Minor code cleanup.

[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 47aac9488de54449f81399f33b29f3e264f55ac7..349bd9acd6540dcddbe0c2bc574706c735e0d05f 100644 (file)
--- a/src/com/hughes/android/dictionary/parser/WikiTokenizer.java
+++ b/src/com/hughes/android/dictionary/parser/WikiTokenizer.java
@@ -33,6 +33,47 @@ public final class WikiTokenizer {
     void onHeading(WikiTokenizer wikiTokenizer);
     void onListItem(WikiTokenizer wikiTokenizer);
     void onComment(WikiTokenizer wikiTokenizer);
+    void onHtml(WikiTokenizer wikiTokenizer);
+  }
+  
+  public static class DoNothingCallback implements Callback {
+
+    @Override
+    public void onPlainText(String text) {
+    }
+
+    @Override
+    public void onMarkup(WikiTokenizer wikiTokenizer) {
+    }
+
+    @Override
+    public void onWikiLink(WikiTokenizer wikiTokenizer) {
+    }
+
+    @Override
+    public void onNewline(WikiTokenizer wikiTokenizer) {
+    }
+
+    @Override
+    public void onFunction(WikiTokenizer tokenizer, String functionName,
+        List<String> functionPositionArgs, Map<String, String> functionNamedArgs) {
+    }
+
+    @Override
+    public void onHeading(WikiTokenizer wikiTokenizer) {
+    }
+
+    @Override
+    public void onListItem(WikiTokenizer wikiTokenizer) {
+    }
+
+    @Override
+    public void onComment(WikiTokenizer wikiTokenizer) {
+    }
+
+    @Override
+    public void onHtml(WikiTokenizer wikiTokenizer) {
+    }
   }
   
   //private static final Pattern wikiTokenEvent = Pattern.compile("($)", Pattern.MULTILINE);
@@ -43,6 +84,9 @@ public final class WikiTokenizer {
         "=|" +  // Need the = because we might have to find unescaped =
                "<!--|" +
                "''|" +
+        "<pre>|" +
+        "<math>|" +
+        "<ref>|" +
                "$)", Pattern.MULTILINE);
   private static final String listChars = "*#:;";
   
@@ -67,6 +111,7 @@ public final class WikiTokenizer {
   private boolean isComment;
   private boolean isFunction;
   private boolean isWikiLink;
+  private boolean isHtml;
   private int firstUnescapedPipePos;
   
   private int lastUnescapedPipePos;
@@ -79,7 +124,9 @@ public final class WikiTokenizer {
     this(wikiText, true);
   }
 
-  public WikiTokenizer(final String wikiText, final boolean isNewline) {
+  public WikiTokenizer(String wikiText, final boolean isNewline) {
+    wikiText = wikiText.replace('\u2028', '\n');
+    wikiText = wikiText.replace('\u0085', '\n');
     this.wikiText = wikiText;
     this.matcher = wikiTokenEvent.matcher(wikiText);
     justReturnedNewline = isNewline;
@@ -97,6 +144,7 @@ public final class WikiTokenizer {
     isComment = false;
     isFunction = false;
     isWikiLink = false;
+    isHtml = false;
     
     firstUnescapedPipePos = -1;
     lastUnescapedPipePos = -1;
@@ -110,6 +158,9 @@ public final class WikiTokenizer {
       "\\[\\[|" +
       "<!--|" +
       "''|" +
+      "<pre>|" +
+      "<math>|" +
+      "<ref>|" +
       "[\n]"
       );
 
@@ -124,7 +175,7 @@ public final class WikiTokenizer {
           callback.onPlainText(tokenizer.token());
         } else if (tokenizer.isMarkup()) {
           callback.onMarkup(tokenizer);
-        } else if (tokenizer.isWikiLink) {
+        } else if (tokenizer.isWikiLink()) {
           callback.onWikiLink(tokenizer);
         } else if (tokenizer.isNewline()) {
           callback.onNewline(tokenizer);
@@ -136,13 +187,21 @@ public final class WikiTokenizer {
           callback.onListItem(tokenizer);
         } else if (tokenizer.isComment()) {
           callback.onComment(tokenizer);
+        } else if (tokenizer.isHtml()) {
+          callback.onHtml(tokenizer);
+        } else if (!tokenizer.errors.isEmpty()) {
+          // Log was already printed....
         } else {
-          throw new IllegalStateException("Unknown wiki state.");
+          throw new IllegalStateException("Unknown wiki state: " + tokenizer.token());
         }
       }
     }
   }
   
+  public List<String> errors() {
+    return errors;
+  }
+  
   public boolean isNewline() {
     return justReturnedNewline;
   }
@@ -182,6 +241,13 @@ public final class WikiTokenizer {
     assert isListItem();
     return wikiText.substring(start, listPrefixEnd);
   }
+  
+  public static String getListTag(char c) {
+    if (c == '#') {
+      return "ol";
+    }
+    return "ul";
+  }
 
   public String listItemWikiText() {
     assert isListItem();
@@ -196,9 +262,10 @@ public final class WikiTokenizer {
     assert isFunction();
     // "{{.."
     if (firstUnescapedPipePos != -1) {
-      return wikiText.substring(start + 2, firstUnescapedPipePos);
+      return trimNewlines(wikiText.substring(start + 2, firstUnescapedPipePos).trim());
     }
-    return wikiText.substring(start + 2, end - 2);
+    final int safeEnd = Math.max(start + 2, end - 2);
+    return trimNewlines(wikiText.substring(start + 2, safeEnd).trim());
   }
   
   public List<String> functionPositionArgs() {
@@ -221,21 +288,25 @@ public final class WikiTokenizer {
     assert isWikiLink();
     // "[[.."
     if (lastUnescapedPipePos != -1) {
-      return wikiText.substring(lastUnescapedPipePos + 1, end - 2);
+      return trimNewlines(wikiText.substring(lastUnescapedPipePos + 1, end - 2));
     }
     assert start + 2 < wikiText.length() && end >= 2: wikiText;
-    return wikiText.substring(start + 2, end - 2);
+    return trimNewlines(wikiText.substring(start + 2, end - 2));
   }
 
   public String wikiLinkDest() {
     assert isWikiLink();
     // "[[.."
     if (firstUnescapedPipePos != -1) {
-      return wikiText.substring(start + 2, firstUnescapedPipePos);
+      return trimNewlines(wikiText.substring(start + 2, firstUnescapedPipePos));
     }
     return null;
   }
   
+  public boolean isHtml() {
+    return isHtml;
+  }
+
   public boolean remainderStartsWith(final String prefix) {
     return wikiText.startsWith(prefix, start);
   }
@@ -274,7 +345,7 @@ public final class WikiTokenizer {
       return this;
     }
     
-    if (justReturnedNewline) {    
+    if (justReturnedNewline) {   
       justReturnedNewline = false;
 
       final char firstChar = wikiText.charAt(end);
@@ -338,11 +409,19 @@ public final class WikiTokenizer {
 
     if (wikiText.startsWith("<pre>", start)) {
       end = safeIndexOf(wikiText, start, "</pre>", "\n");
+      isHtml = true;
       return this;
     }
 
+    if (wikiText.startsWith("<ref>", start)) {
+        end = safeIndexOf(wikiText, start, "</ref>", "\n");
+        isHtml = true;
+        return this;
+      }
+
     if (wikiText.startsWith("<math>", start)) {
       end = safeIndexOf(wikiText, start, "</math>", "\n");
+      isHtml = true;
       return this;
     }
 
@@ -404,9 +483,9 @@ public final class WikiTokenizer {
         final String matchText = matcher.group();
         final int matchStart = matcher.start();
         
-        assert matcher.end() > end || matchText.length() == 0: "Group=" + matcher.group();
+        assert matcher.end() > end || matchText.length() == 0: "Group=" + matchText;
         if (matchText.length() == 0) {
-          assert matchStart == wikiText.length() || wikiText.charAt(matchStart) == '\n';
+          assert matchStart == wikiText.length() || wikiText.charAt(matchStart) == '\n' : wikiText + ", " + matchStart;
           if (firstNewline == -1) {
             firstNewline = matcher.end();
           }
@@ -425,15 +504,15 @@ public final class WikiTokenizer {
         } else if (matchText.equals("]]") || matchText.equals("}}")) {
           if (tokenStack.size() > 0) {
             final String removed = tokenStack.remove(tokenStack.size() - 1);
-            if (removed.equals("{{") && !matcher.group().equals("}}")) {
+            if (removed.equals("{{") && !matchText.equals("}}")) {
               errors.add("Unmatched {{ error: " + wikiText.substring(start));
               return safeIndexOf(wikiText, start, "\n", "\n");
-            } else if (removed.equals("[[") && !matcher.group().equals("]]")) {
+            } else if (removed.equals("[[") && !matchText.equals("]]")) {
               errors.add("Unmatched [[ error: " + wikiText.substring(start));
               return safeIndexOf(wikiText, start, "\n", "\n");
             }
           } else {
-            errors.add("Pop too many error: " + wikiText.substring(start).replaceAll("\n", "\\\\n"));
+            errors.add("Pop too many error: " + wikiText.substring(start).replace("\n", "\\\\n"));
             // If we were looking for a newline
             return safeIndexOf(wikiText, start, "\n", "\n");
           }
@@ -453,7 +532,7 @@ public final class WikiTokenizer {
             errors.add("Unmatched <!-- error: " + wikiText.substring(start));
             return safeIndexOf(wikiText, start, "\n", "\n");
           }
-        } else if (matchText.equals("''")) {
+        } else if (matchText.equals("''") || (matchText.startsWith("<") && matchText.endsWith(">"))) {
           // Don't care.
         } else {
           assert false : "Match text='" + matchText + "'";
@@ -472,8 +551,8 @@ public final class WikiTokenizer {
       // We were looking for the end, we got it.
       return end;
     }
+    errors.add("Couldn't find: " + toFind + ", "+ wikiText.substring(start));
     if (firstNewline != -1) {
-      errors.add("Couldn't find: " + toFind + ", "+ wikiText.substring(start));
       return firstNewline;
     }
     return end;
@@ -486,14 +565,24 @@ public final class WikiTokenizer {
       if (lastUnescapedEqualsPos > lastUnescapedPipePos) {
         final String key = wikiText.substring(lastUnescapedPipePos + 1, lastUnescapedEqualsPos);
         final String value = wikiText.substring(lastUnescapedEqualsPos + 1, matchStart);
-        namedArgs.put(key, value);
+        namedArgs.put(trimNewlines(key), trimNewlines(value));
       } else {
         final String value = wikiText.substring(lastUnescapedPipePos + 1, matchStart);
-        positionArgs.add(value);
+        positionArgs.add(trimNewlines(value));
       }
     }
     lastUnescapedPipePos = matchStart;
   }
+  
+  static final String trimNewlines(String s) {
+    while (s.startsWith("\n")) {
+      s = s.substring(1);
+    }
+    while (s.endsWith("\n")) {
+      s = s.substring(0, s.length() - 1);
+    }
+    return s.replace('\n', ' ');
+  }
 
   static int safeIndexOf(final String s, final int start, final String target, final String backup) {
     int close = s.indexOf(target, start);