/** gets map for all languages */
public Map<String, String> getCaptionToQueryMap(Collection<Document> docs) {
// identify all the langs in the docs, and the corresponding lexicons
Set<String> languages = IndexUtils.allLanguagesInDocs(docs);
Set<Lexicon1Lang> lexicons = new LinkedHashSet<Lexicon1Lang>();
for (String lang : languages) {
Lexicon1Lang lex = languageToLexicon.get(lang);
if (lex != null) lexicons.add(lex);
// this lexicon doesn't know about this language
else log.warn("Warning: no support for " + lang + " in lexicon " + name);
}
Map<String, String> result = new LinkedHashMap<String, String>();
// aggregate results for each lang into result
for (Lexicon1Lang lex : lexicons) {
Map<String, String> resultsForThisLang = lex.captionToExpandedQuery;
for (String caption : resultsForThisLang.keySet()) {
String queryThisLang = resultsForThisLang.get(caption);
String query = result.get(caption);
// if caption doesn't exist already, create a new entry, or else add to the existing set of
// docs that match this caption
if (query == null) result.put(caption, queryThisLang);
else result.put(caption, query + "|" + queryThisLang);
}
}
return result;
}
// identify all the langs in the docs, and the corresponding lexicons
private Collection<Lexicon1Lang> getRelevantLexicon1Langs(Collection<Document> docs) {
if (docs == null) return languageToLexicon.values(); // just return all lexicons
Set<String> languages = IndexUtils.allLanguagesInDocs(docs);
Set<Lexicon1Lang> lexicons = new LinkedHashSet<Lexicon1Lang>();
for (String lang : languages) {
Lexicon1Lang lex = languageToLexicon.get(lang);
if (lex != null) lexicons.add(lex); // this lexicon doesn't know about this language
else log.warn("Warning: no support for " + lang + " in lexicon " + name);
}
return lexicons;
}
