mirror of
https://gitflic.ru/project/openide/openide.git
synced 2025-12-15 11:53:49 +07:00
regexp: fix no leading zero octal character lexing
This commit is contained in:
Bas Leijdekkers
@@ -132,55 +132,55 @@ class _RegExLexer implements FlexLexer {
|
||||
|
||||
/* The ZZ_CMAP_A table has 3056 entries */
|
||||
static final char ZZ_CMAP_A[] = zzUnpackCMap(
|
||||
"\10\0\2\63\1\66\1\67\1\70\1\66\22\0\1\16\1\75\1\0\1\76\1\21\1\0\1\72\1\62"+
|
||||
"\1\4\1\5\1\23\1\24\1\65\1\15\1\3\1\0\1\2\3\55\4\54\2\1\1\71\1\0\1\61\1\74"+
|
||||
"\1\73\1\22\1\0\2\32\1\41\1\37\1\47\1\50\1\33\1\56\1\42\2\13\1\46\1\64\1\57"+
|
||||
"\1\13\1\44\1\51\1\35\1\36\1\13\1\45\3\35\1\13\1\34\1\10\1\12\1\11\1\20\1\14"+
|
||||
"\1\0\1\27\1\30\1\40\1\37\2\27\1\31\1\56\1\42\1\13\1\60\1\45\1\13\1\26\1\13"+
|
||||
"\1\43\1\13\1\26\1\35\1\26\1\53\2\35\1\52\1\13\1\33\1\6\1\25\1\7\7\0\1\67\24"+
|
||||
"\0\1\13\12\0\1\13\4\0\1\13\5\0\27\13\1\0\12\13\4\0\14\13\16\0\5\13\7\0\1\13"+
|
||||
"\1\0\1\13\1\0\5\13\1\0\2\13\2\0\4\13\1\0\1\13\6\0\1\13\1\0\3\13\1\0\1\13\1"+
|
||||
"\0\4\13\1\0\23\13\1\0\13\13\10\0\6\13\1\0\26\13\2\0\1\13\6\0\10\13\10\0\13"+
|
||||
"\13\5\0\3\13\15\0\12\17\4\0\6\13\1\0\1\13\17\0\2\13\7\0\2\13\12\17\3\13\2"+
|
||||
"\0\2\13\1\0\16\13\15\0\11\13\13\0\1\13\16\0\12\17\6\13\4\0\2\13\4\0\1\13\5"+
|
||||
"\0\6\13\4\0\1\13\11\0\1\13\3\0\1\13\7\0\11\13\7\0\5\13\17\0\26\13\3\0\1\13"+
|
||||
"\2\0\1\13\7\0\12\13\4\0\12\17\1\13\4\0\10\13\2\0\2\13\2\0\26\13\1\0\7\13\1"+
|
||||
"\0\1\13\3\0\4\13\3\0\1\13\20\0\1\13\15\0\2\13\1\0\1\13\5\0\6\13\4\0\2\13\1"+
|
||||
"\0\2\13\1\0\2\13\1\0\2\13\17\0\4\13\1\0\1\13\7\0\12\17\2\0\3\13\20\0\11\13"+
|
||||
"\1\0\2\13\1\0\2\13\1\0\5\13\3\0\1\13\2\0\1\13\30\0\1\13\13\0\10\13\2\0\1\13"+
|
||||
"\3\0\1\13\1\0\6\13\3\0\3\13\1\0\4\13\3\0\2\13\1\0\1\13\1\0\2\13\3\0\2\13\3"+
|
||||
"\0\3\13\3\0\14\13\13\0\10\13\1\0\2\13\10\0\3\13\5\0\4\13\1\0\5\13\3\0\1\13"+
|
||||
"\3\0\2\13\15\0\13\13\2\0\1\13\21\0\1\13\12\0\6\13\5\0\22\13\3\0\10\13\1\0"+
|
||||
"\11\13\1\0\1\13\2\0\7\13\11\0\1\13\1\0\2\13\14\0\12\17\7\0\2\13\1\0\1\13\2"+
|
||||
"\0\2\13\1\0\1\13\2\0\1\13\6\0\4\13\1\0\7\13\1\0\3\13\1\0\1\13\1\0\1\13\2\0"+
|
||||
"\2\13\1\0\4\13\1\0\2\13\11\0\1\13\2\0\5\13\1\0\1\13\11\0\12\17\2\0\14\13\1"+
|
||||
"\0\24\13\13\0\5\13\3\0\6\13\4\0\4\13\3\0\1\13\3\0\2\13\7\0\3\13\4\0\15\13"+
|
||||
"\14\0\1\13\1\0\6\13\1\0\1\13\5\0\1\13\2\0\13\13\1\0\15\13\1\0\4\13\2\0\7\13"+
|
||||
"\1\0\1\13\1\0\4\13\2\0\1\13\1\0\4\13\2\0\7\13\1\0\1\13\1\0\4\13\2\0\16\13"+
|
||||
"\2\0\6\13\2\0\15\13\2\0\1\13\1\0\10\13\7\0\15\13\1\0\6\13\23\0\1\13\4\0\1"+
|
||||
"\13\3\0\11\13\1\0\1\13\5\0\17\13\1\0\16\13\2\0\14\13\13\0\1\13\15\0\7\13\7"+
|
||||
"\0\16\13\15\0\2\13\12\17\3\0\3\13\11\0\4\13\1\0\4\13\3\0\2\13\11\0\10\13\1"+
|
||||
"\0\1\13\1\0\1\13\1\0\1\13\1\0\6\13\1\0\7\13\1\0\1\13\3\0\3\13\1\0\7\13\3\0"+
|
||||
"\4\13\2\0\6\13\14\0\2\67\7\0\1\13\15\0\1\13\2\0\1\13\4\0\1\13\2\0\12\13\1"+
|
||||
"\0\1\13\3\0\5\13\6\0\1\13\1\0\1\13\1\0\1\13\1\0\4\13\1\0\13\13\2\0\4\13\5"+
|
||||
"\0\5\13\4\0\1\13\4\0\2\13\13\0\5\13\6\0\4\13\3\0\2\13\14\0\10\13\7\0\10\13"+
|
||||
"\1\0\7\13\6\0\2\13\12\0\5\13\5\0\2\13\3\0\7\13\6\0\3\13\12\17\2\13\13\0\11"+
|
||||
"\13\2\0\27\13\2\0\7\13\1\0\3\13\1\0\4\13\1\0\4\13\2\0\6\13\3\0\1\13\1\0\1"+
|
||||
"\13\2\0\5\13\1\0\12\13\12\17\5\13\1\0\3\13\1\0\10\13\4\0\7\13\3\0\1\13\3\0"+
|
||||
"\2\13\1\0\1\13\3\0\2\13\2\0\5\13\2\0\1\13\1\0\1\13\30\0\3\13\3\0\6\13\2\0"+
|
||||
"\6\13\2\0\6\13\11\0\7\13\4\0\5\13\3\0\5\13\5\0\1\13\1\0\10\13\1\0\5\13\1\0"+
|
||||
"\1\13\1\0\2\13\1\0\2\13\1\0\12\13\6\0\12\13\2\0\6\13\2\0\6\13\2\0\6\13\2\0"+
|
||||
"\3\13\3\0\14\13\1\0\16\13\1\0\2\13\1\0\2\13\1\0\10\13\6\0\4\13\4\0\16\13\2"+
|
||||
"\0\1\13\1\0\14\13\1\0\2\13\3\0\1\13\2\0\4\13\1\0\2\13\12\0\10\13\6\0\6\13"+
|
||||
"\1\0\3\13\1\0\12\13\3\0\1\13\12\0\4\13\13\0\12\17\1\13\1\0\1\13\3\0\7\13\1"+
|
||||
"\0\1\13\1\0\4\13\1\0\17\13\1\0\2\13\14\0\3\13\4\0\2\13\1\0\1\13\20\0\4\13"+
|
||||
"\10\0\1\13\13\0\10\13\5\0\3\13\2\0\1\13\2\0\2\13\2\0\4\13\1\0\14\13\1\0\1"+
|
||||
"\13\1\0\7\13\1\0\21\13\1\0\4\13\2\0\10\13\1\0\7\13\1\0\14\13\1\0\4\13\1\0"+
|
||||
"\5\13\1\0\1\13\3\0\14\13\2\0\13\13\1\0\10\13\2\0\22\17\1\0\2\13\1\0\1\13\2"+
|
||||
"\0\1\13\1\0\12\13\1\0\4\13\1\0\1\13\1\0\1\13\6\0\1\13\4\0\1\13\1\0\1\13\1"+
|
||||
"\0\1\13\1\0\3\13\1\0\2\13\1\0\1\13\2\0\1\13\1\0\1\13\1\0\1\13\1\0\1\13\1\0"+
|
||||
"\1\13\1\0\2\13\1\0\1\13\2\0\4\13\1\0\7\13\1\0\4\13\1\0\4\13\1\0\1\13\1\0\12"+
|
||||
"\13\1\0\5\13\1\0\3\13\1\0\5\13\1\0\5\13");
|
||||
"\10\0\2\63\1\66\1\67\1\70\1\66\22\0\1\14\1\75\1\0\1\76\1\17\1\0\1\72\1\62"+
|
||||
"\1\2\1\3\1\21\1\22\1\65\1\13\1\1\1\0\1\53\3\55\4\54\2\47\1\71\1\0\1\61\1\74"+
|
||||
"\1\73\1\20\1\0\2\30\1\37\1\35\1\45\1\46\1\31\1\56\1\40\2\11\1\44\1\64\1\57"+
|
||||
"\1\11\1\42\1\50\1\33\1\34\1\11\1\43\3\33\1\11\1\32\1\6\1\10\1\7\1\16\1\12"+
|
||||
"\1\0\1\25\1\26\1\36\1\35\2\25\1\27\1\56\1\40\1\11\1\60\1\43\1\11\1\24\1\11"+
|
||||
"\1\41\1\11\1\24\1\33\1\24\1\52\2\33\1\51\1\11\1\31\1\4\1\23\1\5\7\0\1\67\24"+
|
||||
"\0\1\11\12\0\1\11\4\0\1\11\5\0\27\11\1\0\12\11\4\0\14\11\16\0\5\11\7\0\1\11"+
|
||||
"\1\0\1\11\1\0\5\11\1\0\2\11\2\0\4\11\1\0\1\11\6\0\1\11\1\0\3\11\1\0\1\11\1"+
|
||||
"\0\4\11\1\0\23\11\1\0\13\11\10\0\6\11\1\0\26\11\2\0\1\11\6\0\10\11\10\0\13"+
|
||||
"\11\5\0\3\11\15\0\12\15\4\0\6\11\1\0\1\11\17\0\2\11\7\0\2\11\12\15\3\11\2"+
|
||||
"\0\2\11\1\0\16\11\15\0\11\11\13\0\1\11\16\0\12\15\6\11\4\0\2\11\4\0\1\11\5"+
|
||||
"\0\6\11\4\0\1\11\11\0\1\11\3\0\1\11\7\0\11\11\7\0\5\11\17\0\26\11\3\0\1\11"+
|
||||
"\2\0\1\11\7\0\12\11\4\0\12\15\1\11\4\0\10\11\2\0\2\11\2\0\26\11\1\0\7\11\1"+
|
||||
"\0\1\11\3\0\4\11\3\0\1\11\20\0\1\11\15\0\2\11\1\0\1\11\5\0\6\11\4\0\2\11\1"+
|
||||
"\0\2\11\1\0\2\11\1\0\2\11\17\0\4\11\1\0\1\11\7\0\12\15\2\0\3\11\20\0\11\11"+
|
||||
"\1\0\2\11\1\0\2\11\1\0\5\11\3\0\1\11\2\0\1\11\30\0\1\11\13\0\10\11\2\0\1\11"+
|
||||
"\3\0\1\11\1\0\6\11\3\0\3\11\1\0\4\11\3\0\2\11\1\0\1\11\1\0\2\11\3\0\2\11\3"+
|
||||
"\0\3\11\3\0\14\11\13\0\10\11\1\0\2\11\10\0\3\11\5\0\4\11\1\0\5\11\3\0\1\11"+
|
||||
"\3\0\2\11\15\0\13\11\2\0\1\11\21\0\1\11\12\0\6\11\5\0\22\11\3\0\10\11\1\0"+
|
||||
"\11\11\1\0\1\11\2\0\7\11\11\0\1\11\1\0\2\11\14\0\12\15\7\0\2\11\1\0\1\11\2"+
|
||||
"\0\2\11\1\0\1\11\2\0\1\11\6\0\4\11\1\0\7\11\1\0\3\11\1\0\1\11\1\0\1\11\2\0"+
|
||||
"\2\11\1\0\4\11\1\0\2\11\11\0\1\11\2\0\5\11\1\0\1\11\11\0\12\15\2\0\14\11\1"+
|
||||
"\0\24\11\13\0\5\11\3\0\6\11\4\0\4\11\3\0\1\11\3\0\2\11\7\0\3\11\4\0\15\11"+
|
||||
"\14\0\1\11\1\0\6\11\1\0\1\11\5\0\1\11\2\0\13\11\1\0\15\11\1\0\4\11\2\0\7\11"+
|
||||
"\1\0\1\11\1\0\4\11\2\0\1\11\1\0\4\11\2\0\7\11\1\0\1\11\1\0\4\11\2\0\16\11"+
|
||||
"\2\0\6\11\2\0\15\11\2\0\1\11\1\0\10\11\7\0\15\11\1\0\6\11\23\0\1\11\4\0\1"+
|
||||
"\11\3\0\11\11\1\0\1\11\5\0\17\11\1\0\16\11\2\0\14\11\13\0\1\11\15\0\7\11\7"+
|
||||
"\0\16\11\15\0\2\11\12\15\3\0\3\11\11\0\4\11\1\0\4\11\3\0\2\11\11\0\10\11\1"+
|
||||
"\0\1\11\1\0\1\11\1\0\1\11\1\0\6\11\1\0\7\11\1\0\1\11\3\0\3\11\1\0\7\11\3\0"+
|
||||
"\4\11\2\0\6\11\14\0\2\67\7\0\1\11\15\0\1\11\2\0\1\11\4\0\1\11\2\0\12\11\1"+
|
||||
"\0\1\11\3\0\5\11\6\0\1\11\1\0\1\11\1\0\1\11\1\0\4\11\1\0\13\11\2\0\4\11\5"+
|
||||
"\0\5\11\4\0\1\11\4\0\2\11\13\0\5\11\6\0\4\11\3\0\2\11\14\0\10\11\7\0\10\11"+
|
||||
"\1\0\7\11\6\0\2\11\12\0\5\11\5\0\2\11\3\0\7\11\6\0\3\11\12\15\2\11\13\0\11"+
|
||||
"\11\2\0\27\11\2\0\7\11\1\0\3\11\1\0\4\11\1\0\4\11\2\0\6\11\3\0\1\11\1\0\1"+
|
||||
"\11\2\0\5\11\1\0\12\11\12\15\5\11\1\0\3\11\1\0\10\11\4\0\7\11\3\0\1\11\3\0"+
|
||||
"\2\11\1\0\1\11\3\0\2\11\2\0\5\11\2\0\1\11\1\0\1\11\30\0\3\11\3\0\6\11\2\0"+
|
||||
"\6\11\2\0\6\11\11\0\7\11\4\0\5\11\3\0\5\11\5\0\1\11\1\0\10\11\1\0\5\11\1\0"+
|
||||
"\1\11\1\0\2\11\1\0\2\11\1\0\12\11\6\0\12\11\2\0\6\11\2\0\6\11\2\0\6\11\2\0"+
|
||||
"\3\11\3\0\14\11\1\0\16\11\1\0\2\11\1\0\2\11\1\0\10\11\6\0\4\11\4\0\16\11\2"+
|
||||
"\0\1\11\1\0\14\11\1\0\2\11\3\0\1\11\2\0\4\11\1\0\2\11\12\0\10\11\6\0\6\11"+
|
||||
"\1\0\3\11\1\0\12\11\3\0\1\11\12\0\4\11\13\0\12\15\1\11\1\0\1\11\3\0\7\11\1"+
|
||||
"\0\1\11\1\0\4\11\1\0\17\11\1\0\2\11\14\0\3\11\4\0\2\11\1\0\1\11\20\0\4\11"+
|
||||
"\10\0\1\11\13\0\10\11\5\0\3\11\2\0\1\11\2\0\2\11\2\0\4\11\1\0\14\11\1\0\1"+
|
||||
"\11\1\0\7\11\1\0\21\11\1\0\4\11\2\0\10\11\1\0\7\11\1\0\14\11\1\0\4\11\1\0"+
|
||||
"\5\11\1\0\1\11\3\0\14\11\2\0\13\11\1\0\10\11\2\0\22\15\1\0\2\11\1\0\1\11\2"+
|
||||
"\0\1\11\1\0\12\11\1\0\4\11\1\0\1\11\1\0\1\11\6\0\1\11\4\0\1\11\1\0\1\11\1"+
|
||||
"\0\1\11\1\0\3\11\1\0\2\11\1\0\1\11\2\0\1\11\1\0\1\11\1\0\1\11\1\0\1\11\1\0"+
|
||||
"\1\11\1\0\2\11\1\0\1\11\2\0\4\11\1\0\7\11\1\0\4\11\1\0\4\11\1\0\1\11\1\0\12"+
|
||||
"\11\1\0\5\11\1\0\3\11\1\0\5\11\1\0\5\11");
|
||||
|
||||
/**
|
||||
* Translates DFA states to action switch labels.
|
||||
@@ -195,18 +195,18 @@ class _RegExLexer implements FlexLexer {
|
||||
"\1\35\1\12\1\36\1\37\1\2\1\40\1\41\1\24"+
|
||||
"\1\42\1\43\1\44\1\45\1\46\1\47\1\1\1\26"+
|
||||
"\1\50\1\51\2\52\1\53\2\0\1\54\1\0\1\55"+
|
||||
"\1\56\1\57\1\60\1\61\1\62\1\63\1\64\1\12"+
|
||||
"\1\65\1\66\1\67\1\70\1\12\1\70\1\71\2\72"+
|
||||
"\1\56\1\57\1\60\1\61\1\62\1\12\1\63\1\64"+
|
||||
"\1\65\1\66\1\12\1\66\1\67\2\70\1\71\1\72"+
|
||||
"\1\73\1\74\1\75\1\76\1\77\1\100\1\101\1\12"+
|
||||
"\1\102\1\103\1\104\1\0\1\105\1\106\1\107\1\110"+
|
||||
"\1\0\1\111\1\112\1\113\1\114\1\115\1\0\1\116"+
|
||||
"\1\0\1\117\1\0\2\120\1\0\1\121\1\122\1\123"+
|
||||
"\2\76\1\124\1\125\1\126\1\127\1\130\1\131\1\132"+
|
||||
"\1\133\1\0\2\120\1\0\1\134\2\76\1\124\1\135"+
|
||||
"\1\136\1\76\1\124\1\76\1\137\4\76";
|
||||
"\1\0\1\117\2\0\1\120\1\121\1\122\1\73\2\75"+
|
||||
"\1\123\2\124\1\125\1\126\1\127\1\130\1\131\1\132"+
|
||||
"\1\133\2\0\1\73\2\75\1\134\1\123\2\124\1\135"+
|
||||
"\1\75\1\123\1\136\1\75\1\137\4\75";
|
||||
|
||||
private static int [] zzUnpackAction() {
|
||||
int [] result = new int[150];
|
||||
int [] result = new int[152];
|
||||
int offset = 0;
|
||||
offset = zzUnpackAction(ZZ_ACTION_PACKED_0, offset, result);
|
||||
return result;
|
||||
@@ -240,19 +240,19 @@ class _RegExLexer implements FlexLexer {
|
||||
"\0\u042f\0\u042f\0\u06e4\0\u042f\0\u04ad\0\u04ec\0\u042f\0\u042f"+
|
||||
"\0\u042f\0\u0723\0\u0762\0\u042f\0\u042f\0\u07a1\0\u042f\0\u042f"+
|
||||
"\0\u042f\0\u07e0\0\u081f\0\u085e\0\u089d\0\u042f\0\u08dc\0\u091b"+
|
||||
"\0\u042f\0\u095a\0\u0999\0\u042f\0\u042f\0\u042f\0\u042f\0\u042f"+
|
||||
"\0\u042f\0\u042f\0\u042f\0\u09d8\0\u0a17\0\u042f\0\u042f\0\u0a56"+
|
||||
"\0\u042f\0\u042f\0\u042f\0\u042f\0\u0a95\0\u0ad4\0\u042f\0\u042f"+
|
||||
"\0\u042f\0\u042f\0\u042f\0\u042f\0\u042f\0\u042f\0\u042f\0\u042f"+
|
||||
"\0\u042f\0\u095a\0\u0999\0\u042f\0\u042f\0\u09d8\0\u042f\0\u042f"+
|
||||
"\0\u042f\0\u0a17\0\u042f\0\u0a56\0\u0a95\0\u0ad4\0\u042f\0\u042f"+
|
||||
"\0\u0b13\0\u042f\0\u042f\0\u042f\0\u0b52\0\u042f\0\u042f\0\u042f"+
|
||||
"\0\u042f\0\u0b91\0\u0bd0\0\u042f\0\u042f\0\u042f\0\u042f\0\u0c0f"+
|
||||
"\0\u042f\0\u0c4e\0\u042f\0\u0c8d\0\u0ccc\0\u0d0b\0\u0d4a\0\u042f"+
|
||||
"\0\u042f\0\u042f\0\u0d89\0\u0dc8\0\u0e07\0\u042f\0\u042f\0\u042f"+
|
||||
"\0\u042f\0\u042f\0\u042f\0\u042f\0\u0e46\0\u0e85\0\u042f\0\u0ec4"+
|
||||
"\0\u042f\0\u0f03\0\u042f\0\u0f42\0\u042f\0\u042f\0\u0f81\0\u0fc0"+
|
||||
"\0\u0fff\0\u042f\0\u103e\0\u107d\0\u10bc\0\u10fb";
|
||||
"\0\u042f\0\u0c4e\0\u042f\0\u0c8d\0\u0ccc\0\u042f\0\u042f\0\u042f"+
|
||||
"\0\u0d0b\0\u0d4a\0\u0d89\0\u0dc8\0\u0e07\0\u0e46\0\u042f\0\u042f"+
|
||||
"\0\u042f\0\u042f\0\u042f\0\u042f\0\u042f\0\u0e85\0\u0ec4\0\u042f"+
|
||||
"\0\u042f\0\u0f03\0\u042f\0\u0f42\0\u0f81\0\u042f\0\u042f\0\u0fc0"+
|
||||
"\0\u0fff\0\u042f\0\u103e\0\u042f\0\u107d\0\u10bc\0\u10fb\0\u113a";
|
||||
|
||||
private static int [] zzUnpackRowMap() {
|
||||
int [] result = new int[150];
|
||||
int [] result = new int[152];
|
||||
int offset = 0;
|
||||
offset = zzUnpackRowMap(ZZ_ROWMAP_PACKED_0, offset, result);
|
||||
return result;
|
||||
@@ -275,88 +275,86 @@ class _RegExLexer implements FlexLexer {
|
||||
private static final int [] ZZ_TRANS = zzUnpackTrans();
|
||||
|
||||
private static final String ZZ_TRANS_PACKED_0 =
|
||||
"\3\22\1\23\1\24\1\25\1\26\1\27\1\30\1\31"+
|
||||
"\1\22\1\23\1\24\1\25\1\26\1\27\1\30\1\31"+
|
||||
"\1\32\3\22\1\33\1\22\1\34\1\35\1\36\1\37"+
|
||||
"\1\40\1\41\35\22\1\42\2\22\1\42\1\22\1\42"+
|
||||
"\5\22\1\43\12\22\1\44\64\22\7\45\1\46\3\45"+
|
||||
"\1\47\4\45\1\34\5\45\26\47\2\45\3\47\3\45"+
|
||||
"\1\47\13\45\2\50\4\45\1\46\7\45\1\50\34\45"+
|
||||
"\2\50\7\45\1\51\11\45\11\52\1\53\1\54\53\52"+
|
||||
"\3\0\26\52\1\55\45\52\3\0\6\52\6\22\1\26"+
|
||||
"\1\22\1\56\1\57\1\60\2\22\1\61\45\22\1\62"+
|
||||
"\2\22\1\62\1\22\1\62\1\22\1\63\4\22\6\52"+
|
||||
"\1\26\1\52\1\30\1\52\1\32\5\52\1\64\45\52"+
|
||||
"\1\42\1\22\1\42\6\52\6\45\1\65\1\45\1\66"+
|
||||
"\1\45\1\32\21\45\1\67\1\45\1\67\2\45\1\67"+
|
||||
"\2\45\1\67\1\45\1\67\10\45\1\67\4\45\1\67"+
|
||||
"\20\45\1\65\1\45\1\66\1\45\1\32\64\45\5\70"+
|
||||
"\1\71\5\70\1\72\1\70\1\73\10\70\26\72\2\70"+
|
||||
"\3\72\3\70\1\72\4\70\1\74\5\70\66\14\1\75"+
|
||||
"\10\14\13\70\1\76\12\70\26\76\2\70\3\76\3\70"+
|
||||
"\1\76\6\70\1\77\16\70\1\76\12\70\26\76\2\70"+
|
||||
"\3\76\1\70\1\100\1\70\1\76\17\70\1\71\5\70"+
|
||||
"\1\76\12\70\26\76\2\70\3\76\3\70\1\76\13\70"+
|
||||
"\2\50\2\70\1\71\5\70\1\76\3\70\1\50\6\70"+
|
||||
"\26\76\2\50\3\76\3\70\1\76\12\70\13\101\1\47"+
|
||||
"\4\101\1\34\5\101\26\47\2\101\3\47\3\101\1\47"+
|
||||
"\4\101\1\102\5\101\121\0\1\103\55\0\2\104\14\0"+
|
||||
"\1\104\34\0\2\104\7\0\1\105\22\0\1\106\1\107"+
|
||||
"\5\0\1\110\56\0\1\111\1\112\1\113\3\114\1\115"+
|
||||
"\1\116\1\117\1\111\1\117\1\120\1\111\1\121\1\122"+
|
||||
"\1\111\1\117\5\114\2\123\1\124\1\125\3\126\3\127"+
|
||||
"\1\130\2\131\2\132\3\133\1\120\1\134\1\135\1\136"+
|
||||
"\2\112\1\137\1\140\1\141\2\111\1\122\1\120\1\111"+
|
||||
"\1\122\1\111\1\122\6\111\47\0\1\142\30\0\2\47"+
|
||||
"\1\0\2\47\5\0\5\47\6\0\33\47\3\0\1\47"+
|
||||
"\13\0\2\50\14\0\1\50\34\0\2\50\32\0\1\143"+
|
||||
"\76\0\1\144\1\145\5\0\1\110\50\0\1\146\5\0"+
|
||||
"\1\111\1\112\1\113\3\114\1\115\1\116\3\117\1\120"+
|
||||
"\1\111\1\121\1\122\1\111\1\117\5\114\2\123\1\124"+
|
||||
"\1\125\3\126\3\127\1\130\2\131\2\132\3\133\1\120"+
|
||||
"\1\134\1\135\1\136\2\112\1\137\1\140\1\141\2\111"+
|
||||
"\1\122\1\120\1\111\1\122\1\111\1\122\6\111\72\0"+
|
||||
"\1\147\17\0\1\72\12\0\26\72\2\0\3\72\3\0"+
|
||||
"\1\72\25\0\1\73\12\0\26\73\2\0\3\73\3\0"+
|
||||
"\1\73\13\0\2\76\10\0\5\76\6\0\33\76\3\0"+
|
||||
"\1\76\23\0\1\150\71\0\1\151\37\0\1\152\14\0"+
|
||||
"\1\153\1\154\6\0\1\155\1\0\1\155\1\156\1\157"+
|
||||
"\1\160\1\0\2\104\4\0\1\161\7\0\1\104\34\0"+
|
||||
"\2\104\7\0\1\105\12\0\2\162\14\0\1\162\34\0"+
|
||||
"\2\162\32\0\1\106\76\0\1\163\1\164\65\0\2\112"+
|
||||
"\51\0\2\112\23\0\1\165\51\0\1\166\1\165\27\0"+
|
||||
"\1\167\151\0\1\170\1\171\14\0\77\172\1\0\2\173"+
|
||||
"\3\0\1\174\20\0\2\173\1\0\1\173\4\0\3\173"+
|
||||
"\5\0\2\173\3\0\2\173\22\0\2\175\24\0\2\175"+
|
||||
"\1\0\1\175\4\0\3\175\5\0\2\175\3\0\2\175"+
|
||||
"\102\0\1\176\1\177\25\0\1\200\146\0\1\201\12\0"+
|
||||
"\1\202\76\0\1\203\1\204\1\0\5\205\1\0\71\205"+
|
||||
"\1\0\2\162\4\0\1\161\7\0\1\162\34\0\2\162"+
|
||||
"\32\0\1\163\67\0\1\206\51\0\2\206\23\0\1\207"+
|
||||
"\51\0\2\207\52\0\1\210\46\0\2\211\24\0\2\211"+
|
||||
"\1\0\1\211\4\0\3\211\5\0\2\211\3\0\2\211"+
|
||||
"\22\0\2\212\4\0\1\213\17\0\2\212\1\0\1\212"+
|
||||
"\4\0\3\212\5\0\2\212\3\0\2\212\22\0\2\214"+
|
||||
"\24\0\2\214\1\0\1\214\4\0\3\214\5\0\2\214"+
|
||||
"\3\0\2\214\21\0\5\205\1\215\71\205\2\0\1\216"+
|
||||
"\51\0\2\216\30\0\1\126\70\0\2\217\4\0\1\211"+
|
||||
"\17\0\2\217\1\0\1\217\4\0\3\217\5\0\2\217"+
|
||||
"\3\0\2\217\22\0\2\220\24\0\2\220\1\0\1\220"+
|
||||
"\4\0\3\220\5\0\2\220\3\0\2\220\22\0\2\221"+
|
||||
"\4\0\1\211\17\0\2\221\1\0\1\221\4\0\3\221"+
|
||||
"\5\0\2\221\3\0\2\221\22\0\2\222\24\0\2\222"+
|
||||
"\1\0\1\222\4\0\3\222\5\0\2\222\3\0\2\222"+
|
||||
"\22\0\2\223\4\0\1\211\17\0\2\223\1\0\1\223"+
|
||||
"\4\0\3\223\5\0\2\223\3\0\2\223\22\0\2\224"+
|
||||
"\4\0\1\211\17\0\2\224\1\0\1\224\4\0\3\224"+
|
||||
"\5\0\2\224\3\0\2\224\22\0\2\225\4\0\1\211"+
|
||||
"\17\0\2\225\1\0\1\225\4\0\3\225\5\0\2\225"+
|
||||
"\3\0\2\225\22\0\2\226\4\0\1\211\17\0\2\226"+
|
||||
"\1\0\1\226\4\0\3\226\5\0\2\226\3\0\2\226"+
|
||||
"\22\0\2\226\4\0\1\213\17\0\2\226\1\0\1\226"+
|
||||
"\4\0\3\226\5\0\2\226\3\0\2\226\21\0";
|
||||
"\1\40\1\41\37\22\1\42\2\22\1\42\1\22\1\42"+
|
||||
"\5\22\1\43\10\22\1\44\66\22\5\45\1\46\3\45"+
|
||||
"\1\47\4\45\1\34\5\45\23\47\1\45\3\47\3\45"+
|
||||
"\3\47\3\45\1\47\17\45\1\46\7\45\1\50\31\45"+
|
||||
"\1\50\3\45\3\50\7\45\1\51\11\45\7\52\1\53"+
|
||||
"\1\54\55\52\3\0\24\52\1\55\47\52\3\0\6\52"+
|
||||
"\4\22\1\26\1\22\1\56\1\57\1\60\2\22\1\61"+
|
||||
"\47\22\1\62\2\22\1\62\1\22\1\62\1\22\1\63"+
|
||||
"\4\22\4\52\1\26\1\52\1\30\1\52\1\32\5\52"+
|
||||
"\1\64\47\52\1\42\1\22\1\42\6\52\4\45\1\65"+
|
||||
"\1\45\1\66\1\45\1\32\21\45\1\67\1\45\1\67"+
|
||||
"\2\45\1\67\2\45\1\67\1\45\1\67\12\45\1\67"+
|
||||
"\4\45\1\67\16\45\1\65\1\45\1\66\1\45\1\32"+
|
||||
"\66\45\3\70\1\71\5\70\1\72\1\70\1\73\10\70"+
|
||||
"\23\72\1\70\3\72\3\70\3\72\3\70\1\72\4\70"+
|
||||
"\1\74\5\70\66\14\1\75\10\14\11\70\1\76\12\70"+
|
||||
"\23\76\1\70\3\76\3\70\3\76\3\70\1\76\6\70"+
|
||||
"\1\77\14\70\1\76\12\70\23\76\1\70\3\76\3\70"+
|
||||
"\3\76\1\70\1\100\1\70\1\76\15\70\1\71\5\70"+
|
||||
"\1\76\12\70\23\76\1\70\3\76\3\70\3\76\3\70"+
|
||||
"\1\76\15\70\1\71\5\70\1\76\3\70\1\50\6\70"+
|
||||
"\23\76\1\50\3\76\3\50\3\76\3\70\1\76\12\70"+
|
||||
"\11\101\1\47\4\101\1\34\5\101\23\47\1\101\3\47"+
|
||||
"\3\101\3\47\3\101\1\47\4\101\1\102\5\101\117\0"+
|
||||
"\1\103\73\0\1\104\31\0\1\104\3\0\3\104\7\0"+
|
||||
"\1\105\20\0\1\106\1\107\5\0\1\110\60\0\1\111"+
|
||||
"\3\112\1\113\1\114\1\115\1\111\1\115\1\116\1\111"+
|
||||
"\1\117\1\120\1\111\1\115\5\112\2\121\1\122\1\123"+
|
||||
"\3\124\3\125\1\126\2\127\2\130\3\131\1\116\1\132"+
|
||||
"\1\133\1\134\1\135\1\136\2\132\1\137\1\140\1\141"+
|
||||
"\2\111\1\120\1\116\1\111\1\120\1\111\1\120\6\111"+
|
||||
"\45\0\1\142\33\0\2\47\5\0\5\47\6\0\35\47"+
|
||||
"\3\0\1\47\27\0\1\50\31\0\1\50\3\0\3\50"+
|
||||
"\30\0\1\143\76\0\1\144\1\145\5\0\1\110\52\0"+
|
||||
"\1\146\5\0\1\111\3\112\1\113\1\114\3\115\1\116"+
|
||||
"\1\111\1\117\1\120\1\111\1\115\5\112\2\121\1\122"+
|
||||
"\1\123\3\124\3\125\1\126\2\127\2\130\3\131\1\116"+
|
||||
"\1\132\1\133\1\134\1\135\1\136\2\132\1\137\1\140"+
|
||||
"\1\141\2\111\1\120\1\116\1\111\1\120\1\111\1\120"+
|
||||
"\6\111\72\0\1\147\15\0\1\72\12\0\23\72\1\0"+
|
||||
"\3\72\3\0\3\72\3\0\1\72\23\0\1\73\12\0"+
|
||||
"\23\73\1\0\3\73\3\0\3\73\3\0\1\73\23\0"+
|
||||
"\5\76\6\0\35\76\3\0\1\76\21\0\1\150\71\0"+
|
||||
"\1\151\37\0\1\152\16\0\1\153\1\154\6\0\1\155"+
|
||||
"\1\0\1\155\1\156\1\157\1\160\5\0\1\161\7\0"+
|
||||
"\1\104\31\0\1\104\3\0\3\104\7\0\1\105\26\0"+
|
||||
"\1\162\31\0\1\162\3\0\3\162\30\0\1\106\76\0"+
|
||||
"\1\163\1\164\72\0\1\165\153\0\1\166\1\167\14\0"+
|
||||
"\77\170\47\0\1\171\3\0\3\171\25\0\1\172\20\0"+
|
||||
"\2\173\1\0\1\173\4\0\3\173\5\0\3\173\3\0"+
|
||||
"\3\173\46\0\2\174\1\0\1\174\4\0\3\174\5\0"+
|
||||
"\3\174\3\0\3\174\74\0\1\175\1\176\1\175\102\0"+
|
||||
"\1\177\1\200\23\0\1\201\150\0\1\202\12\0\1\203"+
|
||||
"\76\0\1\204\1\205\1\0\3\206\1\0\73\206\5\0"+
|
||||
"\1\161\7\0\1\162\31\0\1\162\3\0\3\162\30\0"+
|
||||
"\1\163\116\0\1\207\116\0\1\210\3\0\3\210\26\0"+
|
||||
"\1\211\17\0\2\212\1\0\1\212\4\0\3\212\5\0"+
|
||||
"\3\212\3\0\3\212\46\0\2\213\1\0\1\213\4\0"+
|
||||
"\3\213\5\0\3\213\3\0\3\213\46\0\2\214\1\0"+
|
||||
"\1\214\4\0\3\214\5\0\3\214\3\0\3\214\74\0"+
|
||||
"\3\215\74\0\3\216\21\0\3\206\1\217\73\206\5\0"+
|
||||
"\1\124\76\0\1\213\17\0\2\220\1\0\1\220\4\0"+
|
||||
"\3\220\5\0\3\220\3\0\3\220\46\0\2\221\1\0"+
|
||||
"\1\221\4\0\3\221\5\0\3\221\3\0\3\221\74\0"+
|
||||
"\3\222\26\0\1\213\17\0\2\223\1\0\1\223\4\0"+
|
||||
"\3\223\5\0\3\223\3\0\3\223\46\0\2\224\1\0"+
|
||||
"\1\224\4\0\3\224\5\0\3\224\3\0\3\224\26\0"+
|
||||
"\1\213\17\0\2\225\1\0\1\225\4\0\3\225\5\0"+
|
||||
"\3\225\3\0\3\225\26\0\1\213\17\0\2\226\1\0"+
|
||||
"\1\226\4\0\3\226\5\0\3\226\3\0\3\226\26\0"+
|
||||
"\1\213\17\0\2\227\1\0\1\227\4\0\3\227\5\0"+
|
||||
"\3\227\3\0\3\227\26\0\1\213\17\0\2\230\1\0"+
|
||||
"\1\230\4\0\3\230\5\0\3\230\3\0\3\230\26\0"+
|
||||
"\1\211\17\0\2\230\1\0\1\230\4\0\3\230\5\0"+
|
||||
"\3\230\3\0\3\230\21\0";
|
||||
|
||||
private static int [] zzUnpackTrans() {
|
||||
int [] result = new int[4410];
|
||||
int [] result = new int[4473];
|
||||
int offset = 0;
|
||||
offset = zzUnpackTrans(ZZ_TRANS_PACKED_0, offset, result);
|
||||
return result;
|
||||
@@ -398,15 +396,14 @@ class _RegExLexer implements FlexLexer {
|
||||
"\1\1\1\11\1\1\11\11\1\1\2\11\2\1\3\11"+
|
||||
"\1\1\1\11\1\1\1\11\1\1\2\11\1\1\1\11"+
|
||||
"\2\1\3\11\2\1\2\11\1\1\3\11\2\1\2\0"+
|
||||
"\1\11\1\0\1\1\1\11\2\1\10\11\2\1\2\11"+
|
||||
"\1\1\4\11\2\1\2\11\1\1\3\11\1\0\4\11"+
|
||||
"\1\0\1\1\4\11\1\0\1\11\1\0\1\11\1\0"+
|
||||
"\2\1\1\0\3\11\3\1\7\11\1\0\1\1\1\11"+
|
||||
"\1\0\1\11\1\1\1\11\1\1\2\11\3\1\1\11"+
|
||||
"\4\1";
|
||||
"\1\11\1\0\1\1\11\11\2\1\2\11\1\1\3\11"+
|
||||
"\1\1\1\11\3\1\2\11\1\1\3\11\1\0\4\11"+
|
||||
"\1\0\1\1\4\11\1\0\1\11\1\0\1\11\2\0"+
|
||||
"\3\11\6\1\7\11\2\0\2\11\1\1\1\11\2\1"+
|
||||
"\2\11\2\1\1\11\1\1\1\11\4\1";
|
||||
|
||||
private static int [] zzUnpackAttribute() {
|
||||
int [] result = new int[150];
|
||||
int [] result = new int[152];
|
||||
int offset = 0;
|
||||
offset = zzUnpackAttribute(ZZ_ATTRIBUTE_PACKED_0, offset, result);
|
||||
return result;
|
||||
@@ -479,6 +476,8 @@ class _RegExLexer implements FlexLexer {
|
||||
private boolean allowHorizontalWhitespaceClass;
|
||||
private boolean allowPosixBracketExpressions;
|
||||
private boolean allowTransformationEscapes;
|
||||
private int maxOctal = 0777;
|
||||
private int minOctalDigits = 1;
|
||||
|
||||
_RegExLexer(EnumSet<RegExpCapability> capabilities) {
|
||||
this((java.io.Reader)null);
|
||||
@@ -493,6 +492,18 @@ class _RegExLexer implements FlexLexer {
|
||||
this.allowEmptyCharacterClass = capabilities.contains(RegExpCapability.ALLOW_EMPTY_CHARACTER_CLASS);
|
||||
this.allowPosixBracketExpressions = capabilities.contains(RegExpCapability.POSIX_BRACKET_EXPRESSIONS);
|
||||
this.allowTransformationEscapes = capabilities.contains(RegExpCapability.TRANSFORMATION_ESCAPES);
|
||||
if (capabilities.contains(RegExpCapability.MAX_OCTAL_177)) {
|
||||
maxOctal = 0177;
|
||||
}
|
||||
else if (capabilities.contains(RegExpCapability.MAX_OCTAL_377)) {
|
||||
maxOctal = 0377;
|
||||
}
|
||||
if (capabilities.contains(RegExpCapability.MIN_OCTAL_2_DIGITS)) {
|
||||
minOctalDigits = 2;
|
||||
}
|
||||
else if (capabilities.contains(RegExpCapability.MIN_OCTAL_3_DIGITS)) {
|
||||
minOctalDigits = 3;
|
||||
}
|
||||
}
|
||||
|
||||
private void yypushstate(int state) {
|
||||
@@ -957,89 +968,110 @@ class _RegExLexer implements FlexLexer {
|
||||
}
|
||||
case 141: break;
|
||||
case 47:
|
||||
{ if (allowOctalNoLeadingZero) {
|
||||
CharSequence s = yytext();
|
||||
int i = 1;
|
||||
for (; i < s.length(); i++) {
|
||||
if (s.charAt(i) > '7') break;
|
||||
}
|
||||
if (i > 1) {
|
||||
yypushback(yylength() - i);
|
||||
return RegExpTT.OCT_CHAR;
|
||||
}
|
||||
}
|
||||
if (yystate() == CLASS2) {
|
||||
yypushback(yylength() - 2);
|
||||
return RegExpTT.REDUNDANT_ESCAPE;
|
||||
}
|
||||
while (yylength() > 2 && Integer.parseInt(yytext().toString().substring(1)) > capturingGroupCount) {
|
||||
yypushback(1);
|
||||
}
|
||||
return RegExpTT.BACKREF;
|
||||
{ return (yystate() == CLASS2) ? RegExpTT.REDUNDANT_ESCAPE : RegExpTT.ESC_CHARACTER;
|
||||
}
|
||||
case 142: break;
|
||||
case 48:
|
||||
{ return (allowOctalNoLeadingZero ? RegExpTT.OCT_CHAR : RegExpTT.BAD_OCT_VALUE);
|
||||
{ return (allowDanglingMetacharacters != Boolean.TRUE && yystate() != CLASS2) ? RegExpTT.ESC_CHARACTER : RegExpTT.REDUNDANT_ESCAPE;
|
||||
}
|
||||
case 143: break;
|
||||
case 49:
|
||||
{ return (yystate() == CLASS2) ? RegExpTT.REDUNDANT_ESCAPE : RegExpTT.ESC_CHARACTER;
|
||||
{ return (allowDanglingMetacharacters == Boolean.FALSE && yystate() != CLASS2) ? RegExpTT.ESC_CHARACTER : RegExpTT.REDUNDANT_ESCAPE;
|
||||
}
|
||||
case 144: break;
|
||||
case 50:
|
||||
{ return (allowDanglingMetacharacters != Boolean.TRUE && yystate() != CLASS2) ? RegExpTT.ESC_CHARACTER : RegExpTT.REDUNDANT_ESCAPE;
|
||||
{ return RegExpTT.ESC_CHARACTER;
|
||||
}
|
||||
case 145: break;
|
||||
case 51:
|
||||
{ return (allowDanglingMetacharacters == Boolean.FALSE && yystate() != CLASS2) ? RegExpTT.ESC_CHARACTER : RegExpTT.REDUNDANT_ESCAPE;
|
||||
{ return (yystate() == CLASS2) ? RegExpTT.ESC_CHARACTER : RegExpTT.REDUNDANT_ESCAPE;
|
||||
}
|
||||
case 146: break;
|
||||
case 52:
|
||||
{ return RegExpTT.ESC_CHARACTER;
|
||||
{ return commentMode ? RegExpTT.CHARACTER : RegExpTT.REDUNDANT_ESCAPE;
|
||||
}
|
||||
case 147: break;
|
||||
case 53:
|
||||
{ return (yystate() == CLASS2) ? RegExpTT.ESC_CHARACTER : RegExpTT.REDUNDANT_ESCAPE;
|
||||
{ return RegExpTT.ESC_CTRL_CHARACTER;
|
||||
}
|
||||
case 148: break;
|
||||
case 54:
|
||||
{ return commentMode ? RegExpTT.CHARACTER : RegExpTT.REDUNDANT_ESCAPE;
|
||||
{ return yystate() != CLASS2 ? RegExpTT.BOUNDARY : RegExpTT.ESC_CHARACTER;
|
||||
}
|
||||
case 149: break;
|
||||
case 55:
|
||||
{ return RegExpTT.ESC_CTRL_CHARACTER;
|
||||
{ return RegExpTT.CHAR_CLASS;
|
||||
}
|
||||
case 150: break;
|
||||
case 56:
|
||||
{ return yystate() != CLASS2 ? RegExpTT.BOUNDARY : RegExpTT.ESC_CHARACTER;
|
||||
{ if (xmlSchemaMode) return RegExpTT.CHAR_CLASS; else return StringEscapesTokenTypes.INVALID_CHARACTER_ESCAPE_TOKEN;
|
||||
}
|
||||
case 151: break;
|
||||
case 57:
|
||||
{ return RegExpTT.CHAR_CLASS;
|
||||
{ yypushstate(PROP); return RegExpTT.PROPERTY;
|
||||
}
|
||||
case 152: break;
|
||||
case 58:
|
||||
{ if (xmlSchemaMode) return RegExpTT.CHAR_CLASS; else return StringEscapesTokenTypes.INVALID_CHARACTER_ESCAPE_TOKEN;
|
||||
{ return allowTransformationEscapes ? RegExpTT.CHAR_CLASS : StringEscapesTokenTypes.INVALID_CHARACTER_ESCAPE_TOKEN;
|
||||
}
|
||||
case 153: break;
|
||||
case 59:
|
||||
{ yypushstate(PROP); return RegExpTT.PROPERTY;
|
||||
{ String text = yytext().toString().substring(1);
|
||||
if (allowOctalNoLeadingZero) {
|
||||
if (Integer.parseInt(text) <= capturingGroupCount && yystate() != CLASS2) return RegExpTT.BACKREF;
|
||||
int i = 0;
|
||||
int value = 0;
|
||||
for (; i < text.length(); i++) {
|
||||
char c = text.charAt(i);
|
||||
if (c > '7') break;
|
||||
value = value * 8 + (c - '0');
|
||||
}
|
||||
if (i > 0) {
|
||||
yypushback(text.length() - i);
|
||||
if (value > maxOctal) {
|
||||
yypushback(1);
|
||||
return RegExpTT.BAD_OCT_VALUE;
|
||||
}
|
||||
if (minOctalDigits > i && yystate() != CLASS2) {
|
||||
return RegExpTT.BAD_OCT_VALUE;
|
||||
}
|
||||
return RegExpTT.OCT_CHAR;
|
||||
}
|
||||
return StringEscapesTokenTypes.INVALID_CHARACTER_ESCAPE_TOKEN;
|
||||
}
|
||||
else {
|
||||
if (yystate() == CLASS2) {
|
||||
yypushback(yylength() - 2);
|
||||
return StringEscapesTokenTypes.INVALID_CHARACTER_ESCAPE_TOKEN;
|
||||
}
|
||||
/* java.util.regex.Pattern says about backrefs:
|
||||
"In this class, \1 through \9 are always interpreted as back references,
|
||||
and a larger number is accepted as a back reference if at least that many
|
||||
subexpressions exist at that point in the regular expression, otherwise the
|
||||
parser will drop digits until the number is smaller or equal to the existing
|
||||
number of groups or it is one digit."
|
||||
*/
|
||||
while (yylength() > 2 && Integer.parseInt(yytext().toString().substring(1)) > capturingGroupCount) {
|
||||
yypushback(1);
|
||||
}
|
||||
return RegExpTT.BACKREF;
|
||||
}
|
||||
}
|
||||
case 154: break;
|
||||
case 60:
|
||||
{ return allowTransformationEscapes ? RegExpTT.CHAR_CLASS : StringEscapesTokenTypes.INVALID_CHARACTER_ESCAPE_TOKEN;
|
||||
{ yypushstate(QUOTED); return RegExpTT.QUOTE_BEGIN;
|
||||
}
|
||||
case 155: break;
|
||||
case 61:
|
||||
{ yypushstate(QUOTED); return RegExpTT.QUOTE_BEGIN;
|
||||
{ return RegExpTT.BAD_HEX_VALUE;
|
||||
}
|
||||
case 156: break;
|
||||
case 62:
|
||||
{ return RegExpTT.BAD_HEX_VALUE;
|
||||
{ return allowTransformationEscapes ? RegExpTT.CHAR_CLASS : StringEscapesTokenTypes.INVALID_UNICODE_ESCAPE_TOKEN;
|
||||
}
|
||||
case 157: break;
|
||||
case 63:
|
||||
{ return allowTransformationEscapes ? RegExpTT.CHAR_CLASS : StringEscapesTokenTypes.INVALID_UNICODE_ESCAPE_TOKEN;
|
||||
{ return (allowOctalNoLeadingZero ? RegExpTT.OCT_CHAR : RegExpTT.BAD_OCT_VALUE);
|
||||
}
|
||||
case 158: break;
|
||||
case 64:
|
||||
@@ -1127,23 +1159,23 @@ class _RegExLexer implements FlexLexer {
|
||||
}
|
||||
case 174: break;
|
||||
case 80:
|
||||
{ return RegExpTT.OCT_CHAR;
|
||||
{ yybegin(NAMED_GROUP); return RegExpTT.RUBY_NAMED_GROUP_CALL;
|
||||
}
|
||||
case 175: break;
|
||||
case 81:
|
||||
{ yybegin(NAMED_GROUP); return RegExpTT.RUBY_NAMED_GROUP_CALL;
|
||||
{ yybegin(QUOTED_NAMED_GROUP); return RegExpTT.RUBY_QUOTED_NAMED_GROUP_CALL;
|
||||
}
|
||||
case 176: break;
|
||||
case 82:
|
||||
{ yybegin(QUOTED_NAMED_GROUP); return RegExpTT.RUBY_QUOTED_NAMED_GROUP_CALL;
|
||||
{ if (xmlSchemaMode) { yypushback(1); return RegExpTT.CHAR_CLASS; } else return RegExpTT.CTRL;
|
||||
}
|
||||
case 177: break;
|
||||
case 83:
|
||||
{ if (xmlSchemaMode) { yypushback(1); return RegExpTT.CHAR_CLASS; } else return RegExpTT.CTRL;
|
||||
{ return StringEscapesTokenTypes.INVALID_UNICODE_ESCAPE_TOKEN;
|
||||
}
|
||||
case 178: break;
|
||||
case 84:
|
||||
{ return StringEscapesTokenTypes.INVALID_UNICODE_ESCAPE_TOKEN;
|
||||
{ return RegExpTT.OCT_CHAR;
|
||||
}
|
||||
case 179: break;
|
||||
case 85:
|
||||
|
||||
@@ -74,5 +74,25 @@ public enum RegExpCapability {
|
||||
/**
|
||||
* \\u, \l, \\U, \L, and \E
|
||||
*/
|
||||
TRANSFORMATION_ESCAPES
|
||||
TRANSFORMATION_ESCAPES,
|
||||
|
||||
/**
|
||||
* \\177 (decimal 127) is maximal octal character
|
||||
*/
|
||||
MAX_OCTAL_177,
|
||||
|
||||
/**
|
||||
* \\377 (decimal 255) is maximal octal character
|
||||
*/
|
||||
MAX_OCTAL_377,
|
||||
|
||||
/**
|
||||
* At least 2 digits needed in octal escape outside character class to be valid (like regexp under ruby)
|
||||
*/
|
||||
MIN_OCTAL_2_DIGITS,
|
||||
|
||||
/**
|
||||
* At least 3 digits needed in octal escape outside character class to be valid (like regexp under python)
|
||||
*/
|
||||
MIN_OCTAL_3_DIGITS,
|
||||
}
|
||||
|
||||
@@ -35,6 +35,8 @@ import java.util.EnumSet;
|
||||
private boolean allowHorizontalWhitespaceClass;
|
||||
private boolean allowPosixBracketExpressions;
|
||||
private boolean allowTransformationEscapes;
|
||||
private int maxOctal = 0777;
|
||||
private int minOctalDigits = 1;
|
||||
|
||||
_RegExLexer(EnumSet<RegExpCapability> capabilities) {
|
||||
this((java.io.Reader)null);
|
||||
@@ -49,6 +51,18 @@ import java.util.EnumSet;
|
||||
this.allowEmptyCharacterClass = capabilities.contains(RegExpCapability.ALLOW_EMPTY_CHARACTER_CLASS);
|
||||
this.allowPosixBracketExpressions = capabilities.contains(RegExpCapability.POSIX_BRACKET_EXPRESSIONS);
|
||||
this.allowTransformationEscapes = capabilities.contains(RegExpCapability.TRANSFORMATION_ESCAPES);
|
||||
if (capabilities.contains(RegExpCapability.MAX_OCTAL_177)) {
|
||||
maxOctal = 0177;
|
||||
}
|
||||
else if (capabilities.contains(RegExpCapability.MAX_OCTAL_377)) {
|
||||
maxOctal = 0377;
|
||||
}
|
||||
if (capabilities.contains(RegExpCapability.MIN_OCTAL_2_DIGITS)) {
|
||||
minOctalDigits = 2;
|
||||
}
|
||||
else if (capabilities.contains(RegExpCapability.MIN_OCTAL_3_DIGITS)) {
|
||||
minOctalDigits = 3;
|
||||
}
|
||||
}
|
||||
|
||||
private void yypushstate(int state) {
|
||||
@@ -90,8 +104,6 @@ import java.util.EnumSet;
|
||||
%xstate PY_COND_REF
|
||||
%xstate BRACKET_EXPRESSION
|
||||
|
||||
DIGITS=[1-9][0-9]*
|
||||
|
||||
DOT="."
|
||||
LPAREN="("
|
||||
RPAREN=")"
|
||||
@@ -151,32 +163,47 @@ HEX_CHAR=[0-9a-fA-F]
|
||||
{ESCAPE} {XML_CLASS} { if (xmlSchemaMode) return RegExpTT.CHAR_CLASS; else return StringEscapesTokenTypes.INVALID_CHARACTER_ESCAPE_TOKEN; }
|
||||
|
||||
|
||||
/* java.util.regex.Pattern says about backrefs:
|
||||
"In this class, \1 through \9 are always interpreted as back references,
|
||||
and a larger number is accepted as a back reference if at least that many
|
||||
subexpressions exist at that point in the regular expression, otherwise the
|
||||
parser will drop digits until the number is smaller or equal to the existing
|
||||
number of groups or it is one digit."
|
||||
*/
|
||||
{ESCAPE} {DIGITS} { if (allowOctalNoLeadingZero) {
|
||||
CharSequence s = yytext();
|
||||
int i = 1;
|
||||
for (; i < s.length(); i++) {
|
||||
if (s.charAt(i) > '7') break;
|
||||
/* 999 back references should be enough for everybody */
|
||||
{ESCAPE} [1-9][0-9]{0,2} { String text = yytext().toString().substring(1);
|
||||
if (allowOctalNoLeadingZero) {
|
||||
if (Integer.parseInt(text) <= capturingGroupCount && yystate() != CLASS2) return RegExpTT.BACKREF;
|
||||
int i = 0;
|
||||
int value = 0;
|
||||
for (; i < text.length(); i++) {
|
||||
char c = text.charAt(i);
|
||||
if (c > '7') break;
|
||||
value = value * 8 + (c - '0');
|
||||
}
|
||||
if (i > 1) {
|
||||
yypushback(yylength() - i);
|
||||
if (i > 0) {
|
||||
yypushback(text.length() - i);
|
||||
if (value > maxOctal) {
|
||||
yypushback(1);
|
||||
return RegExpTT.BAD_OCT_VALUE;
|
||||
}
|
||||
if (minOctalDigits > i && yystate() != CLASS2) {
|
||||
return RegExpTT.BAD_OCT_VALUE;
|
||||
}
|
||||
return RegExpTT.OCT_CHAR;
|
||||
}
|
||||
return StringEscapesTokenTypes.INVALID_CHARACTER_ESCAPE_TOKEN;
|
||||
}
|
||||
if (yystate() == CLASS2) {
|
||||
yypushback(yylength() - 2);
|
||||
return RegExpTT.REDUNDANT_ESCAPE;
|
||||
else {
|
||||
if (yystate() == CLASS2) {
|
||||
yypushback(yylength() - 2);
|
||||
return StringEscapesTokenTypes.INVALID_CHARACTER_ESCAPE_TOKEN;
|
||||
}
|
||||
/* java.util.regex.Pattern says about backrefs:
|
||||
"In this class, \1 through \9 are always interpreted as back references,
|
||||
and a larger number is accepted as a back reference if at least that many
|
||||
subexpressions exist at that point in the regular expression, otherwise the
|
||||
parser will drop digits until the number is smaller or equal to the existing
|
||||
number of groups or it is one digit."
|
||||
*/
|
||||
while (yylength() > 2 && Integer.parseInt(yytext().toString().substring(1)) > capturingGroupCount) {
|
||||
yypushback(1);
|
||||
}
|
||||
return RegExpTT.BACKREF;
|
||||
}
|
||||
while (yylength() > 2 && Integer.parseInt(yytext().toString().substring(1)) > capturingGroupCount) {
|
||||
yypushback(1);
|
||||
}
|
||||
return RegExpTT.BACKREF;
|
||||
}
|
||||
|
||||
{ESCAPE} "-" { return (yystate() == CLASS2) ? RegExpTT.ESC_CHARACTER : RegExpTT.REDUNDANT_ESCAPE; }
|
||||
|
||||
@@ -20,6 +20,8 @@ import com.intellij.testFramework.LexerTestCase;
|
||||
|
||||
import java.util.EnumSet;
|
||||
|
||||
import static org.intellij.lang.regexp.RegExpCapability.*;
|
||||
|
||||
/**
|
||||
* @author Bas Leijdekkers
|
||||
*/
|
||||
@@ -44,8 +46,8 @@ public class RegExpLexerTest extends LexerTestCase {
|
||||
}
|
||||
|
||||
public void testEditorReplacement() {
|
||||
RegExpLexer lexer = new RegExpLexer(EnumSet.of(RegExpCapability.TRANSFORMATION_ESCAPES));
|
||||
String text = "\\U$1\\E\\u$3\\l$4\\L$2\\E";
|
||||
RegExpLexer lexer = new RegExpLexer(EnumSet.of(TRANSFORMATION_ESCAPES));
|
||||
final String text = "\\U$1\\E\\u$3\\l$4\\L$2\\E";
|
||||
doTest(text, "CHAR_CLASS ('\\U')\n" +
|
||||
"DOLLAR ('$')\n" +
|
||||
"CHARACTER ('1')\n" +
|
||||
@@ -80,7 +82,7 @@ public class RegExpLexerTest extends LexerTestCase {
|
||||
}
|
||||
|
||||
public void testIntersection() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(RegExpCapability.NESTED_CHARACTER_CLASSES));
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(NESTED_CHARACTER_CLASSES));
|
||||
doTest("[a&&]", "CLASS_BEGIN ('[')\n" +
|
||||
"CHARACTER ('a')\n" +
|
||||
"ANDAND ('&&')\n" +
|
||||
@@ -98,7 +100,7 @@ public class RegExpLexerTest extends LexerTestCase {
|
||||
}
|
||||
|
||||
public void testPosixBracketExpression() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(RegExpCapability.POSIX_BRACKET_EXPRESSIONS));
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(POSIX_BRACKET_EXPRESSIONS));
|
||||
doTest("[[:xdigit:]]", "CLASS_BEGIN ('[')\n" +
|
||||
"BRACKET_EXPRESSION_BEGIN ('[:')\n" +
|
||||
"NAME ('xdigit')\n" +
|
||||
@@ -107,7 +109,7 @@ public class RegExpLexerTest extends LexerTestCase {
|
||||
}
|
||||
|
||||
public void testNegatedPosixBracketExpression() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(RegExpCapability.POSIX_BRACKET_EXPRESSIONS));
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(POSIX_BRACKET_EXPRESSIONS));
|
||||
doTest("[[:^xdigit:]]", "CLASS_BEGIN ('[')\n" +
|
||||
"BRACKET_EXPRESSION_BEGIN ('[:')\n" +
|
||||
"CARET ('^')\n" +
|
||||
@@ -116,17 +118,142 @@ public class RegExpLexerTest extends LexerTestCase {
|
||||
"CLASS_END (']')", lexer);
|
||||
}
|
||||
|
||||
public void testOctalWithoutLeadingZero() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(RegExpCapability.OCTAL_NO_LEADING_ZERO));
|
||||
doTest("\\0\\123[\\123]", "OCT_CHAR ('\\0')\n" +
|
||||
"OCT_CHAR ('\\123')\n" +
|
||||
"CLASS_BEGIN ('[')\n" +
|
||||
"OCT_CHAR ('\\123')\n" +
|
||||
"CLASS_END (']')", lexer);
|
||||
/**
|
||||
* \\177 is the maximum valid octal character under Ruby.
|
||||
*/
|
||||
public void testMaxOctalNoLeadingZero1() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(OCTAL_NO_LEADING_ZERO, MAX_OCTAL_177));
|
||||
doTest("\\177\\200", "OCT_CHAR ('\\177')\n" +
|
||||
"BAD_OCT_VALUE ('\\20')\n" +
|
||||
"CHARACTER ('0')", lexer);
|
||||
}
|
||||
|
||||
/**
|
||||
* \\377 is the maximum valid octal character under javascript. \\400 is interpreted as \\40 followed by a 0 character.
|
||||
* The BAD_OCT_VALUE token is converted to OCT_CHAR in com.intellij.lang.javascript.inject.JSRegexpParserDefinition
|
||||
*/
|
||||
public void testMaxOctalNoLeadingZero2() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(OCTAL_NO_LEADING_ZERO, MAX_OCTAL_377));
|
||||
doTest("\\177\\200\\377\\400", "OCT_CHAR ('\\177')\n" +
|
||||
"OCT_CHAR ('\\200')\n" +
|
||||
"OCT_CHAR ('\\377')\n" +
|
||||
"BAD_OCT_VALUE ('\\40')\n" +
|
||||
"CHARACTER ('0')", lexer);
|
||||
}
|
||||
|
||||
/**
|
||||
* \\777 is valid octal character in python regex dialect.
|
||||
*/
|
||||
public void testMaxOctalNoLeadingZero3() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(OCTAL_NO_LEADING_ZERO));
|
||||
doTest("\\177\\200\\377\\400\\777", "OCT_CHAR ('\\177')\n" +
|
||||
"OCT_CHAR ('\\200')\n" +
|
||||
"OCT_CHAR ('\\377')\n" +
|
||||
"OCT_CHAR ('\\400')\n" +
|
||||
"OCT_CHAR ('\\777')", lexer);
|
||||
}
|
||||
|
||||
/**
|
||||
* \\1 and \\11 valid under js, both inside and outside character class
|
||||
*/
|
||||
public void testOctalNoLeadingZero1() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(OCTAL_NO_LEADING_ZERO));
|
||||
doTest("\\1()\\1\\11[\\1\\11]", "OCT_CHAR ('\\1')\n" +
|
||||
"GROUP_BEGIN ('(')\n" +
|
||||
"GROUP_END (')')\n" +
|
||||
"BACKREF ('\\1')\n" +
|
||||
"OCT_CHAR ('\\11')\n" +
|
||||
"CLASS_BEGIN ('[')\n" +
|
||||
"OCT_CHAR ('\\1')\n" +
|
||||
"OCT_CHAR ('\\11')\n" +
|
||||
"CLASS_END (']')", lexer);
|
||||
}
|
||||
|
||||
/**
|
||||
* \\1 not valid and \\11 valid under ruby, outside character class
|
||||
*/
|
||||
public void testOctalNoLeadingZero2() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(OCTAL_NO_LEADING_ZERO, MIN_OCTAL_2_DIGITS));
|
||||
doTest("\\1()\\1\\11[\\1\\11]", "BAD_OCT_VALUE ('\\1')\n" +
|
||||
"GROUP_BEGIN ('(')\n" +
|
||||
"GROUP_END (')')\n" +
|
||||
"BACKREF ('\\1')\n" +
|
||||
"OCT_CHAR ('\\11')\n" +
|
||||
"CLASS_BEGIN ('[')\n" +
|
||||
"OCT_CHAR ('\\1')\n" +
|
||||
"OCT_CHAR ('\\11')\n" +
|
||||
"CLASS_END (']')", lexer);
|
||||
}
|
||||
|
||||
/**
|
||||
* \\1 and \\11 not valid under python, outside character class
|
||||
*/
|
||||
public void testOctalNoLeadingZero3() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(OCTAL_NO_LEADING_ZERO, MIN_OCTAL_3_DIGITS));
|
||||
doTest("\\1()\\1\\11\\111[\\1\\11\\111]", "BAD_OCT_VALUE ('\\1')\n" +
|
||||
"GROUP_BEGIN ('(')\n" +
|
||||
"GROUP_END (')')\n" +
|
||||
"BACKREF ('\\1')\n" +
|
||||
"BAD_OCT_VALUE ('\\11')\n" +
|
||||
"OCT_CHAR ('\\111')\n" +
|
||||
"CLASS_BEGIN ('[')\n" +
|
||||
"OCT_CHAR ('\\1')\n" +
|
||||
"OCT_CHAR ('\\11')\n" +
|
||||
"OCT_CHAR ('\\111')\n" +
|
||||
"CLASS_END (']')", lexer);
|
||||
}
|
||||
|
||||
|
||||
/** octal is never a back reference inside a character class, valid under js, ruby, python */
|
||||
public void testOctalInsideCharClass() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(OCTAL_NO_LEADING_ZERO));
|
||||
doTest("()()()()()()()()()()[\\1\\10\\100]", "GROUP_BEGIN ('(')\nGROUP_END (')')\n" +
|
||||
"GROUP_BEGIN ('(')\nGROUP_END (')')\n" +
|
||||
"GROUP_BEGIN ('(')\nGROUP_END (')')\n" +
|
||||
"GROUP_BEGIN ('(')\nGROUP_END (')')\n" +
|
||||
"GROUP_BEGIN ('(')\nGROUP_END (')')\n" +
|
||||
"GROUP_BEGIN ('(')\nGROUP_END (')')\n" +
|
||||
"GROUP_BEGIN ('(')\nGROUP_END (')')\n" +
|
||||
"GROUP_BEGIN ('(')\nGROUP_END (')')\n" +
|
||||
"GROUP_BEGIN ('(')\nGROUP_END (')')\n" +
|
||||
"GROUP_BEGIN ('(')\nGROUP_END (')')\n" +
|
||||
"CLASS_BEGIN ('[')\n" +
|
||||
"OCT_CHAR ('\\1')\n" +
|
||||
"OCT_CHAR ('\\10')\n" +
|
||||
"OCT_CHAR ('\\100')\n" +
|
||||
"CLASS_END (']')", lexer);
|
||||
}
|
||||
|
||||
/** \0 always valid under js, ruby, python regex dialects, never a back reference. */
|
||||
public void testZeroOctalNoLeadingZero() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(OCTAL_NO_LEADING_ZERO));
|
||||
doTest("\\0()\\0[\\0]", "OCT_CHAR ('\\0')\n" +
|
||||
"GROUP_BEGIN ('(')\n" +
|
||||
"GROUP_END (')')\n" +
|
||||
"OCT_CHAR ('\\0')\n" +
|
||||
"CLASS_BEGIN ('[')\n" +
|
||||
"OCT_CHAR ('\\0')\n" +
|
||||
"CLASS_END (']')", lexer);
|
||||
}
|
||||
|
||||
/** three digit octal (\100) always valid, either octal or backreference under js, ruby and python */
|
||||
public void testThreeDigitOctalNoLeadingZero() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(OCTAL_NO_LEADING_ZERO));
|
||||
doTest("\\100" +
|
||||
"()()()()()()()()()()" +
|
||||
"()()()()()()()()()()" +
|
||||
"()()()()()()()()()()" +
|
||||
"()()()()()()()()()()" +
|
||||
"()()()()()()()()()()" +
|
||||
"()()()()()()()()()()" +
|
||||
"()()()()()()()()()()" +
|
||||
"()()()()()()()()()()" +
|
||||
"()()()()()()()()()()" +
|
||||
"()()()()()()()()()()\\100[\\100]", null, lexer);
|
||||
}
|
||||
|
||||
public void testOctalFollowedByDigit() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(RegExpCapability.OCTAL_NO_LEADING_ZERO));
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(OCTAL_NO_LEADING_ZERO));
|
||||
doTest("\\39[\\39]", "OCT_CHAR ('\\3')\n" +
|
||||
"CHARACTER ('9')\n" +
|
||||
"CLASS_BEGIN ('[')\n" +
|
||||
@@ -142,7 +269,7 @@ public class RegExpLexerTest extends LexerTestCase {
|
||||
"CHARACTER ('2')\n" +
|
||||
"CHARACTER ('3')\n" +
|
||||
"CLASS_BEGIN ('[')\n" +
|
||||
"REDUNDANT_ESCAPE ('\\1')\n" +
|
||||
"INVALID_CHARACTER_ESCAPE_TOKEN ('\\1')\n" +
|
||||
"CHARACTER ('2')\n" +
|
||||
"CHARACTER ('3')\n" +
|
||||
"CLASS_END (']')", lexer);
|
||||
@@ -150,14 +277,20 @@ public class RegExpLexerTest extends LexerTestCase {
|
||||
|
||||
public void testOctalWithLeadingZero2() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.noneOf(RegExpCapability.class));
|
||||
doTest("\\08\\01\\012\\0123\\0377\\0400", "BAD_OCT_VALUE ('\\0')\n" +
|
||||
"CHARACTER ('8')\n" +
|
||||
"OCT_CHAR ('\\01')\n" +
|
||||
"OCT_CHAR ('\\012')\n" +
|
||||
"OCT_CHAR ('\\0123')\n" +
|
||||
"OCT_CHAR ('\\0377')\n" +
|
||||
"OCT_CHAR ('\\040')\n" +
|
||||
"CHARACTER ('0')", lexer);
|
||||
doTest("\\08\\01\\00\\012\\0123\\0377\\0400", "BAD_OCT_VALUE ('\\0')\n" +
|
||||
"CHARACTER ('8')\n" +
|
||||
"OCT_CHAR ('\\01')\n" +
|
||||
"OCT_CHAR ('\\00')\n" +
|
||||
"OCT_CHAR ('\\012')\n" +
|
||||
"OCT_CHAR ('\\0123')\n" +
|
||||
"OCT_CHAR ('\\0377')\n" +
|
||||
"OCT_CHAR ('\\040')\n" +
|
||||
"CHARACTER ('0')", lexer);
|
||||
}
|
||||
|
||||
public void testBackReference() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.noneOf(RegExpCapability.class));
|
||||
doTest("(a)(b)(c)(d)(e)(f)(g)(h)(i)(j)\\105", null, lexer);
|
||||
}
|
||||
|
||||
public void testNoNestedCharacterClasses1() {
|
||||
@@ -188,7 +321,7 @@ public class RegExpLexerTest extends LexerTestCase {
|
||||
}
|
||||
|
||||
public void testNestedCharacterClasses1() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(RegExpCapability.NESTED_CHARACTER_CLASSES));
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(NESTED_CHARACTER_CLASSES));
|
||||
doTest("[a-z&&[^aeuoi]]", "CLASS_BEGIN ('[')\n" +
|
||||
"CHARACTER ('a')\n" +
|
||||
"MINUS ('-')\n" +
|
||||
@@ -206,7 +339,7 @@ public class RegExpLexerTest extends LexerTestCase {
|
||||
}
|
||||
|
||||
public void testNestedCharacterClasses2() {
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(RegExpCapability.NESTED_CHARACTER_CLASSES));
|
||||
final RegExpLexer lexer = new RegExpLexer(EnumSet.of(NESTED_CHARACTER_CLASSES));
|
||||
doTest("[]]", "CLASS_BEGIN ('[')\n" +
|
||||
"CHARACTER (']')\n" +
|
||||
"CLASS_END (']')", lexer);
|
||||
@@ -219,6 +352,6 @@ public class RegExpLexerTest extends LexerTestCase {
|
||||
|
||||
@Override
|
||||
protected String getDirPath() {
|
||||
return "";
|
||||
return "community/RegExpSupport/testData/lexer";
|
||||
}
|
||||
}
|
||||
|
||||
32
RegExpSupport/testData/lexer/backReference.txt
Normal file
32
RegExpSupport/testData/lexer/backReference.txt
Normal file
@@ -0,0 +1,32 @@
|
||||
GROUP_BEGIN ('(')
|
||||
CHARACTER ('a')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
CHARACTER ('b')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
CHARACTER ('c')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
CHARACTER ('d')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
CHARACTER ('e')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
CHARACTER ('f')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
CHARACTER ('g')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
CHARACTER ('h')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
CHARACTER ('i')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
CHARACTER ('j')
|
||||
GROUP_END (')')
|
||||
BACKREF ('\10')
|
||||
CHARACTER ('5')
|
||||
205
RegExpSupport/testData/lexer/threeDigitOctalNoLeadingZero.txt
Normal file
205
RegExpSupport/testData/lexer/threeDigitOctalNoLeadingZero.txt
Normal file
@@ -0,0 +1,205 @@
|
||||
OCT_CHAR ('\100')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
GROUP_BEGIN ('(')
|
||||
GROUP_END (')')
|
||||
BACKREF ('\100')
|
||||
CLASS_BEGIN ('[')
|
||||
OCT_CHAR ('\100')
|
||||
CLASS_END (']')
|
||||
@@ -33,8 +33,8 @@ public class PythonRegexpParserDefinition extends RegExpParserDefinition {
|
||||
public static final IFileElementType PYTHON_REGEXP_FILE = new IFileElementType("PYTHON_REGEXP_FILE", PythonRegexpLanguage.INSTANCE);
|
||||
protected final EnumSet<RegExpCapability> CAPABILITIES = EnumSet.of(RegExpCapability.DANGLING_METACHARACTERS,
|
||||
RegExpCapability.OCTAL_NO_LEADING_ZERO,
|
||||
RegExpCapability.OMIT_NUMBERS_IN_QUANTIFIERS);
|
||||
|
||||
RegExpCapability.OMIT_NUMBERS_IN_QUANTIFIERS,
|
||||
RegExpCapability.MIN_OCTAL_3_DIGITS);
|
||||
@NotNull
|
||||
public Lexer createLexer(Project project) {
|
||||
return new RegExpLexer(CAPABILITIES);
|
||||
|
||||
Reference in New Issue
Block a user