2 package com.hughes.android.dictionary.parser.wiktionary;
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;
10 import org.apache.commons.lang3.StringEscapeUtils;
12 import java.util.ArrayList;
13 import java.util.LinkedHashMap;
14 import java.util.List;
16 import java.util.regex.Pattern;
18 public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
20 interface LangConfig {
21 boolean skipSection(final String name);
22 boolean skipWikiLink(final WikiTokenizer wikiTokenizer);
23 String adjustWikiLink(String wikiLinkDest);
25 static final Map<String,LangConfig> isoToLangConfig = new LinkedHashMap<String,LangConfig>();
27 final Pattern enSkipSections = Pattern.compile(".*Translations|Anagrams|References.*");
28 isoToLangConfig.put("EN", new LangConfig() {
30 public boolean skipSection(String headingText) {
31 return enSkipSections.matcher(headingText).matches();
35 public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
36 final String wikiText = wikiTokenizer.wikiLinkText();
37 if (wikiText.startsWith("Category:")) {
43 public String adjustWikiLink(String wikiLinkDest) {
44 if (wikiLinkDest.startsWith("w:") || wikiLinkDest.startsWith("Image:")) {
51 public static final String NAME = "WholeSectionToHtmlParser";
53 final IndexBuilder titleIndexBuilder;
54 final LangConfig langConfig;
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);
63 void parseSection(String heading, String text) {
64 HtmlEntry htmlEntry = new HtmlEntry(entrySource, StringEscapeUtils.escapeHtml3(title));
65 IndexedEntry indexedEntry = new IndexedEntry(htmlEntry);
67 final AppendAndIndexWikiCallback<WholeSectionToHtmlParser> callback = new AppendCallback(
70 callback.builder = new StringBuilder();
71 callback.indexedEntry = indexedEntry;
72 callback.dispatch(text, null);
74 htmlEntry.html = callback.builder.toString();
75 indexedEntry.isValid = true;
77 final TokenData tokenData = titleIndexBuilder.getOrCreateTokenData(title);
79 htmlEntry.addToDictionary(titleIndexBuilder.index.dict);
80 tokenData.htmlEntries.add(htmlEntry);
81 // titleIndexBuilder.addEntryWithString(indexedEntry, title,
82 // EntryTypeName.WIKTIONARY_TITLE_MULTI_DETAIL);
86 void removeUselessArgs(Map<String, String> namedArgs) {
89 class AppendCallback extends AppendAndIndexWikiCallback<WholeSectionToHtmlParser> {
90 public AppendCallback(WholeSectionToHtmlParser parser) {
95 public void onPlainText(String plainText) {
96 super.onPlainText(StringEscapeUtils.escapeHtml3(plainText));
100 public void onWikiLink(WikiTokenizer wikiTokenizer) {
101 if (wikiTokenizer.wikiLinkText().endsWith(":" + title)) {
102 // Skips wikilinks like: [[en::dick]]
105 if (langConfig.skipWikiLink(wikiTokenizer)) {
109 if (wikiTokenizer.wikiLinkDest() != null) {
110 linkDest = langConfig.adjustWikiLink(wikiTokenizer.wikiLinkDest());
112 linkDest = wikiTokenizer.wikiLinkText();
114 if (linkDest != null) {
115 builder.append(String.format("<a href=\"%s\">", linkDest));
116 super.onWikiLink(wikiTokenizer);
117 builder.append(String.format("</a>"));
119 super.onWikiLink(wikiTokenizer);
124 public void onFunction(WikiTokenizer wikiTokenizer, String name,
125 List<String> args, Map<String, String> namedArgs) {
126 super.onFunction(wikiTokenizer, name, args, namedArgs);
130 public void onHtml(WikiTokenizer wikiTokenizer) {
131 super.onHtml(wikiTokenizer);
135 public void onNewline(WikiTokenizer wikiTokenizer) {
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();
151 builder.append(String.format("\n<h%d>", depth));
152 dispatch(headingText, null);
153 builder.append(String.format("</h%d>\n", depth));
156 final List<Character> listPrefixStack = new ArrayList<Character>();
159 public void onListItem(WikiTokenizer wikiTokenizer) {
160 if (builder.length() != 0 && builder.charAt(builder.length() - 1) != '\n') {
161 builder.append("\n");
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()));
169 builder.append("<li>");
170 dispatch(wikiTokenizer.listItemWikiText(), null);
171 builder.append("</li>\n");
173 WikiTokenizer nextToken = wikiTokenizer.nextToken();
174 boolean returnToLineStart = false;
175 if (nextToken != null && nextToken.isNewline()) {
176 nextToken = nextToken.nextToken();
177 returnToLineStart = true;
179 final String nextListHeader;
180 if (nextToken == null || !nextToken.isListItem()) {
183 nextListHeader = nextToken.listItemPrefix();
185 if (returnToLineStart) {
186 wikiTokenizer.returnToLineStart();
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)));
194 boolean boldOn = false;
195 boolean italicOn = false;
198 public void onMarkup(WikiTokenizer wikiTokenizer) {
199 if ("'''".equals(wikiTokenizer.token())) {
201 builder.append("<b>");
203 builder.append("</b>");
206 } else if ("''".equals(wikiTokenizer.token())) {
208 builder.append("<em>");
210 builder.append("</em>");
212 italicOn = !italicOn;