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 public static final String NAME = "WholeSectionToHtmlParser";
22 interface LangConfig {
23 boolean skipSection(final String name);
24 boolean skipWikiLink(final WikiTokenizer wikiTokenizer);
25 String adjustWikiLink(String wikiLinkDest);
27 static final Map<String,LangConfig> isoToLangConfig = new LinkedHashMap<String,LangConfig>();
29 final Pattern enSkipSections = Pattern.compile(".*Translations|Anagrams|References.*");
30 isoToLangConfig.put("EN", new LangConfig() {
32 public boolean skipSection(String headingText) {
33 return enSkipSections.matcher(headingText).matches();
37 public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
38 final String wikiText = wikiTokenizer.wikiLinkText();
39 if (wikiText.startsWith("Category:")) {
45 public String adjustWikiLink(String wikiLinkDest) {
46 if (wikiLinkDest.startsWith("w:") || wikiLinkDest.startsWith("Image:")) {
53 final IndexBuilder titleIndexBuilder;
54 final String skipLangIso;
55 final LangConfig langConfig;
57 public WholeSectionToHtmlParser(final IndexBuilder titleIndexBuilder, final String wiktionaryIso, final String skipLangIso) {
58 this.titleIndexBuilder = titleIndexBuilder;
59 assert isoToLangConfig.containsKey(wiktionaryIso): wiktionaryIso;
60 this.langConfig = isoToLangConfig.get(wiktionaryIso);
61 this.skipLangIso = skipLangIso;
65 void parseSection(String heading, String text) {
66 HtmlEntry htmlEntry = new HtmlEntry(entrySource, StringEscapeUtils.escapeHtml3(title));
67 IndexedEntry indexedEntry = new IndexedEntry(htmlEntry);
69 final AppendAndIndexWikiCallback<WholeSectionToHtmlParser> callback = new AppendCallback(
72 callback.builder = new StringBuilder();
73 callback.indexedEntry = indexedEntry;
74 callback.dispatch(text, null);
76 htmlEntry.html = callback.builder.toString();
77 indexedEntry.isValid = true;
79 final TokenData tokenData = titleIndexBuilder.getOrCreateTokenData(title);
81 htmlEntry.addToDictionary(titleIndexBuilder.index.dict);
82 tokenData.htmlEntries.add(htmlEntry);
83 // titleIndexBuilder.addEntryWithString(indexedEntry, title,
84 // EntryTypeName.WIKTIONARY_TITLE_MULTI_DETAIL);
88 void removeUselessArgs(Map<String, String> namedArgs) {
91 class AppendCallback extends AppendAndIndexWikiCallback<WholeSectionToHtmlParser> {
92 public AppendCallback(WholeSectionToHtmlParser parser) {
97 public void onPlainText(String plainText) {
98 super.onPlainText(StringEscapeUtils.escapeHtml3(plainText));
102 public void onWikiLink(WikiTokenizer wikiTokenizer) {
103 if (wikiTokenizer.wikiLinkText().endsWith(":" + title)) {
104 // Skips wikilinks like: [[en::dick]]
107 if (langConfig.skipWikiLink(wikiTokenizer)) {
111 if (wikiTokenizer.wikiLinkDest() != null) {
112 linkDest = langConfig.adjustWikiLink(wikiTokenizer.wikiLinkDest());
114 linkDest = wikiTokenizer.wikiLinkText();
116 if (linkDest != null) {
117 builder.append(String.format("<a href=\"%s\">", linkDest));
118 super.onWikiLink(wikiTokenizer);
119 builder.append(String.format("</a>"));
121 super.onWikiLink(wikiTokenizer);
126 public void onFunction(WikiTokenizer wikiTokenizer, String name,
127 List<String> args, Map<String, String> namedArgs) {
128 if (skipLangIso.equalsIgnoreCase(namedArgs.get("lang"))) {
129 namedArgs.remove("lang");
131 super.onFunction(wikiTokenizer, name, args, namedArgs);
135 public void onHtml(WikiTokenizer wikiTokenizer) {
136 super.onHtml(wikiTokenizer);
140 public void onNewline(WikiTokenizer wikiTokenizer) {
144 public void onHeading(WikiTokenizer wikiTokenizer) {
145 final String headingText = wikiTokenizer.headingWikiText();
146 final int depth = wikiTokenizer.headingDepth();
147 if (langConfig.skipSection(headingText)) {
148 while ((wikiTokenizer = wikiTokenizer.nextToken()) != null) {
149 if (wikiTokenizer.isHeading() && wikiTokenizer.headingDepth() <= depth) {
150 wikiTokenizer.returnToLineStart();
156 builder.append(String.format("\n<h%d>", depth));
157 dispatch(headingText, null);
158 builder.append(String.format("</h%d>\n", depth));
161 final List<Character> listPrefixStack = new ArrayList<Character>();
164 public void onListItem(WikiTokenizer wikiTokenizer) {
165 if (builder.length() != 0 && builder.charAt(builder.length() - 1) != '\n') {
166 builder.append("\n");
168 final String prefix = wikiTokenizer.listItemPrefix();
169 while (listPrefixStack.size() < prefix.length()) {
170 builder.append(String.format("<%s>",
171 WikiTokenizer.getListTag(prefix.charAt(listPrefixStack.size()))));
172 listPrefixStack.add(prefix.charAt(listPrefixStack.size()));
174 builder.append("<li>");
175 dispatch(wikiTokenizer.listItemWikiText(), null);
176 builder.append("</li>\n");
178 WikiTokenizer nextToken = wikiTokenizer.nextToken();
179 boolean returnToLineStart = false;
180 if (nextToken != null && nextToken.isNewline()) {
181 nextToken = nextToken.nextToken();
182 returnToLineStart = true;
184 final String nextListHeader;
185 if (nextToken == null || !nextToken.isListItem()) {
188 nextListHeader = nextToken.listItemPrefix();
190 if (returnToLineStart) {
191 wikiTokenizer.returnToLineStart();
193 while (listPrefixStack.size() > nextListHeader.length()) {
194 final char prefixChar = listPrefixStack.remove(listPrefixStack.size() - 1);
195 builder.append(String.format("</%s>\n", WikiTokenizer.getListTag(prefixChar)));
199 boolean boldOn = false;
200 boolean italicOn = false;
203 public void onMarkup(WikiTokenizer wikiTokenizer) {
204 if ("'''".equals(wikiTokenizer.token())) {
206 builder.append("<b>");
208 builder.append("</b>");
211 } else if ("''".equals(wikiTokenizer.token())) {
213 builder.append("<em>");
215 builder.append("</em>");
217 italicOn = !italicOn;