diff options
author | Andreas Gohr <andi@splitbrain.org> | 2013-02-03 22:57:45 +0100 |
---|---|---|
committer | Andreas Gohr <andi@splitbrain.org> | 2013-02-03 22:57:45 +0100 |
commit | 3da7921f08ecdda929466921ecc50698f1adf99e (patch) | |
tree | 0e179e504399e874bfd785d0a95eec44b76d5952 /inc/indexer.php | |
parent | 6cf2bbfa12b776cf47cb69ae40fb8862f715ad01 (diff) | |
parent | cc4bb766fdac23358d7b586aa3830b9650eed7a8 (diff) | |
download | rpg-3da7921f08ecdda929466921ecc50698f1adf99e.tar.gz rpg-3da7921f08ecdda929466921ecc50698f1adf99e.tar.bz2 |
Merge branch 'master' into future
* master: (162 commits)
fixed revision JS for images
upgraded SimplePie to 1.3.1 FS#2708
removed obsolete browser plugin (migrate does it)
adjust spacing to match standard 1.4em grid
added comment on use of whitelist vs blacklist
Updated idfilter() function for IIS
use var and remove suggestions when needed Use variable for maximum number of suggestions for quicksearch. And hide suggestions when search field is emptied, or when no suggestion are found.
added 'home' class to first link in hierarchical breadcrumbs
reduced required max width to go into tablet mode
re-added linear gradients for firefox
added missing styling for disabled form elements (FS#2705)
fixed acronyms in italics (FS#2684)
improved print styles (includes fixes for FS#2645 and FS#2707)
basic styles improvements
Greek language update
Use list in acl help text, for more structure
Galician language update
touch the config on save, even if no changes were made
unwind the width narrowing commit
put some whitespace between form submit button and fieldset bottom border
...
Conflicts:
lib/plugins/config/admin.php
lib/plugins/config/settings/config.class.php
Diffstat (limited to 'inc/indexer.php')
-rw-r--r-- | inc/indexer.php | 230 |
1 files changed, 220 insertions, 10 deletions
diff --git a/inc/indexer.php b/inc/indexer.php index f22aee3a0..e518907d7 100644 --- a/inc/indexer.php +++ b/inc/indexer.php @@ -102,6 +102,10 @@ function wordlen($w){ * @author Tom N Harris <tnharris@whoopdedo.org> */ class Doku_Indexer { + /** + * @var array $pidCache Cache for getPID() + */ + protected $pidCache = array(); /** * Adds the contents of a page to the fulltext index @@ -120,7 +124,7 @@ class Doku_Indexer { return "locked"; // load known documents - $pid = $this->addIndexKey('page', '', $page); + $pid = $this->getPIDNoLock($page); if ($pid === false) { $this->unlock(); return false; @@ -256,7 +260,7 @@ class Doku_Indexer { return "locked"; // load known documents - $pid = $this->addIndexKey('page', '', $page); + $pid = $this->getPIDNoLock($page); if ($pid === false) { $this->unlock(); return false; @@ -335,6 +339,109 @@ class Doku_Indexer { } /** + * Rename a page in the search index without changing the indexed content. This function doesn't check if the + * old or new name exists in the filesystem. It returns an error if the old page isn't in the page list of the + * indexer and it deletes all previously indexed content of the new page. + * + * @param string $oldpage The old page name + * @param string $newpage The new page name + * @return string|bool If the page was successfully renamed, can be a message in the case of an error + */ + public function renamePage($oldpage, $newpage) { + if (!$this->lock()) return 'locked'; + + $pages = $this->getPages(); + + $id = array_search($oldpage, $pages); + if ($id === false) { + $this->unlock(); + return 'page is not in index'; + } + + $new_id = array_search($newpage, $pages); + if ($new_id !== false) { + // make sure the page is not in the index anymore + if ($this->deletePageNoLock($newpage) !== true) { + return false; + } + + $pages[$new_id] = 'deleted:'.time().rand(0, 9999); + } + + $pages[$id] = $newpage; + + // update index + if (!$this->saveIndex('page', '', $pages)) { + $this->unlock(); + return false; + } + + // reset the pid cache + $this->pidCache = array(); + + $this->unlock(); + return true; + } + + /** + * Renames a meta value in the index. This doesn't change the meta value in the pages, it assumes that all pages + * will be updated. + * + * @param string $key The metadata key of which a value shall be changed + * @param string $oldvalue The old value that shall be renamed + * @param string $newvalue The new value to which the old value shall be renamed, can exist (then values will be merged) + * @return bool|string If renaming the value has been successful, false or error message on error. + */ + public function renameMetaValue($key, $oldvalue, $newvalue) { + if (!$this->lock()) return 'locked'; + + // change the relation references index + $metavalues = $this->getIndex($key, '_w'); + $oldid = array_search($oldvalue, $metavalues); + if ($oldid !== false) { + $newid = array_search($newvalue, $metavalues); + if ($newid !== false) { + // free memory + unset ($metavalues); + + // okay, now we have two entries for the same value. we need to merge them. + $indexline = $this->getIndexKey($key, '_i', $oldid); + if ($indexline != '') { + $newindexline = $this->getIndexKey($key, '_i', $newid); + $pagekeys = $this->getIndex($key, '_p'); + $parts = explode(':', $indexline); + foreach ($parts as $part) { + list($id, $count) = explode('*', $part); + $newindexline = $this->updateTuple($newindexline, $id, $count); + + $keyline = explode(':', $pagekeys[$id]); + // remove old meta value + $keyline = array_diff($keyline, array($oldid)); + // add new meta value when not already present + if (!in_array($newid, $keyline)) { + array_push($keyline, $newid); + } + $pagekeys[$id] = implode(':', $keyline); + } + $this->saveIndex($key, '_p', $pagekeys); + unset($pagekeys); + $this->saveIndexKey($key, '_i', $oldid, ''); + $this->saveIndexKey($key, '_i', $newid, $newindexline); + } + } else { + $metavalues[$oldid] = $newvalue; + if (!$this->saveIndex($key, '_w', $metavalues)) { + $this->unlock(); + return false; + } + } + } + + $this->unlock(); + return true; + } + + /** * Remove a page from the index * * Erases entries in all known indexes. @@ -347,10 +454,26 @@ class Doku_Indexer { if (!$this->lock()) return "locked"; + $result = $this->deletePageNoLock($page); + + $this->unlock(); + + return $result; + } + + /** + * Remove a page from the index without locking the index, only use this function if the index is already locked + * + * Erases entries in all known indexes. + * + * @param string $page a page name + * @return boolean the function completed successfully + * @author Tom N Harris <tnharris@whoopdedo.org> + */ + protected function deletePageNoLock($page) { // load known documents - $pid = $this->addIndexKey('page', '', $page); + $pid = $this->getPIDNoLock($page); if ($pid === false) { - $this->unlock(); return false; } @@ -376,7 +499,6 @@ class Doku_Indexer { } // Save the reverse index if (!$this->saveIndexKey('pageword', '', $pid, "")) { - $this->unlock(); return false; } @@ -393,6 +515,37 @@ class Doku_Indexer { $this->saveIndexKey($metaname.'_p', '', $pid, ''); } + return true; + } + + /** + * Clear the whole index + * + * @return bool If the index has been cleared successfully + */ + public function clear() { + global $conf; + + if (!$this->lock()) return false; + + @unlink($conf['indexdir'].'/page.idx'); + @unlink($conf['indexdir'].'/title.idx'); + @unlink($conf['indexdir'].'/pageword.idx'); + @unlink($conf['indexdir'].'/metadata.idx'); + $dir = @opendir($conf['indexdir']); + if($dir!==false){ + while(($f = readdir($dir)) !== false){ + if(substr($f,-4)=='.idx' && + (substr($f,0,1)=='i' || substr($f,0,1)=='w' + || substr($f,-6)=='_w.idx' || substr($f,-6)=='_i.idx' || substr($f,-6)=='_p.idx')) + @unlink($conf['indexdir']."/$f"); + } + } + @unlink($conf['indexdir'].'/lengths.idx'); + + // clear the pid cache + $this->pidCache = array(); + $this->unlock(); return true; } @@ -454,6 +607,58 @@ class Doku_Indexer { } /** + * Get the numeric PID of a page + * + * @param string $page The page to get the PID for + * @return bool|int The page id on success, false on error + */ + public function getPID($page) { + // return PID without locking when it is in the cache + if (isset($this->pidCache[$page])) return $this->pidCache[$page]; + + if (!$this->lock()) + return false; + + // load known documents + $pid = $this->getPIDNoLock($page); + if ($pid === false) { + $this->unlock(); + return false; + } + + $this->unlock(); + return $pid; + } + + /** + * Get the numeric PID of a page without locking the index. + * Only use this function when the index is already locked. + * + * @param string $page The page to get the PID for + * @return bool|int The page id on success, false on error + */ + protected function getPIDNoLock($page) { + // avoid expensive addIndexKey operation for the most recently requested pages by using a cache + if (isset($this->pidCache[$page])) return $this->pidCache[$page]; + $pid = $this->addIndexKey('page', '', $page); + // limit cache to 10 entries by discarding the oldest element as in DokuWiki usually only the most recently + // added item will be requested again + if (count($this->pidCache) > 10) array_shift($this->pidCache); + $this->pidCache[$page] = $pid; + return $pid; + } + + /** + * Get the page id of a numeric PID + * + * @param int $pid The PID to get the page id for + * @return string The page id + */ + public function getPageFromPID($pid) { + return $this->getIndexKey('page', '', $pid); + } + + /** * Find pages in the fulltext index containing the words, * * The search words must be pre-tokenized, meaning only letters and @@ -946,7 +1151,7 @@ class Doku_Indexer { * @param string $idx name of the index * @param string $suffix subpart identifier * @param string $value line to find in the index - * @return int line number of the value in the index + * @return int|bool line number of the value in the index or false if writing the index failed * @author Tom N Harris <tnharris@whoopdedo.org> */ protected function addIndexKey($idx, $suffix, $value) { @@ -1140,8 +1345,8 @@ class Doku_Indexer { * @author Tom N Harris <tnharris@whoopdedo.org> */ function idx_get_indexer() { - static $Indexer = null; - if (is_null($Indexer)) { + static $Indexer; + if (!isset($Indexer)) { $Indexer = new Doku_Indexer(); } return $Indexer; @@ -1223,6 +1428,12 @@ function idx_addPage($page, $verbose=false, $force=false) { return $result; } + $Indexer = idx_get_indexer(); + $pid = $Indexer->getPID($page); + if ($pid === false) { + if ($verbose) print("Indexer: getting the PID failed for $page".DOKU_LF); + return false; + } $body = ''; $metadata = array(); $metadata['title'] = p_get_metadata($page, 'title', METADATA_RENDER_UNLIMITED); @@ -1230,14 +1441,13 @@ function idx_addPage($page, $verbose=false, $force=false) { $metadata['relation_references'] = array_keys($references); else $metadata['relation_references'] = array(); - $data = compact('page', 'body', 'metadata'); + $data = compact('page', 'body', 'metadata', 'pid'); $evt = new Doku_Event('INDEXER_PAGE_ADD', $data); if ($evt->advise_before()) $data['body'] = $data['body'] . " " . rawWiki($page); $evt->advise_after(); unset($evt); extract($data); - $Indexer = idx_get_indexer(); $result = $Indexer->addPageWords($page, $body); if ($result === "locked") { if ($verbose) print("Indexer: locked".DOKU_LF); |