| Index: trunk/extensions/Translate/scripts/groupStatistics.php |
| — | — | @@ -1,21 +1,29 @@ |
| 2 | 2 | <?php |
| 3 | 3 | /** |
| 4 | | - * Statistics about message groups. |
| | 4 | + * Commandline script to general statistics about the localisation level of |
| | 5 | + * localisation for one or more message groups. |
| 5 | 6 | * |
| | 7 | + * @file |
| | 8 | + * @ingroup Script Stats |
| 6 | 9 | * @author Niklas Laxstrom |
| 7 | 10 | * @author Siebrand Mazeland |
| 8 | 11 | * @copyright Copyright © 2007-2010, Niklas Laxström, Siebrand Mazeland |
| 9 | 12 | * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License 2.0 or later |
| 10 | | - * @file |
| 11 | | - * @ingroup Script Stats |
| 12 | 13 | */ |
| 13 | 14 | |
| 14 | 15 | /** |
| 15 | | - * @todo Needs documentation. |
| | 16 | + * Array of the most spoken languages in the world. |
| | 17 | + * Source: http://stats.wikimedia.org/EN/Sitemap.htm. |
| | 18 | + * |
| | 19 | + * Key value pairs of: |
| | 20 | + * [MediaWiki localisation code] => array( |
| | 21 | + * [position in top 50], |
| | 22 | + * [speakers in millions], |
| | 23 | + * [continent where localisation is spoken] |
| | 24 | + * ) |
| | 25 | + * |
| 16 | 26 | */ |
| 17 | 27 | $mostSpokenLanguages = array( |
| 18 | | - // 'language code' => array( position, speakers in millions, continent ), |
| 19 | | - // Source: http://stats.wikimedia.org/EN/Sitemap.htm |
| 20 | 28 | 'en' => array( 1, 1500, 'multiple' ), |
| 21 | 29 | 'zh-hans' => array( 2, 1300, 'asia' ), |
| 22 | 30 | 'zh-hant' => array( 2, 1300, 'asia' ), |
| — | — | @@ -73,16 +81,21 @@ |
| 74 | 82 | ); |
| 75 | 83 | |
| 76 | 84 | /** |
| 77 | | - * @todo Needs documentation. |
| | 85 | + * Variable with key-value pairs with a named index and an array of key-value |
| | 86 | + * pairs where the key is a MessageGroup ID and the value is a weight of the |
| | 87 | + * group in the sum of the values for all the groups in the array. |
| | 88 | + * |
| | 89 | + * Definitions in this variable can be used to report weighted meta localisation |
| | 90 | + * scores. |
| 78 | 91 | */ |
| 79 | 92 | $localisedWeights = array( |
| 80 | 93 | 'wikimedia' => array( |
| 81 | | - 'core-0-mostused' => 40, |
| | 94 | + 'core-0-mostused' => 40, |
| 82 | 95 | 'core' => 30, |
| 83 | 96 | 'ext-0-wikimedia' => 30 |
| 84 | 97 | ), |
| 85 | 98 | 'mediawiki' => array( |
| 86 | | - 'core-0-mostused' => 30, |
| | 99 | + 'core-0-mostused' => 30, |
| 87 | 100 | 'core' => 30, |
| 88 | 101 | 'ext-0-wikimedia' => 20, |
| 89 | 102 | 'ext-0-all' => 20 |
| — | — | @@ -91,7 +104,7 @@ |
| 92 | 105 | |
| 93 | 106 | /** |
| 94 | 107 | * Code map to map localisation codes to Wikimedia project codes. Only |
| 95 | | - * exclusions and remapping is defined here. It is assumed that the first part |
| | 108 | + * exclusion and remapping is defined here. It is assumed that the first part |
| 96 | 109 | * of the localisation code is the WMF project name otherwise (zh-hans -> zh). |
| 97 | 110 | */ |
| 98 | 111 | $wikimediaCodeMap = array( |
| — | — | @@ -163,26 +176,10 @@ |
| 164 | 177 | $optionsWithArgs = array( 'groups', 'output', 'skiplanguages', 'legenddetail', 'legendsummary' ); |
| 165 | 178 | require( dirname( __FILE__ ) . '/cli.inc' ); |
| 166 | 179 | |
| 167 | | -/** |
| 168 | | - * @todo Needs documentation. |
| 169 | | - * @ingroup Stats |
| 170 | | - */ |
| 171 | | -class TranslateStatsOutput extends wikiStatsOutput { |
| 172 | | - function heading() { |
| 173 | | - echo '{| class="sortable wikitable" border="2" cellpadding="4" cellspacing="0" style="background-color: #F9F9F9; border: 1px #AAAAAA solid; border-collapse: collapse; clear:both;" width="100%"' . "\n"; |
| 174 | | - } |
| 175 | | - |
| 176 | | - function summaryheading() { |
| 177 | | - echo "\n" . '{| class="sortable wikitable" border="2" cellpadding="4" cellspacing="0" style="background-color: #F9F9F9; border: 1px #AAAAAA solid; border-collapse: collapse; clear:both;"' . "\n"; |
| 178 | | - } |
| 179 | | - |
| 180 | | - function addFreeText( $freeText ) { |
| 181 | | - echo $freeText; |
| 182 | | - } |
| | 180 | +if ( isset( $options['help'] ) ) { |
| | 181 | + showUsage(); |
| 183 | 182 | } |
| 184 | 183 | |
| 185 | | -if ( isset( $options['help'] ) ) showUsage(); |
| 186 | | - |
| 187 | 184 | // Show help and exit if '--most' does not have a valid value and no groups set |
| 188 | 185 | if ( isset( $options['most'] ) && !isset( $localisedWeights[$options['most']] ) && !isset( $options['groups'] ) ) { |
| 189 | 186 | showUsage(); |
| — | — | @@ -192,54 +189,6 @@ |
| 193 | 190 | $options['output'] = 'default'; |
| 194 | 191 | } |
| 195 | 192 | |
| 196 | | -/** |
| 197 | | - * Print a usage message. |
| 198 | | - */ |
| 199 | | -function showUsage() { |
| 200 | | - $msg = <<<END |
| 201 | | - --help : this help message |
| 202 | | - --groups LIST: comma separated list of groups |
| 203 | | - --skiplanguages LIST: comma separated list of skipped languages |
| 204 | | - --skipzero : skip languages that do not have any localisation at all |
| 205 | | - --fuzzy : add column for fuzzy counts |
| 206 | | - --output TYPE: select an another output engine |
| 207 | | - * 'csv' : Comma Separated Values. |
| 208 | | - * 'wiki' : MediaWiki syntax. |
| 209 | | - * 'text' : Text with tabs. |
| 210 | | - --most : [SCOPE]: report on the 50 most spoken languages. Skipzero is |
| 211 | | - ignored. If a valid scope is defined, the group list |
| 212 | | - and fuzzy are ignored and the localisation levels are |
| 213 | | - weighted and reported. |
| 214 | | - * mediawiki: |
| 215 | | - core-0-mostused (30%) |
| 216 | | - core (30%) |
| 217 | | - ext-0-wikimedia (20%) |
| 218 | | - ext-0-all (20%) |
| 219 | | - * wikimedia: |
| 220 | | - core-0-mostused (40%) |
| 221 | | - core (30%) |
| 222 | | - ext-0-wikimedia (30%) |
| 223 | | - --speakers : add column for number of speakers (est.). Only valid when |
| 224 | | - combined with --most. |
| 225 | | - --nol10n : do not add localised language name if I18ntags is installed. |
| 226 | | - --continent : add a continent column. Only available when output is |
| 227 | | - 'wiki' or not specified. |
| 228 | | - --summary : add a summary with counts and scores per continent category |
| 229 | | - and totals. Only available for a valid 'most' value. |
| 230 | | - --legenddetail : Page name for legend to be transcluded at the top of |
| 231 | | - the details table |
| 232 | | - --legendsummary : Page name for legend to be transcluded at the top of |
| 233 | | - the summary table |
| 234 | | - --wmfscore : Only output WMF language code and weighted score for all |
| 235 | | - language codes for weighing group 'wikimedia' in CSV. This |
| 236 | | - report must keep a stable layout as it is used/will be |
| 237 | | - used in the Wikimedia statistics. |
| 238 | | - |
| 239 | | -END; |
| 240 | | - STDERR( $msg ); |
| 241 | | - exit( 1 ); |
| 242 | | -} |
| 243 | | - |
| 244 | 193 | # Select an output engine |
| 245 | 194 | switch ( $options['output'] ) { |
| 246 | 195 | case 'wiki': |
| — | — | @@ -480,11 +429,15 @@ |
| 481 | 430 | |
| 482 | 431 | $allZero = true; |
| 483 | 432 | foreach ( $columns as $fields ) { |
| 484 | | - if ( intval( $fields[1] ) !== 0 ) $allZero = false; |
| | 433 | + if ( intval( $fields[1] ) !== 0 ) { |
| | 434 | + $allZero = false; |
| | 435 | + } |
| 485 | 436 | } |
| 486 | 437 | |
| 487 | 438 | // Skip dummy languages if requested |
| 488 | | - if ( $allZero && isset( $options['skipzero'] ) ) continue; |
| | 439 | + if ( $allZero && isset( $options['skipzero'] ) ) { |
| | 440 | + continue; |
| | 441 | + } |
| 489 | 442 | |
| 490 | 443 | // Output the the row |
| 491 | 444 | if ( !$wmfscore ) { |
| — | — | @@ -646,3 +599,69 @@ |
| 647 | 600 | echo $code . ';' . number_format( $stats['score'] ) . ";\n"; |
| 648 | 601 | } |
| 649 | 602 | } |
| | 603 | + |
| | 604 | +/** |
| | 605 | + * Print a usage message. |
| | 606 | + */ |
| | 607 | +function showUsage() { |
| | 608 | + $msg = <<<PHP |
| | 609 | + --help : this help message |
| | 610 | + --groups LIST: comma separated list of groups |
| | 611 | + --skiplanguages LIST: comma separated list of skipped languages |
| | 612 | + --skipzero : skip languages that do not have any localisation at all |
| | 613 | + --fuzzy : add column for fuzzy counts |
| | 614 | + --output TYPE: select an another output engine |
| | 615 | + * 'csv' : Comma Separated Values. |
| | 616 | + * 'wiki' : MediaWiki syntax. |
| | 617 | + * 'text' : Text with tabs. |
| | 618 | + --most : [SCOPE]: report on the 50 most spoken languages. Skipzero is |
| | 619 | + ignored. If a valid scope is defined, the group list |
| | 620 | + and fuzzy are ignored and the localisation levels are |
| | 621 | + weighted and reported. |
| | 622 | + * mediawiki: |
| | 623 | + core-0-mostused (30%) |
| | 624 | + core (30%) |
| | 625 | + ext-0-wikimedia (20%) |
| | 626 | + ext-0-all (20%) |
| | 627 | + * wikimedia: |
| | 628 | + core-0-mostused (40%) |
| | 629 | + core (30%) |
| | 630 | + ext-0-wikimedia (30%) |
| | 631 | + --speakers : add column for number of speakers (est.). Only valid when |
| | 632 | + combined with --most. |
| | 633 | + --nol10n : do not add localised language name if I18ntags is installed. |
| | 634 | + --continent : add a continent column. Only available when output is |
| | 635 | + 'wiki' or not specified. |
| | 636 | + --summary : add a summary with counts and scores per continent category |
| | 637 | + and totals. Only available for a valid 'most' value. |
| | 638 | + --legenddetail : Page name for legend to be transcluded at the top of |
| | 639 | + the details table |
| | 640 | + --legendsummary : Page name for legend to be transcluded at the top of |
| | 641 | + the summary table |
| | 642 | + --wmfscore : Only output WMF language code and weighted score for all |
| | 643 | + language codes for weighing group 'wikimedia' in CSV. This |
| | 644 | + report must keep a stable layout as it is used/will be |
| | 645 | + used in the Wikimedia statistics. |
| | 646 | + |
| | 647 | +PHP; |
| | 648 | + STDERR( $msg ); |
| | 649 | + exit( 1 ); |
| | 650 | +} |
| | 651 | + |
| | 652 | +/** |
| | 653 | + * @todo Needs documentation. |
| | 654 | + * @ingroup Stats |
| | 655 | + */ |
| | 656 | +class TranslateStatsOutput extends wikiStatsOutput { |
| | 657 | + function heading() { |
| | 658 | + echo '{| class="sortable wikitable" border="2" cellpadding="4" cellspacing="0" style="background-color: #F9F9F9; border: 1px #AAAAAA solid; border-collapse: collapse; clear:both;" width="100%"' . "\n"; |
| | 659 | + } |
| | 660 | + |
| | 661 | + function summaryheading() { |
| | 662 | + echo "\n" . '{| class="sortable wikitable" border="2" cellpadding="4" cellspacing="0" style="background-color: #F9F9F9; border: 1px #AAAAAA solid; border-collapse: collapse; clear:both;"' . "\n"; |
| | 663 | + } |
| | 664 | + |
| | 665 | + function addFreeText( $freeText ) { |
| | 666 | + echo $freeText; |
| | 667 | + } |
| | 668 | +} |