | + |
| | 103 | + |
| 57 | 104 | def downloader_launcher(properties, settings, logger): |
| 58 | 105 | ''' |
| 59 | 106 | This launcher calls the dump downloader to download a Wikimedia dump file. |
| Index: trunk/tools/editor_trends/classes/wikiprojects.py |
| — | — | @@ -66,8 +66,9 @@ |
| 67 | 67 | self.short_project = 'wiki' |
| 68 | 68 | self.long_project = 'wikipedia' if self.short_project == 'wiki' else \ |
| 69 | 69 | self.projects.get(self.short_project, None) |
| 70 | | - self.language_code = determine_default_language() |
| 71 | | - self.language = self.get_english_language_name() |
| | 70 | + |
| | 71 | + self.language = determine_default_language() |
| | 72 | + self.language_code = MAPPING[self.language] |
| 72 | 73 | self.valid_languages = self.project_supports_language() |
| 73 | 74 | |
| 74 | 75 | if args: |
| — | — | @@ -85,7 +86,7 @@ |
| 86 | 87 | self.collection = self.get_value('collection') |
| 87 | 88 | self.ignore = self.get_value('except') |
| 88 | 89 | self.clean = self.get_value('new') |
| 89 | | - |
| | 90 | + self.force = self.get_value('force') |
| 90 | 91 | self.project = self.get_projectname() |
| 91 | 92 | self.location = self.get_project_location() |
| 92 | 93 | self.filename = self.generate_wikidump_filename() |
| — | — | @@ -317,7 +318,8 @@ |
| 318 | 319 | Wikipedia project is most likely of interest |
| 319 | 320 | ''' |
| 320 | 321 | language_code = locale.getdefaultlocale()[0] |
| 321 | | - return language_code.split('_')[0] |
| | 322 | + language_code = language_code.split('_')[0] |
| | 323 | + return get_language(language_code) |
| 322 | 324 | |
| 323 | 325 | |
| 324 | 326 | def get_language(language_code): |