public void optimizeWords() {
String correctedWord;
java.util.List<String> v = new ArrayList<String>(words.keySet());
MsgTextPane.write("Applying stem correction to Dictionary...");
matchInfo = false;
Collections.sort(v);
int success = 0;
int failed = 0;
for (String str : v) {
// str is lowercase and deturkified because it is in keySet
correctedWord = correctWordByRules(str);
if (!correctedWord.equals(words.get(str))) {
failed++;
} else {
if (success < 100) {
dictFrame.writeDictArea("Redundant word removed : " + correctedWord + "\n", false);
} else if (success == 100) {
dictFrame.writeDictArea("...", false);
}
words.remove(str);
success++;
}
}
MsgTextPane.write("Purge on dictionary : " + success + " words removed\n");
matchInfo = true;
}
private void printEntryList(HashSet set, HashMap map) {
Iterator iter = set.iterator();
while (iter.hasNext()) {
String key = (String) iter.next();
String entry = (String) map.get(key);
if (entry != null) {
dictFrame.writeDictArea(key + " >> " + entry + "\n", false);
} else {
dictFrame.writeDictArea(key + "\n", false);
}
}
}
public void removeStem(String word) {
String key = language.removeDiacritics(word);
stems.remove(key);
addedstems.remove(key);
removedstems.add(key);
dictFrame.writeDictArea("Stem removed: ", false);
dictFrame.writeSelectDictArea(word);
dictFrame.writeDictArea("\n", false);
dictFrame.scrollEnd();
dictFrame.isModified(true);
}
public void removeWord(String word) {
String key = language.removeDiacritics(word);
words.remove(key);
addedwords.remove(key);
removedwords.add(key);
dictFrame.writeDictArea("Word removed: ", false);
dictFrame.writeDictArea(word, false);
dictFrame.writeDictArea("\n", false);
dictFrame.scrollEnd();
dictFrame.isModified(true);
}
public String correctWordByDictionary(String word) {
if (words.containsKey(word)) {
String correctedWord = words.get(word);
if (matchInfo) {
dictFrame.writeDictArea(word + " >> ", false);
dictFrame.writeSelectDictArea(correctedWord);
dictFrame.writeDictArea("\n", false);
dictFrame.scrollEnd();
}
return correctedWord;
} else {
return word;
}
}
public void addStem(String word) {
String key = language.removeDiacritics(word);
if (validCharacters(key)) {
stems.put(key, word);
removedstems.remove(key);
addedstems.add(key);
dictFrame.writeDictArea("Stem added: ", false);
dictFrame.writeSelectDictArea(word);
dictFrame.writeDictArea("\n", false);
dictFrame.scrollEnd();
dictFrame.isModified(true);
}
}
public void optimizeStems() {
WordTree w = new WordTree(language);
for (String key : words.keySet()) {
String word = words.get(key);
w.addWord(word, 1);
}
stems.clear();
ArrayList<String> stemList;
stemList = w.scanStems();
for (String stem : stemList) {
stems.put(language.removeDiacritics(stem), stem);
}
MsgTextPane.write(stemList.size() + " stems extracted");
if (true) {
String correctedWord;
String correctStem;
java.util.List<String> v = new ArrayList<String>(stems.keySet());
// MsgTextPane.write("Applying stem reduction to Dictionary...");
matchInfo = false;
Collections.sort(v);
int success = 0;
int failed = 0;
for (String str : v) {
correctStem = stems.get(str);
stems.remove(str);
correctedWord = correctWordByRules(str);
// no dictionary lookup because otherwise if stem happens to be in words it is removed
if (correctedWord.equals(correctStem)) {
if (success < 100) {
dictFrame.writeDictArea("Redundant stem removed : " + correctedWord + "\n", false);
} else if (success == 100) {
dictFrame.writeDictArea("...", false);
}
success++;
} else { // put it back
stems.put(str, correctStem);
failed++;
}
}
MsgTextPane.write(success + " redundant stems removed\n");
matchInfo = true;
}
}
public boolean saveDictionary(String fileName) {
boolean confirm;
ConfirmDialog cd = new ConfirmDialog();
cd.popUp(
null,
words.size() + "/" + stems.size() + " entries will be written to " + fileName,
"Continue",
"Cancel");
confirm = cd.confirm;
if (confirm) {
try {
File initialFile = new File(fileName);
OutputStream is = new FileOutputStream(initialFile);
OutputStreamWriter isr = new OutputStreamWriter(is, "UTF-8");
MsgTextPane.write("write encoding = " + isr.getEncoding());
BufferedWriter outputStream = new BufferedWriter(isr);
java.util.List<String> v;
v = new ArrayList<String>(words.keySet());
for (String str : v) {
outputStream.write((String) words.get(str));
outputStream.newLine();
}
v = new ArrayList<String>(stems.keySet());
for (String str : v) {
outputStream.write("[" + (String) stems.get(str) + "]");
outputStream.newLine();
}
outputStream.close();
addedwords.clear();
addedstems.clear();
removedwords.clear();
removedstems.clear();
dictFrame.isModified(false);
return true;
} catch (IOException io) {
MsgTextPane.write(" io exception during save dictionary");
return false;
}
} else {
return false;
}
}
public void printModifications() {
dictFrame.writeDictArea("\nDICTIONARY MODIFICATIONS\n", false);
if (!addedwords.isEmpty()) {
dictFrame.writeDictArea("WORDS ADDED\n", false);
printEntryList(addedwords, words);
}
if (!removedwords.isEmpty()) {
dictFrame.writeDictArea("WORDS REMOVED\n", false);
printEntryList(removedwords, words);
}
if (!addedstems.isEmpty()) {
dictFrame.writeDictArea("STEMS ADDED\n", false);
printEntryList(addedstems, stems);
}
if (!removedstems.isEmpty()) {
dictFrame.writeDictArea("STEMS REMOVED\n", false);
printEntryList(removedstems, stems);
}
dictFrame.scrollEnd();
}
public void printAll(String dictionaryPattern) {
Pattern pattern;
Matcher matcher;
java.util.List<String> v;
int count;
if (dictionaryPattern.equals("")) {
dictionaryPattern = "^.*$";
}
pattern = Pattern.compile(dictionaryPattern);
v = new ArrayList<>(words.keySet());
Collections.sort(v);
dictFrame.writeDictArea("Searching Dictionary..." + "\n", true);
count = 1;
for (String str : v) {
matcher = pattern.matcher((String) str); // search the keys
Boolean found = false;
if (matcher.find()) {
dictFrame.writeDictArea(count + " " + (String) words.get(str) + "\n", false);
count++;
}
}
dictFrame.writeDictArea(count - 1 + " entries in dictionary" + "\n", true);
v = new ArrayList<>(stems.keySet());
Collections.sort(v);
dictFrame.writeDictArea("Searching stems..." + "\n", true);
count = 1;
for (String str : v) {
matcher = pattern.matcher((String) str); // search the keys
Boolean found = false;
if (matcher.find()) {
dictFrame.writeDictArea(count + " " + str + " " + (String) stems.get(str) + "\n", false);
count++;
}
}
dictFrame.writeDictArea(count - 1 + " stems" + "\n", true);
}
public void dictionaryWindowVisible(boolean b) {
dictFrame.setVisible(b);
}
public GenericDictionary(Language l) {
language = l;
dictFrame = new DictionaryFrame(language);
dictFrame.setVisible(false);
}
