Better {{form of}} handling, remove "lang=..."

[DictionaryPC.git] / src / com / hughes / android / dictionary / parser / WikiTokenizer.java

diff --git a/src/com/hughes/android/dictionary/parser/WikiTokenizer.java b/src/com/hughes/android/dictionary/parser/WikiTokenizer.java
index 4377d8a2c9fabc01111bc628672fe9115d539bf9..e2d59c45e85f90706cb51a330992d7d4f4222cad 100644 (file)
--- a/src/com/hughes/android/dictionary/parser/WikiTokenizer.java
+++ b/src/com/hughes/android/dictionary/parser/WikiTokenizer.java
@@ -64,10 +64,15 @@ public final class WikiTokenizer {
   
 
   public WikiTokenizer(final String wikiText) {
+    this(wikiText, true);
+  }
+
+  public WikiTokenizer(final String wikiText, final boolean isNewline) {
     this.wikiText = wikiText;
     this.matcher = wikiTokenEvent.matcher(wikiText);
+    justReturnedNewline = isNewline;
   }
-    
+
   private void clear() {
     errors.clear();
     tokenStack.clear();
@@ -168,6 +173,7 @@ public final class WikiTokenizer {
     if (lastUnescapedPipePos != -1) {
       return wikiText.substring(lastUnescapedPipePos + 1, end - 2);
     }
+    assert start + 2 < wikiText.length() && end >= 2: wikiText;
     return wikiText.substring(start + 2, end - 2);
   }
 
@@ -211,7 +217,7 @@ public final class WikiTokenizer {
     }
     
     // Eat a newline if we're looking at one:
-    final boolean atNewline = wikiText.charAt(end) == '\n';
+    final boolean atNewline = wikiText.charAt(end) == '\n' || wikiText.charAt(end) == '\u2028';
     if (atNewline) {
       justReturnedNewline = true;
       ++end;
@@ -332,7 +338,7 @@ public final class WikiTokenizer {
   
   public String token() {
     final String token = wikiText.substring(start, end);
-    assert token.equals("\n") || !token.endsWith("\n") : token;
+    assert token.equals("\n") || !token.endsWith("\n") : "token='" + token + "'";
     return token;
   }
   
@@ -342,6 +348,7 @@ public final class WikiTokenizer {
     final boolean insideFunction = toFind.equals("}}");
     
     int end = start;
+    int firstNewline = -1;
     while (end < wikiText.length()) {
       if (matcher.find(end)) {
         final String matchText = matcher.group();
@@ -350,6 +357,9 @@ public final class WikiTokenizer {
         assert matcher.end() > end || matchText.length() == 0: "Group=" + matcher.group();
         if (matchText.length() == 0) {
           assert matchStart == wikiText.length() || wikiText.charAt(matchStart) == '\n';
+          if (firstNewline == -1) {
+            firstNewline = matcher.end();
+          }
           if (tokenStack.isEmpty() && toFind.equals("\n")) {
             return matchStart;
           }
@@ -408,6 +418,14 @@ public final class WikiTokenizer {
       // Inside the while loop.  Just go forward.
       end = Math.max(end, matcher.end());
     }
+    if (toFind.equals("\n") && tokenStack.isEmpty()) {
+      // We were looking for the end, we got it.
+      return end;
+    }
+    if (firstNewline != -1) {
+      errors.add("Couldn't find: " + toFind + ", "+ wikiText.substring(start));
+      return firstNewline;
+    }
     return end;
   }
 
@@ -440,8 +458,8 @@ public final class WikiTokenizer {
     return s.length();
   }
 
-  public static String toPlainText(String sense) {
-    final WikiTokenizer wikiTokenizer = new WikiTokenizer(sense);
+  public static String toPlainText(final String wikiText) {
+    final WikiTokenizer wikiTokenizer = new WikiTokenizer(wikiText);
     final StringBuilder builder = new StringBuilder();
     while (wikiTokenizer.nextToken() != null) {
       if (wikiTokenizer.isPlainText()) {
@@ -457,4 +475,16 @@ public final class WikiTokenizer {
     return builder.toString();
   }
 
+  public static StringBuilder appendFunction(final StringBuilder builder, final String name, List<String> args,
+      final Map<String, String> namedArgs) {
+    builder.append(name);
+    for (final String arg : args) {
+      builder.append("|").append(arg);
+    }
+    for (final Map.Entry<String, String> entry : namedArgs.entrySet()) {
+      builder.append("|").append(entry.getKey()).append("=").append(entry.getValue());
+    }
+    return builder;
+  }
+
 }