private void recordSurrogateError(StringBuilder inputData, int index, char codePoint)
throws SnuggleParseException {
sessionContext.registerError(
new InputError(
CoreErrorCode.TTEG05, null, Integer.toHexString(codePoint), Integer.valueOf(index)));
inputData.setCharAt(index, ' ');
}
private int[] calculateNewlineIndicesAndCheckCodePoints(StringBuilder inputData)
throws SnuggleParseException {
List<Integer> newlineIndicesBuilder = new ArrayList<Integer>();
newlineIndicesBuilder.add(Integer.valueOf(-1));
char lastChar = 0;
char thisChar; /* (16 bit char only) */
int codePoint; /* (Full Unicode code point */
for (int i = 0, length = inputData.length(); i < length; i++, lastChar = thisChar) {
thisChar = inputData.charAt(i);
if (thisChar == '\n') {
newlineIndicesBuilder.add(Integer.valueOf(i));
}
if (Character.isHighSurrogate(lastChar)) {
if (Character.isLowSurrogate(thisChar)) {
codePoint = Character.toCodePoint(lastChar, thisChar);
} else {
/* Error: last was bad surrogate character */
recordSurrogateError(inputData, i - 1, lastChar);
continue;
}
} else if (Character.isLowSurrogate(thisChar)) {
/* Error: this is bad surrogate character */
recordSurrogateError(inputData, i, thisChar);
continue;
} else {
codePoint = thisChar;
}
/* Check that we allow this codepoint */
if (Character.isISOControl(codePoint)
&& !(codePoint == '\r' || codePoint == '\n' || codePoint == '\t')) {
sessionContext.registerError(
new InputError(
CoreErrorCode.TTEG02, null, Integer.toHexString(codePoint), Integer.valueOf(i)));
inputData.setCharAt(i, ' ');
}
}
/* Make sure last character wasn't surrogate pair starter */
if (Character.isHighSurrogate(lastChar)) {
recordSurrogateError(inputData, inputData.length() - 1, lastChar);
}
/* Finally store newline information */
int[] calculatedNewlineIndices = new int[newlineIndicesBuilder.size()];
for (int i = 0; i < calculatedNewlineIndices.length; i++) {
calculatedNewlineIndices[i] = newlineIndicesBuilder.get(i);
}
return calculatedNewlineIndices;
}
