// ---------------------------------------------------------------------------------
//
// scanSet Construct a UnicodeSet from the text at the current scan
// position. Advance the scan position to the first character
// after the set.
//
// A new RBBI setref node referring to the set is pushed onto the node
// stack.
//
// The scan position is normally under the control of the state machine
// that controls rule parsing. UnicodeSets, however, are parsed by
// the UnicodeSet constructor, not by the RBBI rule parser.
//
// ---------------------------------------------------------------------------------
internal void ScanSet() {
UnicodeSet uset = null;
int startPos;
ILOG.J2CsMapping.Text.ParsePosition pos = new ILOG.J2CsMapping.Text.ParsePosition(fScanIndex);
int i;
startPos = fScanIndex;
try {
uset = new UnicodeSet(fRB.fRules, pos, fSymbolTable,
IBM.ICU.Text.UnicodeSet.IGNORE_SPACE);
} catch (Exception e) { // TODO: catch fewer exception types.
// Repackage UnicodeSet errors as RBBI rule builder errors, with
// location info.
Error(IBM.ICU.Text.RBBIRuleBuilder.U_BRK_MALFORMED_SET);
}
// Verify that the set contains at least one code point.
//
if (uset.IsEmpty()) {
// This set is empty.
// Make it an error, because it almost certainly is not what the
// user wanted.
// Also, avoids having to think about corner cases in the tree
// manipulation code
// that occurs later on.
// TODO: this shouldn't be an error; it does happen.
Error(IBM.ICU.Text.RBBIRuleBuilder.U_BRK_RULE_EMPTY_SET);
}
// Advance the RBBI parse postion over the UnicodeSet pattern.
// Don't just set fScanIndex because the line/char positions maintained
// for error reporting would be thrown off.
i = pos.GetIndex();
for (;;) {
if (fNextIndex >= i) {
break;
}
NextCharLL();
}
RBBINode n;
n = PushNewNode(IBM.ICU.Text.RBBINode.setRef);
n.fFirstPos = startPos;
n.fLastPos = fNextIndex;
n.fText = fRB.fRules.Substring(n.fFirstPos,(n.fLastPos)-(n.fFirstPos));
// findSetFor() serves several purposes here:
// - Adopts storage for the UnicodeSet, will be responsible for
// deleting.
// - Mantains collection of all sets in use, needed later for
// establishing
// character categories for run time engine.
// - Eliminates mulitiple instances of the same set.
// - Creates a new uset node if necessary (if this isn't a duplicate.)
FindSetFor(n.fText, n, uset);
}
/// <exclude/>
/// <summary>
/// Return the set of all characters that may be modified by this
/// Transliterator, ignoring the effect of our filter.
/// </summary>
///
protected internal override UnicodeSet HandleGetSourceSet()
{
UnicodeSet set = new UnicodeSet();
for (int i = 0; i < trans.Length; ++i)
{
set.AddAll(trans[i].GetSourceSet());
// Take the example of Hiragana-Latin. This is really
// Hiragana-Katakana; Katakana-Latin. The source set of
// these two is roughly [:Hiragana:] and [:Katakana:].
// But the source set for the entire transliterator is
// actually [:Hiragana:] ONLY -- that is, the first
// non-empty source set.
// This is a heuristic, and not 100% reliable.
if (!set.IsEmpty())
{
break;
}
}
return(set);
}