X-Git-Url: http://gitweb.fperrin.net/?a=blobdiff_plain;f=src%2Fcom%2Fhughes%2Fandroid%2Fdictionary%2Fengine%2FWiktionarySplitter.java;h=5935df8f51646f28d6462f9d59bf679ec65cabc5;hb=7254095347927bc766f1dffc2ba4b27816180d6b;hp=3cee85da6a2a89743f53be952239e4aea8f87d0f;hpb=4c7701228aff69ceb4dffd23ed319382b5f0a55c;p=DictionaryPC.git

diff --git a/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java b/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java
index 3cee85d..5935df8 100644
--- a/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java
+++ b/src/com/hughes/android/dictionary/engine/WiktionarySplitter.java
@@ -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:") ||
@@ -144,7 +146,6 @@ 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:") ||
@@ -186,18 +187,26 @@ 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:") ||
+
                 // sentinel
                 false
-           ) {
-            return;
-        }
-        if (title.contains(":")) {
+               ) return;
             if (!title.startsWith("Sign gloss:")) {
                 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);