T257236: Performance improvement for page initialization

WPCleaner · Feb 8, 2021 · 311eddb · 311eddb
1 parent c39ea32
commit 311eddb
Show file tree

Hide file tree

Showing 44 changed files with 180 additions and 116 deletions.
diff --git a/WikipediaCleaner/src/org/wikipediacleaner/Bot.java b/WikipediaCleaner/src/org/wikipediacleaner/Bot.java
@@ -374,7 +374,8 @@ private BasicWorker executeFixCheckWiki(Action actionConfig) {
   private BasicWorker executeFixListCheckWiki(Action actionConfig) {
     Page page = null;
     if (actionConfig.actionArgs.length > 0) {
-      page = DataManager.getPage(wiki, actionConfig.actionArgs[0], null, null, null);
+      page = DataManager.createSimplePage(
+          wiki, actionConfig.actionArgs[0], null, null, null);
     }
     List<CheckErrorAlgorithm> algorithms = new ArrayList<>();
     List<CheckErrorAlgorithm> allAlgorithms = new ArrayList<>();

diff --git a/WikipediaCleaner/src/org/wikipediacleaner/api/algorithm/AlgorithmError.java b/WikipediaCleaner/src/org/wikipediacleaner/api/algorithm/AlgorithmError.java
@@ -400,7 +400,8 @@ public Page getPage(int index) {
    * @param pageId Page id.
    */
   private void addPage(String page, Integer pageId) {
-    Page tmpPage = DataManager.getPage(wikipedia, page, pageId, null, null);
+    Page tmpPage = DataManager.createSimplePage(
+        wikipedia, page, pageId, null, null);
     if (!errors.contains(tmpPage)) {
       errors.add(tmpPage);
     }

diff --git a/WikipediaCleaner/src/org/wikipediacleaner/api/check/CheckWiki.java b/WikipediaCleaner/src/org/wikipediacleaner/api/check/CheckWiki.java
@@ -296,7 +296,7 @@ public void retrieveConfiguration(
     try {
       String translationPage = wpcConfiguration.getString(WPCConfigurationString.CW_TRANSLATION_PAGE);
       if (translationPage != null) {
-        Page page = DataManager.getPage(
+        Page page = DataManager.createSimplePage(
             wiki, translationPage,
             null, null, null);
         api.retrieveContents(wiki, Collections.singleton(page), false, false);
@@ -313,7 +313,7 @@ public void retrieveConfiguration(
     for (int i = 0; i < CWConfiguration.MAX_ERROR_NUMBER; i++) {
       CWConfigurationError error = cwConfiguration.getErrorConfiguration(i);
       if ((error != null) && (error.getWhiteListPageName() != null)) {
-        Page page = DataManager.getPage(
+        Page page = DataManager.createSimplePage(
             wiki, error.getWhiteListPageName(), null, null, null);
         whiteListPages.put(error.getWhiteListPageName(), page);
       }

diff --git a/...r/src/org/wikipediacleaner/api/check/algorithm/a0xx/a09x/a090/CheckErrorAlgorithm090.java b/...r/src/org/wikipediacleaner/api/check/algorithm/a0xx/a09x/a090/CheckErrorAlgorithm090.java
@@ -823,7 +823,7 @@ public void computeLinkInformation() {
           endIndex++;
         }
       }
-      Page articlePage = DataManager.getPage(
+      Page articlePage = DataManager.createSimplePage(
           analysis.getWikipedia(), article, null, null, null);
       needColon = Boolean.FALSE;
       if (articlePage.getNamespace() != null) {

diff --git a/...r/src/org/wikipediacleaner/api/check/algorithm/a0xx/a09x/a095/CheckErrorAlgorithm095.java b/...r/src/org/wikipediacleaner/api/check/algorithm/a0xx/a09x/a095/CheckErrorAlgorithm095.java
@@ -61,7 +61,7 @@ public boolean analyze(
       if ((linkDest != null) &&
           (linkDest.trim().length() > 0) &&
           (linkDest.indexOf(':') >= 0)) {
-        Page page = DataManager.getPage(wiki, linkDest, null, null, null);
+        Page page = DataManager.createSimplePage(wiki, linkDest, null, null, null);
         Integer namespace = page.getNamespace();
         if ((namespace != null) &&
             ((namespace.intValue() == Namespace.USER) ||

diff --git a/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a52x/a524/CheckErrorAlgorithm524.java b/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a52x/a524/CheckErrorAlgorithm524.java
@@ -308,7 +308,7 @@ public List<Page> getSpecialList(EnumWikipedia wiki, int limit) {
     if (category != null) {
       API api = APIFactory.getAPI();
       String title = wiki.getWikiConfiguration().getPageTitle(Namespace.CATEGORY, category);
-      Page categoryPage = DataManager.getPage(wiki, title, null, null, null);
+      Page categoryPage = DataManager.createSimplePage(wiki, title, null, null, Namespace.CATEGORY);
       try {
         api.retrieveCategoryMembers(wiki, categoryPage, 0, false, limit);
         result = categoryPage.getRelatedPages(RelatedPages.CATEGORY_MEMBERS);

diff --git a/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a52x/a526/CheckErrorAlgorithm526.java b/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a52x/a526/CheckErrorAlgorithm526.java
@@ -256,7 +256,7 @@ public List<Page> getSpecialList(EnumWikipedia wiki, int limit) {
     // Use internal links
     if (dumpAnalysis != null) {
       API api = APIFactory.getAPI();
-      Page page = DataManager.getPage(wiki, dumpAnalysis, null, null, null);
+      Page page = DataManager.createSimplePage(wiki, dumpAnalysis, null, null, null);
       try {
         api.retrieveLinks(wiki, page, null, null, false, false);
         if (page.getLinks() != null) {

diff --git a/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a52x/a527/CheckErrorAlgorithm527.java b/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a52x/a527/CheckErrorAlgorithm527.java
@@ -217,7 +217,7 @@ public List<Page> getSpecialList(EnumWikipedia wiki, int limit) {
     if (category != null) {
       API api = APIFactory.getAPI();
       String title = wiki.getWikiConfiguration().getPageTitle(Namespace.CATEGORY, category);
-      Page categoryPage = DataManager.getPage(wiki, title, null, null, null);
+      Page categoryPage = DataManager.createSimplePage(wiki, title, null, null, Namespace.CATEGORY);
       try {
         api.retrieveCategoryMembers(wiki, categoryPage, 0, false, limit);
         result = categoryPage.getRelatedPages(RelatedPages.CATEGORY_MEMBERS);

diff --git a/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a52x/a528/CheckErrorAlgorithm528.java b/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a52x/a528/CheckErrorAlgorithm528.java
@@ -165,7 +165,7 @@ public List<Page> getSpecialList(EnumWikipedia wiki, int limit) {
     if (category != null) {
       API api = APIFactory.getAPI();
       String title = wiki.getWikiConfiguration().getPageTitle(Namespace.CATEGORY, category);
-      Page categoryPage = DataManager.getPage(wiki, title, null, null, null);
+      Page categoryPage = DataManager.createSimplePage(wiki, title, null, null, Namespace.CATEGORY);
       try {
         api.retrieveCategoryMembers(wiki, categoryPage, 0, false, limit);
         result = categoryPage.getRelatedPages(RelatedPages.CATEGORY_MEMBERS);

diff --git a/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a52x/a529/CheckErrorAlgorithm529.java b/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a52x/a529/CheckErrorAlgorithm529.java
@@ -270,7 +270,7 @@ public List<Page> getSpecialList(EnumWikipedia wiki, int limit) {
     if (category != null) {
       API api = APIFactory.getAPI();
       String title = wiki.getWikiConfiguration().getPageTitle(Namespace.CATEGORY, category);
-      Page categoryPage = DataManager.getPage(wiki, title, null, null, null);
+      Page categoryPage = DataManager.createSimplePage(wiki, title, null, null, Namespace.CATEGORY);
       try {
         api.retrieveCategoryMembers(wiki, categoryPage, 0, false, limit);
         result = categoryPage.getRelatedPages(RelatedPages.CATEGORY_MEMBERS);

diff --git a/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a53x/a530/CheckErrorAlgorithm530.java b/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a53x/a530/CheckErrorAlgorithm530.java
@@ -172,7 +172,7 @@ public List<Page> getSpecialList(EnumWikipedia wiki, int limit) {
     if (category != null) {
       API api = APIFactory.getAPI();
       String title = wiki.getWikiConfiguration().getPageTitle(Namespace.CATEGORY, category);
-      Page categoryPage = DataManager.getPage(wiki, title, null, null, null);
+      Page categoryPage = DataManager.createSimplePage(wiki, title, null, null, Namespace.CATEGORY);
       try {
         api.retrieveCategoryMembers(wiki, categoryPage, 0, false, limit);
         result = categoryPage.getRelatedPages(RelatedPages.CATEGORY_MEMBERS);

diff --git a/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a54x/a545/CheckErrorAlgorithm545.java b/...r/src/org/wikipediacleaner/api/check/algorithm/a5xx/a54x/a545/CheckErrorAlgorithm545.java
@@ -163,7 +163,7 @@ public List<Page> getSpecialList(EnumWikipedia wiki, int limit) {
     API api = APIFactory.getAPI();
     for (String categoryName : categoriesName) {
       String title = wiki.getWikiConfiguration().getPageTitle(Namespace.CATEGORY, categoryName);
-      Page category = DataManager.getPage(wiki, title, null, null, null);
+      Page category = DataManager.createSimplePage(wiki, title, null, null, Namespace.CATEGORY);
       try {
         api.retrieveCategoryMembers(wiki, category, 0, false, limit);
         List<Page> tmp = category.getRelatedPages(RelatedPages.CATEGORY_MEMBERS);

diff --git a/WikipediaCleaner/src/org/wikipediacleaner/api/configuration/WPCConfiguration.java b/WikipediaCleaner/src/org/wikipediacleaner/api/configuration/WPCConfiguration.java
@@ -628,7 +628,7 @@ public void initSuggestions(API api, boolean forceInit) {
             if (elements.length >= 4) {
               String pageName = elements[0];
               if (!pages.containsKey(pageName)) {
-                pages.put(pageName, DataManager.getPage(wiki, pageName, null, null, null));
+                pages.put(pageName, DataManager.createSimplePage(wiki, pageName, null, null, null));
               }
             }
           }
@@ -637,7 +637,7 @@ public void initSuggestions(API api, boolean forceInit) {
         if (suggestionTypoPages != null) {
           for (String suggestionPage : suggestionTypoPages) {
             if (!pages.containsKey(suggestionPage)) {
-              pages.put(suggestionPage, DataManager.getPage(wiki, suggestionPage, null, null, null));
+              pages.put(suggestionPage, DataManager.createSimplePage(wiki, suggestionPage, null, null, null));
             }
           }
         }
@@ -996,7 +996,7 @@ private void setDisambiguationCategories(String value) {
       this.disambiguationCategories = new ArrayList<>(tmp.size());
       for (String category : tmp) {
         this.disambiguationCategories.add(
-            DataManager.getPage(wiki, category, null, null, null));
+            DataManager.createSimplePage(wiki, category, null, null, null));
       }
     } else {
       this.disambiguationCategories = null;
@@ -1356,7 +1356,7 @@ public List<Page> getTemplatesForHelpRequested() {
       for (String template : templatesForHelpRequested) {
         String title = wiki.getWikiConfiguration().getPageTitle(
             Namespace.TEMPLATE, template);
-        tmp.add(DataManager.getPage(wiki, title, null, null, null));
+        tmp.add(DataManager.createSimplePage(wiki, title, null, null, Namespace.TEMPLATE));
       }
       return tmp;
     }

diff --git a/WikipediaCleaner/src/org/wikipediacleaner/api/constants/EnumWikipedia.java b/WikipediaCleaner/src/org/wikipediacleaner/api/constants/EnumWikipedia.java
@@ -458,7 +458,7 @@ public Boolean isDisambiguationPage(Page page) {
   public void initDisambiguationTemplates(API api) {
     if (disambiguationTemplates == null) {
       synchronized (api) {
-        Page page = DataManager.getPage(
+        Page page = DataManager.createSimplePage(
             this, "Mediawiki:Disambiguationspage",
             null, null, null);
         try {

diff --git a/WikipediaCleaner/src/org/wikipediacleaner/api/data/DataManager.java b/WikipediaCleaner/src/org/wikipediacleaner/api/data/DataManager.java
@@ -54,47 +54,79 @@ public static Page getExistingPage(
   }
 
   /**
+   * Create a page.
+   * 
    * @param wiki Wiki.
-   * @param title Page title.
-   * @param pageId Page id.
-   * @param revisionId Revision id.
-   * @param knownPages Already known pages.
-   * @return The requested page.
+   * @param title Page title (full title including the optional namespace)
+   * @param pageId Page identifier.
+   * @param revisionId Revision identifier.
+   * @param namespaceHint Suggestion for the namespace.
+   * @return Page.
    */
-  public static Page getPage(
+  public static Page createSimplePage(
       EnumWikipedia wiki,
       String title, Integer pageId, String revisionId,
-      List<Page> knownPages) {
+      Integer namespaceHint) {
 
-    // Check in the known pages
-    Page page = getExistingPage(wiki, title, revisionId, knownPages);
-    if (page != null) {
-      return page;
-    }
-
-    // Retrieve page
-    page = new Page(wiki, title);
+    // Create page
+    Page page = new Page(wiki, title);
     page.setPageId(pageId);
     page.setRevisionId(revisionId);
 
     // Manage namespace
+    Integer namespaceId = null;
     if (page.getTitle() != null) {
       int colonIndex = page.getTitle().indexOf(':');
       if (colonIndex > 0) {
         String namespaceText = page.getTitle().substring(0, colonIndex);
-        List<Namespace> namespaces = wiki.getWikiConfiguration().getNamespaces();
-        if (namespaces != null) {
-          for (Namespace namespace : namespaces) {
-            if (namespace.isPossibleName(namespaceText)) {
-              page.setNamespace(namespace.getId());
+        if (namespaceHint != null) {
+          Namespace namespace = wiki.getWikiConfiguration().getNamespace(namespaceHint);
+          if ((namespace != null) && (namespace.isPossibleName(namespaceText))) {
+            namespaceId = namespace.getId();
+          }
+        }
+        if (namespaceId == null) {
+          List<Namespace> namespaces = wiki.getWikiConfiguration().getNamespaces();
+          if (namespaces != null) {
+            for (Namespace namespace : namespaces) {
+              if (namespace.isPossibleName(namespaceText)) {
+                namespaceId = namespace.getId();
+                break;
+              }
             }
           }
         }
       }
-      if (page.getNamespace() == null) {
-        page.setNamespace(Namespace.MAIN);
-      }
     }
+    if (namespaceId == null) {
+      namespaceId = Namespace.MAIN;
+    }
+    page.setNamespace(namespaceId);
+
+    return page;
+  }
+
+  /**
+   * @param wiki Wiki.
+   * @param title Page title.
+   * @param pageId Page id.
+   * @param revisionId Revision id.
+   * @param knownPages Already known pages.
+   * @return The requested page.
+   */
+  public static Page getPage(
+      EnumWikipedia wiki,
+      String title, Integer pageId, String revisionId,
+      List<Page> knownPages) {
+
+    // Check in the known pages
+    Page page = getExistingPage(wiki, title, revisionId, knownPages);
+    if (page != null) {
+      return page;
+    }
+
+    // Create page
+    page = createSimplePage(wiki, title, pageId, revisionId, null);
 
     // Manage comments
     Configuration config = Configuration.getConfiguration();

diff --git a/WikipediaCleaner/src/org/wikipediacleaner/api/data/Page.java b/WikipediaCleaner/src/org/wikipediacleaner/api/data/Page.java
@@ -471,7 +471,7 @@ public Page getArticlePage() {
     if (articlePageName == null) {
       return null;
     }
-    Page articlePage = DataManager.getPage(
+    Page articlePage = DataManager.createSimplePage(
         getWikipedia(), articlePageName, null, null, null);
     return articlePage;
   }
@@ -514,7 +514,7 @@ public Page getTalkPage() {
     if (talkPageName == null) {
       return null;
     }
-    Page talkPage = DataManager.getPage(
+    Page talkPage = DataManager.createSimplePage(
         getWikipedia(), talkPageName, null, null, null);
     return talkPage;
   }
@@ -553,8 +553,9 @@ public String getTalkPageName() {
    * @return Subpage.
    */
   public Page getSubPage(String subpage) {
-    Page subPage = DataManager.getPage(
-        getWikipedia(), getTitle() + "/" + subpage, null, null, null);
+    Page subPage = DataManager.createSimplePage(
+        getWikipedia(), getTitle() + "/" + subpage,
+        null, null, getNamespace());
     return subPage;
   }
 

diff --git a/WikipediaCleaner/src/org/wikipediacleaner/api/dataaccess/StaticPageListProvider.java b/WikipediaCleaner/src/org/wikipediacleaner/api/dataaccess/StaticPageListProvider.java
@@ -29,7 +29,7 @@ public class StaticPageListProvider implements PageListProvider {
    * @param title Title of the page to provide.
    */
   public StaticPageListProvider(EnumWikipedia wiki, String title) {
-    this.pages = Collections.singletonList(DataManager.getPage(wiki, title, null, null, null));
+    this.pages = Collections.singletonList(DataManager.createSimplePage(wiki, title, null, null, null));
   }
 
   /**

diff --git a/WikipediaCleaner/src/org/wikipediacleaner/api/dump/PageHandler.java b/WikipediaCleaner/src/org/wikipediacleaner/api/dump/PageHandler.java
@@ -184,21 +184,19 @@ public void endElement(String uri, String localName, String qName) throws SAXExc
       if (processor != null) {
         increasePageCount();
         try {
-          Integer namespaceNum = Integer.valueOf(namespace.toString());
-          if (processor.isForNamespace(namespaceNum)) {
-            Page page = DataManager.getPage(
-                processor.getWiki(), title.toString(),
-                Integer.valueOf(pageId.toString(), 10), revisionId.toString(),
-                null);
-            page.setNamespace(namespaceNum);
-            page.setContents(revisionText.toString());
-            if (redirect.length() > 0) {
-              PageRedirect redirects = page.getRedirects();
-              redirects.isRedirect(true);
-              redirects.add(DataManager.getPage(processor.getWiki(), redirect.toString(), null, null, null), null);
-            }
-            processor.processPage(page);
+          Integer namespaceId = Integer.valueOf(namespace.toString());
+          Page currentPage = DataManager.createSimplePage(
+              processor.getWiki(), title.toString(),
+              Integer.valueOf(pageId.toString(), 10), revisionId.toString(),
+              namespaceId);
+          currentPage.setNamespace(namespaceId);
+          currentPage.setContents(revisionText.toString());
+          if (redirect.length() > 0) {
+            PageRedirect redirects = currentPage.getRedirects();
+            redirects.isRedirect(true);
+            redirects.add(DataManager.createSimplePage(processor.getWiki(), redirect.toString(), null, null, null), null);
           }
+          processor.processPage(currentPage);
         } catch (NumberFormatException e) {
           log.error("Problem in endElement: " + e.getMessage());
         }

diff --git a/WikipediaCleaner/src/org/wikipediacleaner/api/impl/MediaWikiAPI.java b/WikipediaCleaner/src/org/wikipediacleaner/api/impl/MediaWikiAPI.java
@@ -237,13 +237,13 @@ public void loadConfiguration(
 
       // Decide which pages to be retrieved
       String configPageName = wiki.getConfigurationPage();
-      Page page = DataManager.getPage(
+      Page page = DataManager.createSimplePage(
           wiki, configPageName, null, null, null);
       Page userConfigPage = null;
       if ((userName != null) && (userName.trim().length() > 0) &&
           (wiki.getUserConfigurationPage(userName) != null) &&
           (!Page.areSameTitle(wiki.getUserConfigurationPage(userName), configPageName))) {
-        userConfigPage = DataManager.getPage(
+        userConfigPage = DataManager.createSimplePage(
             wiki,
             wiki.getUserConfigurationPage(userName),
             null, null, null);
@@ -1168,7 +1168,7 @@ public String getLanguageLink(EnumWikipedia from, EnumWikipedia to, String title
       throws APIException {
     ApiLanguageLinksResult result = new ApiXmlLanguageLinksResult(from, httpClient);
     ApiLanguageLinksRequest request = new ApiLanguageLinksRequest(from, result);
-    return request.getLanguageLink(DataManager.getPage(from, title, null, null, null), to);
+    return request.getLanguageLink(DataManager.createSimplePage(from, title, null, null, null), to);
   }
 
   // ==========================================================================

diff --git a/WikipediaCleaner/src/org/wikipediacleaner/gui/swing/MonitorRCWindow.java b/WikipediaCleaner/src/org/wikipediacleaner/gui/swing/MonitorRCWindow.java
@@ -218,7 +218,8 @@ public void recentChanges(List<RecentChange> newRC, Date currentTime) {
     // Check if an update has been made on a monitored page
     for (RecentChange rc : filteredNewRC) {
       if (monitoredPages.containsKey(rc.getTitle())) {
-        Page page = DataManager.getPage(getWikipedia(), rc.getTitle(), null, null, null);
+        Page page = DataManager.createSimplePage(
+            getWikipedia(), rc.getTitle(), null, null, null);
         try {
           updateDabWarning.updateWarning(
               Collections.singletonList(page), null, null, null);
@@ -295,7 +296,7 @@ public void recentChanges(List<RecentChange> newRC, Date currentTime) {
       if (oldEnough) {
         modelRCInteresting.removeRecentChanges(title);
         if (!redirect) {
-          Page page = DataManager.getPage(getWikipedia(), title, null, null, null);
+          Page page = DataManager.createSimplePage(getWikipedia(), title, null, null, null);
           pages.add(page);
           creators.put(title, creator);
           modifiers.put(title, pageModifiers);

diff --git a/WikipediaCleaner/src/org/wikipediacleaner/gui/swing/OnePageWindow.java b/WikipediaCleaner/src/org/wikipediacleaner/gui/swing/OnePageWindow.java
@@ -184,7 +184,7 @@ protected void clean() {
     if (getTextContents() != null) {
       getTextContents().setText(null);
     }
-    page = DataManager.getPage(getWikipedia(), getTextPageName(), null, null, null);
+    page = DataManager.createSimplePage(getWikipedia(), getTextPageName(), null, null, null);
     updateComponentState();
   }