2 package com.hughes.android.dictionary.parser.wiktionary;
4 import com.hughes.android.dictionary.engine.EntryTypeName;
5 import com.hughes.android.dictionary.engine.HtmlEntry;
6 import com.hughes.android.dictionary.engine.IndexBuilder;
7 import com.hughes.android.dictionary.engine.IndexBuilder.TokenData;
8 import com.hughes.android.dictionary.engine.IndexedEntry;
9 import com.hughes.android.dictionary.parser.WikiTokenizer;
10 import com.hughes.util.StringUtil;
11 import com.sun.xml.internal.rngom.util.Uri;
13 import org.apache.commons.lang3.StringEscapeUtils;
15 import java.util.ArrayList;
16 import java.util.LinkedHashMap;
17 import java.util.List;
19 import java.util.regex.Pattern;
21 public class WholeSectionToHtmlParser extends AbstractWiktionaryParser {
23 public static final String NAME = "WholeSectionToHtmlParser";
25 interface LangConfig {
26 boolean skipSection(final String name);
27 EntryTypeName sectionNameToEntryType(String sectionName);
28 boolean skipWikiLink(final WikiTokenizer wikiTokenizer);
29 String adjustWikiLink(String wikiLinkDest, final String wikiLinkText);
30 void addFunctionCallbacks(
31 Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks);
33 static final Map<String,LangConfig> isoToLangConfig = new LinkedHashMap<String,LangConfig>();
35 final Pattern enSkipSections = Pattern.compile(".*Translations|Anagrams|References.*");
36 isoToLangConfig.put("EN", new LangConfig() {
38 public boolean skipSection(String headingText) {
39 return enSkipSections.matcher(headingText).matches();
43 public EntryTypeName sectionNameToEntryType(String sectionName) {
44 if (sectionName.equalsIgnoreCase("Synonyms")) {
45 return EntryTypeName.SYNONYM_MULTI;
47 if (sectionName.equalsIgnoreCase("Antonyms")) {
48 return EntryTypeName.ANTONYM_MULTI;
50 if (EnParser.partOfSpeechHeader.matcher(sectionName).matches()) {
51 // We need to put it in the other index, too.
54 if (sectionName.equalsIgnoreCase("Derived Terms")) {
61 public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
62 final String wikiText = wikiTokenizer.wikiLinkText();
63 if (wikiText.startsWith("Category:")) {
69 public String adjustWikiLink(String wikiLinkDest, String wikiLinkText) {
70 if (wikiLinkDest.startsWith("w:") || wikiLinkDest.startsWith("Image:")) {
73 final int hashPos = wikiLinkDest.indexOf("#");
75 wikiLinkDest = wikiLinkDest.substring(0, hashPos);
76 if (wikiLinkDest.isEmpty()) {
77 wikiLinkDest = wikiLinkText;
84 public void addFunctionCallbacks(
85 Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks) {
86 EnFunctionCallbacks.addGenericCallbacks(functionCallbacks);
90 final LangConfig basicLangConfig = new LangConfig() {
92 public boolean skipSection(String headingText) {
96 public EntryTypeName sectionNameToEntryType(String sectionName) {
97 return EntryTypeName.WIKTIONARY_MENTIONED;
100 public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
101 final String wikiText = wikiTokenizer.wikiLinkText();
102 if (wikiText.startsWith("Category:")) {
108 public String adjustWikiLink(String wikiLinkDest, final String wikiLinkText) {
113 public void addFunctionCallbacks(
114 Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks) {
117 isoToLangConfig.put("FR", basicLangConfig);
118 isoToLangConfig.put("DE", basicLangConfig);
119 isoToLangConfig.put("IT", basicLangConfig);
122 final IndexBuilder titleIndexBuilder;
123 final IndexBuilder defIndexBuilder;
124 final String skipLangIso;
125 final LangConfig langConfig;
126 final String webUrlTemplate;
129 public WholeSectionToHtmlParser(final IndexBuilder titleIndexBuilder, final IndexBuilder defIndexBuilder, final String wiktionaryIso, final String skipLangIso,
130 final String webUrlTemplate) {
131 this.titleIndexBuilder = titleIndexBuilder;
132 this.defIndexBuilder = defIndexBuilder;
133 assert isoToLangConfig.containsKey(wiktionaryIso): wiktionaryIso;
134 this.langConfig = isoToLangConfig.get(wiktionaryIso);
135 this.skipLangIso = skipLangIso;
136 this.webUrlTemplate = webUrlTemplate;
139 IndexedEntry indexedEntry = null;
142 public void parseSection(String heading, String text) {
143 assert entrySource != null;
144 final HtmlEntry htmlEntry = new HtmlEntry(entrySource, title);
145 indexedEntry = new IndexedEntry(htmlEntry);
147 final AppendAndIndexWikiCallback<WholeSectionToHtmlParser> callback = new AppendCallback(
149 langConfig.addFunctionCallbacks(callback.functionCallbacks);
151 callback.builder = new StringBuilder();
152 callback.indexedEntry = indexedEntry;
153 callback.dispatch(text, null);
155 if (webUrlTemplate != null) {
156 final String webUrl = String.format(webUrlTemplate, title);
157 callback.builder.append(String.format("<p> <a href=\"%s\">%s</a>", Uri.escapeDisallowedChars(webUrl), escapeHtmlLiteral(webUrl)));
159 htmlEntry.html = callback.builder.toString();
160 indexedEntry.isValid = true;
162 final TokenData tokenData = titleIndexBuilder.getOrCreateTokenData(title);
164 htmlEntry.addToDictionary(titleIndexBuilder.index.dict);
165 tokenData.htmlEntries.add(htmlEntry);
166 // titleIndexBuilder.addEntryWithString(indexedEntry, title,
167 // EntryTypeName.WIKTIONARY_TITLE_MULTI_DETAIL);
173 void removeUselessArgs(Map<String, String> namedArgs) {
177 public void addLinkToCurrentEntry(String token, EntryTypeName entryTypeName) {
178 titleIndexBuilder.addEntryWithString(indexedEntry, token, entryTypeName);
181 public static String escapeHtmlLiteral(final String plainText) {
182 final String htmlEscaped = StringEscapeUtils.escapeHtml3(plainText);
183 if (StringUtil.isAscii(htmlEscaped)) {
186 return StringUtil.escapeToPureHtmlUnicode(plainText);
193 class AppendCallback extends AppendAndIndexWikiCallback<WholeSectionToHtmlParser> {
194 public AppendCallback(WholeSectionToHtmlParser parser) {
199 public void onPlainText(String plainText) {
200 super.onPlainText(escapeHtmlLiteral(plainText));
204 public void onWikiLink(WikiTokenizer wikiTokenizer) {
205 if (wikiTokenizer.wikiLinkText().endsWith(":" + title)) {
206 // Skips wikilinks like: [[en::dick]]
209 if (langConfig.skipWikiLink(wikiTokenizer)) {
213 if (wikiTokenizer.wikiLinkDest() != null) {
214 linkDest = langConfig.adjustWikiLink(wikiTokenizer.wikiLinkDest(), wikiTokenizer.wikiLinkText());
216 linkDest = wikiTokenizer.wikiLinkText();
218 if (sectionEntryTypeName != null) {
219 // TODO: inside a definition, this could be the wrong language.
220 titleIndexBuilder.addEntryWithString(indexedEntry, wikiTokenizer.wikiLinkText(), sectionEntryTypeName);
222 if (linkDest != null) {
223 builder.append(String.format("<a href=\"%s\">", HtmlEntry.formatQuickdicUrl("", linkDest)));
224 super.onWikiLink(wikiTokenizer);
225 builder.append(String.format("</a>"));
227 super.onWikiLink(wikiTokenizer);
232 public void onFunction(WikiTokenizer wikiTokenizer, String name,
233 List<String> args, Map<String, String> namedArgs) {
234 if (skipLangIso.equalsIgnoreCase(namedArgs.get("lang"))) {
235 namedArgs.remove("lang");
237 super.onFunction(wikiTokenizer, name, args, namedArgs);
241 public void onHtml(WikiTokenizer wikiTokenizer) {
242 super.onHtml(wikiTokenizer);
246 public void onNewline(WikiTokenizer wikiTokenizer) {
249 EntryTypeName sectionEntryTypeName;
250 IndexBuilder currentIndexBuilder;
253 public void onHeading(WikiTokenizer wikiTokenizer) {
254 final String headingText = wikiTokenizer.headingWikiText();
255 sectionEntryTypeName = langConfig.sectionNameToEntryType(headingText);
256 final int depth = wikiTokenizer.headingDepth();
257 if (langConfig.skipSection(headingText)) {
258 while ((wikiTokenizer = wikiTokenizer.nextToken()) != null) {
259 if (wikiTokenizer.isHeading() && wikiTokenizer.headingDepth() <= depth) {
260 wikiTokenizer.returnToLineStart();
266 builder.append(String.format("\n<h%d>", depth));
267 dispatch(headingText, null);
268 builder.append(String.format("</h%d>\n", depth));
271 final List<Character> listPrefixStack = new ArrayList<Character>();
274 public void onListItem(WikiTokenizer wikiTokenizer) {
275 if (builder.length() != 0 && builder.charAt(builder.length() - 1) != '\n') {
276 builder.append("\n");
278 final String prefix = wikiTokenizer.listItemPrefix();
279 while (listPrefixStack.size() < prefix.length()) {
280 builder.append(String.format("<%s>",
281 WikiTokenizer.getListTag(prefix.charAt(listPrefixStack.size()))));
282 listPrefixStack.add(prefix.charAt(listPrefixStack.size()));
284 builder.append("<li>");
285 dispatch(wikiTokenizer.listItemWikiText(), null);
286 builder.append("</li>\n");
288 WikiTokenizer nextToken = wikiTokenizer.nextToken();
289 boolean returnToLineStart = false;
290 if (nextToken != null && nextToken.isNewline()) {
291 nextToken = nextToken.nextToken();
292 returnToLineStart = true;
294 final String nextListHeader;
295 if (nextToken == null || !nextToken.isListItem()) {
298 nextListHeader = nextToken.listItemPrefix();
300 if (returnToLineStart) {
301 wikiTokenizer.returnToLineStart();
303 while (listPrefixStack.size() > nextListHeader.length()) {
304 final char prefixChar = listPrefixStack.remove(listPrefixStack.size() - 1);
305 builder.append(String.format("</%s>\n", WikiTokenizer.getListTag(prefixChar)));
309 boolean boldOn = false;
310 boolean italicOn = false;
313 public void onMarkup(WikiTokenizer wikiTokenizer) {
314 if ("'''".equals(wikiTokenizer.token())) {
316 builder.append("<b>");
318 builder.append("</b>");
321 } else if ("''".equals(wikiTokenizer.token())) {
323 builder.append("<em>");
325 builder.append("</em>");
327 italicOn = !italicOn;