// ----------------------------------------------------------------------------------------
//
// Constructor.
//
// ----------------------------------------------------------------------------------------
internal RBBIRuleScanner(RBBIRuleBuilder rb) {
this.kStackSize = 100;
this.fC = new RBBIRuleScanner.RBBIRuleChar ();
this.fStack = new short[kStackSize];
this.fNodeStack = new RBBINode[kStackSize];
this.fSetTable = new Hashtable();
this.fRuleSets = new UnicodeSet[10];
fRB = rb;
fLineNum = 1;
//
// Set up the constant Unicode Sets.
// Note: These could be made static and shared among
// all instances of RBBIRuleScanners.
fRuleSets[IBM.ICU.Text.RBBIRuleParseTable.kRuleSet_rule_char - 128] = new UnicodeSet(
gRuleSet_rule_char_pattern);
fRuleSets[IBM.ICU.Text.RBBIRuleParseTable.kRuleSet_white_space - 128] = new UnicodeSet(
gRuleSet_white_space_pattern);
fRuleSets[IBM.ICU.Text.RBBIRuleParseTable.kRuleSet_name_char - 128] = new UnicodeSet(
gRuleSet_name_char_pattern);
fRuleSets[IBM.ICU.Text.RBBIRuleParseTable.kRuleSet_name_start_char - 128] = new UnicodeSet(
gRuleSet_name_start_char_pattern);
fRuleSets[IBM.ICU.Text.RBBIRuleParseTable.kRuleSet_digit_char - 128] = new UnicodeSet(
gRuleSet_digit_char_pattern);
fSymbolTable = new RBBISymbolTable(this, rb.fRules);
}
// ---------------------------------------------------------------------------------
//
// nextChar for rules scanning. At this level, we handle stripping
// out comments and processing backslash character escapes.
// The rest of the rules grammar is handled at the next level up.
//
// ---------------------------------------------------------------------------------
internal void NextChar(RBBIRuleScanner.RBBIRuleChar c) {
// Unicode Character constants needed for the processing done by
// nextChar(),
// in hex because literals wont work on EBCDIC machines.
fScanIndex = fNextIndex;
c.fChar = NextCharLL();
c.fEscaped = false;
//
// check for '' sequence.
// These are recognized in all contexts, whether in quoted text or not.
//
if (c.fChar == '\'') {
if (IBM.ICU.Text.UTF16.CharAt(fRB.fRules, fNextIndex) == '\'') {
c.fChar = NextCharLL(); // get nextChar officially so character
// counts
c.fEscaped = true; // stay correct.
} else {
// Single quote, by itself.
// Toggle quoting mode.
// Return either '(' or ')', because quotes cause a grouping of
// the quoted text.
fQuoteMode = !fQuoteMode;
if (fQuoteMode == true) {
c.fChar = '(';
} else {
c.fChar = ')';
}
c.fEscaped = false; // The paren that we return is not escaped.
return;
}
}
if (fQuoteMode) {
c.fEscaped = true;
} else {
// We are not in a 'quoted region' of the source.
//
if (c.fChar == '#') {
// Start of a comment. Consume the rest of it.
// The new-line char that terminates the comment is always
// returned.
// It will be treated as white-space, and serves to break up
// anything
// that might otherwise incorrectly clump together with a
// comment in
// the middle (a variable name, for example.)
for (;;) {
c.fChar = NextCharLL();
if (c.fChar == (int) -1
|| // EOF
c.fChar == '\r' || c.fChar == '\n'
|| c.fChar == chNEL || c.fChar == chLS) {
break;
}
}
}
if (c.fChar == (int) -1) {
return;
}
//
// check for backslash escaped characters.
// Use String.unescapeAt() to handle them.
//
if (c.fChar == '\\') {
c.fEscaped = true;
int[] unescapeIndex = new int[1];
unescapeIndex[0] = fNextIndex;
c.fChar = IBM.ICU.Impl.Utility.UnescapeAt(fRB.fRules, unescapeIndex);
if (unescapeIndex[0] == fNextIndex) {
Error(IBM.ICU.Text.RBBIRuleBuilder.U_BRK_HEX_DIGITS_EXPECTED);
}
fCharNum += unescapeIndex[0] - fNextIndex;
fNextIndex = unescapeIndex[0];
}
}
// putc(c.fChar, stdout);
}
