da5b4f17b7
* new dictionary format that supports syllabaries * optimized the dictionary build cache significantly to truly build only the changed language files * code style fixes
65 lines
2.5 KiB
Groovy
65 lines
2.5 KiB
Groovy
class Wrapper {
|
|
static def getDictionaryLineData(String line, String delimiter) {
|
|
String[] parts = line.split(delimiter, 2)
|
|
String word = parts[0]
|
|
String transcription = parts.length > 1 && parts[1] =~ "^[a-zA-Z]+\$" ? parts[1] : ""
|
|
|
|
int frequency
|
|
try {
|
|
int partsElement = transcription.isEmpty() ? 1 : 2
|
|
frequency = (parts.length > partsElement ? parts[partsElement] : "0") as int
|
|
} catch (Exception ignored) {
|
|
frequency = -1
|
|
}
|
|
|
|
return [word, transcription, frequency]
|
|
}
|
|
|
|
|
|
static def wordToDigitSequence(Locale locale, String word, HashMap<String, String> sounds, boolean isTranscribed) {
|
|
String sequence = ""
|
|
|
|
final String normalizedWord = isTranscribed ? word : word.toUpperCase(locale)
|
|
String currentSound = ""
|
|
|
|
for (int i = 0, end = normalizedWord.length() - 1; i <= end; i++) {
|
|
char currentChar = normalizedWord.charAt(i)
|
|
char nextChar = i < end ? normalizedWord.charAt(i + 1) : 0
|
|
int nextCharType = Character.getType(nextChar)
|
|
|
|
currentSound += currentChar
|
|
|
|
// charAt(i) returns "ΐ" as three separate characters, but they must be treated as one.
|
|
if (
|
|
locale.getLanguage() == "el"
|
|
&& (nextCharType == Character.NON_SPACING_MARK || nextCharType == Character.ENCLOSING_MARK || nextCharType == Character.COMBINING_SPACING_MARK)
|
|
) {
|
|
continue
|
|
}
|
|
|
|
if (!isTranscribed || i == end || Character.isUpperCase(nextChar)) {
|
|
if (!sounds.containsKey(currentSound)) {
|
|
throw new IllegalArgumentException("Sound or layout entry '${currentSound}' does not belong to the language sound list: ${sounds}.")
|
|
} else {
|
|
sequence += sounds.get(currentSound)
|
|
currentSound = ""
|
|
}
|
|
}
|
|
}
|
|
|
|
if (sequence.isEmpty()) {
|
|
throw new IllegalArgumentException("The word does not contain any valid sounds.")
|
|
}
|
|
|
|
return sequence
|
|
}
|
|
|
|
|
|
static def getLanguageHash(File definitionFile, File dictionaryFile) {
|
|
def definitionHash = definitionFile != null && definitionFile.exists() ? definitionFile.text.digest("SHA-256") : ""
|
|
def dictionaryHash = dictionaryFile != null && dictionaryFile.exists() ? dictionaryFile.text.digest("SHA-256") : ""
|
|
return definitionHash + dictionaryHash
|
|
}
|
|
}
|
|
|
|
ext.DictionaryTools = Wrapper
|