src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java

   1
   2 package com.hughes.android.dictionary.parser.wiktionary;
   3
   4 import com.hughes.android.dictionary.engine.HtmlEntry;
   5 import com.hughes.android.dictionary.engine.IndexBuilder;
   6 import com.hughes.android.dictionary.engine.IndexBuilder.TokenData;
   7 import com.hughes.android.dictionary.engine.IndexedEntry;
   8 import com.hughes.android.dictionary.parser.WikiTokenizer;
   9
  10 import org.apache.commons.lang3.StringEscapeUtils;
  11
  12 import java.util.ArrayList;
  13 import java.util.List;
  14 import java.util.Map;
  15 import java.util.regex.Pattern;
  16
  17 public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
  18
  19     public static final String NAME = "WholeSectionToHtmlParser";
  20     public static final Pattern skipSections = Pattern.compile(".*Translations.*");
  21
  22     final IndexBuilder titleIndexBuilder;
  23
  24     public WholeSectionToHtmlParser(final IndexBuilder titleIndexBuilder) {
  25         this.titleIndexBuilder = titleIndexBuilder;
  26
  27     }
  28
  29     @Override
  30     void parseSection(String heading, String text) {
  31         HtmlEntry htmlEntry = new HtmlEntry(entrySource, StringEscapeUtils.escapeHtml3(title));
  32         IndexedEntry indexedEntry = new IndexedEntry(htmlEntry);
  33
  34         final AppendAndIndexWikiCallback<WholeSectionToHtmlParser> callback = new AppendCallback(
  35                 this);
  36
  37         callback.builder = new StringBuilder();
  38         callback.indexedEntry = indexedEntry;
  39         callback.dispatch(text, null);
  40
  41         htmlEntry.html = callback.builder.toString();
  42         indexedEntry.isValid = true;
  43
  44         final TokenData tokenData = titleIndexBuilder.getOrCreateTokenData(title);
  45
  46         htmlEntry.addToDictionary(titleIndexBuilder.index.dict);
  47         tokenData.htmlEntries.add(htmlEntry);
  48         // titleIndexBuilder.addEntryWithString(indexedEntry, title,
  49         // EntryTypeName.WIKTIONARY_TITLE_MULTI_DETAIL);
  50     }
  51
  52     @Override
  53     void removeUselessArgs(Map<String, String> namedArgs) {
  54     }
  55
  56     class AppendCallback extends AppendAndIndexWikiCallback<WholeSectionToHtmlParser> {
  57         public AppendCallback(WholeSectionToHtmlParser parser) {
  58             super(parser);
  59         }
  60
  61         @Override
  62         public void onPlainText(String plainText) {
  63             super.onPlainText(StringEscapeUtils.escapeHtml3(plainText));
  64         }
  65
  66         @Override
  67         public void onWikiLink(WikiTokenizer wikiTokenizer) {
  68             if (wikiTokenizer.wikiLinkText().endsWith(":" + title)) {
  69                 // Skips wikilinks like: [[en::dick]]
  70                 return;
  71             }
  72             super.onWikiLink(wikiTokenizer);
  73         }
  74
  75         @Override
  76         public void onFunction(WikiTokenizer wikiTokenizer, String name,
  77                 List<String> args, Map<String, String> namedArgs) {
  78             super.onFunction(wikiTokenizer, name, args, namedArgs);
  79         }
  80
  81         @Override
  82         public void onHtml(WikiTokenizer wikiTokenizer) {
  83             super.onHtml(wikiTokenizer);
  84         }
  85
  86         @Override
  87         public void onNewline(WikiTokenizer wikiTokenizer) {
  88         }
  89
  90         @Override
  91         public void onHeading(WikiTokenizer wikiTokenizer) {
  92             final String headingText = wikiTokenizer.headingWikiText();
  93             final int depth = wikiTokenizer.headingDepth();
  94             if (skipSections.matcher(headingText).matches()) {
  95                 while ((wikiTokenizer = wikiTokenizer.nextToken()) != null) {
  96                     if (wikiTokenizer.isHeading() && wikiTokenizer.headingDepth() <= depth) {
  97                         wikiTokenizer.returnToLineStart();
  98                         return;
  99                     }
 100                 }
 101                 return;
 102             }
 103             builder.append(String.format("\n<h%d>", depth));
 104             dispatch(headingText, null);
 105             builder.append(String.format("</h%d>\n", depth));
 106         }
 107
 108         final List<Character> listPrefixStack = new ArrayList<Character>();
 109
 110         @Override
 111         public void onListItem(WikiTokenizer wikiTokenizer) {
 112             if (builder.length() != 0 && builder.charAt(builder.length() - 1) != '\n') {
 113                 builder.append("\n");
 114             }
 115             final String prefix = wikiTokenizer.listItemPrefix();
 116             while (listPrefixStack.size() < prefix.length()) {
 117                 builder.append(String.format("<%s>",
 118                         WikiTokenizer.getListTag(prefix.charAt(listPrefixStack.size()))));
 119                 listPrefixStack.add(prefix.charAt(listPrefixStack.size()));
 120             }
 121             builder.append("<li>");
 122             dispatch(wikiTokenizer.listItemWikiText(), null);
 123             builder.append("</li>\n");
 124
 125             WikiTokenizer nextToken = wikiTokenizer.nextToken();
 126             boolean returnToLineStart = false;
 127             if (nextToken != null && nextToken.isNewline()) {
 128                 nextToken = nextToken.nextToken();
 129                 returnToLineStart = true;
 130             }
 131             final String nextListHeader;
 132             if (nextToken == null || !nextToken.isListItem()) {
 133                 nextListHeader = "";
 134             } else {
 135                 nextListHeader = nextToken.listItemPrefix();
 136             }
 137             if (returnToLineStart) {
 138                 wikiTokenizer.returnToLineStart();
 139             }
 140             while (listPrefixStack.size() > nextListHeader.length()) {
 141                 final char prefixChar = listPrefixStack.remove(listPrefixStack.size() - 1);
 142                 builder.append(String.format("</%s>\n", WikiTokenizer.getListTag(prefixChar)));
 143             }
 144         }
 145
 146         boolean boldOn = false;
 147         boolean italicOn = false;
 148
 149         @Override
 150         public void onMarkup(WikiTokenizer wikiTokenizer) {
 151             if ("'''".equals(wikiTokenizer.token())) {
 152                 if (!boldOn) {
 153                     builder.append("<b>");
 154                 } else {
 155                     builder.append("</b>");
 156                 }
 157                 boldOn = !boldOn;
 158             } else if ("''".equals(wikiTokenizer.token())) {
 159                 if (!italicOn) {
 160                     builder.append("<em>");
 161                 } else {
 162                     builder.append("</em>");
 163                 }
 164                 italicOn = !italicOn;
 165             } else {
 166                 assert false;
 167             }
 168         }
 169
 170     }
 171
 172 }