1
0
mirror of https://github.com/gnosygnu/xowa.git synced 2024-10-27 20:34:16 +00:00

Xomw: Convert XomwLinkHolderArray

This commit is contained in:
gnosygnu 2017-02-20 08:48:31 -05:00
parent 7fe4ed59ac
commit dec8dbc713
6 changed files with 676 additions and 624 deletions

View File

@ -50,7 +50,7 @@ public class HashByInt {
int len = hash.Len(); int len = hash.Len();
for (int i = 0; i < len; i++) { for (int i = 0; i < len; i++) {
HashByIntItem item = (HashByIntItem)hash.Get_at(i); HashByIntItem item = (HashByIntItem)hash.Get_at(i);
rv.Add(item.key, item); rv.Add(item.key, item.val);
} }
return rv; return rv;
} }

View File

@ -17,15 +17,19 @@ along with this program. If not, see <http://www.gnu.org/licenses/>.
*/ */
package gplx.xowa.mediawiki.includes.parsers; import gplx.*; import gplx.xowa.*; import gplx.xowa.mediawiki.*; import gplx.xowa.mediawiki.includes.*; package gplx.xowa.mediawiki.includes.parsers; import gplx.*; import gplx.xowa.*; import gplx.xowa.mediawiki.*; import gplx.xowa.mediawiki.includes.*;
import gplx.langs.htmls.*; import gplx.langs.htmls.*;
import gplx.xowa.mediawiki.includes.htmls.*;
import gplx.xowa.mediawiki.includes.linkers.*;
/** /**
* Holder of replacement pairs for wiki links * Holder of replacement pairs for wiki links
*/ */
public class XomwLinkHolderArray { public class XomwLinkHolderArray {
private final Bry_bfr tmp = Bry_bfr_.New(); private final XomwLinkHolderList internals = new XomwLinkHolderList();
private final Xomw_link_holder_list internals = new Xomw_link_holder_list();
// public $interwikis = []; // public $interwikis = [];
// public $size = 0; // private int size = 0;
//
private final Bry_bfr tmp = Bry_bfr_.New();
private final Xomw_atr_mgr extraAtrs = new Xomw_atr_mgr();
private final Xomw_qry_mgr query = new Xomw_qry_mgr();
/** /**
* @var Parser * @var Parser
*/ */
@ -39,25 +43,25 @@ public class XomwLinkHolderArray {
this.parent = parent; this.parent = parent;
} }
// /** /**
// * Reduce memory usage to reduce the impact of circular references * Reduce memory usage to reduce the impact of circular references
// */ */
// public function __destruct() { // public function __destruct() {
// foreach ( $this as $name => $value ) { // foreach ( $this as $name => $value ) {
// unset( $this->$name ); // unset( this.$name );
// } // }
// } // }
//
// /** /**
// * Don't serialize the parent Object, it is big, and not needed when it is * Don't serialize the parent Object, it is big, and not needed when it is
// * a parameter to mergeForeign(), which is the only application of * a parameter to mergeForeign(), which is the only application of
// * serializing at present. * serializing at present.
// * *
// * Compact the titles, only serialize the text form. * Compact the titles, only serialize the text form.
// * @return array * @return array
// */ */
// public function __sleep() { // public function __sleep() {
// foreach ( $this->internals as &$nsLinks ) { // foreach ( this.internals as &$nsLinks ) {
// foreach ( $nsLinks as &$entry ) { // foreach ( $nsLinks as &$entry ) {
// unset( $entry['title'] ); // unset( $entry['title'] );
// } // }
@ -65,19 +69,19 @@ public class XomwLinkHolderArray {
// unset( $nsLinks ); // unset( $nsLinks );
// unset( $entry ); // unset( $entry );
// //
// foreach ( $this->interwikis as &$entry ) { // foreach ( this.interwikis as &$entry ) {
// unset( $entry['title'] ); // unset( $entry['title'] );
// } // }
// unset( $entry ); // unset( $entry );
// //
// return [ 'internals', 'interwikis', 'size' ]; // return [ 'internals', 'interwikis', 'size' ];
// } // }
//
// /** /**
// * Recreate the Title objects * Recreate the Title objects
// */ */
// public function __wakeup() { // public function __wakeup() {
// foreach ( $this->internals as &$nsLinks ) { // foreach ( this.internals as &$nsLinks ) {
// foreach ( $nsLinks as &$entry ) { // foreach ( $nsLinks as &$entry ) {
// $entry['title'] = Title::newFromText( $entry['pdbk'] ); // $entry['title'] = Title::newFromText( $entry['pdbk'] );
// } // }
@ -85,26 +89,26 @@ public class XomwLinkHolderArray {
// unset( $nsLinks ); // unset( $nsLinks );
// unset( $entry ); // unset( $entry );
// //
// foreach ( $this->interwikis as &$entry ) { // foreach ( this.interwikis as &$entry ) {
// $entry['title'] = Title::newFromText( $entry['pdbk'] ); // $entry['title'] = Title::newFromText( $entry['pdbk'] );
// } // }
// unset( $entry ); // unset( $entry );
// } // }
//
// /** // /**
// * Merge another LinkHolderArray into this one // * Merge another LinkHolderArray into this one
// * @param LinkHolderArray $other // * @param LinkHolderArray $other
// */ // */
// public function merge( $other ) { // public function merge( $other ) {
// foreach ( $other->internals as $ns => $entries ) { // foreach ( $other->internals as $ns => $entries ) {
// $this->size += count( $entries ); // this.size += count( $entries );
// if ( !isset( $this->internals[$ns] ) ) { // if ( !isset( this.internals[$ns] ) ) {
// $this->internals[$ns] = $entries; // this.internals[$ns] = $entries;
// } else { // } else {
// $this->internals[$ns] += $entries; // this.internals[$ns] += $entries;
// } // }
// } // }
// $this->interwikis += $other->interwikis; // this.interwikis += $other->interwikis;
// } // }
// //
// /** // /**
@ -120,14 +124,14 @@ public class XomwLinkHolderArray {
// * @return array // * @return array
// */ // */
// public function mergeForeign( $other, $texts ) { // public function mergeForeign( $other, $texts ) {
// $this->tempIdOffset = $idOffset = $this->parent->nextLinkID(); // this.tempIdOffset = $idOffset = this.parent->nextLinkID();
// $maxId = 0; // $maxId = 0;
// //
// # Renumber @gplx.Internal protected links // # Renumber @gplx.Internal protected links
// foreach ( $other->internals as $ns => $nsLinks ) { // foreach ( $other->internals as $ns => $nsLinks ) {
// foreach ( $nsLinks as $key => $entry ) { // foreach ( $nsLinks as $key => $entry ) {
// $newKey = $idOffset + $key; // $newKey = $idOffset + $key;
// $this->internals[$ns][$newKey] = $entry; // this.internals[$ns][$newKey] = $entry;
// $maxId = $newKey > $maxId ? $newKey : $maxId; // $maxId = $newKey > $maxId ? $newKey : $maxId;
// } // }
// } // }
@ -137,15 +141,15 @@ public class XomwLinkHolderArray {
// # Renumber interwiki links // # Renumber interwiki links
// foreach ( $other->interwikis as $key => $entry ) { // foreach ( $other->interwikis as $key => $entry ) {
// $newKey = $idOffset + $key; // $newKey = $idOffset + $key;
// $this->interwikis[$newKey] = $entry; // this.interwikis[$newKey] = $entry;
// $maxId = $newKey > $maxId ? $newKey : $maxId; // $maxId = $newKey > $maxId ? $newKey : $maxId;
// } // }
// $texts = preg_replace_callback( '/(<!--IWLINK )(\d+)(-->)/', // $texts = preg_replace_callback( '/(<!--IWLINK )(\d+)(-->)/',
// [ $this, 'mergeForeignCallback' ], $texts ); // [ $this, 'mergeForeignCallback' ], $texts );
// //
// # Set the parent link ID to be beyond the highest used ID // # Set the parent link ID to be beyond the highest used ID
// $this->parent->setLinkID( $maxId + 1 ); // this.parent->setLinkID( $maxId + 1 );
// $this->tempIdOffset = null; // this.tempIdOffset = null;
// return $texts; // return $texts;
// } // }
// //
@ -154,7 +158,7 @@ public class XomwLinkHolderArray {
// * @return String // * @return String
// */ // */
// protected function mergeForeignCallback( $m ) { // protected function mergeForeignCallback( $m ) {
// return $m[1] . ( $m[2] + $this->tempIdOffset ) . $m[3]; // return $m[1] . ( $m[2] + this.tempIdOffset ) . $m[3];
// } // }
// //
// /** // /**
@ -164,7 +168,7 @@ public class XomwLinkHolderArray {
// * @return LinkHolderArray // * @return LinkHolderArray
// */ // */
// public function getSubArray( $text ) { // public function getSubArray( $text ) {
// $sub = new LinkHolderArray( $this->parent ); // $sub = new LinkHolderArray( this.parent );
// //
// # Internal links // # Internal links
// $pos = 0; // $pos = 0;
@ -176,7 +180,7 @@ public class XomwLinkHolderArray {
// } // }
// $ns = $m[1][0]; // $ns = $m[1][0];
// $key = $m[2][0]; // $key = $m[2][0];
// $sub->internals[$ns][$key] = $this->internals[$ns][$key]; // $sub->internals[$ns][$key] = this.internals[$ns][$key];
// $pos = $m[0][1] + strlen( $m[0][0] ); // $pos = $m[0][1] + strlen( $m[0][0] );
// } // }
// //
@ -187,7 +191,7 @@ public class XomwLinkHolderArray {
// break; // break;
// } // }
// $key = $m[1][0]; // $key = $m[1][0];
// $sub->interwikis[$key] = $this->interwikis[$key]; // $sub->interwikis[$key] = this.interwikis[$key];
// $pos = $m[0][1] + strlen( $m[0][0] ); // $pos = $m[0][1] + strlen( $m[0][0] );
// } // }
// return $sub; // return $sub;
@ -199,19 +203,19 @@ public class XomwLinkHolderArray {
// */ // */
// public function isBig() { // public function isBig() {
// global $wgLinkHolderBatchSize; // global $wgLinkHolderBatchSize;
// return $this->size > $wgLinkHolderBatchSize; // return this.size > $wgLinkHolderBatchSize;
// }
//
// /**
// * Clear all stored link holders.
// * Make sure you don't have any text left using these link holders, before you call this
// */
// public function clear() {
// $this->internals = [];
// $this->interwikis = [];
// $this->size = 0;
// } // }
/**
* Clear all stored link holders.
* Make sure you don't have any text left using these link holders, before you call this
*/
public void clear() {
this.internals.Clear();//
// this.interwikis = [];
// this.size = 0;
}
/** /**
* Make a link placeholder. The text returned can be later resolved to a real link with * Make a link placeholder. The text returned can be later resolved to a real link with
* replaceLinkHolders(). This is done for two reasons: firstly to avoid further * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
@ -235,15 +239,15 @@ public class XomwLinkHolderArray {
// list( $inside, $trail ) = Linker::splitTrail( $trail ); // list( $inside, $trail ) = Linker::splitTrail( $trail );
byte[] inside = Bry_.Empty; byte[] inside = Bry_.Empty;
Xomw_link_holder_item entry = new Xomw_link_holder_item XomwLinkHolderItem entry = new XomwLinkHolderItem
( nt ( nt
, tmp.Add_bry_many(prefix, text, inside).To_bry_and_clear() , tmp.Add_bry_many(prefix, text, inside).To_bry_and_clear()
, query); , query);
if (nt.isExternal()) { if (nt.isExternal()) {
// Use a globally unique ID to keep the objects mergable // Use a globally unique ID to keep the objects mergable
// $key = $this->parent->nextLinkID(); // $key = this.parent->nextLinkID();
// $this->interwikis[$key] = $entry; // this.interwikis[$key] = $entry;
// $retVal = "<!--IWLINK $key-->{$trail}"; // $retVal = "<!--IWLINK $key-->{$trail}";
} }
else { else {
@ -253,37 +257,38 @@ public class XomwLinkHolderArray {
} }
} }
} }
//
// /** /**
// * Replace <!--LINK--> link placeholders with actual links, in the buffer * Replace <!--LINK--> link placeholders with actual links, in the buffer
// * *
// * @param String $text * @param String $text
// */ */
// public function replace( &$text ) { public void Replace(Xomw_parser_ctx pctx, Xomw_parser_bfr pbfr) {
// $this->replaceInternal( $text ); this.replaceInternal(pbfr);
// $this->replaceInterwiki( $text ); // $this->replaceInterwiki( $text );
// } }
//
// /** /**
// * Replace @gplx.Internal protected links * Replace @gplx.Internal protected links
// * @param String $text * @param String $text
// */ */
// protected function replaceInternal( &$text ) { private void replaceInternal(Xomw_parser_bfr pbfr) {
// if ( !$this->internals ) { if (internals.Len() == 0) {
// return; return;
// } }
//
// SKIP:Replace_internals does db lookup to identify redlinks;
// global $wgContLang; // global $wgContLang;
// //
// $colours = []; // $colours = [];
// $linkCache = LinkCache::singleton(); // $linkCache = LinkCache::singleton();
// $output = $this->parent->getOutput(); // $output = this.parent->getOutput();
// $linkRenderer = $this->parent->getLinkRenderer(); Xomw_link_renderer linkRenderer = this.parent.getLinkRenderer();
// //
// $dbr = wfGetDB( DB_REPLICA ); // $dbr = wfGetDB( DB_REPLICA );
// //
// # Sort by namespace // # Sort by namespace
// ksort( $this->internals ); // ksort( this.internals );
// //
// $linkcolour_ids = []; // $linkcolour_ids = [];
// //
@ -291,6 +296,378 @@ public class XomwLinkHolderArray {
// $lb = new LinkBatch(); // $lb = new LinkBatch();
// $lb->setCaller( __METHOD__ ); // $lb->setCaller( __METHOD__ );
// //
// foreach ( this.internals as $ns => $entries ) {
// foreach ( $entries as $entry ) {
// /** @var Title $title */
// $title = $entry['title'];
// $pdbk = $entry['pdbk'];
//
// # Skip invalid entries.
// # Result will be ugly, but prevents crash.
// if ( is_null( $title ) ) {
// continue;
// }
//
// # Check if it's a static known link, e.g. interwiki
// if ( $title->isAlwaysKnown() ) {
// $colours[$pdbk] = '';
// } elseif ( $ns == NS_SPECIAL ) {
// $colours[$pdbk] = 'new';
// } else {
// $id = $linkCache->getGoodLinkID( $pdbk );
// if ( $id != 0 ) {
// $colours[$pdbk] = $linkRenderer->getLinkClasses( $title );
// $output->addLink( $title, $id );
// $linkcolour_ids[$id] = $pdbk;
// } elseif ( $linkCache->isBadLink( $pdbk ) ) {
// $colours[$pdbk] = 'new';
// } else {
// # Not in the link cache, add it to the query
// $lb->addObj( $title );
// }
// }
// }
// }
// if ( !$lb->isEmpty() ) {
// $fields = array_merge(
// LinkCache::getSelectFields(),
// [ 'page_namespace', 'page_title' ]
// );
//
// $res = $dbr->select(
// 'page',
// $fields,
// $lb->constructSet( 'page', $dbr ),
// __METHOD__
// );
//
// # Fetch data and form into an associative array
// # non-existent = broken
// foreach ( $res as $s ) {
// $title = Title::makeTitle( $s->page_namespace, $s->page_title );
// $pdbk = $title->getPrefixedDBkey();
// $linkCache->addGoodLinkObjFromRow( $title, $s );
// $output->addLink( $title, $s->page_id );
// $colours[$pdbk] = $linkRenderer->getLinkClasses( $title );
// // add id to the extension todolist
// $linkcolour_ids[$s->page_id] = $pdbk;
// }
// unset( $res );
// }
// if ( count( $linkcolour_ids ) ) {
// // pass an array of page_ids to an extension
// Hooks::run( 'GetLinkColours', [ $linkcolour_ids, &$colours ] );
// }
//
// # Do a second query for different language variants of links and categories
// if ( $wgContLang->hasVariants() ) {
// this.doVariants( $colours );
// }
// Construct search and replace arrays
Bry_bfr src_bfr = pbfr.Src();
byte[] src = src_bfr.Bfr();
int src_bgn = 0;
int src_end = src_bfr.Len();
Bry_bfr bfr = pbfr.Trg();
pbfr.Switch();
int cur = src_bgn;
int prv = 0;
while (true) {
int link_bgn = Bry_find_.Find_fwd(src, Bry__link__bgn, cur, src_end);
if (link_bgn == Bry_find_.Not_found) {
bfr.Add_mid(src, prv, src_end);
break;
}
int key_bgn = link_bgn + Bry__link__bgn.length;
int key_end = Bry_find_.Find_fwd_while_num(src, key_bgn, src_end);
int link_key = Bry_.To_int_or(src, key_bgn, key_end, -1);
XomwLinkHolderItem item = internals.Get_by(link_key);
// $pdbk = $entry['pdbk'];
// $title = $entry['title'];
// $query = isset( $entry['query'] ) ? $entry['query'] : [];
// $key = "$ns:$index";
// $searchkey = "<!--LINK $key-->";
// $displayText = $entry['text'];
// if ( isset( $entry['selflink'] ) ) {
// $replacePairs[$searchkey] = Linker::makeSelfLinkObj( $title, $displayText, $query );
// continue;
// }
// if ( $displayText === '' ) {
// $displayText = null;
// } else {
// $displayText = new HtmlArmor( $displayText );
// }
// if ( !isset( $colours[$pdbk] ) ) {
// $colours[$pdbk] = 'new';
// }
// $attribs = [];
// if ( $colours[$pdbk] == 'new' ) {
// $linkCache->addBadLinkObj( $title );
// $output->addLink( $title, 0 );
// $link = $linkRenderer->makeBrokenLink(
// $title, $displayText, $attribs, $query
// );
// } else {
// $link = $linkRenderer->makePreloadedLink(
// $title, $displayText, $colours[$pdbk], $attribs, $query
// );
// }
//
// $replacePairs[$searchkey] = $link;
// }
// }
bfr.Add_mid(src, prv, link_bgn);
linkRenderer.Make_preloaded_link(bfr, item.Title(), item.Text(), Bry_.Empty, extraAtrs, query.Clear());
cur = key_end + Gfh_tag_.Comm_end_len;
prv = cur;
}
// $replacer = new HashtableReplacer( $replacePairs, 1 );
//
// # Do the thing
// $text = preg_replace_callback(
// '/(<!--LINK .*?-->)/',
// $replacer->cb(),
// $text
// );
}
// /**
// * Replace interwiki links
// * @param String $text
// */
// protected function replaceInterwiki( &$text ) {
// if ( empty( this.interwikis ) ) {
// return;
// }
//
// # Make interwiki link HTML
// $output = this.parent->getOutput();
// $replacePairs = [];
// $linkRenderer = this.parent->getLinkRenderer();
// foreach ( this.interwikis as $key => $link ) {
// $replacePairs[$key] = $linkRenderer->makeLink(
// $link['title'],
// new HtmlArmor( $link['text'] )
// );
// $output->addInterwikiLink( $link['title'] );
// }
// $replacer = new HashtableReplacer( $replacePairs, 1 );
//
// $text = preg_replace_callback(
// '/<!--IWLINK (.*?)-->/',
// $replacer->cb(),
// $text );
// }
//
// /**
// * Modify this.internals and $colours according to language variant linking rules
// * @param array $colours
// */
// protected function doVariants( &$colours ) {
// global $wgContLang;
// $linkBatch = new LinkBatch();
// $variantMap = []; // maps $pdbkey_Variant => $keys (of link holders)
// $output = this.parent->getOutput();
// $linkCache = LinkCache::singleton();
// $titlesToBeConverted = '';
// $titlesAttrs = [];
//
// // Concatenate titles to a single String, thus we only need auto convert the
// // single String to all variants. This would improve parser's performance
// // significantly.
// foreach ( this.internals as $ns => $entries ) {
// if ( $ns == NS_SPECIAL ) {
// continue;
// }
// foreach ( $entries as $index => $entry ) {
// $pdbk = $entry['pdbk'];
// // we only deal with new links (in its first query)
// if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] === 'new' ) {
// $titlesAttrs[] = [ $index, $entry['title'] ];
// // separate titles with \0 because it would never appears
// // in a valid title
// $titlesToBeConverted .= $entry['title']->getText() . "\0";
// }
// }
// }
//
// // Now do the conversion and explode String to text of titles
// $titlesAllVariants = $wgContLang->autoConvertToAllVariants( rtrim( $titlesToBeConverted, "\0" ) );
// $allVariantsName = array_keys( $titlesAllVariants );
// foreach ( $titlesAllVariants as &$titlesVariant ) {
// $titlesVariant = explode( "\0", $titlesVariant );
// }
//
// // Then add variants of links to link batch
// $parentTitle = this.parent->getTitle();
// foreach ( $titlesAttrs as $i => $attrs ) {
// /** @var Title $title */
// list( $index, $title ) = $attrs;
// $ns = $title->getNamespace();
// $text = $title->getText();
//
// foreach ( $allVariantsName as $variantName ) {
// $textVariant = $titlesAllVariants[$variantName][$i];
// if ( $textVariant === $text ) {
// continue;
// }
//
// $variantTitle = Title::makeTitle( $ns, $textVariant );
//
// // Self-link checking for mixed/different variant titles. At this point, we
// // already know the exact title does not exist, so the link cannot be to a
// // variant of the current title that exists as a separate page.
// if ( $variantTitle->equals( $parentTitle ) && !$title->hasFragment() ) {
// this.internals[$ns][$index]['selflink'] = true;
// continue 2;
// }
//
// $linkBatch->addObj( $variantTitle );
// $variantMap[$variantTitle->getPrefixedDBkey()][] = "$ns:$index";
// }
// }
//
// // process categories, check if a category exists in some variant
// $categoryMap = []; // maps $category_variant => $category (dbkeys)
// $varCategories = []; // category replacements oldDBkey => newDBkey
// foreach ( $output->getCategoryLinks() as $category ) {
// $categoryTitle = Title::makeTitleSafe( NS_CATEGORY, $category );
// $linkBatch->addObj( $categoryTitle );
// $variants = $wgContLang->autoConvertToAllVariants( $category );
// foreach ( $variants as $variant ) {
// if ( $variant !== $category ) {
// $variantTitle = Title::makeTitleSafe( NS_CATEGORY, $variant );
// if ( is_null( $variantTitle ) ) {
// continue;
// }
// $linkBatch->addObj( $variantTitle );
// $categoryMap[$variant] = [ $category, $categoryTitle ];
// }
// }
// }
//
// if ( !$linkBatch->isEmpty() ) {
// // construct query
// $dbr = wfGetDB( DB_REPLICA );
// $fields = array_merge(
// LinkCache::getSelectFields(),
// [ 'page_namespace', 'page_title' ]
// );
//
// $varRes = $dbr->select( 'page',
// $fields,
// $linkBatch->constructSet( 'page', $dbr ),
// __METHOD__
// );
//
// $linkcolour_ids = [];
// $linkRenderer = this.parent->getLinkRenderer();
//
// // for each found variants, figure out link holders and replace
// foreach ( $varRes as $s ) {
// $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
// $varPdbk = $variantTitle->getPrefixedDBkey();
// $vardbk = $variantTitle->getDBkey();
//
// $holderKeys = [];
// if ( isset( $variantMap[$varPdbk] ) ) {
// $holderKeys = $variantMap[$varPdbk];
// $linkCache->addGoodLinkObjFromRow( $variantTitle, $s );
// $output->addLink( $variantTitle, $s->page_id );
// }
//
// // loop over link holders
// foreach ( $holderKeys as $key ) {
// list( $ns, $index ) = explode( ':', $key, 2 );
// $entry =& this.internals[$ns][$index];
// $pdbk = $entry['pdbk'];
//
// if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] === 'new' ) {
// // found link in some of the variants, replace the link holder data
// $entry['title'] = $variantTitle;
// $entry['pdbk'] = $varPdbk;
//
// // set pdbk and colour
// $colours[$varPdbk] = $linkRenderer->getLinkClasses( $variantTitle );
// $linkcolour_ids[$s->page_id] = $pdbk;
// }
// }
//
// // check if the Object is a variant of a category
// if ( isset( $categoryMap[$vardbk] ) ) {
// list( $oldkey, $oldtitle ) = $categoryMap[$vardbk];
// if ( !isset( $varCategories[$oldkey] ) && !$oldtitle->exists() ) {
// $varCategories[$oldkey] = $vardbk;
// }
// }
// }
// Hooks::run( 'GetLinkColours', [ $linkcolour_ids, &$colours ] );
//
// // rebuild the categories in original order (if there are replacements)
// if ( count( $varCategories ) > 0 ) {
// $newCats = [];
// $originalCats = $output->getCategories();
// foreach ( $originalCats as $cat => $sortkey ) {
// // make the replacement
// if ( array_key_exists( $cat, $varCategories ) ) {
// $newCats[$varCategories[$cat]] = $sortkey;
// } else {
// $newCats[$cat] = $sortkey;
// }
// }
// $output->setCategoryLinks( $newCats );
// }
// }
// }
//
// /**
// * Replace <!--LINK--> link placeholders with plain text of links
// * (not HTML-formatted).
// *
// * @param String $text
// * @return String
// */
// public function replaceText( $text ) {
// $text = preg_replace_callback(
// '/<!--(LINK|IWLINK) (.*?)-->/',
// [ &$this, 'replaceTextCallback' ],
// $text );
//
// return $text;
// }
//
// /**
// * Callback for replaceText()
// *
// * @param array $matches
// * @return String
// * @private
// */
// public function replaceTextCallback( $matches ) {
// $type = $matches[1];
// $key = $matches[2];
// if ( $type == 'LINK' ) {
// list( $ns, $index ) = explode( ':', $key, 2 );
// if ( isset( this.internals[$ns][$index]['text'] ) ) {
// return this.internals[$ns][$index]['text'];
// }
// } elseif ( $type == 'IWLINK' ) {
// if ( isset( this.interwikis[$key]['text'] ) ) {
// return this.interwikis[$key]['text'];
// }
// }
// return $matches[0];
// }
// private void Replace_internal__db() {
// // Generate query
// $lb = new LinkBatch();
// $lb->setCaller( __METHOD__ );
//
// foreach ( $this->internals as $ns => $entries ) { // foreach ( $this->internals as $ns => $entries ) {
// foreach ( $entries as $entry ) { // foreach ( $entries as $entry ) {
// /** @var Title $title */ // /** @var Title $title */
@ -358,281 +735,42 @@ public class XomwLinkHolderArray {
// if ( $wgContLang->hasVariants() ) { // if ( $wgContLang->hasVariants() ) {
// $this->doVariants( $colours ); // $this->doVariants( $colours );
// } // }
//
// # Construct search and replace arrays
// $replacePairs = [];
// foreach ( $this->internals as $ns => $entries ) {
// foreach ( $entries as $index => $entry ) {
// $pdbk = $entry['pdbk'];
// $title = $entry['title'];
// $query = isset( $entry['query'] ) ? $entry['query'] : [];
// $key = "$ns:$index";
// $searchkey = "<!--LINK $key-->";
// $displayText = $entry['text'];
// if ( isset( $entry['selflink'] ) ) {
// $replacePairs[$searchkey] = Linker::makeSelfLinkObj( $title, $displayText, $query );
// continue;
// }
// if ( $displayText === '' ) {
// $displayText = null;
// } else {
// $displayText = new HtmlArmor( $displayText );
// }
// if ( !isset( $colours[$pdbk] ) ) {
// $colours[$pdbk] = 'new';
// }
// $attribs = [];
// if ( $colours[$pdbk] == 'new' ) {
// $linkCache->addBadLinkObj( $title );
// $output->addLink( $title, 0 );
// $link = $linkRenderer->makeBrokenLink(
// $title, $displayText, $attribs, $query
// );
// } else {
// $link = $linkRenderer->makePreloadedLink(
// $title, $displayText, $colours[$pdbk], $attribs, $query
// );
// }
//
// $replacePairs[$searchkey] = $link;
// }
// }
// $replacer = new HashtableReplacer( $replacePairs, 1 );
//
// # Do the thing
// $text = preg_replace_callback(
// '/(<!--LINK .*?-->)/',
// $replacer->cb(),
// $text
// );
// }
//
// /**
// * Replace interwiki links
// * @param String $text
// */
// protected function replaceInterwiki( &$text ) {
// if ( empty( $this->interwikis ) ) {
// return;
// }
//
// # Make interwiki link HTML
// $output = $this->parent->getOutput();
// $replacePairs = [];
// $linkRenderer = $this->parent->getLinkRenderer();
// foreach ( $this->interwikis as $key => $link ) {
// $replacePairs[$key] = $linkRenderer->makeLink(
// $link['title'],
// new HtmlArmor( $link['text'] )
// );
// $output->addInterwikiLink( $link['title'] );
// }
// $replacer = new HashtableReplacer( $replacePairs, 1 );
//
// $text = preg_replace_callback(
// '/<!--IWLINK (.*?)-->/',
// $replacer->cb(),
// $text );
// }
//
// /**
// * Modify $this->internals and $colours according to language variant linking rules
// * @param array $colours
// */
// protected function doVariants( &$colours ) {
// global $wgContLang;
// $linkBatch = new LinkBatch();
// $variantMap = []; // maps $pdbkey_Variant => $keys (of link holders)
// $output = $this->parent->getOutput();
// $linkCache = LinkCache::singleton();
// $titlesToBeConverted = '';
// $titlesAttrs = [];
//
// // Concatenate titles to a single String, thus we only need auto convert the
// // single String to all variants. This would improve parser's performance
// // significantly.
// foreach ( $this->internals as $ns => $entries ) {
// if ( $ns == NS_SPECIAL ) {
// continue;
// }
// foreach ( $entries as $index => $entry ) {
// $pdbk = $entry['pdbk'];
// // we only deal with new links (in its first query)
// if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] === 'new' ) {
// $titlesAttrs[] = [ $index, $entry['title'] ];
// // separate titles with \0 because it would never appears
// // in a valid title
// $titlesToBeConverted .= $entry['title']->getText() . "\0";
// }
// }
// }
//
// // Now do the conversion and explode String to text of titles
// $titlesAllVariants = $wgContLang->autoConvertToAllVariants( rtrim( $titlesToBeConverted, "\0" ) );
// $allVariantsName = array_keys( $titlesAllVariants );
// foreach ( $titlesAllVariants as &$titlesVariant ) {
// $titlesVariant = explode( "\0", $titlesVariant );
// }
//
// // Then add variants of links to link batch
// $parentTitle = $this->parent->getTitle();
// foreach ( $titlesAttrs as $i => $attrs ) {
// /** @var Title $title */
// list( $index, $title ) = $attrs;
// $ns = $title->getNamespace();
// $text = $title->getText();
//
// foreach ( $allVariantsName as $variantName ) {
// $textVariant = $titlesAllVariants[$variantName][$i];
// if ( $textVariant === $text ) {
// continue;
// }
//
// $variantTitle = Title::makeTitle( $ns, $textVariant );
//
// // Self-link checking for mixed/different variant titles. At this point, we
// // already know the exact title does not exist, so the link cannot be to a
// // variant of the current title that exists as a separate page.
// if ( $variantTitle->equals( $parentTitle ) && !$title->hasFragment() ) {
// $this->internals[$ns][$index]['selflink'] = true;
// continue 2;
// }
//
// $linkBatch->addObj( $variantTitle );
// $variantMap[$variantTitle->getPrefixedDBkey()][] = "$ns:$index";
// }
// }
//
// // process categories, check if a category exists in some variant
// $categoryMap = []; // maps $category_variant => $category (dbkeys)
// $varCategories = []; // category replacements oldDBkey => newDBkey
// foreach ( $output->getCategoryLinks() as $category ) {
// $categoryTitle = Title::makeTitleSafe( NS_CATEGORY, $category );
// $linkBatch->addObj( $categoryTitle );
// $variants = $wgContLang->autoConvertToAllVariants( $category );
// foreach ( $variants as $variant ) {
// if ( $variant !== $category ) {
// $variantTitle = Title::makeTitleSafe( NS_CATEGORY, $variant );
// if ( is_null( $variantTitle ) ) {
// continue;
// }
// $linkBatch->addObj( $variantTitle );
// $categoryMap[$variant] = [ $category, $categoryTitle ];
// }
// }
// }
//
// if ( !$linkBatch->isEmpty() ) {
// // construct query
// $dbr = wfGetDB( DB_REPLICA );
// $fields = array_merge(
// LinkCache::getSelectFields(),
// [ 'page_namespace', 'page_title' ]
// );
//
// $varRes = $dbr->select( 'page',
// $fields,
// $linkBatch->constructSet( 'page', $dbr ),
// __METHOD__
// );
//
// $linkcolour_ids = [];
// $linkRenderer = $this->parent->getLinkRenderer();
//
// // for each found variants, figure out link holders and replace
// foreach ( $varRes as $s ) {
// $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
// $varPdbk = $variantTitle->getPrefixedDBkey();
// $vardbk = $variantTitle->getDBkey();
//
// $holderKeys = [];
// if ( isset( $variantMap[$varPdbk] ) ) {
// $holderKeys = $variantMap[$varPdbk];
// $linkCache->addGoodLinkObjFromRow( $variantTitle, $s );
// $output->addLink( $variantTitle, $s->page_id );
// }
//
// // loop over link holders
// foreach ( $holderKeys as $key ) {
// list( $ns, $index ) = explode( ':', $key, 2 );
// $entry =& $this->internals[$ns][$index];
// $pdbk = $entry['pdbk'];
//
// if ( !isset( $colours[$pdbk] ) || $colours[$pdbk] === 'new' ) {
// // found link in some of the variants, replace the link holder data
// $entry['title'] = $variantTitle;
// $entry['pdbk'] = $varPdbk;
//
// // set pdbk and colour
// $colours[$varPdbk] = $linkRenderer->getLinkClasses( $variantTitle );
// $linkcolour_ids[$s->page_id] = $pdbk;
// }
// }
//
// // check if the Object is a variant of a category
// if ( isset( $categoryMap[$vardbk] ) ) {
// list( $oldkey, $oldtitle ) = $categoryMap[$vardbk];
// if ( !isset( $varCategories[$oldkey] ) && !$oldtitle->exists() ) {
// $varCategories[$oldkey] = $vardbk;
// }
// }
// }
// Hooks::run( 'GetLinkColours', [ $linkcolour_ids, &$colours ] );
//
// // rebuild the categories in original order (if there are replacements)
// if ( count( $varCategories ) > 0 ) {
// $newCats = [];
// $originalCats = $output->getCategories();
// foreach ( $originalCats as $cat => $sortkey ) {
// // make the replacement
// if ( array_key_exists( $cat, $varCategories ) ) {
// $newCats[$varCategories[$cat]] = $sortkey;
// } else {
// $newCats[$cat] = $sortkey;
// }
// }
// $output->setCategoryLinks( $newCats );
// }
// }
// }
//
// /**
// * Replace <!--LINK--> link placeholders with plain text of links
// * (not HTML-formatted).
// *
// * @param String $text
// * @return String
// */
// public function replaceText( $text ) {
// $text = preg_replace_callback(
// '/<!--(LINK|IWLINK) (.*?)-->/',
// [ &$this, 'replaceTextCallback' ],
// $text );
//
// return $text;
// }
//
// /**
// * Callback for replaceText()
// *
// * @param array $matches
// * @return String
// * @private
// */
// public function replaceTextCallback( $matches ) {
// $type = $matches[1];
// $key = $matches[2];
// if ( $type == 'LINK' ) {
// list( $ns, $index ) = explode( ':', $key, 2 );
// if ( isset( $this->internals[$ns][$index]['text'] ) ) {
// return $this->internals[$ns][$index]['text'];
// }
// } elseif ( $type == 'IWLINK' ) {
// if ( isset( $this->interwikis[$key]['text'] ) ) {
// return $this->interwikis[$key]['text'];
// }
// }
// return $matches[0];
// } // }
public void Test__add(XomwTitle ttl, byte[] capt) {
int key = parent.nextLinkID();
XomwLinkHolderItem item = new XomwLinkHolderItem(ttl, capt, Bry_.Ary_empty);
internals.Add(key, item);
}
private static final byte[] Bry__link__bgn = Bry_.new_a7("<!--LINK "); private static final byte[] Bry__link__bgn = Bry_.new_a7("<!--LINK ");
} }
class XomwLinkHolderList {
private int ary_len = 0, ary_max = 128;
private XomwLinkHolderItem[] ary = new XomwLinkHolderItem[128];
public int Len() {return ary_len;}
public void Clear() {
ary_len = 0;
if (ary_max > 128)
ary = new XomwLinkHolderItem[128];
}
public void Add(int key, XomwLinkHolderItem item) {
if (key >= ary_max) {
int new_max = ary_max * 2;
ary = (XomwLinkHolderItem[])Array_.Resize(ary, new_max);
ary_max = new_max;
}
ary[key] = item;
ary_len++;
}
public XomwLinkHolderItem Get_by(int key) {return ary[key];}
}
class XomwLinkHolderItem {
public XomwLinkHolderItem(XomwTitle title, byte[] text, byte[][] query) {
this.title = title;
this.text = text;
this.query = query;
}
public XomwTitle Title() {return title;} private final XomwTitle title;
public byte[] Text() {return text;} private final byte[] text;
public byte[] Pdbk() {return title.getPrefixedDBkey();}
public byte[][] Query() {return query;} private final byte[][] query;
}

View File

@ -15,237 +15,134 @@ GNU Affero General Public License for more details.
You should have received a copy of the GNU Affero General Public License You should have received a copy of the GNU Affero General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>. along with this program. If not, see <http://www.gnu.org/licenses/>.
*/ */
package gplx.xowa.mediawiki.includes.parsers; import gplx.*; import gplx.xowa.*; import gplx.xowa.mediawiki.*; import gplx.xowa.mediawiki.includes.*; //namespace gplx.xowa.mediawiki.includes.parsers {
import gplx.langs.htmls.*; // using gplx.langs.htmls;
import gplx.xowa.mediawiki.includes.*; // using gplx.xowa.mediawiki.includes;
import gplx.xowa.mediawiki.includes.htmls.*; // using gplx.xowa.mediawiki.includes.htmls;
import gplx.xowa.mediawiki.includes.linkers.*; // using gplx.xowa.mediawiki.includes.linkers;
public class Xomw_link_holders { // public class Xomw_link_holders {
private final Xomw_link_renderer link_renderer; // private final Xomw_link_renderer link_renderer;
private final Bry_bfr tmp; // private final Bry_bfr tmp;
private int link_id = 0; // MOVED:Parser.php // private int link_id = 0; // MOVED:Parser.php
private final Xomw_link_holder_list internals = new Xomw_link_holder_list(); // private final XomwLinkHolderList internals = new XomwLinkHolderList();
private final Xomw_atr_mgr extra_atrs = new Xomw_atr_mgr(); // private final Xomw_atr_mgr extra_atrs = new Xomw_atr_mgr();
private final Xomw_qry_mgr query = new Xomw_qry_mgr(); // private final Xomw_qry_mgr query = new Xomw_qry_mgr();
public Xomw_link_holders(Xomw_link_renderer link_renderer, Bry_bfr tmp) { // public Xomw_link_holders(Xomw_link_renderer link_renderer, Bry_bfr tmp) {
this.link_renderer = link_renderer; // this.link_renderer = link_renderer;
this.tmp = tmp; // this.tmp = tmp;
}
public void Clear() {
internals.Clear();
link_id = 0;
}
public void Make_holder(Bry_bfr bfr, XomwTitle nt, byte[] text, byte[][] query, byte[] trail, byte[] prefix) {
if (nt == null) {
// Fail gracefully
bfr.Add_str_a7("<!-- ERROR -->").Add(prefix).Add(text).Add(trail);
}
else {
// Separate the link trail from the rest of the link
// list( $inside, $trail ) = Linker::splitTrail( $trail );
byte[] inside = Bry_.Empty;
Xomw_link_holder_item entry = new Xomw_link_holder_item(nt, tmp.Add_bry_many(prefix, text, inside).To_bry_and_clear(), query);
boolean is_external = false; // $nt->isExternal()
if (is_external) {
// Use a globally unique ID to keep the objects mergable
// $key = $this->parent->nextLinkID();
// $this->interwikis[$key] = $entry;
// $retVal = "<!--IWLINK $key-->{$trail}";
}
else {
int key = link_id++;
internals.Add(key, entry);
bfr.Add(Bry__link__bgn).Add_int_variable(key).Add(Gfh_tag_.Comm_end).Add(trail); // "<!--LINK $ns:$key-->{$trail}";
}
}
}
public void Test__add(XomwTitle ttl, byte[] capt) {
int key = link_id++;
Xomw_link_holder_item item = new Xomw_link_holder_item(ttl, capt, Bry_.Ary_empty);
internals.Add(key, item);
}
public void Replace(Xomw_parser_ctx pctx, Xomw_parser_bfr pbfr) {
this.Replace_internal(pbfr);
// $this->replaceInterwiki( $text );
}
private void Replace_internal(Xomw_parser_bfr pbfr) {
if (internals.Len() == 0)
return;
// $colours = [];
// $linkCache = LinkCache::singleton();
// $output = $this->parent->getOutput();
// $linkRenderer = $this->parent->getLinkRenderer();
// $linkcolour_ids = [];
// SKIP:Replace_internals does db lookup to identify redlinks;
// Construct search and replace arrays
Bry_bfr src_bfr = pbfr.Src();
byte[] src = src_bfr.Bfr();
int src_bgn = 0;
int src_end = src_bfr.Len();
Bry_bfr bfr = pbfr.Trg();
pbfr.Switch();
int cur = src_bgn;
int prv = 0;
while (true) {
int link_bgn = Bry_find_.Find_fwd(src, Bry__link__bgn, cur, src_end);
if (link_bgn == Bry_find_.Not_found) {
bfr.Add_mid(src, prv, src_end);
break;
}
int key_bgn = link_bgn + Bry__link__bgn.length;
int key_end = Bry_find_.Find_fwd_while_num(src, key_bgn, src_end);
int link_key = Bry_.To_int_or(src, key_bgn, key_end, -1);
Xomw_link_holder_item item = internals.Get_by(link_key);
// $pdbk = $entry['pdbk'];
// $title = $entry['title'];
// $query = isset( $entry['query'] ) ? $entry['query'] : [];
// $key = "$ns:$index";
// $searchkey = "<!--LINK $key-->";
// $displayText = $entry['text'];
// if ( isset( $entry['selflink'] ) ) {
// $replacePairs[$searchkey] = Linker::makeSelfLinkObj( $title, $displayText, $query );
// continue;
// } // }
// if ( $displayText === '' ) { // public void Clear() {
// $displayText = null; // internals.Clear();
// } else {
// $displayText = new HtmlArmor( $displayText );
// }
// if ( !isset( $colours[$pdbk] ) ) {
// $colours[$pdbk] = 'new';
// }
// $attribs = [];
// if ( $colours[$pdbk] == 'new' ) {
// $linkCache->addBadLinkObj( $title );
// $output->addLink( $title, 0 );
// $link = $linkRenderer->makeBrokenLink(
// $title, $displayText, $attribs, $query
// );
// } else {
// $link = $linkRenderer->makePreloadedLink(
// $title, $displayText, $colours[$pdbk], $attribs, $query
// );
// }
bfr.Add_mid(src, prv, link_bgn);
link_renderer.Make_preloaded_link(bfr, item.Title(), item.Text(), Bry_.Empty, extra_atrs, query.Clear());
cur = key_end + Gfh_tag_.Comm_end_len;
prv = cur;
}
}
// private void Replace_internal__db() {
// // Generate query
// $lb = new LinkBatch();
// $lb->setCaller( __METHOD__ );
// //
// foreach ( $this->internals as $ns => $entries ) { // link_id = 0;
// foreach ( $entries as $entry ) { // }
// /** @var Title $title */ // public void Make_holder(Bry_bfr bfr, XomwTitle nt, byte[] text, byte[][] query, byte[] trail, byte[] prefix) {
// $title = $entry['title']; // if (nt == null) {
// $pdbk = $entry['pdbk']; // // Fail gracefully
// bfr.Add_str_a7("<!-- ERROR -->").Add(prefix).Add(text).Add(trail);
// }
// else {
// // Separate the link trail from the rest of the link
//// list( $inside, $trail ) = Linker::splitTrail( $trail );
// byte[] inside = Bry_.Empty;
// //
// # Skip invalid entries. // XomwLinkHolderItem entry = new XomwLinkHolderItem(nt, tmp.Add_bry_many(prefix, text, inside).To_bry_and_clear(), query);
// # Result will be ugly, but prevents crash. //
// if ( is_null( $title ) ) { // boolean is_external = false; // $nt->isExternal()
// continue; // if (is_external) {
// // Use a globally unique ID to keep the objects mergable
//// $key = $this->parent->nextLinkID();
//// $this->interwikis[$key] = $entry;
//// $retVal = "<!--IWLINK $key-->{$trail}";
// }
// else {
// int key = link_id++;
// internals.Add(key, entry);
// bfr.Add(Bry__link__bgn).Add_int_variable(key).Add(Gfh_tag_.Comm_end).Add(trail); // "<!--LINK $ns:$key-->{$trail}";
// }
// }
// }
// public void Test__add(XomwTitle ttl, byte[] capt) {
// int key = link_id++;
// XomwLinkHolderItem item = new XomwLinkHolderItem(ttl, capt, Bry_.Ary_empty);
// internals.Add(key, item);
// }
// public void Replace(Xomw_parser_ctx pctx, Xomw_parser_bfr pbfr) {
// this.Replace_internal(pbfr);
//// $this->replaceInterwiki( $text );
// }
// private void Replace_internal(Xomw_parser_bfr pbfr) {
// if (internals.Len() == 0)
// return;
//
//// $colours = [];
//// $linkCache = LinkCache::singleton();
//// $output = $this->parent->getOutput();
//// $linkRenderer = $this->parent->getLinkRenderer();
//
//// $linkcolour_ids = [];
//
// // SKIP:Replace_internals does db lookup to identify redlinks;
//
// // Construct search and replace arrays
// Bry_bfr src_bfr = pbfr.Src();
// byte[] src = src_bfr.Bfr();
// int src_bgn = 0;
// int src_end = src_bfr.Len();
// Bry_bfr bfr = pbfr.Trg();
// pbfr.Switch();
//
// int cur = src_bgn;
// int prv = 0;
// while (true) {
// int link_bgn = Bry_find_.Find_fwd(src, Bry__link__bgn, cur, src_end);
// if (link_bgn == Bry_find_.Not_found) {
// bfr.Add_mid(src, prv, src_end);
// break;
// }
// int key_bgn = link_bgn + Bry__link__bgn.length;
// int key_end = Bry_find_.Find_fwd_while_num(src, key_bgn, src_end);
// int link_key = Bry_.To_int_or(src, key_bgn, key_end, -1);
// XomwLinkHolderItem item = internals.Get_by(link_key);
//
//// $pdbk = $entry['pdbk'];
//// $title = $entry['title'];
//// $query = isset( $entry['query'] ) ? $entry['query'] : [];
//// $key = "$ns:$index";
//// $searchkey = "<!--LINK $key-->";
//// $displayText = $entry['text'];
//// if ( isset( $entry['selflink'] ) ) {
//// $replacePairs[$searchkey] = Linker::makeSelfLinkObj( $title, $displayText, $query );
//// continue;
//// }
//// if ( $displayText === '' ) {
//// $displayText = null;
//// } else {
//// $displayText = new HtmlArmor( $displayText );
//// }
//// if ( !isset( $colours[$pdbk] ) ) {
//// $colours[$pdbk] = 'new';
//// }
//// $attribs = [];
//// if ( $colours[$pdbk] == 'new' ) {
//// $linkCache->addBadLinkObj( $title );
//// $output->addLink( $title, 0 );
//// $link = $linkRenderer->makeBrokenLink(
//// $title, $displayText, $attribs, $query
//// );
//// } else {
//// $link = $linkRenderer->makePreloadedLink(
//// $title, $displayText, $colours[$pdbk], $attribs, $query
//// );
//// }
//
// bfr.Add_mid(src, prv, link_bgn);
// link_renderer.Make_preloaded_link(bfr, item.Title(), item.Text(), Bry_.Empty, extra_atrs, query.Clear());
// cur = key_end + Gfh_tag_.Comm_end_len;
// prv = cur;
// }
// } // }
// //
// # Check if it's a static known link, e.g. interwiki // private static final byte[] Bry__link__bgn = Bry_.new_a7("<!--LINK ");
// if ( $title->isAlwaysKnown() ) {
// $colours[$pdbk] = '';
// } elseif ( $ns == NS_SPECIAL ) {
// $colours[$pdbk] = 'new';
// } else {
// $id = $linkCache->getGoodLinkID( $pdbk );
// if ( $id != 0 ) {
// $colours[$pdbk] = $linkRenderer->getLinkClasses( $title );
// $output->addLink( $title, $id );
// $linkcolour_ids[$id] = $pdbk;
// } elseif ( $linkCache->isBadLink( $pdbk ) ) {
// $colours[$pdbk] = 'new';
// } else {
// # Not in the link cache, add it to the query
// $lb->addObj( $title );
// } // }
//} //}
// }
// }
// if ( !$lb->isEmpty() ) {
// $fields = array_merge(
// LinkCache::getSelectFields(),
// [ 'page_namespace', 'page_title' ]
// );
//
// $res = $dbr->select(
// 'page',
// $fields,
// $lb->constructSet( 'page', $dbr ),
// __METHOD__
// );
//
// # Fetch data and form into an associative array
// # non-existent = broken
// foreach ( $res as $s ) {
// $title = Title::makeTitle( $s->page_namespace, $s->page_title );
// $pdbk = $title->getPrefixedDBkey();
// $linkCache->addGoodLinkObjFromRow( $title, $s );
// $output->addLink( $title, $s->page_id );
// $colours[$pdbk] = $linkRenderer->getLinkClasses( $title );
// // add id to the extension todolist
// $linkcolour_ids[$s->page_id] = $pdbk;
// }
// unset( $res );
// }
// if ( count( $linkcolour_ids ) ) {
// // pass an array of page_ids to an extension
// Hooks::run( 'GetLinkColours', [ $linkcolour_ids, &$colours ] );
// }
//
// # Do a second query for different language variants of links and categories
// if ( $wgContLang->hasVariants() ) {
// $this->doVariants( $colours );
// }
// }
private static final byte[] Bry__link__bgn = Bry_.new_a7("<!--LINK ");
}
class Xomw_link_holder_list {
private int ary_len = 0, ary_max = 128;
private Xomw_link_holder_item[] ary = new Xomw_link_holder_item[128];
public int Len() {return ary_len;}
public void Clear() {
ary_len = 0;
if (ary_max > 128)
ary = new Xomw_link_holder_item[128];
}
public void Add(int key, Xomw_link_holder_item item) {
if (key >= ary_max) {
int new_max = ary_max * 2;
ary = (Xomw_link_holder_item[])Array_.Resize(ary, new_max);
ary_max = new_max;
}
ary[key] = item;
ary_len++;
}
public Xomw_link_holder_item Get_by(int key) {return ary[key];}
}
class Xomw_link_holder_item {
public Xomw_link_holder_item(XomwTitle title, byte[] text, byte[][] query) {
this.title = title;
this.text = text;
this.query = query;
}
public XomwTitle Title() {return title;} private final XomwTitle title;
public byte[] Text() {return text;} private final byte[] text;
public byte[] Pdbk() {return title.getPrefixedDBkey();}
public byte[][] Query() {return query;} private final byte[][] query;
}

View File

@ -26,7 +26,7 @@ public class Xomw_link_holders__tst {
} }
} }
class Xomw_link_holders__fxt { class Xomw_link_holders__fxt {
private final Xomw_link_holders holders = new Xomw_link_holders(new Xomw_link_renderer(new XomwSanitizer()), Bry_bfr_.New()); private final XomwLinkHolderArray holders = new XomwLinkHolderArray(new Xomw_parser());
private final Xomw_parser_bfr pbfr = new Xomw_parser_bfr(); private final Xomw_parser_bfr pbfr = new Xomw_parser_bfr();
private boolean apos = true; private boolean apos = true;
public Xomw_link_holders__fxt() { public Xomw_link_holders__fxt() {

View File

@ -33,7 +33,7 @@ public class Xomw_parser {
private final Xomw_magiclinks_wkr magiclinks_wkr; private final Xomw_magiclinks_wkr magiclinks_wkr;
private final Xomw_doubleunder_wkr doubleunder_wkr = new Xomw_doubleunder_wkr(); private final Xomw_doubleunder_wkr doubleunder_wkr = new Xomw_doubleunder_wkr();
private final Xomw_link_renderer link_renderer; private final Xomw_link_renderer link_renderer;
private final Xomw_link_holders holders; private final XomwLinkHolderArray holders;
private final Xomw_heading_cbk__html heading_wkr_cbk; private final Xomw_heading_cbk__html heading_wkr_cbk;
private final Btrie_slim_mgr protocols_trie; private final Btrie_slim_mgr protocols_trie;
private final Xomw_doubleunder_data doubleunder_data = new Xomw_doubleunder_data(); private final Xomw_doubleunder_data doubleunder_data = new Xomw_doubleunder_data();
@ -63,6 +63,23 @@ public class Xomw_parser {
} }
/**
* Get a LinkRenderer instance to make links with
*
* @since 1.28
* @return LinkRenderer
*/
public Xomw_link_renderer getLinkRenderer() {
// if ( !$this->mLinkRenderer ) {
// $this->mLinkRenderer = MediaWikiServices::getInstance()
// ->getLinkRendererFactory()->create();
// $this->mLinkRenderer->setStubThreshold(
// $this->getOptions()->getStubThreshold()
// );
// }
return link_renderer;
}
public Xomw_parser() { public Xomw_parser() {
if (regex_space == null) { if (regex_space == null) {
synchronized (Type_adp_.ClassOf_obj(this)) { synchronized (Type_adp_.ClassOf_obj(this)) {
@ -77,7 +94,7 @@ public class Xomw_parser {
this.link_renderer = new Xomw_link_renderer(sanitizer); this.link_renderer = new Xomw_link_renderer(sanitizer);
this.linker = new XomwLinker(link_renderer); this.linker = new XomwLinker(link_renderer);
this.protocols_trie = Xomw_parser.Protocols__dflt(); this.protocols_trie = Xomw_parser.Protocols__dflt();
this.holders = new Xomw_link_holders(link_renderer, tmp); this.holders = new XomwLinkHolderArray(this);
this.table_wkr = new Xomw_table_wkr(this); this.table_wkr = new Xomw_table_wkr(this);
this.quote_wkr = new Xomw_quote_wkr(this); this.quote_wkr = new Xomw_quote_wkr(this);
this.lnke_wkr = new Xomw_lnke_wkr(this); this.lnke_wkr = new Xomw_lnke_wkr(this);

View File

@ -37,7 +37,7 @@ import gplx.xowa.parsers.uniqs.*;
* P1: link_prefix; EX: b[[A]]; [not enabled on enwiki] * P1: link_prefix; EX: b[[A]]; [not enabled on enwiki]
*/ */
public class Xomw_lnki_wkr {// THREAD.UNSAFE: caching for repeated calls public class Xomw_lnki_wkr {// THREAD.UNSAFE: caching for repeated calls
private final Xomw_link_holders holders; private final XomwLinkHolderArray holders;
private final XomwLinker linker; private final XomwLinker linker;
private final Xomw_link_renderer link_renderer; private final Xomw_link_renderer link_renderer;
// private final Btrie_slim_mgr protocols_trie; // private final Btrie_slim_mgr protocols_trie;
@ -55,7 +55,7 @@ public class Xomw_lnki_wkr {// THREAD.UNSAFE: caching for repeated calls
private final List_adp tmp_list = List_adp_.New(); private final List_adp tmp_list = List_adp_.New();
private final Hash_adp mImageParams = Hash_adp_bry.cs(); private final Hash_adp mImageParams = Hash_adp_bry.cs();
private final Hash_adp mImageParamsMagicArray = Hash_adp_bry.cs(); private final Hash_adp mImageParamsMagicArray = Hash_adp_bry.cs();
public Xomw_lnki_wkr(Xomw_parser parser, Xomw_link_holders holders, Xomw_link_renderer link_renderer, Btrie_slim_mgr protocols_trie) { public Xomw_lnki_wkr(Xomw_parser parser, XomwLinkHolderArray holders, Xomw_link_renderer link_renderer, Btrie_slim_mgr protocols_trie) {
this.parser = parser; this.parser = parser;
this.holders = holders; this.holders = holders;
this.link_renderer = link_renderer; this.link_renderer = link_renderer;
@ -77,7 +77,7 @@ public class Xomw_lnki_wkr {// THREAD.UNSAFE: caching for repeated calls
} }
} }
public void Clear_state() { public void Clear_state() {
holders.Clear(); holders.clear();
} }
public void Replace_internal_links(Xomw_parser_ctx pctx, Xomw_parser_bfr pbfr) { public void Replace_internal_links(Xomw_parser_ctx pctx, Xomw_parser_bfr pbfr) {
// XO.PBFR // XO.PBFR
@ -440,11 +440,11 @@ public class Xomw_lnki_wkr {// THREAD.UNSAFE: caching for repeated calls
} }
else { else {
// Links will be added to the output link list after checking // Links will be added to the output link list after checking
holders.Make_holder(bfr, nt, text, Bry_.Ary_empty, trail, prefix); holders.makeHolder(bfr, nt, text, Bry_.Ary_empty, trail, prefix);
} }
} }
} }
public void makeImage(Xomw_parser_ctx pctx, Bry_bfr bfr, XomwTitle title, byte[] options_at_link, Xomw_link_holders holders) { public void makeImage(Xomw_parser_ctx pctx, Bry_bfr bfr, XomwTitle title, byte[] options_at_link, XomwLinkHolderArray holders) {
// Check if the options text is of the form "options|alt text" // Check if the options text is of the form "options|alt text"
// Options are: // Options are:
// * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang // * thumbnail make a thumbnail with enlarge-icon and caption, alignment depends on lang