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 esSkipSections = Pattern.compile(".*(Traducciones|Locuciones).*");
91 isoToLangConfig.put("ES", new LangConfig() {
93 public boolean skipSection(String headingText) {
94 return esSkipSections.matcher(headingText).matches();
98 public EntryTypeName sectionNameToEntryType(String sectionName) {
99 if (sectionName.equalsIgnoreCase("sinónimo") || sectionName.equalsIgnoreCase("sinónimos")) {
100 return EntryTypeName.SYNONYM_MULTI;
102 if (sectionName.equalsIgnoreCase("antónimo") || sectionName.equalsIgnoreCase("antónimos")) {
103 return EntryTypeName.ANTONYM_MULTI;
109 public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
110 final String wikiText = wikiTokenizer.wikiLinkText();
111 if (wikiText.startsWith("Categoría:")) {
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 // TODO: need Spanish variant
135 EnFunctionCallbacks.addGenericCallbacks(functionCallbacks);
139 final Pattern deSkipSections = Pattern.compile(".*(Übersetzungen|Referenzen|Quellen).*");
140 isoToLangConfig.put("DE", new LangConfig() {
142 public boolean skipSection(String headingText) {
143 return deSkipSections.matcher(headingText).matches();
147 public EntryTypeName sectionNameToEntryType(String sectionName) {
148 if (sectionName.equalsIgnoreCase("Synonyme")) {
149 return EntryTypeName.SYNONYM_MULTI;
151 if (sectionName.equalsIgnoreCase("Gegenwörter")) {
152 return EntryTypeName.ANTONYM_MULTI;
158 public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
159 final String wikiText = wikiTokenizer.wikiLinkText();
160 if (wikiText.startsWith("???Category:")) {
166 public String adjustWikiLink(String wikiLinkDest, String wikiLinkText) {
167 if (wikiLinkDest.startsWith("w:") || wikiLinkDest.startsWith("Image:")) {
170 final int hashPos = wikiLinkDest.indexOf("#");
172 wikiLinkDest = wikiLinkDest.substring(0, hashPos);
173 if (wikiLinkDest.isEmpty()) {
174 wikiLinkDest = wikiLinkText;
181 public void addFunctionCallbacks(
182 Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks) {
183 DeFunctionCallbacks.addGenericCallbacks(functionCallbacks);
187 final Pattern itSkipSections = Pattern.compile(".*(Traduzione|Note / Riferimenti).*");
188 isoToLangConfig.put("IT", new LangConfig() {
190 public boolean skipSection(String headingText) {
191 return itSkipSections.matcher(headingText).matches();
195 public EntryTypeName sectionNameToEntryType(String sectionName) {
196 if (sectionName.equalsIgnoreCase("Sinonimi")) {
197 return EntryTypeName.SYNONYM_MULTI;
199 if (sectionName.equalsIgnoreCase("Antonimi/Contrari")) {
200 return EntryTypeName.ANTONYM_MULTI;
206 public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
207 final String wikiText = wikiTokenizer.wikiLinkText();
208 if (wikiText.startsWith("???Category:")) {
214 public String adjustWikiLink(String wikiLinkDest, String wikiLinkText) {
215 if (wikiLinkDest.startsWith("w:") || wikiLinkDest.startsWith("Image:")) {
218 final int hashPos = wikiLinkDest.indexOf("#");
220 wikiLinkDest = wikiLinkDest.substring(0, hashPos);
221 if (wikiLinkDest.isEmpty()) {
222 wikiLinkDest = wikiLinkText;
229 public void addFunctionCallbacks(
230 Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks) {
231 ItFunctionCallbacks.addGenericCallbacks(functionCallbacks);
236 final Pattern frSkipSections = Pattern.compile(".*(Traductions).*");
237 isoToLangConfig.put("FR", new LangConfig() {
239 public boolean skipSection(String headingText) {
240 return frSkipSections.matcher(headingText).matches();
244 public EntryTypeName sectionNameToEntryType(String sectionName) {
245 if (sectionName.equalsIgnoreCase("Synonymes")) {
246 return EntryTypeName.SYNONYM_MULTI;
252 public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
256 public String adjustWikiLink(String wikiLinkDest, String wikiLinkText) {
257 if (wikiLinkDest.startsWith("w:") || wikiLinkDest.startsWith("Image:")) {
260 final int hashPos = wikiLinkDest.indexOf("#");
262 wikiLinkDest = wikiLinkDest.substring(0, hashPos);
263 if (wikiLinkDest.isEmpty()) {
264 wikiLinkDest = wikiLinkText;
271 public void addFunctionCallbacks(
272 Map<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks) {
273 FrFunctionCallbacks.addGenericCallbacks(functionCallbacks);
278 final IndexBuilder titleIndexBuilder;
279 final IndexBuilder defIndexBuilder;
280 final String skipLangIso;
281 final LangConfig langConfig;
282 final String webUrlTemplate;
285 public WholeSectionToHtmlParser(final IndexBuilder titleIndexBuilder, final IndexBuilder defIndexBuilder, final String wiktionaryIso, final String skipLangIso,
286 final String webUrlTemplate) {
287 this.titleIndexBuilder = titleIndexBuilder;
288 this.defIndexBuilder = defIndexBuilder;
289 assert isoToLangConfig.containsKey(wiktionaryIso): wiktionaryIso;
290 this.langConfig = isoToLangConfig.get(wiktionaryIso);
291 this.skipLangIso = skipLangIso;
292 this.webUrlTemplate = webUrlTemplate;
295 IndexedEntry indexedEntry = null;
298 public void parseSection(String heading, String text) {
299 assert entrySource != null;
300 final HtmlEntry htmlEntry = new HtmlEntry(entrySource, title);
301 indexedEntry = new IndexedEntry(htmlEntry);
303 final AppendAndIndexWikiCallback<WholeSectionToHtmlParser> callback = new AppendCallback(
305 langConfig.addFunctionCallbacks(callback.functionCallbacks);
307 callback.builder = new StringBuilder();
308 callback.indexedEntry = indexedEntry;
309 callback.dispatch(text, null);
311 if (webUrlTemplate != null) {
312 final String webUrl = String.format(webUrlTemplate, title);
313 // URI.create can raise an exception e.g. if webUrl contains %, just ignore those cases.
315 callback.builder.append(String.format("<p> <a href=\"%s\">%s</a>", URI.create(webUrl).toString(), escapeHtmlLiteral(webUrl)));
316 } catch (Exception e)
319 htmlEntry.html = callback.builder.toString();
320 indexedEntry.isValid = true;
322 final TokenData tokenData = titleIndexBuilder.getOrCreateTokenData(title);
323 tokenData.hasMainEntry = true;
325 htmlEntry.addToDictionary(titleIndexBuilder.index.dict);
326 tokenData.htmlEntries.add(htmlEntry);
327 // titleIndexBuilder.addEntryWithString(indexedEntry, title,
328 // EntryTypeName.WIKTIONARY_TITLE_MULTI_DETAIL);
334 void removeUselessArgs(Map<String, String> namedArgs) {
338 public void addLinkToCurrentEntry(String token, final String lang, EntryTypeName entryTypeName) {
339 if (lang == null || lang.equals(skipLangIso)) {
340 titleIndexBuilder.addEntryWithString(indexedEntry, token, entryTypeName);
344 public static String escapeHtmlLiteral(final String plainText) {
345 final String htmlEscaped = StringEscapeUtils.escapeHtml3(plainText);
346 if (StringUtil.isAscii(htmlEscaped)) {
349 return StringUtil.escapeUnicodeToPureHtml(plainText);
356 class AppendCallback extends AppendAndIndexWikiCallback<WholeSectionToHtmlParser> {
357 public AppendCallback(WholeSectionToHtmlParser parser) {
362 public void onPlainText(String plainText) {
363 super.onPlainText(escapeHtmlLiteral(plainText));
367 public void onWikiLink(WikiTokenizer wikiTokenizer) {
368 if (wikiTokenizer.wikiLinkText().endsWith(":" + title)) {
369 // Skips wikilinks like: [[en::dick]]
372 if (langConfig.skipWikiLink(wikiTokenizer)) {
376 if (wikiTokenizer.wikiLinkDest() != null) {
377 linkDest = langConfig.adjustWikiLink(wikiTokenizer.wikiLinkDest(), wikiTokenizer.wikiLinkText());
379 linkDest = wikiTokenizer.wikiLinkText();
381 if (sectionEntryTypeName != null) {
382 // TODO: inside a definition, this could be the wrong language.
383 titleIndexBuilder.addEntryWithString(indexedEntry, wikiTokenizer.wikiLinkText(), sectionEntryTypeName);
385 if (!StringUtil.isNullOrEmpty(linkDest)) {
386 builder.append(String.format("<a href=\"%s\">", HtmlEntry.formatQuickdicUrl("", linkDest)));
387 super.onWikiLink(wikiTokenizer);
388 builder.append(String.format("</a>"));
390 super.onWikiLink(wikiTokenizer);
395 public void onFunction(WikiTokenizer wikiTokenizer, String name,
396 List<String> args, Map<String, String> namedArgs) {
397 if (skipLangIso.equalsIgnoreCase(namedArgs.get("lang"))) {
398 namedArgs.remove("lang");
400 super.onFunction(wikiTokenizer, name, args, namedArgs);
404 public void onHtml(WikiTokenizer wikiTokenizer) {
405 super.onHtml(wikiTokenizer);
409 public void onNewline(WikiTokenizer wikiTokenizer) {
412 EntryTypeName sectionEntryTypeName;
413 IndexBuilder currentIndexBuilder;
416 public void onHeading(WikiTokenizer wikiTokenizer) {
417 final String headingText = wikiTokenizer.headingWikiText();
418 sectionEntryTypeName = langConfig.sectionNameToEntryType(headingText);
419 final int depth = wikiTokenizer.headingDepth();
420 if (langConfig.skipSection(headingText)) {
421 //System.out.println("Skipping section:" + headingText);
422 while ((wikiTokenizer = wikiTokenizer.nextToken()) != null) {
423 if (wikiTokenizer.isHeading() && wikiTokenizer.headingDepth() <= depth) {
424 // System.out.println("Resume on: " + wikiTokenizer.token());
425 wikiTokenizer.returnToLineStart();
428 // System.out.println("Skipped: " + wikiTokenizer.token());
433 builder.append(String.format("\n<h%d>", depth));
434 dispatch(headingText, null);
435 builder.append(String.format("</h%d>\n", depth));
438 final List<Character> listPrefixStack = new ArrayList<Character>();
441 public void onListItem(WikiTokenizer wikiTokenizer) {
442 if (builder.length() != 0 && builder.charAt(builder.length() - 1) != '\n') {
443 builder.append("\n");
445 final String prefix = wikiTokenizer.listItemPrefix();
446 while (listPrefixStack.size() < prefix.length()) {
447 builder.append(String.format("<%s>",
448 WikiTokenizer.getListTag(prefix.charAt(listPrefixStack.size()))));
449 listPrefixStack.add(prefix.charAt(listPrefixStack.size()));
451 builder.append("<li>");
452 dispatch(wikiTokenizer.listItemWikiText(), null);
453 builder.append("</li>\n");
455 WikiTokenizer nextToken = wikiTokenizer.nextToken();
456 boolean returnToLineStart = false;
457 if (nextToken != null && nextToken.isNewline()) {
458 nextToken = nextToken.nextToken();
459 returnToLineStart = true;
461 final String nextListHeader;
462 if (nextToken == null || !nextToken.isListItem()) {
465 nextListHeader = nextToken.listItemPrefix();
467 if (returnToLineStart) {
468 wikiTokenizer.returnToLineStart();
470 while (listPrefixStack.size() > nextListHeader.length()) {
471 final char prefixChar = listPrefixStack.remove(listPrefixStack.size() - 1);
472 builder.append(String.format("</%s>\n", WikiTokenizer.getListTag(prefixChar)));
476 boolean boldOn = false;
477 boolean italicOn = false;
480 public void onMarkup(WikiTokenizer wikiTokenizer) {
481 if ("'''".equals(wikiTokenizer.token())) {
483 builder.append("<b>");
485 builder.append("</b>");
488 } else if ("''".equals(wikiTokenizer.token())) {
490 builder.append("<em>");
492 builder.append("</em>");
494 italicOn = !italicOn;