1 package com.hughes.android.dictionary.parser.wiktionary;
3 import com.hughes.android.dictionary.engine.HtmlEntry;
4 import com.hughes.android.dictionary.engine.IndexBuilder;
5 import com.hughes.android.dictionary.engine.IndexBuilder.TokenData;
6 import com.hughes.android.dictionary.engine.IndexedEntry;
7 import com.hughes.android.dictionary.parser.WikiTokenizer;
9 import org.apache.commons.lang3.StringEscapeUtils;
11 import java.util.ArrayList;
12 import java.util.List;
14 import java.util.regex.Pattern;
16 public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
18 public static final String NAME = "WholeSectionToHtmlParser";
19 public static final Pattern skipSections = Pattern.compile(".*Translations.*");
21 final IndexBuilder titleIndexBuilder;
23 public WholeSectionToHtmlParser(final IndexBuilder titleIndexBuilder) {
24 this.titleIndexBuilder = titleIndexBuilder;
28 void parseSection(String heading, String text) {
29 HtmlEntry htmlEntry = new HtmlEntry(entrySource, title);
30 IndexedEntry indexedEntry = new IndexedEntry(htmlEntry);
32 final AppendAndIndexWikiCallback<WholeSectionToHtmlParser> callback = new AppendCallback(this);
33 callback.builder = new StringBuilder();
34 callback.indexedEntry = indexedEntry;
35 callback.dispatch(text, null);
37 htmlEntry.html = callback.builder.toString();
38 indexedEntry.isValid = true;
40 final TokenData tokenData = titleIndexBuilder.getOrCreateTokenData(title);
42 htmlEntry.addToDictionary(titleIndexBuilder.index.dict);
43 tokenData.htmlEntries.add(htmlEntry);
44 //titleIndexBuilder.addEntryWithString(indexedEntry, title, EntryTypeName.WIKTIONARY_TITLE_MULTI_DETAIL);
48 void removeUselessArgs(Map<String, String> namedArgs) {
51 class AppendCallback extends AppendAndIndexWikiCallback<WholeSectionToHtmlParser> {
52 public AppendCallback(WholeSectionToHtmlParser parser) {
57 public void onPlainText(String plainText) {
58 super.onPlainText(StringEscapeUtils.escapeHtml3(plainText));
62 public void onWikiLink(WikiTokenizer wikiTokenizer) {
63 super.onWikiLink(wikiTokenizer);
67 public void onFunction(WikiTokenizer wikiTokenizer, String name,
68 List<String> args, Map<String, String> namedArgs) {
69 super.onFunction(wikiTokenizer, name, args, namedArgs);
73 public void onHtml(WikiTokenizer wikiTokenizer) {
74 super.onHtml(wikiTokenizer);
78 public void onNewline(WikiTokenizer wikiTokenizer) {
82 public void onHeading(WikiTokenizer wikiTokenizer) {
83 final String headingText = wikiTokenizer.headingWikiText();
84 final int depth = wikiTokenizer.headingDepth();
85 if (skipSections.matcher(headingText).matches()) {
86 while ((wikiTokenizer = wikiTokenizer.nextToken()) != null) {
87 if (wikiTokenizer.isHeading() && wikiTokenizer.headingDepth() <= depth) {
88 wikiTokenizer.returnToLineStart();
94 builder.append(String.format("\n<h%d>", depth));
95 dispatch(headingText, null);
96 builder.append(String.format("</h%d>\n", depth));
99 final List<Character> listPrefixStack = new ArrayList<Character>();
101 public void onListItem(WikiTokenizer wikiTokenizer) {
102 if (builder.length() != 0 && builder.charAt(builder.length() - 1) != '\n') {
103 builder.append("\n");
105 final String prefix = wikiTokenizer.listItemPrefix();
106 while (listPrefixStack.size() < prefix.length()) {
107 builder.append(String.format("<%s>", WikiTokenizer.getListTag(prefix.charAt(listPrefixStack.size()))));
108 listPrefixStack.add(prefix.charAt(listPrefixStack.size()));
110 builder.append("<li>");
111 dispatch(wikiTokenizer.listItemWikiText(), null);
112 builder.append("</li>\n");
114 WikiTokenizer nextToken = wikiTokenizer.nextToken();
115 boolean returnToLineStart = false;
116 if (nextToken != null && nextToken.isNewline()) {
117 nextToken = nextToken.nextToken();
118 returnToLineStart = true;
120 final String nextListHeader;
121 if (nextToken == null || !nextToken.isListItem()) {
124 nextListHeader = nextToken.listItemPrefix();
126 if (returnToLineStart) {
127 wikiTokenizer.returnToLineStart();
129 while (listPrefixStack.size() > nextListHeader.length()) {
130 final char prefixChar = listPrefixStack.remove(listPrefixStack.size() - 1);
131 builder.append(String.format("</%s>\n", WikiTokenizer.getListTag(prefixChar)));
135 boolean boldOn = false;
136 boolean italicOn = false;
138 public void onMarkup(WikiTokenizer wikiTokenizer) {
139 if ("'''".equals(wikiTokenizer.token())) {
141 builder.append("<b>");
143 builder.append("</b>");
146 } else if ("''".equals(wikiTokenizer.token())) {
148 builder.append("<em>");
150 builder.append("</em>");
152 italicOn = !italicOn;