protected static void convertAbbreviations(
BaseReader reader, BaseWriter writer, Observer observer, ProgressInfo progress)
throws Exception {
Set<String> keys = reader.getAbbreviationKeys();
if (keys == null) {
log.info("No abbreviations available");
return;
}
log.info("Number of abbreviations: {}", keys.size());
// Update progress
observer.update(progress.setMessage("Sorting Abbreviations"), null);
ArrayList<String> sortedKeys = new ArrayList<String>(keys.size());
for (String key : keys) {
sortedKeys.add(key);
}
// Create Collator and sort abbreviations
Collator collator = createCollator(writer);
if (collator != null) {
Collections.sort(sortedKeys, collator);
} else {
log.warn("Collator couldn't be created, sorting of Abbreviations is skipped");
}
// Update progress
observer.update(progress.setMessage("Converting Abbreviations"), null);
for (Iterator<String> iterator = sortedKeys.iterator(); iterator.hasNext(); ) {
String abbr = iterator.next();
AbbreviationInfo abbrInfo = reader.getAbbreviationInfo(abbr);
String definition = abbrInfo.getAbbreviation();
if (abbr == null
|| definition == null
|| abbr.trim().equals("")
|| definition.trim().equals("")) {
log.info("Abbreviation {} is excluded", abbrInfo);
continue;
}
abbrInfo.setAbbreviation(abbrInfo.getAbbreviation().trim());
abbrInfo.setDefinition(abbrInfo.getDefinition().trim());
writer.saveAbbreviationInfo(reader.getAbbreviationInfo(abbr));
}
writer.flush();
}
protected static void convertMediaResources(
BaseReader reader, BaseWriter writer, Observer observer, ProgressInfo progress)
throws Exception {
Set<String> keys = reader.getMediaResourceKeys();
if (keys == null) {
log.info("No media resources available");
return;
}
log.info("Number of media resources: {}", keys.size());
// Update progress
observer.update(progress.setMessage("Sorting Media Resources"), null);
ArrayList<String> sortedKeys = new ArrayList<String>(keys.size());
for (String key : keys) {
sortedKeys.add(key);
}
// Create Collator and sort articles
Collator collator = createCollator(writer);
if (collator != null) {
Collections.sort(sortedKeys, collator);
} else {
log.warn("Collator couldn't be created, sorting of Media Resources is skipped");
}
// Update progress
observer.update(progress.setMessage("Converting Media Resources"), null);
for (Iterator<String> iterator = sortedKeys.iterator(); iterator.hasNext(); ) {
String sortedKey = iterator.next();
MediaResourceInfo mediaInfo = reader.getMediaResourceInfo(new MediaResourceKey(sortedKey));
mediaInfo.getKey().setResourceKey(mediaInfo.getKey().getResourceKey());
writer.saveMediaResourceInfo(mediaInfo);
}
writer.flush();
}
protected static void convertArticles(
BaseReader reader, BaseWriter writer, Observer observer, ProgressInfo progress)
throws Exception {
log.info("Started converting articles: Size {}", reader.getWords().size());
// Update progress
observer.update(progress.setMessage("Mapping and Sorting Articles"), null);
// Clone words to speed up the process (it can be based on buffered list)
List<String> srcWords = new ArrayList<String>(reader.getWords());
// Mappings and redirects
Map<Integer, String> srcMappings = reader.getAdaptedWordsMappings();
Map<Integer, Integer> srcRedirects = reader.getWordsRedirects();
// Create Collator and sort articles
Collator collator = createCollator(writer);
// Create and initialize words mapper
ConversionWordsMapper wordsMapper =
new ConversionWordsMapper(collator, srcWords, srcMappings, srcRedirects);
wordsMapper.init();
List<String> normWords = wordsMapper.getNormalizedWords();
// Update progress
observer.update(progress.setMessage("Converting Articles"), null);
for (int i = 0; i < normWords.size(); i++) {
String curWord = normWords.get(i);
int oldWordId = wordsMapper.getOldWordId(curWord);
// Word info to retrieve articles. Can't be moved outside because
// it' overridden every time article is retrieved.
WordInfo oldWordInfo = new WordInfo(oldWordId);
oldWordInfo.setId(oldWordId);
ArticleInfo inArticleInfo = reader.getAdaptedArticleInfo(oldWordInfo);
if (inArticleInfo == null || inArticleInfo.getArticle() == null) {
log.warn("Couldn't find article for {}", oldWordInfo);
throw new IllegalStateException("Couldn't find article for " + oldWordInfo);
}
WordInfo newWordInfo = new WordInfo(i, curWord);
// Add mapping if any
String newMapping = wordsMapper.getNewWordMappingByOldWordId(oldWordId);
if (newMapping != null) {
newWordInfo.setWordMapping(newMapping);
}
// Add redirect if any
int newRedirectToId = wordsMapper.getNewWordRedirect(oldWordId);
if (newRedirectToId >= 0 && newRedirectToId != i) { // Don't allow redirects to itself
newWordInfo.setRedirectToId(newRedirectToId);
}
ArticleInfo outArticleInfo = new ArticleInfo(newWordInfo, inArticleInfo.getArticle());
writer.saveAdaptedArticleInfo(outArticleInfo);
// if (i > 10000) break;
}
writer.flush();
}
