* Initially, when the client passes in titles=, pageids=, or revisions= parameter,
* an instance of the ApiPageSet class will normalize titles,
* determine if the pages/revisions exist, and prefetch any additional data page data requested.
- *
+ *
* When generator is used, the result of the generator will become the input for the
* second instance of this class, and all subsequent actions will go use the second instance
- * for all their work.
- *
- * @addtogroup API
+ * for all their work.
+ *
+ * @ingroup API
*/
class ApiPageSet extends ApiQueryBase {
- private $mAllPages; // [ns][dbkey] => page_id or 0 when missing
- private $mTitles, $mGoodTitles, $mMissingTitles, $mMissingPageIDs, $mRedirectTitles;
+ private $mAllPages; // [ns][dbkey] => page_id or negative when missing
+ private $mTitles, $mGoodTitles, $mMissingTitles, $mInvalidTitles;
+ private $mMissingPageIDs, $mRedirectTitles;
private $mNormalizedTitles, $mInterwikiTitles;
private $mResolveRedirects, $mPendingRedirectIDs;
private $mGoodRevIDs, $mMissingRevIDs;
private $mRequestedPageFields;
public function __construct($query, $resolveRedirects = false) {
- parent :: __construct($query, __CLASS__);
+ parent :: __construct($query, 'query');
$this->mAllPages = array ();
$this->mTitles = array();
$this->mGoodTitles = array ();
$this->mMissingTitles = array ();
+ $this->mInvalidTitles = array ();
$this->mMissingPageIDs = array ();
$this->mRedirectTitles = array ();
$this->mNormalizedTitles = array ();
$this->mResolveRedirects = $resolveRedirects;
if($resolveRedirects)
$this->mPendingRedirectIDs = array();
-
+
$this->mFakePageId = -1;
}
}
/**
- * Returns an array [ns][dbkey] => page_id for all requested titles
- * page_id is a unique negative number in case title was not found
+ * Returns an array [ns][dbkey] => page_id for all requested titles.
+ * page_id is a unique negative number in case title was not found.
+ * Invalid titles will also have negative page IDs and will be in namespace 0
*/
public function getAllTitlesByNamespace() {
return $this->mAllPages;
return $this->mMissingTitles;
}
+ /**
+ * Titles that were deemed invalid by Title::newFromText()
+ * The array's index will be unique and negative for each item
+ * @return array of strings (not Title objects)
+ */
+ public function getInvalidTitles() {
+ return $this->mInvalidTitles;
+ }
+
/**
* Page IDs that were not found in the database
* @return array of page IDs
}
/**
- * Get a list of title normalizations - maps the title given
+ * Get a list of title normalizations - maps the title given
* with its normalized version.
- * @return array raw_prefixed_title (string) => prefixed_title (string)
+ * @return array raw_prefixed_title (string) => prefixed_title (string)
*/
public function getNormalizedTitles() {
return $this->mNormalizedTitles;
}
/**
- * Get a list of interwiki titles - maps the title given
+ * Get a list of interwiki titles - maps the title given
* with to the interwiki prefix.
- * @return array raw_prefixed_title (string) => interwiki_prefix (string)
+ * @return array raw_prefixed_title (string) => interwiki_prefix (string)
*/
public function getInterwikiTitles() {
return $this->mInterwikiTitles;
break;
case 'revids' :
if($this->mResolveRedirects)
- $this->dieUsage('revids may not be used with redirect resolution', 'params');
+ $this->setWarning('Redirect resolution cannot be used together with the revids= parameter. '.
+ 'Any redirects the revids= point to have not been resolved.');
+ $this->mResolveRedirects = false;
$this->initFromRevIDs($revids);
break;
default :
*/
public function populateFromPageIDs($pageIDs) {
$this->profileIn();
- $pageIDs = array_map('intval', $pageIDs); // paranoia
$this->initFromPageIds($pageIDs);
$this->profileOut();
}
* Extract all requested fields from the row received from the database
*/
public function processDbRow($row) {
-
+
// Store Title object in various data structures
$title = Title :: makeTitle($row->page_namespace, $row->page_title);
-
- // skip any pages that user has no rights to read
- if ($title->userCanRead()) {
-
- $pageId = intval($row->page_id);
- $this->mAllPages[$row->page_namespace][$row->page_title] = $pageId;
- $this->mTitles[] = $title;
-
- if ($this->mResolveRedirects && $row->page_is_redirect == '1') {
- $this->mPendingRedirectIDs[$pageId] = $title;
- } else {
- $this->mGoodTitles[$pageId] = $title;
- }
-
- foreach ($this->mRequestedPageFields as $fieldName => & $fieldValues)
- $fieldValues[$pageId] = $row-> $fieldName;
+
+ $pageId = intval($row->page_id);
+ $this->mAllPages[$row->page_namespace][$row->page_title] = $pageId;
+ $this->mTitles[] = $title;
+
+ if ($this->mResolveRedirects && $row->page_is_redirect == '1') {
+ $this->mPendingRedirectIDs[$pageId] = $title;
+ } else {
+ $this->mGoodTitles[$pageId] = $title;
}
+
+ foreach ($this->mRequestedPageFields as $fieldName => & $fieldValues)
+ $fieldValues[$pageId] = $row-> $fieldName;
}
-
+
public function finishPageSetGeneration() {
$this->profileIn();
$this->resolvePendingRedirects();
$this->profileOut();
}
-
+
/**
* This method populates internal variables with page information
* based on the given array of title strings.
- *
+ *
* Steps:
* #1 For each title, get data from `page` table
* #2 If page was not found in the DB, store it as missing
- *
+ *
* Additionally, when resolving redirects:
* #3 If no more redirects left, stop.
* #4 For each redirect, get its links from `pagelinks` table.
* #5 Substitute the original LinkBatch object with the new list
- * #6 Repeat from step #1
+ * #6 Repeat from step #1
*/
private function initFromTitles($titles) {
$linkBatch = $this->processTitlesArray($titles);
if($linkBatch->isEmpty())
return;
-
+
$db = $this->getDB();
$set = $linkBatch->constructSet('page', $db);
}
private function initFromPageIds($pageids) {
- if(empty($pageids))
+ if(!count($pageids))
return;
-
+
+ $pageids = array_map('intval', $pageids); // paranoia
$set = array (
'page_id' => $pageids
);
$this->profileDBIn();
$res = $db->select('page', $this->getPageTableFields(), $set, __METHOD__);
$this->profileDBOut();
-
- $this->initFromQueryResult($db, $res, array_flip($pageids), false); // process PageIDs
+
+ $remaining = array_flip($pageids);
+ $this->initFromQueryResult($db, $res, $remaining, false); // process PageIDs
// Resolve any found redirects
$this->resolvePendingRedirects();
}
-
+
/**
* Iterate through the result of the query on 'page' table,
* and for each row create and store title object and save any extra fields requested.
private function initFromQueryResult($db, $res, &$remaining = null, $processTitles = null) {
if (!is_null($remaining) && is_null($processTitles))
ApiBase :: dieDebug(__METHOD__, 'Missing $processTitles parameter when $remaining is provided');
-
+
while ($row = $db->fetchObject($res)) {
$pageId = intval($row->page_id);
else
unset ($remaining[$pageId]);
}
-
+
// Store any extra fields requested by modules
$this->processDbRow($row);
}
$db->freeResult($res);
-
+
if(isset($remaining)) {
// Any items left in the $remaining list are added as missing
if($processTitles) {
else
{
// The remaining pageids do not exist
- if(empty($this->mMissingPageIDs))
+ if(!$this->mMissingPageIDs)
$this->mMissingPageIDs = array_keys($remaining);
else
$this->mMissingPageIDs = array_merge($this->mMissingPageIDs, array_keys($remaining));
private function initFromRevIDs($revids) {
- if(empty($revids))
+ if(!count($revids))
return;
-
+
$db = $this->getDB();
$pageids = array();
$remaining = array_flip($revids);
-
+
$tables = array('revision');
$fields = array('rev_id','rev_page');
$where = array('rev_deleted' => 0, 'rev_id' => $revids);
-
+
// Get pageIDs data from the `page` table
$this->profileDBIn();
$res = $db->select( $tables, $fields, $where, __METHOD__ );
$this->mMissingRevIDs = array_keys($remaining);
// Populate all the page information
- if($this->mResolveRedirects)
- ApiBase :: dieDebug(__METHOD__, 'revids may not be used with redirect resolution');
$this->initFromPageIds(array_keys($pageids));
}
if($this->mResolveRedirects) {
$db = $this->getDB();
$pageFlds = $this->getPageTableFields();
-
+
// Repeat until all redirects have been resolved
// The infinite loop is prevented by keeping all known pages in $this->mAllPages
- while (!empty ($this->mPendingRedirectIDs)) {
-
+ while ($this->mPendingRedirectIDs) {
+
// Resolve redirects by querying the pagelinks table, and repeat the process
// Create a new linkBatch object for the next pass
$linkBatch = $this->getRedirectTargets();
-
+
if ($linkBatch->isEmpty())
break;
-
+
$set = $linkBatch->constructSet('page', $db);
if(false === $set)
break;
-
+
// Get pageIDs data from the `page` table
$this->profileDBIn();
$res = $db->select('page', $pageFlds, $set, __METHOD__);
$this->profileDBOut();
-
+
// Hack: get the ns:titles stored in array(ns => array(titles)) format
$this->initFromQueryResult($db, $res, $linkBatch->data, true);
}
}
private function getRedirectTargets() {
-
- $linkBatch = new LinkBatch();
+ $lb = new LinkBatch();
$db = $this->getDB();
- // find redirect targets for all redirect pages
$this->profileDBIn();
- $res = $db->select('pagelinks', array (
- 'pl_from',
- 'pl_namespace',
- 'pl_title'
- ), array (
- 'pl_from' => array_keys($this->mPendingRedirectIDs
- )), __METHOD__);
+ $res = $db->select('redirect', array(
+ 'rd_from',
+ 'rd_namespace',
+ 'rd_title'
+ ), array('rd_from' => array_keys($this->mPendingRedirectIDs)),
+ __METHOD__
+ );
$this->profileDBOut();
- while ($row = $db->fetchObject($res)) {
-
- $plfrom = intval($row->pl_from);
-
- // Bug 7304 workaround
- // ( http://bugzilla.wikipedia.org/show_bug.cgi?id=7304 )
- // A redirect page may have more than one link.
- // This code will only use the first link returned.
- if (isset ($this->mPendingRedirectIDs[$plfrom])) { // remove line when bug 7304 is fixed
-
- $titleStrFrom = $this->mPendingRedirectIDs[$plfrom]->getPrefixedText();
- $titleStrTo = Title :: makeTitle($row->pl_namespace, $row->pl_title)->getPrefixedText();
- unset ($this->mPendingRedirectIDs[$plfrom]); // remove line when bug 7304 is fixed
-
- // Avoid an infinite loop by checking if we have already processed this target
- if (!isset ($this->mAllPages[$row->pl_namespace][$row->pl_title])) {
- $linkBatch->add($row->pl_namespace, $row->pl_title);
- }
- } else {
- // This redirect page has more than one link.
- // This is very slow, but safer until bug 7304 is resolved
- $title = Title :: newFromID($plfrom);
- $titleStrFrom = $title->getPrefixedText();
-
+ while($row = $db->fetchObject($res))
+ {
+ $rdfrom = intval($row->rd_from);
+ $from = $this->mPendingRedirectIDs[$rdfrom]->getPrefixedText();
+ $to = Title::makeTitle($row->rd_namespace, $row->rd_title)->getPrefixedText();
+ unset($this->mPendingRedirectIDs[$rdfrom]);
+ if(!isset($this->mAllPages[$row->rd_namespace][$row->rd_title]))
+ $lb->add($row->rd_namespace, $row->rd_title);
+ $this->mRedirectTitles[$from] = $to;
+ }
+ $db->freeResult($res);
+ if($this->mPendingRedirectIDs)
+ {
+ # We found pages that aren't in the redirect table
+ # Add them
+ foreach($this->mPendingRedirectIDs as $id => $title)
+ {
$article = new Article($title);
- $text = $article->getContent();
- $titleTo = Title :: newFromRedirect($text);
- $titleStrTo = $titleTo->getPrefixedText();
-
- if (is_null($titleStrTo))
- ApiBase :: dieDebug(__METHOD__, 'Bug7304 workaround: redir target from {$title->getPrefixedText()} not found');
-
- // Avoid an infinite loop by checking if we have already processed this target
- if (!isset ($this->mAllPages[$titleTo->getNamespace()][$titleTo->getDBkey()])) {
- $linkBatch->addObj($titleTo);
- }
+ $rt = $article->insertRedirect();
+ if(!$rt)
+ # What the hell. Let's just ignore this
+ continue;
+ $lb->addObj($rt);
+ $this->mRedirectTitles[$title->getPrefixedText()] = $rt->getPrefixedText();
+ unset($this->mPendingRedirectIDs[$id]);
}
-
- $this->mRedirectTitles[$titleStrFrom] = $titleStrTo;
}
- $db->freeResult($res);
-
- // All IDs must exist in the page table
- if (!empty($this->mPendingRedirectIDs[$plfrom]))
- ApiBase :: dieDebug(__METHOD__, 'Invalid redirect IDs were found');
-
- return $linkBatch;
+ return $lb;
}
/**
* Given an array of title strings, convert them into Title objects.
* Alternativelly, an array of Title objects may be given.
- * This method validates access rights for the title,
+ * This method validates access rights for the title,
* and appends normalization values to the output.
- *
+ *
* @return LinkBatch of title objects.
*/
private function processTitlesArray($titles) {
$linkBatch = new LinkBatch();
foreach ($titles as $title) {
-
+
$titleObj = is_string($title) ? Title :: newFromText($title) : $title;
if (!$titleObj)
- $this->dieUsage("bad title $titleString", 'invalidtitle');
-
+ {
+ # Handle invalid titles gracefully
+ $this->mAllpages[0][$title] = $this->mFakePageId;
+ $this->mInvalidTitles[$this->mFakePageId] = $title;
+ $this->mFakePageId--;
+ continue; // There's nothing else we can do
+ }
$iw = $titleObj->getInterwiki();
- if (!empty($iw)) {
+ if (strval($iw) !== '') {
// This title is an interwiki link.
$this->mInterwikiTitles[$titleObj->getPrefixedText()] = $iw;
- } else {
+ } else {
// Validation
if ($titleObj->getNamespace() < 0)
- $this->dieUsage("No support for special page $titleString has been implemented", 'unsupportednamespace');
- if (!$titleObj->userCanRead())
- $this->dieUsage("No read permission for $titleString", 'titleaccessdenied');
+ $this->dieUsage("No support for special pages has been implemented", 'unsupportednamespace');
$linkBatch->addObj($titleObj);
}
-
+
// Make sure we remember the original title that was given to us
// This way the caller can correlate new titles with the originally requested,
// i.e. namespace is localized or capitalization is different
return $linkBatch;
}
-
+
protected function getAllowedParams() {
return array (
'titles' => array (
return __CLASS__ . ': $Id$';
}
}
-
dépôts / lhc / web / wiklou.git / blobdiff