import com.hughes.android.dictionary.engine.IndexedEntry;
import com.hughes.android.dictionary.parser.WikiTokenizer;
import com.hughes.util.StringUtil;
-import com.sun.xml.internal.rngom.util.Uri;
import org.apache.commons.lang3.StringEscapeUtils;
+import java.net.URI;
import java.util.ArrayList;
import java.util.LinkedHashMap;
import java.util.List;
}
});
+ final Pattern esSkipSections = Pattern.compile(".*(Traducciones|Locuciones).*");
+ isoToLangConfig.put("ES", new LangConfig() {
+ @Override
+ public boolean skipSection(String headingText) {
+ return esSkipSections.matcher(headingText).matches();
+ }
+
+ @Override
+ public EntryTypeName sectionNameToEntryType(String sectionName) {
+ if (sectionName.equalsIgnoreCase("sinónimo") || sectionName.equalsIgnoreCase("sinónimos")) {
+ return EntryTypeName.SYNONYM_MULTI;
+ }
+ if (sectionName.equalsIgnoreCase("antónimo") || sectionName.equalsIgnoreCase("antónimos")) {
+ 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<String, FunctionCallback<WholeSectionToHtmlParser>> functionCallbacks) {
+ // TODO: need Spanish variant
+ }
+ });
+
final Pattern deSkipSections = Pattern.compile(".*(Übersetzungen|Referenzen|Quellen).*");
isoToLangConfig.put("DE", new LangConfig() {
@Override
@Override
public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
final String wikiText = wikiTokenizer.wikiLinkText();
- if (wikiText.startsWith("???Category:")) {
+ if (wikiText.startsWith("Kategorie:")) {
return true;
}
return false;
@Override
public boolean skipWikiLink(WikiTokenizer wikiTokenizer) {
final String wikiText = wikiTokenizer.wikiLinkText();
- if (wikiText.startsWith("???Category:")) {
+ if (wikiText.startsWith("Categoria:")) {
return true;
}
return false;
});
- final Pattern frSkipSections = Pattern.compile(".*(Traductions).*");
+ final Pattern frSkipSections = Pattern.compile(".*(Traductions|[Aa]nagrammes).*");
isoToLangConfig.put("FR", new LangConfig() {
@Override
public boolean skipSection(String headingText) {
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
if (webUrlTemplate != null) {
final String webUrl = String.format(webUrlTemplate, title);
- callback.builder.append(String.format("<p> <a href=\"%s\">%s</a>", Uri.escapeDisallowedChars(webUrl), escapeHtmlLiteral(webUrl)));
+ // URI.create can raise an exception e.g. if webUrl contains %, just ignore those cases.
+ try {
+ callback.builder.append(String.format("<p> <a href=\"%s\">%s</a>", URI.create(webUrl).toString(), escapeHtmlLiteral(webUrl)));
+ } catch (Exception e)
+ {}
}
htmlEntry.html = callback.builder.toString();
indexedEntry.isValid = true;
sectionEntryTypeName = langConfig.sectionNameToEntryType(headingText);
final int depth = wikiTokenizer.headingDepth();
if (langConfig.skipSection(headingText)) {
- System.out.println("Skipping section:" + headingText);
+ //System.out.println("Skipping section:" + headingText);
while ((wikiTokenizer = wikiTokenizer.nextToken()) != null) {
if (wikiTokenizer.isHeading() && wikiTokenizer.headingDepth() <= depth) {
- System.out.println("Resume on: " + wikiTokenizer.token());
+ // System.out.println("Resume on: " + wikiTokenizer.token());
wikiTokenizer.returnToLineStart();
return;
} else {
- System.out.println("Skipped: " + wikiTokenizer.token());
+ // System.out.println("Skipped: " + wikiTokenizer.token());
}
}
return;