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;
12 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 (probably)
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 Pattern deSkipSections = Pattern.compile(".*(Übersetzungen|Referenzen|Quellen).*");
91 isoToLangConfig.put("DE", new LangConfig() {
93 public boolean skipSection(String headingText) {
94 return deSkipSections.matcher(headingText).matches();
98 public EntryTypeName sectionNameToEntryType(String sectionName) {
99 if (sectionName.equalsIgnoreCase("Synonyme")) {
100 return EntryTypeName.SYNONYM_MULTI;
102 if (sectionName.equalsIgnoreCase("Gegenwörter")) {
103 return EntryTypeName.ANTONYM_MULTI;
109 public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
110 final String wikiText = wikiTokenizer.wikiLinkText();
111 if (wikiText.startsWith("???Category:")) {
117 public String adjustWikiLink(String wikiLinkDest, String wikiLinkText) {
118 if (wikiLinkDest.startsWith("w:") || wikiLinkDest.startsWith("Image:")) {
121 final int hashPos = wikiLinkDest.indexOf("#");
123 wikiLinkDest = wikiLinkDest.substring(0, hashPos);
124 if (wikiLinkDest.isEmpty()) {
125 wikiLinkDest = wikiLinkText;
132 public void addFunctionCallbacks(
133 Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks) {
134 DeFunctionCallbacks.addGenericCallbacks(functionCallbacks);
138 final Pattern itSkipSections = Pattern.compile(".*(Traduzione|Note / Riferimenti).*");
139 isoToLangConfig.put("IT", new LangConfig() {
141 public boolean skipSection(String headingText) {
142 return itSkipSections.matcher(headingText).matches();
146 public EntryTypeName sectionNameToEntryType(String sectionName) {
147 if (sectionName.equalsIgnoreCase("Sinonimi")) {
148 return EntryTypeName.SYNONYM_MULTI;
150 if (sectionName.equalsIgnoreCase("Antonimi/Contrari")) {
151 return EntryTypeName.ANTONYM_MULTI;
157 public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
158 final String wikiText = wikiTokenizer.wikiLinkText();
159 if (wikiText.startsWith("???Category:")) {
165 public String adjustWikiLink(String wikiLinkDest, String wikiLinkText) {
166 if (wikiLinkDest.startsWith("w:") || wikiLinkDest.startsWith("Image:")) {
169 final int hashPos = wikiLinkDest.indexOf("#");
171 wikiLinkDest = wikiLinkDest.substring(0, hashPos);
172 if (wikiLinkDest.isEmpty()) {
173 wikiLinkDest = wikiLinkText;
180 public void addFunctionCallbacks(
181 Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks) {
182 ItFunctionCallbacks.addGenericCallbacks(functionCallbacks);
187 final Pattern frSkipSections = Pattern.compile(".*(Traductions).*");
188 isoToLangConfig.put("FR", new LangConfig() {
190 public boolean skipSection(String headingText) {
191 return frSkipSections.matcher(headingText).matches();
195 public EntryTypeName sectionNameToEntryType(String sectionName) {
196 if (sectionName.equalsIgnoreCase("Synonymes")) {
197 return EntryTypeName.SYNONYM_MULTI;
203 public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
207 public String adjustWikiLink(String wikiLinkDest, String wikiLinkText) {
208 if (wikiLinkDest.startsWith("w:") || wikiLinkDest.startsWith("Image:")) {
211 final int hashPos = wikiLinkDest.indexOf("#");
213 wikiLinkDest = wikiLinkDest.substring(0, hashPos);
214 if (wikiLinkDest.isEmpty()) {
215 wikiLinkDest = wikiLinkText;
222 public void addFunctionCallbacks(
223 Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks) {
224 FrFunctionCallbacks.addGenericCallbacks(functionCallbacks);
229 final IndexBuilder titleIndexBuilder;
230 final IndexBuilder defIndexBuilder;
231 final String skipLangIso;
232 final LangConfig langConfig;
233 final String webUrlTemplate;
236 public WholeSectionToHtmlParser(final IndexBuilder titleIndexBuilder, final IndexBuilder defIndexBuilder, final String wiktionaryIso, final String skipLangIso,
237 final String webUrlTemplate) {
238 this.titleIndexBuilder = titleIndexBuilder;
239 this.defIndexBuilder = defIndexBuilder;
240 assert isoToLangConfig.containsKey(wiktionaryIso): wiktionaryIso;
241 this.langConfig = isoToLangConfig.get(wiktionaryIso);
242 this.skipLangIso = skipLangIso;
243 this.webUrlTemplate = webUrlTemplate;
246 IndexedEntry indexedEntry = null;
249 public void parseSection(String heading, String text) {
250 assert entrySource != null;
251 final HtmlEntry htmlEntry = new HtmlEntry(entrySource, title);
252 indexedEntry = new IndexedEntry(htmlEntry);
254 final AppendAndIndexWikiCallback<WholeSectionToHtmlParser> callback = new AppendCallback(
256 langConfig.addFunctionCallbacks(callback.functionCallbacks);
258 callback.builder = new StringBuilder();
259 callback.indexedEntry = indexedEntry;
260 callback.dispatch(text, null);
262 if (webUrlTemplate != null) {
263 final String webUrl = String.format(webUrlTemplate, title);
264 // URI.create can raise an exception e.g. if webUrl contains %, just ignore those cases.
266 callback.builder.append(String.format("<p> <a href=\"%s\">%s</a>", URI.create(webUrl).toString(), escapeHtmlLiteral(webUrl)));
267 } catch (Exception e)
270 htmlEntry.html = callback.builder.toString();
271 indexedEntry.isValid = true;
273 final TokenData tokenData = titleIndexBuilder.getOrCreateTokenData(title);
274 tokenData.hasMainEntry = true;
276 htmlEntry.addToDictionary(titleIndexBuilder.index.dict);
277 tokenData.htmlEntries.add(htmlEntry);
278 // titleIndexBuilder.addEntryWithString(indexedEntry, title,
279 // EntryTypeName.WIKTIONARY_TITLE_MULTI_DETAIL);
285 void removeUselessArgs(Map<String, String> namedArgs) {
289 public void addLinkToCurrentEntry(String token, final String lang, EntryTypeName entryTypeName) {
290 if (lang == null || lang.equals(skipLangIso)) {
291 titleIndexBuilder.addEntryWithString(indexedEntry, token, entryTypeName);
295 public static String escapeHtmlLiteral(final String plainText) {
296 final String htmlEscaped = StringEscapeUtils.escapeHtml3(plainText);
297 if (StringUtil.isAscii(htmlEscaped)) {
300 return StringUtil.escapeUnicodeToPureHtml(plainText);
307 class AppendCallback extends AppendAndIndexWikiCallback<WholeSectionToHtmlParser> {
308 public AppendCallback(WholeSectionToHtmlParser parser) {
313 public void onPlainText(String plainText) {
314 super.onPlainText(escapeHtmlLiteral(plainText));
318 public void onWikiLink(WikiTokenizer wikiTokenizer) {
319 if (wikiTokenizer.wikiLinkText().endsWith(":" + title)) {
320 // Skips wikilinks like: [[en::dick]]
323 if (langConfig.skipWikiLink(wikiTokenizer)) {
327 if (wikiTokenizer.wikiLinkDest() != null) {
328 linkDest = langConfig.adjustWikiLink(wikiTokenizer.wikiLinkDest(), wikiTokenizer.wikiLinkText());
330 linkDest = wikiTokenizer.wikiLinkText();
332 if (sectionEntryTypeName != null) {
333 // TODO: inside a definition, this could be the wrong language.
334 titleIndexBuilder.addEntryWithString(indexedEntry, wikiTokenizer.wikiLinkText(), sectionEntryTypeName);
336 if (!StringUtil.isNullOrEmpty(linkDest)) {
337 builder.append(String.format("<a href=\"%s\">", HtmlEntry.formatQuickdicUrl("", linkDest)));
338 super.onWikiLink(wikiTokenizer);
339 builder.append(String.format("</a>"));
341 super.onWikiLink(wikiTokenizer);
346 public void onFunction(WikiTokenizer wikiTokenizer, String name,
347 List<String> args, Map<String, String> namedArgs) {
348 if (skipLangIso.equalsIgnoreCase(namedArgs.get("lang"))) {
349 namedArgs.remove("lang");
351 super.onFunction(wikiTokenizer, name, args, namedArgs);
355 public void onHtml(WikiTokenizer wikiTokenizer) {
356 super.onHtml(wikiTokenizer);
360 public void onNewline(WikiTokenizer wikiTokenizer) {
363 EntryTypeName sectionEntryTypeName;
364 IndexBuilder currentIndexBuilder;
367 public void onHeading(WikiTokenizer wikiTokenizer) {
368 final String headingText = wikiTokenizer.headingWikiText();
369 sectionEntryTypeName = langConfig.sectionNameToEntryType(headingText);
370 final int depth = wikiTokenizer.headingDepth();
371 if (langConfig.skipSection(headingText)) {
372 //System.out.println("Skipping section:" + headingText);
373 while ((wikiTokenizer = wikiTokenizer.nextToken()) != null) {
374 if (wikiTokenizer.isHeading() && wikiTokenizer.headingDepth() <= depth) {
375 // System.out.println("Resume on: " + wikiTokenizer.token());
376 wikiTokenizer.returnToLineStart();
379 // System.out.println("Skipped: " + wikiTokenizer.token());
384 builder.append(String.format("\n<h%d>", depth));
385 dispatch(headingText, null);
386 builder.append(String.format("</h%d>\n", depth));
389 final List<Character> listPrefixStack = new ArrayList<Character>();
392 public void onListItem(WikiTokenizer wikiTokenizer) {
393 if (builder.length() != 0 && builder.charAt(builder.length() - 1) != '\n') {
394 builder.append("\n");
396 final String prefix = wikiTokenizer.listItemPrefix();
397 while (listPrefixStack.size() < prefix.length()) {
398 builder.append(String.format("<%s>",
399 WikiTokenizer.getListTag(prefix.charAt(listPrefixStack.size()))));
400 listPrefixStack.add(prefix.charAt(listPrefixStack.size()));
402 builder.append("<li>");
403 dispatch(wikiTokenizer.listItemWikiText(), null);
404 builder.append("</li>\n");
406 WikiTokenizer nextToken = wikiTokenizer.nextToken();
407 boolean returnToLineStart = false;
408 if (nextToken != null && nextToken.isNewline()) {
409 nextToken = nextToken.nextToken();
410 returnToLineStart = true;
412 final String nextListHeader;
413 if (nextToken == null || !nextToken.isListItem()) {
416 nextListHeader = nextToken.listItemPrefix();
418 if (returnToLineStart) {
419 wikiTokenizer.returnToLineStart();
421 while (listPrefixStack.size() > nextListHeader.length()) {
422 final char prefixChar = listPrefixStack.remove(listPrefixStack.size() - 1);
423 builder.append(String.format("</%s>\n", WikiTokenizer.getListTag(prefixChar)));
427 boolean boldOn = false;
428 boolean italicOn = false;
431 public void onMarkup(WikiTokenizer wikiTokenizer) {
432 if ("'''".equals(wikiTokenizer.token())) {
434 builder.append("<b>");
436 builder.append("</b>");
439 } else if ("''".equals(wikiTokenizer.token())) {
441 builder.append("<em>");
443 builder.append("</em>");
445 italicOn = !italicOn;