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.LinkedHashMap;
  14 import java.util.List;
  15 import java.util.Map;
  16 import java.util.regex.Pattern;
  17
  18 public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
  19
  20     interface LangConfig {
  21         boolean skipSection(final String name);
  22         boolean skipWikiLink(final WikiTokenizer wikiTokenizer);
  23         String adjustWikiLink(String wikiLinkDest);
  24     }
  25     static final Map<String,LangConfig> isoToLangConfig = new LinkedHashMap<String,LangConfig>();
  26     static {
  27         final Pattern enSkipSections = Pattern.compile(".*Translations|Anagrams|References.*");
  28         isoToLangConfig.put("EN", new LangConfig() {
  29             @Override
  30             public boolean skipSection(String headingText) {
  31                 return enSkipSections.matcher(headingText).matches();
  32             }
  33
  34             @Override
  35             public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
  36                 final String wikiText = wikiTokenizer.wikiLinkText();
  37                 if (wikiText.startsWith("Category:")) {
  38                     return true;
  39                 }
  40                 return false;
  41             }
  42             @Override
  43             public String adjustWikiLink(String wikiLinkDest) {
  44                 if (wikiLinkDest.startsWith("w:") || wikiLinkDest.startsWith("Image:")) {
  45                     return null;
  46                 }
  47                 return wikiLinkDest;
  48             }});
  49     }
  50
  51     public static final String NAME = "WholeSectionToHtmlParser";
  52
  53     final IndexBuilder titleIndexBuilder;
  54     final LangConfig langConfig;
  55
  56     public WholeSectionToHtmlParser(final IndexBuilder titleIndexBuilder, final String wiktionaryIso) {
  57         this.titleIndexBuilder = titleIndexBuilder;
  58         assert isoToLangConfig.containsKey(wiktionaryIso): wiktionaryIso;
  59         this.langConfig = isoToLangConfig.get(wiktionaryIso);
  60     }
  61
  62     @Override
  63     void parseSection(String heading, String text) {
  64         HtmlEntry htmlEntry = new HtmlEntry(entrySource, StringEscapeUtils.escapeHtml3(title));
  65         IndexedEntry indexedEntry = new IndexedEntry(htmlEntry);
  66
  67         final AppendAndIndexWikiCallback<WholeSectionToHtmlParser> callback = new AppendCallback(
  68                 this);
  69
  70         callback.builder = new StringBuilder();
  71         callback.indexedEntry = indexedEntry;
  72         callback.dispatch(text, null);
  73
  74         htmlEntry.html = callback.builder.toString();
  75         indexedEntry.isValid = true;
  76
  77         final TokenData tokenData = titleIndexBuilder.getOrCreateTokenData(title);
  78
  79         htmlEntry.addToDictionary(titleIndexBuilder.index.dict);
  80         tokenData.htmlEntries.add(htmlEntry);
  81         // titleIndexBuilder.addEntryWithString(indexedEntry, title,
  82         // EntryTypeName.WIKTIONARY_TITLE_MULTI_DETAIL);
  83     }
  84
  85     @Override
  86     void removeUselessArgs(Map<String, String> namedArgs) {
  87     }
  88
  89     class AppendCallback extends AppendAndIndexWikiCallback<WholeSectionToHtmlParser> {
  90         public AppendCallback(WholeSectionToHtmlParser parser) {
  91             super(parser);
  92         }
  93
  94         @Override
  95         public void onPlainText(String plainText) {
  96             super.onPlainText(StringEscapeUtils.escapeHtml3(plainText));
  97         }
  98
  99         @Override
 100         public void onWikiLink(WikiTokenizer wikiTokenizer) {
 101             if (wikiTokenizer.wikiLinkText().endsWith(":" + title)) {
 102                 // Skips wikilinks like: [[en::dick]]
 103                 return;
 104             }
 105             if (langConfig.skipWikiLink(wikiTokenizer)) {
 106                 return;
 107             }
 108             String linkDest;
 109             if (wikiTokenizer.wikiLinkDest() != null) {
 110                 linkDest = langConfig.adjustWikiLink(wikiTokenizer.wikiLinkDest());
 111             } else {
 112                 linkDest = wikiTokenizer.wikiLinkText();
 113             }
 114             if (linkDest != null) {
 115                 builder.append(String.format("<a href=\"%s\">", linkDest));
 116                 super.onWikiLink(wikiTokenizer);
 117                 builder.append(String.format("</a>"));
 118             } else {
 119                 super.onWikiLink(wikiTokenizer);
 120             }
 121         }
 122
 123         @Override
 124         public void onFunction(WikiTokenizer wikiTokenizer, String name,
 125                 List<String> args, Map<String, String> namedArgs) {
 126             super.onFunction(wikiTokenizer, name, args, namedArgs);
 127         }
 128
 129         @Override
 130         public void onHtml(WikiTokenizer wikiTokenizer) {
 131             super.onHtml(wikiTokenizer);
 132         }
 133
 134         @Override
 135         public void onNewline(WikiTokenizer wikiTokenizer) {
 136         }
 137
 138         @Override
 139         public void onHeading(WikiTokenizer wikiTokenizer) {
 140             final String headingText = wikiTokenizer.headingWikiText();
 141             final int depth = wikiTokenizer.headingDepth();
 142             if (langConfig.skipSection(headingText)) {
 143                 while ((wikiTokenizer = wikiTokenizer.nextToken()) != null) {
 144                     if (wikiTokenizer.isHeading() && wikiTokenizer.headingDepth() <= depth) {
 145                         wikiTokenizer.returnToLineStart();
 146                         return;
 147                     }
 148                 }
 149                 return;
 150             }
 151             builder.append(String.format("\n<h%d>", depth));
 152             dispatch(headingText, null);
 153             builder.append(String.format("</h%d>\n", depth));
 154         }
 155
 156         final List<Character> listPrefixStack = new ArrayList<Character>();
 157
 158         @Override
 159         public void onListItem(WikiTokenizer wikiTokenizer) {
 160             if (builder.length() != 0 && builder.charAt(builder.length() - 1) != '\n') {
 161                 builder.append("\n");
 162             }
 163             final String prefix = wikiTokenizer.listItemPrefix();
 164             while (listPrefixStack.size() < prefix.length()) {
 165                 builder.append(String.format("<%s>",
 166                         WikiTokenizer.getListTag(prefix.charAt(listPrefixStack.size()))));
 167                 listPrefixStack.add(prefix.charAt(listPrefixStack.size()));
 168             }
 169             builder.append("<li>");
 170             dispatch(wikiTokenizer.listItemWikiText(), null);
 171             builder.append("</li>\n");
 172
 173             WikiTokenizer nextToken = wikiTokenizer.nextToken();
 174             boolean returnToLineStart = false;
 175             if (nextToken != null && nextToken.isNewline()) {
 176                 nextToken = nextToken.nextToken();
 177                 returnToLineStart = true;
 178             }
 179             final String nextListHeader;
 180             if (nextToken == null || !nextToken.isListItem()) {
 181                 nextListHeader = "";
 182             } else {
 183                 nextListHeader = nextToken.listItemPrefix();
 184             }
 185             if (returnToLineStart) {
 186                 wikiTokenizer.returnToLineStart();
 187             }
 188             while (listPrefixStack.size() > nextListHeader.length()) {
 189                 final char prefixChar = listPrefixStack.remove(listPrefixStack.size() - 1);
 190                 builder.append(String.format("</%s>\n", WikiTokenizer.getListTag(prefixChar)));
 191             }
 192         }
 193
 194         boolean boldOn = false;
 195         boolean italicOn = false;
 196
 197         @Override
 198         public void onMarkup(WikiTokenizer wikiTokenizer) {
 199             if ("'''".equals(wikiTokenizer.token())) {
 200                 if (!boldOn) {
 201                     builder.append("<b>");
 202                 } else {
 203                     builder.append("</b>");
 204                 }
 205                 boldOn = !boldOn;
 206             } else if ("''".equals(wikiTokenizer.token())) {
 207                 if (!italicOn) {
 208                     builder.append("<em>");
 209                 } else {
 210                     builder.append("</em>");
 211                 }
 212                 italicOn = !italicOn;
 213             } else {
 214                 assert false;
 215             }
 216         }
 217
 218     }
 219
 220 }