| Index: branches/wmf/1.18wmf1/includes/Export.php |
| — | — | @@ -41,6 +41,7 @@ |
| 42 | 42 | const CURRENT = 2; |
| 43 | 43 | const STABLE = 4; // extension defined |
| 44 | 44 | const LOGS = 8; |
| | 45 | + const RANGE = 16; |
| 45 | 46 | |
| 46 | 47 | const BUFFER = 0; |
| 47 | 48 | const STREAM = 1; |
| — | — | @@ -56,7 +57,8 @@ |
| 57 | 58 | * main query is still running. |
| 58 | 59 | * |
| 59 | 60 | * @param $db Database |
| 60 | | - * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT, |
| | 61 | + * @param $history Mixed: one of WikiExporter::FULL, WikiExporter::CURRENT, |
| | 62 | + * WikiExporter::RANGE or WikiExporter::STABLE, |
| 61 | 63 | * or an associative array: |
| 62 | 64 | * offset: non-inclusive offset at which to start the query |
| 63 | 65 | * limit: maximum number of rows to return |
| — | — | @@ -120,6 +122,21 @@ |
| 121 | 123 | } |
| 122 | 124 | |
| 123 | 125 | /** |
| | 126 | + * Dumps a series of page and revision records for those pages |
| | 127 | + * in the database with revisions falling within the rev_id range given. |
| | 128 | + * @param $start Int: inclusive lower limit (this id is included) |
| | 129 | + * @param $end Int: Exclusive upper limit (this id is not included) |
| | 130 | + * If 0, no upper limit. |
| | 131 | + */ |
| | 132 | + public function revsByRange( $start, $end ) { |
| | 133 | + $condition = 'rev_id >= ' . intval( $start ); |
| | 134 | + if ( $end ) { |
| | 135 | + $condition .= ' AND rev_id < ' . intval( $end ); |
| | 136 | + } |
| | 137 | + return $this->dumpFrom( $condition ); |
| | 138 | + } |
| | 139 | + |
| | 140 | + /** |
| 124 | 141 | * @param $title Title |
| 125 | 142 | */ |
| 126 | 143 | public function pageByTitle( $title ) { |
| — | — | @@ -259,6 +276,10 @@ |
| 260 | 277 | wfProfileOut( __METHOD__ ); |
| 261 | 278 | throw new MWException( __METHOD__ . " given invalid history dump type." ); |
| 262 | 279 | } |
| | 280 | + } elseif ( $this->history & WikiExporter::RANGE ) { |
| | 281 | + # Dump of revisions within a specified range |
| | 282 | + $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' ); |
| | 283 | + $opts['ORDER BY'] = 'rev_page ASC, rev_id ASC'; |
| 263 | 284 | } else { |
| 264 | 285 | # Uknown history specification parameter? |
| 265 | 286 | wfProfileOut( __METHOD__ ); |
| — | — | @@ -354,10 +375,6 @@ |
| 355 | 376 | * @ingroup Dump |
| 356 | 377 | */ |
| 357 | 378 | class XmlDumpWriter { |
| 358 | | - var $firstPageWritten = 0; |
| 359 | | - var $lastPageWritten = 0; |
| 360 | | - var $pageInProgress = 0; |
| 361 | | - |
| 362 | 379 | /** |
| 363 | 380 | * Returns the export schema version. |
| 364 | 381 | * @return string |
| — | — | @@ -461,7 +478,6 @@ |
| 462 | 479 | $title = Title::makeTitle( $row->page_namespace, $row->page_title ); |
| 463 | 480 | $out .= ' ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n"; |
| 464 | 481 | $out .= ' ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n"; |
| 465 | | - $this->pageInProgress = $row->page_id; |
| 466 | 482 | if ( $row->page_is_redirect ) { |
| 467 | 483 | $out .= ' ' . Xml::element( 'redirect', array() ) . "\n"; |
| 468 | 484 | } |
| — | — | @@ -482,10 +498,6 @@ |
| 483 | 499 | */ |
| 484 | 500 | function closePage() { |
| 485 | 501 | return " </page>\n"; |
| 486 | | - if (! $this->firstPageWritten) { |
| 487 | | - $this->firstPageWritten = $this->pageInProgress; |
| 488 | | - } |
| 489 | | - $this->lastPageWritten = $this->pageInProgress; |
| 490 | 502 | } |
| 491 | 503 | |
| 492 | 504 | /** |
| — | — | @@ -632,7 +644,7 @@ |
| 633 | 645 | */ |
| 634 | 646 | function writeUpload( $file, $dumpContents = false ) { |
| 635 | 647 | if ( $file->isOld() ) { |
| 636 | | - $archiveName = " " . |
| | 648 | + $archiveName = " " . |
| 637 | 649 | Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n"; |
| 638 | 650 | } else { |
| 639 | 651 | $archiveName = ''; |
| — | — | @@ -640,7 +652,7 @@ |
| 641 | 653 | if ( $dumpContents ) { |
| 642 | 654 | # Dump file as base64 |
| 643 | 655 | # Uses only XML-safe characters, so does not need escaping |
| 644 | | - $contents = ' <contents encoding="base64">' . |
| | 656 | + $contents = ' <contents encoding="base64">' . |
| 645 | 657 | chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) . |
| 646 | 658 | " </contents>\n"; |
| 647 | 659 | } else { |
| — | — | @@ -651,7 +663,7 @@ |
| 652 | 664 | $this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) . |
| 653 | 665 | " " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" . |
| 654 | 666 | " " . Xml::element( 'filename', null, $file->getName() ) . "\n" . |
| 655 | | - $archiveName . |
| | 667 | + $archiveName . |
| 656 | 668 | " " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" . |
| 657 | 669 | " " . Xml::element( 'size', null, $file->getSize() ) . "\n" . |
| 658 | 670 | " " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" . |
| — | — | @@ -700,19 +712,33 @@ |
| 701 | 713 | print $string; |
| 702 | 714 | } |
| 703 | 715 | |
| | 716 | + /** |
| | 717 | + * Close the old file, move it to a specified name, |
| | 718 | + * and reopen new file with the old name. Use this |
| | 719 | + * for writing out a file in multiple pieces |
| | 720 | + * at specified checkpoints (e.g. every n hours). |
| | 721 | + * @param $newname mixed File name. May be a string or an array with one element |
| | 722 | + */ |
| 704 | 723 | function closeRenameAndReopen( $newname ) { |
| 705 | 724 | return; |
| 706 | 725 | } |
| 707 | 726 | |
| 708 | | - function closeAndRename( $newname ) { |
| | 727 | + /** |
| | 728 | + * Close the old file, and move it to a specified name. |
| | 729 | + * Use this for the last piece of a file written out |
| | 730 | + * at specified checkpoints (e.g. every n hours). |
| | 731 | + * @param $newname mixed File name. May be a string or an array with one element |
| | 732 | + * @param $open bool If true, a new file with the old filename will be opened again for writing (default: false) |
| | 733 | + */ |
| | 734 | + function closeAndRename( $newname, $open = false ) { |
| 709 | 735 | return; |
| 710 | 736 | } |
| 711 | 737 | |
| 712 | | - function rename( $newname ) { |
| 713 | | - return; |
| 714 | | - } |
| 715 | | - |
| 716 | | - function getFilename() { |
| | 738 | + /** |
| | 739 | + * Returns the name of the file or files which are |
| | 740 | + * being written to, if there are any. |
| | 741 | + */ |
| | 742 | + function getFilenames() { |
| 717 | 743 | return NULL; |
| 718 | 744 | } |
| 719 | 745 | } |
| — | — | @@ -722,8 +748,7 @@ |
| 723 | 749 | * @ingroup Dump |
| 724 | 750 | */ |
| 725 | 751 | class DumpFileOutput extends DumpOutput { |
| 726 | | - var $handle; |
| 727 | | - var $filename; |
| | 752 | + protected $handle, $filename; |
| 728 | 753 | |
| 729 | 754 | function __construct( $file ) { |
| 730 | 755 | $this->handle = fopen( $file, "wt" ); |
| — | — | @@ -734,58 +759,39 @@ |
| 735 | 760 | fputs( $this->handle, $string ); |
| 736 | 761 | } |
| 737 | 762 | |
| 738 | | - /** |
| 739 | | - * Close the old file, move it to a specified name, |
| 740 | | - * and reopen new file with the old name. Use this |
| 741 | | - * for writing out a file in multiple pieces |
| 742 | | - * at specified checkpoints (e.g. every n hours). |
| 743 | | - */ |
| 744 | 763 | function closeRenameAndReopen( $newname ) { |
| 745 | | - if ( is_array($newname) ) { |
| 746 | | - if (count($newname) > 1) { |
| 747 | | - WfDie("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n"); |
| 748 | | - } |
| 749 | | - else { |
| 750 | | - $newname = $newname[0]; |
| 751 | | - } |
| 752 | | - } |
| 753 | | - if ( $newname ) { |
| 754 | | - fclose( $this->handle ); |
| 755 | | - rename( $this->filename, $newname ); |
| 756 | | - $this->handle = fopen( $this->filename, "wt" ); |
| 757 | | - } |
| | 764 | + $this->closeAndRename( $newname, true ); |
| 758 | 765 | } |
| 759 | 766 | |
| 760 | | - function closeAndRename( $newname ) { |
| 761 | | - if ( is_array($newname) ) { |
| 762 | | - if (count($newname) > 1) { |
| 763 | | - throw new MWException("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n"); |
| | 767 | + function renameOrException( $newname ) { |
| | 768 | + if (! rename( $this->filename, $newname ) ) { |
| | 769 | + throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" ); |
| 764 | 770 | } |
| 765 | | - else { |
| | 771 | + } |
| | 772 | + |
| | 773 | + function checkRenameArgCount( $newname ) { |
| | 774 | + if ( is_array( $newname ) ) { |
| | 775 | + if ( count( $newname ) > 1 ) { |
| | 776 | + throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" ); |
| | 777 | + } else { |
| 766 | 778 | $newname = $newname[0]; |
| 767 | 779 | } |
| 768 | 780 | } |
| 769 | | - if ( $newname ) { |
| 770 | | - fclose( $this->handle ); |
| 771 | | - rename( $this->filename, $newname ); |
| 772 | | - } |
| | 781 | + return $newname; |
| 773 | 782 | } |
| 774 | 783 | |
| 775 | | - function rename( $newname ) { |
| 776 | | - if ( is_array($newname) ) { |
| 777 | | - if (count($newname) > 1) { |
| 778 | | - WfDie("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n"); |
| | 784 | + function closeAndRename( $newname, $open = false ) { |
| | 785 | + $newname = $this->checkRenameArgCount( $newname ); |
| | 786 | + if ( $newname ) { |
| | 787 | + fclose( $this->handle ); |
| | 788 | + $this->renameOrException( $newname ); |
| | 789 | + if ( $open ) { |
| | 790 | + $this->handle = fopen( $this->filename, "wt" ); |
| 779 | 791 | } |
| 780 | | - else { |
| 781 | | - $newname = $newname[0]; |
| 782 | | - } |
| 783 | 792 | } |
| 784 | | - if ( $newname ) { |
| 785 | | - rename( $this->filename, $newname ); |
| 786 | | - } |
| 787 | 793 | } |
| 788 | 794 | |
| 789 | | - function getFilename() { |
| | 795 | + function getFilenames() { |
| 790 | 796 | return $this->filename; |
| 791 | 797 | } |
| 792 | 798 | } |
| — | — | @@ -797,19 +803,19 @@ |
| 798 | 804 | * @ingroup Dump |
| 799 | 805 | */ |
| 800 | 806 | class DumpPipeOutput extends DumpFileOutput { |
| 801 | | - var $command; |
| | 807 | + protected $command, $filename; |
| 802 | 808 | |
| 803 | 809 | function __construct( $command, $file = null ) { |
| 804 | 810 | if ( !is_null( $file ) ) { |
| 805 | 811 | $command .= " > " . wfEscapeShellArg( $file ); |
| 806 | 812 | } |
| 807 | | - |
| 808 | | - $this->startCommand($command); |
| | 813 | + |
| | 814 | + $this->startCommand( $command ); |
| 809 | 815 | $this->command = $command; |
| 810 | 816 | $this->filename = $file; |
| 811 | 817 | } |
| 812 | 818 | |
| 813 | | - function startCommand($command) { |
| | 819 | + function startCommand( $command ) { |
| 814 | 820 | $spec = array( |
| 815 | 821 | 0 => array( "pipe", "r" ), |
| 816 | 822 | ); |
| — | — | @@ -818,59 +824,24 @@ |
| 819 | 825 | $this->handle = $pipes[0]; |
| 820 | 826 | } |
| 821 | 827 | |
| 822 | | - /** |
| 823 | | - * Close the old file, move it to a specified name, |
| 824 | | - * and reopen new file with the old name. |
| 825 | | - */ |
| 826 | 828 | function closeRenameAndReopen( $newname ) { |
| 827 | | - if ( is_array($newname) ) { |
| 828 | | - if (count($newname) > 1) { |
| 829 | | - WfDie("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n"); |
| 830 | | - } |
| 831 | | - else { |
| 832 | | - $newname = $newname[0]; |
| 833 | | - } |
| 834 | | - } |
| 835 | | - if ( $newname ) { |
| 836 | | - fclose( $this->handle ); |
| 837 | | - proc_close($this->procOpenResource); |
| 838 | | - rename( $this->filename, $newname ); |
| 839 | | - $command = $this->command; |
| 840 | | - $command .= " > " . wfEscapeShellArg( $this->filename ); |
| 841 | | - $this->startCommand($command); |
| 842 | | - } |
| | 829 | + $this->closeAndRename( $newname, true ); |
| 843 | 830 | } |
| 844 | 831 | |
| 845 | | - function closeAndRename( $newname ) { |
| 846 | | - if ( is_array($newname) ) { |
| 847 | | - if (count($newname) > 1) { |
| 848 | | - throw new MWException("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n"); |
| 849 | | - } |
| 850 | | - else { |
| 851 | | - $newname = $newname[0]; |
| 852 | | - } |
| 853 | | - } |
| | 832 | + function closeAndRename( $newname, $open = false ) { |
| | 833 | + $newname = $this->checkRenameArgCount( $newname ); |
| 854 | 834 | if ( $newname ) { |
| 855 | | -# pclose( $this->handle ); |
| 856 | 835 | fclose( $this->handle ); |
| 857 | | - proc_close($this->procOpenResource); |
| 858 | | - rename( $this->filename, $newname ); |
| 859 | | - } |
| 860 | | - } |
| 861 | | - |
| 862 | | - function rename( $newname ) { |
| 863 | | - if ( is_array($newname) ) { |
| 864 | | - if (count($newname) > 1) { |
| 865 | | - WfDie("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n"); |
| | 836 | + proc_close( $this->procOpenResource ); |
| | 837 | + $this->renameOrException( $newname ); |
| | 838 | + if ( $open ) { |
| | 839 | + $command = $this->command; |
| | 840 | + $command .= " > " . wfEscapeShellArg( $this->filename ); |
| | 841 | + $this->startCommand( $command ); |
| 866 | 842 | } |
| 867 | | - else { |
| 868 | | - $newname = $newname[0]; |
| 869 | | - } |
| 870 | 843 | } |
| 871 | | - if ( $newname ) { |
| 872 | | - rename( $this->filename, $newname ); |
| 873 | | - } |
| 874 | 844 | } |
| | 845 | + |
| 875 | 846 | } |
| 876 | 847 | |
| 877 | 848 | /** |
| — | — | @@ -898,64 +869,37 @@ |
| 899 | 870 | * @ingroup Dump |
| 900 | 871 | */ |
| 901 | 872 | class Dump7ZipOutput extends DumpPipeOutput { |
| 902 | | - var $filename; |
| | 873 | + protected $filename; |
| 903 | 874 | |
| 904 | 875 | function __construct( $file ) { |
| | 876 | + $command = $this->setup7zCommand( $file ); |
| | 877 | + parent::__construct( $command ); |
| | 878 | + $this->filename = $file; |
| | 879 | + } |
| | 880 | + |
| | 881 | + function setup7zCommand( $file ) { |
| 905 | 882 | $command = "7za a -bd -si " . wfEscapeShellArg( $file ); |
| 906 | 883 | // Suppress annoying useless crap from p7zip |
| 907 | 884 | // Unfortunately this could suppress real error messages too |
| 908 | 885 | $command .= ' >' . wfGetNull() . ' 2>&1'; |
| 909 | | - parent::__construct( $command ); |
| 910 | | - $this->filename = $file; |
| | 886 | + return( $command ); |
| 911 | 887 | } |
| 912 | 888 | |
| 913 | 889 | function closeRenameAndReopen( $newname ) { |
| 914 | | - if ( is_array($newname) ) { |
| 915 | | - if (count($newname) > 1) { |
| 916 | | - WfDie("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n"); |
| 917 | | - } |
| 918 | | - else { |
| 919 | | - $newname = $newname[0]; |
| 920 | | - } |
| 921 | | - } |
| 922 | | - if ( $newname ) { |
| 923 | | - fclose( $this->handle ); |
| 924 | | - proc_close($this->procOpenResource); |
| 925 | | - rename( $this->filename, $newname ); |
| 926 | | - $command = "7za a -bd -si " . wfEscapeShellArg( $file ); |
| 927 | | - $command .= ' >' . wfGetNull() . ' 2>&1'; |
| 928 | | - $this->startCommand($command); |
| 929 | | - } |
| | 890 | + $this->closeAndRename( $newname, true ); |
| 930 | 891 | } |
| 931 | 892 | |
| 932 | | - function closeAndRename( $newname ) { |
| 933 | | - if ( is_array($newname) ) { |
| 934 | | - if (count($newname) > 1) { |
| 935 | | - throw new MWException("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n"); |
| 936 | | - } |
| 937 | | - else { |
| 938 | | - $newname = $newname[0]; |
| 939 | | - } |
| 940 | | - } |
| | 893 | + function closeAndRename( $newname, $open = false ) { |
| | 894 | + $newname = $this->checkRenameArgCount( $newname ); |
| 941 | 895 | if ( $newname ) { |
| 942 | 896 | fclose( $this->handle ); |
| 943 | | - proc_close($this->procOpenResource); |
| 944 | | - rename( $this->filename, $newname ); |
| 945 | | - } |
| 946 | | - } |
| 947 | | - |
| 948 | | - function rename( $newname ) { |
| 949 | | - if ( is_array($newname) ) { |
| 950 | | - if (count($newname) > 1) { |
| 951 | | - WfDie("Export closeRenameAndReopen: passed multiple argumnts for rename of single file\n"); |
| | 897 | + proc_close( $this->procOpenResource ); |
| | 898 | + $this->renameOrException( $newname ); |
| | 899 | + if ( $open ) { |
| | 900 | + $command = $this->setup7zCommand( $file ); |
| | 901 | + $this->startCommand( $command ); |
| 952 | 902 | } |
| 953 | | - else { |
| 954 | | - $newname = $newname[0]; |
| 955 | | - } |
| 956 | 903 | } |
| 957 | | - if ( $newname ) { |
| 958 | | - rename( $this->filename, $newname ); |
| 959 | | - } |
| 960 | 904 | } |
| 961 | 905 | } |
| 962 | 906 | |
| — | — | @@ -1008,18 +952,14 @@ |
| 1009 | 953 | $this->sink->closeRenameAndReopen( $newname ); |
| 1010 | 954 | } |
| 1011 | 955 | |
| 1012 | | - function closeAndRename( $newname ) { |
| 1013 | | - $this->sink->closeAndRename( $newname ); |
| | 956 | + function closeAndRename( $newname, $open = false ) { |
| | 957 | + $this->sink->closeAndRename( $newname, $open ); |
| 1014 | 958 | } |
| 1015 | 959 | |
| 1016 | | - function rename( $newname ) { |
| 1017 | | - $this->sink->rename( $newname ); |
| | 960 | + function getFilenames() { |
| | 961 | + return $this->sink->getFilenames(); |
| 1018 | 962 | } |
| 1019 | 963 | |
| 1020 | | - function getFilename() { |
| 1021 | | - return $this->sink->getFilename(); |
| 1022 | | - } |
| 1023 | | - |
| 1024 | 964 | /** |
| 1025 | 965 | * Override for page-based filter types. |
| 1026 | 966 | * @return bool |
| — | — | @@ -1169,26 +1109,19 @@ |
| 1170 | 1110 | } |
| 1171 | 1111 | |
| 1172 | 1112 | function closeRenameAndReopen( $newnames ) { |
| 1173 | | - for( $i = 0; $i < $this->count; $i++ ) { |
| 1174 | | - $this->sinks[$i]->closeRenameAndReopen( $newnames[$i] ); |
| 1175 | | - } |
| | 1113 | + $this->closeAndRename( $newnames, true ); |
| 1176 | 1114 | } |
| 1177 | 1115 | |
| 1178 | | - function closeAndRename( $newname ) { |
| 1179 | | - for( $i = 0; $i < $this->count; $i++ ) { |
| 1180 | | - $this->sinks[$i]->closeAndRename( $newnames[$i] ); |
| | 1116 | + function closeAndRename( $newnames, $open = false ) { |
| | 1117 | + for ( $i = 0; $i < $this->count; $i++ ) { |
| | 1118 | + $this->sinks[$i]->closeAndRename( $newnames[$i], $open ); |
| 1181 | 1119 | } |
| 1182 | 1120 | } |
| 1183 | | - function rename( $newnames ) { |
| 1184 | | - for( $i = 0; $i < $this->count; $i++ ) { |
| 1185 | | - $this->sinks[$i]->rename( $newnames[$i] ); |
| 1186 | | - } |
| 1187 | | - } |
| 1188 | 1121 | |
| 1189 | | - function getFilename() { |
| | 1122 | + function getFilenames() { |
| 1190 | 1123 | $filenames = array(); |
| 1191 | | - for( $i = 0; $i < $this->count; $i++ ) { |
| 1192 | | - $filenames[] = $this->sinks[$i]->getFilename(); |
| | 1124 | + for ( $i = 0; $i < $this->count; $i++ ) { |
| | 1125 | + $filenames[] = $this->sinks[$i]->getFilenames(); |
| 1193 | 1126 | } |
| 1194 | 1127 | return $filenames; |
| 1195 | 1128 | } |
| Property changes on: branches/wmf/1.18wmf1/includes |
| ___________________________________________________________________ |
| Modified: svn:mergeinfo |
| 1196 | 1129 | Merged /branches/REL1_18/phase3/includes:r103054 |
| Property changes on: branches/wmf/1.18wmf1 |
| ___________________________________________________________________ |
| Modified: svn:mergeinfo |
| 1197 | 1130 | Merged /branches/REL1_18/phase3:r103054 |