X-Git-Url: http://gitweb.fperrin.net/?a=blobdiff_plain;f=src%2Fcom%2Fhughes%2Fandroid%2Fdictionary%2Fparser%2Fwiktionary%2FWholeSectionToHtmlParser.java;h=7109b9c34b9ebefd199a5a532292919d5b54bda2;hb=a2f20ffce65f9a46036e6b34ea9a788aedc2ff54;hp=e4bbb1634cf2bd2e8258470550eaa01f250ba47f;hpb=7f71228c408684245054f96b90ea5f956784240b;p=DictionaryPC.git diff --git a/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java b/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java index e4bbb16..7109b9c 100644 --- a/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java +++ b/src/com/hughes/android/dictionary/parser/wiktionary/WholeSectionToHtmlParser.java @@ -28,7 +28,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { boolean skipWikiLink(final WikiTokenizer wikiTokenizer); String adjustWikiLink(String wikiLinkDest, final String wikiLinkText); void addFunctionCallbacks( - Map> functionCallbacks); + Map> functionCallbacks); } static final Map isoToLangConfig = new LinkedHashMap(); static { @@ -38,7 +38,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { public boolean skipSection(String headingText) { return enSkipSections.matcher(headingText).matches(); } - + @Override public EntryTypeName sectionNameToEntryType(String sectionName) { if (sectionName.equalsIgnoreCase("Synonyms")) { @@ -56,7 +56,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { } return null; } - + @Override public boolean skipWikiLink(WikiTokenizer wikiTokenizer) { final String wikiText = wikiTokenizer.wikiLinkText(); @@ -82,11 +82,11 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { @Override public void addFunctionCallbacks( - Map> functionCallbacks) { + Map> functionCallbacks) { EnFunctionCallbacks.addGenericCallbacks(functionCallbacks); } }); - + final Pattern esSkipSections = Pattern.compile(".*(Traducciones|Locuciones).*"); isoToLangConfig.put("ES", new LangConfig() { @Override @@ -130,9 +130,56 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { @Override public void addFunctionCallbacks( - Map> functionCallbacks) { + Map> functionCallbacks) { // TODO: need Spanish variant - EnFunctionCallbacks.addGenericCallbacks(functionCallbacks); + } + }); + + final Pattern ptSkipSections = Pattern.compile(".*Tradução.*"); + isoToLangConfig.put("PT", new LangConfig() { + @Override + public boolean skipSection(String headingText) { + return esSkipSections.matcher(headingText).matches(); + } + + @Override + public EntryTypeName sectionNameToEntryType(String sectionName) { + if (sectionName.equalsIgnoreCase("Sinônimo") || sectionName.equalsIgnoreCase("Sinônimos")) { + return EntryTypeName.SYNONYM_MULTI; + } + if (sectionName.equalsIgnoreCase("Antônimo") || sectionName.equalsIgnoreCase("Antônimos")) { + return EntryTypeName.ANTONYM_MULTI; + } + return null; + } + + @Override + public boolean skipWikiLink(WikiTokenizer wikiTokenizer) { + final String wikiText = wikiTokenizer.wikiLinkText(); + if (wikiText.startsWith("Categoría:")) { + return true; + } + return false; + } + @Override + public String adjustWikiLink(String wikiLinkDest, String wikiLinkText) { + if (wikiLinkDest.startsWith("w:") || wikiLinkDest.startsWith("Image:")) { + return null; + } + final int hashPos = wikiLinkDest.indexOf("#"); + if (hashPos != -1) { + wikiLinkDest = wikiLinkDest.substring(0, hashPos); + if (wikiLinkDest.isEmpty()) { + wikiLinkDest = wikiLinkText; + } + } + return wikiLinkDest; + } + + @Override + public void addFunctionCallbacks( + Map> functionCallbacks) { + // TODO: need Portuguese variant } }); @@ -142,7 +189,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { public boolean skipSection(String headingText) { return deSkipSections.matcher(headingText).matches(); } - + @Override public EntryTypeName sectionNameToEntryType(String sectionName) { if (sectionName.equalsIgnoreCase("Synonyme")) { @@ -153,11 +200,11 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { } return null; } - + @Override public boolean skipWikiLink(WikiTokenizer wikiTokenizer) { final String wikiText = wikiTokenizer.wikiLinkText(); - if (wikiText.startsWith("???Category:")) { + if (wikiText.startsWith("Kategorie:")) { return true; } return false; @@ -179,18 +226,18 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { @Override public void addFunctionCallbacks( - Map> functionCallbacks) { + Map> functionCallbacks) { DeFunctionCallbacks.addGenericCallbacks(functionCallbacks); } }); - + final Pattern itSkipSections = Pattern.compile(".*(Traduzione|Note / Riferimenti).*"); isoToLangConfig.put("IT", new LangConfig() { @Override public boolean skipSection(String headingText) { return itSkipSections.matcher(headingText).matches(); } - + @Override public EntryTypeName sectionNameToEntryType(String sectionName) { if (sectionName.equalsIgnoreCase("Sinonimi")) { @@ -201,11 +248,11 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { } return null; } - + @Override public boolean skipWikiLink(WikiTokenizer wikiTokenizer) { final String wikiText = wikiTokenizer.wikiLinkText(); - if (wikiText.startsWith("???Category:")) { + if (wikiText.startsWith("Categoria:")) { return true; } return false; @@ -227,29 +274,36 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { @Override public void addFunctionCallbacks( - Map> functionCallbacks) { + Map> functionCallbacks) { ItFunctionCallbacks.addGenericCallbacks(functionCallbacks); } }); - final Pattern frSkipSections = Pattern.compile(".*(Traductions|[Aa]nagrammes).*"); + final Pattern frSkipSections = Pattern.compile(".*([Tt]raductions|[Aa]nagrammes).*"); isoToLangConfig.put("FR", new LangConfig() { @Override public boolean skipSection(String headingText) { return frSkipSections.matcher(headingText).matches(); } - + @Override public EntryTypeName sectionNameToEntryType(String sectionName) { if (sectionName.equalsIgnoreCase("Synonymes")) { return EntryTypeName.SYNONYM_MULTI; } + if (sectionName.equalsIgnoreCase("Antonymes")) { + return EntryTypeName.ANTONYM_MULTI; + } return null; } - + @Override public boolean skipWikiLink(WikiTokenizer wikiTokenizer) { + final String wikiText = wikiTokenizer.wikiLinkText(); + if (wikiText.startsWith("Catégorie:")) { + return true; + } return false; } @Override @@ -269,7 +323,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { @Override public void addFunctionCallbacks( - Map> functionCallbacks) { + Map> functionCallbacks) { FrFunctionCallbacks.addGenericCallbacks(functionCallbacks); } }); @@ -280,10 +334,10 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { final String skipLangIso; final LangConfig langConfig; final String webUrlTemplate; - + public WholeSectionToHtmlParser(final IndexBuilder titleIndexBuilder, final IndexBuilder defIndexBuilder, final String wiktionaryIso, final String skipLangIso, - final String webUrlTemplate) { + final String webUrlTemplate) { this.titleIndexBuilder = titleIndexBuilder; this.defIndexBuilder = defIndexBuilder; assert isoToLangConfig.containsKey(wiktionaryIso): wiktionaryIso; @@ -291,7 +345,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { this.skipLangIso = skipLangIso; this.webUrlTemplate = webUrlTemplate; } - + IndexedEntry indexedEntry = null; @Override @@ -301,7 +355,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { indexedEntry = new IndexedEntry(htmlEntry); final AppendAndIndexWikiCallback callback = new AppendCallback( - this); + this); langConfig.addFunctionCallbacks(callback.functionCallbacks); callback.builder = new StringBuilder(); @@ -310,11 +364,11 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { if (webUrlTemplate != null) { final String webUrl = String.format(webUrlTemplate, title); - // URI.create can raise an exception e.g. if webUrl contains %, just ignore those cases. - try { - callback.builder.append(String.format("

%s", URI.create(webUrl).toString(), escapeHtmlLiteral(webUrl))); - } catch (Exception e) - {} + // URI.create can raise an exception e.g. if webUrl contains %, just ignore those cases. + try { + callback.builder.append(String.format("

%s", URI.create(webUrl).toASCIIString(), escapeHtmlLiteral(webUrl))); + } catch (Exception e) { + } } htmlEntry.html = callback.builder.toString(); indexedEntry.isValid = true; @@ -326,26 +380,26 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { tokenData.htmlEntries.add(htmlEntry); // titleIndexBuilder.addEntryWithString(indexedEntry, title, // EntryTypeName.WIKTIONARY_TITLE_MULTI_DETAIL); - + indexedEntry = null; } @Override void removeUselessArgs(Map namedArgs) { } - + @Override public void addLinkToCurrentEntry(String token, final String lang, EntryTypeName entryTypeName) { if (lang == null || lang.equals(skipLangIso)) { titleIndexBuilder.addEntryWithString(indexedEntry, token, entryTypeName); } } - + public static String escapeHtmlLiteral(final String plainText) { final String htmlEscaped = StringEscapeUtils.escapeHtml3(plainText); if (StringUtil.isAscii(htmlEscaped)) { return htmlEscaped; - } else { + } else { return StringUtil.escapeUnicodeToPureHtml(plainText); } @@ -393,7 +447,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { @Override public void onFunction(WikiTokenizer wikiTokenizer, String name, - List args, Map namedArgs) { + List args, Map namedArgs) { if (skipLangIso.equalsIgnoreCase(namedArgs.get("lang"))) { namedArgs.remove("lang"); } @@ -408,7 +462,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { @Override public void onNewline(WikiTokenizer wikiTokenizer) { } - + EntryTypeName sectionEntryTypeName; IndexBuilder currentIndexBuilder; @@ -445,7 +499,7 @@ public class WholeSectionToHtmlParser extends AbstractWiktionaryParser { final String prefix = wikiTokenizer.listItemPrefix(); while (listPrefixStack.size() < prefix.length()) { builder.append(String.format("<%s>", - WikiTokenizer.getListTag(prefix.charAt(listPrefixStack.size())))); + WikiTokenizer.getListTag(prefix.charAt(listPrefixStack.size())))); listPrefixStack.add(prefix.charAt(listPrefixStack.size())); } builder.append("

  • ");