Get rid of xerces dependency.

[DictionaryPC.git] / src / com / hughes / android / dictionary / engine / WiktionarySplitter.java

diff --git a/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java b/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java
index 28b11bc669f3cbdb146a4a41c4dc2afc802ae2a9..cc9a3b1a0bc6ee56110dfded92cec93a50b83301 100644 (file)
--- a/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java
+++ b/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java
@@ -20,8 +20,8 @@ import java.io.DataOutputStream;
 import java.io.File;
 import java.io.FileInputStream;
 import java.io.FileOutputStream;
-import java.io.InputStream;
 import java.io.IOException;
+import java.io.InputStream;
 import java.io.OutputStream;
 import java.util.ArrayList;
 import java.util.LinkedHashMap;
@@ -32,8 +32,8 @@ import java.util.regex.Pattern;
 
 import javax.xml.parsers.ParserConfigurationException;
 import javax.xml.parsers.SAXParser;
+import javax.xml.parsers.SAXParserFactory;
 
-import org.apache.xerces.jaxp.SAXParserFactoryImpl;
 import org.apache.commons.compress.compressors.CompressorStreamFactory;
 import org.xml.sax.Attributes;
 import org.xml.sax.SAXException;
@@ -75,7 +75,7 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
     }
 
     private void go() throws Exception {
-        final SAXParser parser = SAXParserFactoryImpl.newInstance().newSAXParser();
+        final SAXParser parser = SAXParserFactory.newInstance().newSAXParser();
 
         // Configure things.
         for (final Map.Entry<String, List<Selector>> pathToSelectorsEntry : pathToSelectors.entrySet()) {
@@ -105,7 +105,7 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
                     parser.parse(new BufferedInputStream(in), this);
                 }
             } catch (Exception e) {
-                System.err.println("Exception during parse, lastPageTitle=" + lastPageTitle + ", titleBuilder=" + titleBuilder.toString());
+                System.err.println("Exception during parse, lastPageTitle=" + lastPageTitle + ", titleBuilder=" + titleBuilder.toString() + " of file " + pathToSelectorsEntry.getKey());
                 throw e;
             }
 
@@ -130,10 +130,12 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
     private void endPage() {
         final String title = titleBuilder.toString();
         lastPageTitle = title;
-        if (++pageCount % 1000 == 0) {
+        if (++pageCount % 100000 == 0) {
             System.out.println("endPage: " + title + ", count=" + pageCount);
         }
-        if (title.startsWith("Wiktionary:") ||
+        if (title.startsWith("Unsupported titles/")) return;
+        if (title.contains(":")) {
+            if (title.startsWith("Wiktionary:") ||
                 title.startsWith("Appendix:") ||
                 title.startsWith("Help:") ||
                 title.startsWith("Index:") ||
@@ -144,13 +146,13 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
                 title.startsWith("Rhymes:") ||
                 title.startsWith("Category:") ||
                 title.startsWith("Wikisaurus:") ||
-                title.startsWith("Unsupported titles/") ||
                 title.startsWith("Transwiki:") ||
                 title.startsWith("File:") ||
                 title.startsWith("Thread:") ||
                 title.startsWith("Template:") ||
                 title.startsWith("Summary:") ||
                 title.startsWith("Module:") ||
+                title.startsWith("Reconstruction:") ||
                 // DE
                 title.startsWith("Datei:") ||
                 title.startsWith("Verzeichnis:") ||
@@ -159,6 +161,7 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
                 title.startsWith("Kategorie:") ||
                 title.startsWith("Hilfe:") ||
                 title.startsWith("Reim:") ||
+                title.startsWith("Modul:") ||
                 // FR:
                 title.startsWith("Annexe:") ||
                 title.startsWith("Catégori:") ||
@@ -168,16 +171,20 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
                 title.startsWith("Aide:") ||
                 title.startsWith("Fichier:") ||
                 title.startsWith("Wiktionnaire:") ||
+                title.startsWith("Translations:Wiktionnaire:") ||
+                title.startsWith("Translations:Projet:") ||
                 title.startsWith("Catégorie:") ||
                 title.startsWith("Portail:") ||
                 title.startsWith("utiliusateur:") ||
                 title.startsWith("Kategorio:") ||
+                title.startsWith("Tutoriel:") ||
                 // IT
                 title.startsWith("Wikizionario:") ||
                 title.startsWith("Appendice:") ||
                 title.startsWith("Categoria:") ||
                 title.startsWith("Aiuto:") ||
                 title.startsWith("Portail:") ||
+                title.startsWith("Modulo:") ||
                 // ES
                 title.startsWith("Apéndice:") ||
                 title.startsWith("Archivo:") ||
@@ -186,27 +193,38 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
                 title.startsWith("Plantilla:") ||
                 title.startsWith("Wikcionario:") ||
 
+                // PT
+                title.startsWith("Ajuda:") ||
+                title.startsWith("Apêndice:") ||
+                title.startsWith("Citações:") ||
+                title.startsWith("Portal:") ||
+                title.startsWith("Predefinição:") ||
+                title.startsWith("Vocabulário:") ||
+                title.startsWith("Wikcionário:") ||
+                title.startsWith("Módulo:") ||
+
                 // sentinel
                 false
-           ) {
-            return;
-        }
-        if (title.contains(":")) {
-            if (!title.startsWith("Sign gloss:")) {
+               ) return;
+            // leave the Flexion: pages in for now and do not warn about them
+            if (!title.startsWith("Sign gloss:") && !title.startsWith("Flexion:")) {
                 System.err.println("title with colon: " + title);
             }
         }
 
         String text = textBuilder.toString();
+        // Workaround for Spanish wiktionary {{ES}} and {{ES|word}} patterns
+        text = text.replaceAll("\\{\\{ES(\\|[^{}=]*)?}}", "== {{lengua|es}} ==");
         String translingual = "";
+        int start = 0;
+        final Matcher startMatcher = headingStart.matcher(text);
 
-        while (text.length() > 0) {
+        while (start < text.length()) {
             // Find start.
-            final Matcher startMatcher = headingStart.matcher(text);
-            if (!startMatcher.find()) {
+            if (!startMatcher.find(start)) {
                 return;
             }
-            text = text.substring(startMatcher.end());
+            start = startMatcher.end();
 
             final String heading = startMatcher.group();
 
@@ -218,10 +236,10 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
                 final Pattern endPattern = getEndPattern(depth);
 
                 final Matcher endMatcher = endPattern.matcher(text);
-                if (endMatcher.find()) {
+                if (endMatcher.find(start)) {
                     int end = endMatcher.start();
-                    translingual = text.substring(0, endMatcher.start());
-                    text = text.substring(end);
+                    translingual = text.substring(start, end);
+                    start = end;
                     continue;
                 }
             }
@@ -234,13 +252,13 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
 
                     final Matcher endMatcher = endPattern.matcher(text);
                     final int end;
-                    if (endMatcher.find()) {
+                    if (endMatcher.find(start)) {
                         end = endMatcher.start();
                     } else {
                         end = text.length();
                     }
 
-                    String sectionText = text.substring(0, end);
+                    String sectionText = text.substring(start, end);
                     // Hack to remove empty dummy section from French
                     if (sectionText.startsWith("\n=== {{S|étymologie}} ===\n: {{ébauche-étym")) {
                         int dummy_end = sectionText.indexOf("}}", 41) + 2;
@@ -262,7 +280,7 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
                         throw new RuntimeException(e);
                     }
 
-                    text = text.substring(end);
+                    start = end;
                     break;
                 }
             }
@@ -336,7 +354,7 @@ public class WiktionarySplitter extends org.xml.sax.helpers.DefaultHandler {
 
     public void parse(final File file) throws ParserConfigurationException,
         SAXException, IOException {
-        final SAXParser parser = SAXParserFactoryImpl.newInstance().newSAXParser();
+        final SAXParser parser = SAXParserFactory.newInstance().newSAXParser();
         parser.parse(file, this);
     }