const CURRENT = 2;
const STABLE = 4; // extension defined
const LOGS = 8;
+ const RANGE = 16;
const BUFFER = 0;
const STREAM = 1;
* make additional queries to pull source data while the
* main query is still running.
*
- * @param $db Database
- * @param $history Mixed: one of WikiExporter::FULL or WikiExporter::CURRENT,
+ * @param $db DatabaseBase
+ * @param $history Mixed: one of WikiExporter::FULL, WikiExporter::CURRENT,
+ * WikiExporter::RANGE or WikiExporter::STABLE,
* or an associative array:
* offset: non-inclusive offset at which to start the query
* limit: maximum number of rows to return
return $this->dumpFrom( $condition );
}
+ /**
+ * Dumps a series of page and revision records for those pages
+ * in the database with revisions falling within the rev_id range given.
+ * @param $start Int: inclusive lower limit (this id is included)
+ * @param $end Int: Exclusive upper limit (this id is not included)
+ * If 0, no upper limit.
+ */
+ public function revsByRange( $start, $end ) {
+ $condition = 'rev_id >= ' . intval( $start );
+ if ( $end ) {
+ $condition .= ' AND rev_id < ' . intval( $end );
+ }
+ return $this->dumpFrom( $condition );
+ }
+
/**
* @param $title Title
*/
wfProfileOut( __METHOD__ );
throw new MWException( __METHOD__ . " given invalid history dump type." );
}
+ } elseif ( $this->history & WikiExporter::RANGE ) {
+ # Dump of revisions within a specified range
+ $join['revision'] = array( 'INNER JOIN', 'page_id=rev_page' );
+ $opts['ORDER BY'] = 'rev_page ASC, rev_id ASC';
} else {
# Uknown history specification parameter?
wfProfileOut( __METHOD__ );
* @ingroup Dump
*/
class XmlDumpWriter {
- protected $firstPageWritten = 0, $lastPageWritten = 0, $pageInProgress = 0;
-
/**
* Returns the export schema version.
* @return string
*/
function schemaVersion() {
- return "0.5";
+ return "0.6";
}
/**
function openPage( $row ) {
$out = " <page>\n";
$title = Title::makeTitle( $row->page_namespace, $row->page_title );
- $out .= ' ' . Xml::elementClean( 'title', array(), $title->getPrefixedText() ) . "\n";
+ $out .= ' ' . Xml::elementClean( 'title', array(), self::canonicalTitle( $title ) ) . "\n";
+ $out .= ' ' . Xml::element( 'ns', array(), strval( $row->page_namespace) ) . "\n";
$out .= ' ' . Xml::element( 'id', array(), strval( $row->page_id ) ) . "\n";
- $this->pageInProgress = $row->page_id;
if ( $row->page_is_redirect ) {
- $out .= ' ' . Xml::element( 'redirect', array() ) . "\n";
+ $page = WikiPage::factory( $title );
+ $redirect = $page->getRedirectTarget();
+ if ( $redirect instanceOf Title && $redirect->isValidRedirectTarget() ) {
+ $out .= ' ' . Xml::element( 'redirect', array( 'title' => self::canonicalTitle( $redirect ) ) ) . "\n";
+ }
+ }
+
+ if ( $row->rev_sha1 ) {
+ $out .= " " . Xml::element('sha1', null, strval( $row->rev_sha1 ) ) . "\n";
+ } else {
+ $out .= " <sha1/>\n";
}
+
if ( $row->page_restrictions != '' ) {
$out .= ' ' . Xml::element( 'restrictions', array(),
strval( $row->page_restrictions ) ) . "\n";
* Closes a <page> section on the output stream.
*
* @access private
+ * @return string
*/
function closePage() {
return " </page>\n";
- if ( !$this->firstPageWritten ) {
- $this->firstPageWritten = $this->pageInProgress;
- }
- $this->lastPageWritten = $this->pageInProgress;
}
/**
// Raw text from the database may have invalid chars
$text = strval( Revision::getRevisionText( $row ) );
$out .= " " . Xml::elementClean( 'text',
- array( 'xml:space' => 'preserve', 'bytes' => $row->rev_len ),
+ array( 'xml:space' => 'preserve', 'bytes' => intval( $row->rev_len ) ),
strval( $text ) ) . "\n";
} else {
// Stub output
$out .= " " . Xml::element( 'text',
- array( 'id' => $row->rev_text_id, 'bytes' => $row->rev_len ),
+ array( 'id' => $row->rev_text_id, 'bytes' => intval( $row->rev_len ) ),
"" ) . "\n";
}
$out .= " " . Xml::element( 'text', array( 'deleted' => 'deleted' ) ) . "\n";
} else {
$title = Title::makeTitle( $row->log_namespace, $row->log_title );
- $out .= " " . Xml::elementClean( 'logtitle', null, $title->getPrefixedText() ) . "\n";
+ $out .= " " . Xml::elementClean( 'logtitle', null, self::canonicalTitle( $title ) ) . "\n";
$out .= " " . Xml::elementClean( 'params',
array( 'xml:space' => 'preserve' ),
strval( $row->log_params ) ) . "\n";
function writeContributor( $id, $text ) {
$out = " <contributor>\n";
- if ( $id ) {
+ if ( $id || !IP::isValid( $text ) ) {
$out .= " " . Xml::elementClean( 'username', null, strval( $text ) ) . "\n";
$out .= " " . Xml::element( 'id', null, strval( $id ) ) . "\n";
} else {
/**
* Warning! This data is potentially inconsistent. :(
+ * @return string
*/
function writeUploads( $row, $dumpContents = false ) {
if ( $row->page_namespace == NS_IMAGE ) {
*/
function writeUpload( $file, $dumpContents = false ) {
if ( $file->isOld() ) {
- $archiveName = " " .
+ $archiveName = " " .
Xml::element( 'archivename', null, $file->getArchiveName() ) . "\n";
} else {
$archiveName = '';
if ( $dumpContents ) {
# Dump file as base64
# Uses only XML-safe characters, so does not need escaping
- $contents = ' <contents encoding="base64">' .
+ $contents = ' <contents encoding="base64">' .
chunk_split( base64_encode( file_get_contents( $file->getPath() ) ) ) .
" </contents>\n";
} else {
$this->writeContributor( $file->getUser( 'id' ), $file->getUser( 'text' ) ) .
" " . Xml::elementClean( 'comment', null, $file->getDescription() ) . "\n" .
" " . Xml::element( 'filename', null, $file->getName() ) . "\n" .
- $archiveName .
+ $archiveName .
" " . Xml::element( 'src', null, $file->getCanonicalUrl() ) . "\n" .
" " . Xml::element( 'size', null, $file->getSize() ) . "\n" .
" " . Xml::element( 'sha1base36', null, $file->getSha1() ) . "\n" .
" </upload>\n";
}
+ /**
+ * Return prefixed text form of title, but using the content language's
+ * canonical namespace. This skips any special-casing such as gendered
+ * user namespaces -- which while useful, are not yet listed in the
+ * XML <siteinfo> data so are unsafe in export.
+ *
+ * @param Title $title
+ * @return string
+ * @since 1.18
+ */
+ public static function canonicalTitle( Title $title ) {
+ if ( $title->getInterwiki() ) {
+ return $title->getPrefixedText();
+ }
+
+ global $wgContLang;
+ $prefix = str_replace( '_', ' ', $wgContLang->getNsText( $title->getNamespace() ) );
+
+ if ( $prefix !== '' ) {
+ $prefix .= ':';
+ }
+
+ return $prefix . $title->getText();
+ }
}
}
/**
- * Close the old file, move it to a specified name,
+ * Close the old file, move it to a specified name,
* and reopen new file with the old name. Use this
* for writing out a file in multiple pieces
* at specified checkpoints (e.g. every n hours).
return;
}
- // TODO: document
- function closeAndRename( $newname ) {
- return;
- }
-
- // TODO: document
- function rename( $newname ) {
+ /**
+ * Close the old file, and move it to a specified name.
+ * Use this for the last piece of a file written out
+ * at specified checkpoints (e.g. every n hours).
+ * @param $newname mixed File name. May be a string or an array with one element
+ * @param $open bool If true, a new file with the old filename will be opened again for writing (default: false)
+ */
+ function closeAndRename( $newname, $open = false ) {
return;
}
- // TODO: document
- function getFilename() {
+ /**
+ * Returns the name of the file or files which are
+ * being written to, if there are any.
+ * @return null
+ */
+ function getFilenames() {
return NULL;
}
}
* @ingroup Dump
*/
class DumpFileOutput extends DumpOutput {
- var $handle;
- var $filename;
+ protected $handle, $filename;
function __construct( $file ) {
$this->handle = fopen( $file, "wt" );
}
function closeRenameAndReopen( $newname ) {
- if ( is_array($newname) ) {
- if (count($newname) > 1) {
- throw new MWException(__METHOD__ . ": passed multiple arguments for rename of single file\n");
- }
- else {
- $newname = $newname[0];
- }
- }
- if ( $newname ) {
- fclose( $this->handle );
- rename( $this->filename, $newname );
- $this->handle = fopen( $this->filename, "wt" );
- }
+ $this->closeAndRename( $newname, true );
}
- function closeAndRename( $newname ) {
- if ( is_array($newname) ) {
- if (count($newname) > 1) {
- throw new MWException(__METHOD__ . ": passed multiple arguments for rename of single file\n");
+ function renameOrException( $newname ) {
+ if (! rename( $this->filename, $newname ) ) {
+ throw new MWException( __METHOD__ . ": rename of file {$this->filename} to $newname failed\n" );
}
- else {
- $newname = $newname[0];
- }
- }
- if ( $newname ) {
- fclose( $this->handle );
- rename( $this->filename, $newname );
- }
}
- function rename( $newname ) {
- if ( is_array($newname) ) {
- if (count($newname) > 1) {
- throw new MWException(__METHOD__ . ": passed multiple arguments for rename of single file\n");
- }
- else {
+ function checkRenameArgCount( $newname ) {
+ if ( is_array( $newname ) ) {
+ if ( count( $newname ) > 1 ) {
+ throw new MWException( __METHOD__ . ": passed multiple arguments for rename of single file\n" );
+ } else {
$newname = $newname[0];
}
}
+ return $newname;
+ }
+
+ function closeAndRename( $newname, $open = false ) {
+ $newname = $this->checkRenameArgCount( $newname );
if ( $newname ) {
- rename( $this->filename, $newname );
+ fclose( $this->handle );
+ $this->renameOrException( $newname );
+ if ( $open ) {
+ $this->handle = fopen( $this->filename, "wt" );
+ }
}
}
- function getFilename() {
+ function getFilenames() {
return $this->filename;
}
}
if ( !is_null( $file ) ) {
$command .= " > " . wfEscapeShellArg( $file );
}
-
- $this->startCommand($command);
+
+ $this->startCommand( $command );
$this->command = $command;
$this->filename = $file;
}
- function startCommand($command) {
+ function startCommand( $command ) {
$spec = array(
0 => array( "pipe", "r" ),
);
$this->handle = $pipes[0];
}
- /**
- * Close the old file, move it to a specified name,
- * and reopen new file with the old name.
- */
function closeRenameAndReopen( $newname ) {
- if ( is_array($newname) ) {
- if (count($newname) > 1) {
- throw new MWException(__METHOD__ . ": passed multiple arguments for rename of single file\n");
- }
- else {
- $newname = $newname[0];
- }
- }
- if ( $newname ) {
- fclose( $this->handle );
- proc_close($this->procOpenResource);
- rename( $this->filename, $newname );
- $command = $this->command;
- $command .= " > " . wfEscapeShellArg( $this->filename );
- $this->startCommand($command);
- }
+ $this->closeAndRename( $newname, true );
}
- function closeAndRename( $newname ) {
- if ( is_array($newname) ) {
- if (count($newname) > 1) {
- throw new MWException(__METHOD__ . ": passed multiple arguments for rename of single file\n");
- }
- else {
- $newname = $newname[0];
- }
- }
+ function closeAndRename( $newname, $open = false ) {
+ $newname = $this->checkRenameArgCount( $newname );
if ( $newname ) {
-# pclose( $this->handle );
fclose( $this->handle );
- proc_close($this->procOpenResource);
- rename( $this->filename, $newname );
- }
- }
-
- function rename( $newname ) {
- if ( is_array($newname) ) {
- if (count($newname) > 1) {
- throw new MWException(__METHOD__ . ": passed multiple arguments for rename of single file\n");
+ proc_close( $this->procOpenResource );
+ $this->renameOrException( $newname );
+ if ( $open ) {
+ $command = $this->command;
+ $command .= " > " . wfEscapeShellArg( $this->filename );
+ $this->startCommand( $command );
}
- else {
- $newname = $newname[0];
- }
- }
- if ( $newname ) {
- rename( $this->filename, $newname );
}
}
+
}
/**
* @ingroup Dump
*/
class Dump7ZipOutput extends DumpPipeOutput {
- protected $filename;
-
function __construct( $file ) {
- $command = "7za a -bd -si " . wfEscapeShellArg( $file );
- // Suppress annoying useless crap from p7zip
- // Unfortunately this could suppress real error messages too
- $command .= ' >' . wfGetNull() . ' 2>&1';
+ $command = $this->setup7zCommand( $file );
parent::__construct( $command );
$this->filename = $file;
}
- function closeRenameAndReopen( $newname ) {
- if ( is_array($newname) ) {
- if (count($newname) > 1) {
- throw new MWException(__METHOD__ . ": passed multiple arguments for rename of single file\n");
- }
- else {
- $newname = $newname[0];
- }
- }
- if ( $newname ) {
- fclose( $this->handle );
- proc_close($this->procOpenResource);
- rename( $this->filename, $newname );
- $command = "7za a -bd -si " . wfEscapeShellArg( $file );
- $command .= ' >' . wfGetNull() . ' 2>&1';
- $this->startCommand($command);
- }
+ function setup7zCommand( $file ) {
+ $command = "7za a -bd -si " . wfEscapeShellArg( $file );
+ // Suppress annoying useless crap from p7zip
+ // Unfortunately this could suppress real error messages too
+ $command .= ' >' . wfGetNull() . ' 2>&1';
+ return( $command );
}
- function closeAndRename( $newname ) {
- if ( is_array($newname) ) {
- if (count($newname) > 1) {
- throw new MWException(__METHOD__ . ": passed multiple arguments for rename of single file\n");
- }
- else {
- $newname = $newname[0];
- }
- }
+ function closeAndRename( $newname, $open = false ) {
+ $newname = $this->checkRenameArgCount( $newname );
if ( $newname ) {
fclose( $this->handle );
- proc_close($this->procOpenResource);
- rename( $this->filename, $newname );
- }
- }
-
- function rename( $newname ) {
- if ( is_array($newname) ) {
- if (count($newname) > 1) {
- throw new MWException(__METHOD__ . ": passed multiple arguments for rename of single file\n");
+ proc_close( $this->procOpenResource );
+ $this->renameOrException( $newname );
+ if ( $open ) {
+ $command = $this->setup7zCommand( $this->filename );
+ $this->startCommand( $command );
}
- else {
- $newname = $newname[0];
- }
- }
- if ( $newname ) {
- rename( $this->filename, $newname );
}
}
}
$this->sink->closeRenameAndReopen( $newname );
}
- function closeAndRename( $newname ) {
- $this->sink->closeAndRename( $newname );
+ function closeAndRename( $newname, $open = false ) {
+ $this->sink->closeAndRename( $newname, $open );
}
- function rename( $newname ) {
- $this->sink->rename( $newname );
- }
-
- function getFilename() {
- return $this->sink->getFilename();
+ function getFilenames() {
+ return $this->sink->getFilenames();
}
/**
}
function closeRenameAndReopen( $newnames ) {
- for( $i = 0; $i < $this->count; $i++ ) {
- $this->sinks[$i]->closeRenameAndReopen( $newnames[$i] );
- }
+ $this->closeAndRename( $newnames, true );
}
- function closeAndRename( $newname ) {
- for( $i = 0; $i < $this->count; $i++ ) {
- $this->sinks[$i]->closeAndRename( $newnames[$i] );
- }
- }
- function rename( $newnames ) {
- for( $i = 0; $i < $this->count; $i++ ) {
- $this->sinks[$i]->rename( $newnames[$i] );
+ function closeAndRename( $newnames, $open = false ) {
+ for ( $i = 0; $i < $this->count; $i++ ) {
+ $this->sinks[$i]->closeAndRename( $newnames[$i], $open );
}
}
- function getFilename() {
+ function getFilenames() {
$filenames = array();
- for( $i = 0; $i < $this->count; $i++ ) {
- $filenames[] = $this->sinks[$i]->getFilename();
+ for ( $i = 0; $i < $this->count; $i++ ) {
+ $filenames[] = $this->sinks[$i]->getFilenames();
}
return $filenames;
}
dépôts / lhc / web / wiklou.git / blobdiff