/**
* Core sentiment detection method. doNota = none of the above
*
* @param captions (null/none = all)
*/
public Map<String, Collection<Document>> getEmotions(
Indexer indexer,
Collection<Document> docs,
boolean doNota,
boolean originalContentOnly,
String... captions) {
Collection<Lexicon1Lang> lexicons = getRelevantLexicon1Langs(docs);
Map<String, Collection<Document>> result = new LinkedHashMap<>();
Set<Document> docs_set = Util.castOrCloneAsSet(docs);
// aggregate results for each lang into result
for (Lexicon1Lang lex : lexicons) {
Map<String, Collection<Document>> resultsForThisLang =
(doNota
? lex.getEmotionsWithNOTA(indexer, docs_set, originalContentOnly)
: lex.getEmotions(indexer, docs_set, originalContentOnly, captions));
if (resultsForThisLang == null) continue;
for (String caption : resultsForThisLang.keySet()) {
Collection<Document> resultDocsThisLang = resultsForThisLang.get(caption);
Collection<Document> resultDocs = 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 (resultDocs == null) result.put(caption, resultDocsThisLang);
else resultDocs.addAll(resultDocsThisLang);
}
}
// TODO: the result can be cached at server to avoid redundant computation (by concurrent users,
// which are few for now)
return result;
}
/**
* updates the map for a given language
*
* @throws IOException
* @throws FileNotFoundException
*/
public boolean update(String language, Map<String, String> map) throws IOException {
language = language.toLowerCase();
Lexicon1Lang langLex = languageToLexicon.get(language);
if (langLex == null) {
langLex = new Lexicon1Lang();
languageToLexicon.put(language, langLex);
}
langLex.setRawQueryMap(map);
return true;
}
/**
* returns whether it succeeded
*
* @throws Exception
*/
public boolean save(String dir, String language) throws Exception {
language = language.toLowerCase();
Lexicon1Lang langLex = languageToLexicon.get(language);
if (langLex == null) return false;
langLex.save(
dir
+ File.separator
+ name
+ "."
+ language
+ LEXICON_SUFFIX); // LEXICON_SUFFIX already has a .
return true;
}
// accumulates counts returned by lexicons in each language
// TODO: It is possible to write a generic accumulator that accumulates sum over all the languages
public Map<String, Integer> getLexiconCounts(Indexer indexer, boolean originalContentOnly) {
List<Document> docs = indexer.docs;
Collection<Lexicon1Lang> lexicons = getRelevantLexicon1Langs(docs);
Map<String, Integer> result = new LinkedHashMap<String, Integer>();
Set<Document> docs_set = Util.castOrCloneAsSet(docs);
// aggregate results for each lang into result
for (Lexicon1Lang lex : lexicons) {
Map<String, Integer> resultsForThisLang = lex.getLexiconCounts(indexer, originalContentOnly);
if (resultsForThisLang == null) continue;
for (String caption : resultsForThisLang.keySet()) {
Integer resultCountsThisLang = resultsForThisLang.get(caption);
Integer resultCounts = 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 (resultCounts == null) result.put(caption, resultCountsThisLang);
else result.put(caption, resultCounts + resultCountsThisLang);
}
}
return result;
}