From 80ea3d052a019d119156832aaaf436e08e461754 Mon Sep 17 00:00:00 2001 From: Dmitry Jemerov Date: Sun, 21 Feb 2010 21:18:21 +0300 Subject: [PATCH] support for Python named groups in regexp lexer and parser --- .../lang/regexp/RegExpBraceMatcher.java | 1 + .../lang/regexp/RegExpLanguageHost.java | 1 + .../org/intellij/lang/regexp/RegExpTT.java | 6 +- .../org/intellij/lang/regexp/_RegExLexer.java | 419 +++++++++--------- .../intellij/lang/regexp/psi/RegExpGroup.java | 2 + .../lang/regexp/psi/impl/RegExpGroupImpl.java | 4 + .../intellij/lang/regexp/regexp-lexer.flex | 1 + .../regexp/validation/RegExpAnnotator.java | 9 +- 8 files changed, 234 insertions(+), 209 deletions(-) diff --git a/RegExpSupport/src/org/intellij/lang/regexp/RegExpBraceMatcher.java b/RegExpSupport/src/org/intellij/lang/regexp/RegExpBraceMatcher.java index e51b63fa22ff..387c548c4d21 100644 --- a/RegExpSupport/src/org/intellij/lang/regexp/RegExpBraceMatcher.java +++ b/RegExpSupport/src/org/intellij/lang/regexp/RegExpBraceMatcher.java @@ -28,6 +28,7 @@ public class RegExpBraceMatcher implements PairedBraceMatcher { new BracePair(RegExpTT.SET_OPTIONS, RegExpTT.GROUP_END, true), new BracePair(RegExpTT.NON_CAPT_GROUP, RegExpTT.GROUP_END, true), new BracePair(RegExpTT.POS_LOOKAHEAD, RegExpTT.GROUP_END, true), new BracePair(RegExpTT.NEG_LOOKAHEAD, RegExpTT.GROUP_END, true), new BracePair(RegExpTT.POS_LOOKBEHIND, RegExpTT.GROUP_END, true), new BracePair(RegExpTT.NEG_LOOKBEHIND, RegExpTT.GROUP_END, true), + new BracePair(RegExpTT.PYTHON_NAMED_GROUP, RegExpTT.GROUP_END, true), new BracePair(RegExpTT.CLASS_BEGIN, RegExpTT.CLASS_END, false), new BracePair(RegExpTT.LBRACE, RegExpTT.RBRACE, false), new BracePair(RegExpTT.QUOTE_BEGIN, RegExpTT.QUOTE_END, false),}; } diff --git a/RegExpSupport/src/org/intellij/lang/regexp/RegExpLanguageHost.java b/RegExpSupport/src/org/intellij/lang/regexp/RegExpLanguageHost.java index f1d344564c22..a3251e6d1127 100644 --- a/RegExpSupport/src/org/intellij/lang/regexp/RegExpLanguageHost.java +++ b/RegExpSupport/src/org/intellij/lang/regexp/RegExpLanguageHost.java @@ -22,4 +22,5 @@ public interface RegExpLanguageHost { boolean characterNeedsEscaping(char c); boolean supportsPerl5EmbeddedComments(); boolean supportsPossessiveQuantifiers(); + boolean supportsPythonNamedGroups(); } diff --git a/RegExpSupport/src/org/intellij/lang/regexp/RegExpTT.java b/RegExpSupport/src/org/intellij/lang/regexp/RegExpTT.java index 59eafbca65d7..54f03f7bf886 100644 --- a/RegExpSupport/src/org/intellij/lang/regexp/RegExpTT.java +++ b/RegExpSupport/src/org/intellij/lang/regexp/RegExpTT.java @@ -97,6 +97,9 @@ public interface RegExpTT { IElementType OPTIONS_ON = new RegExpElementType("OPTIONS_ON"); IElementType OPTIONS_OFF = new RegExpElementType("OPTIONS_OFF"); + // (?P... + IElementType PYTHON_NAMED_GROUP = new RegExpElementType("PYTHON_NAMED_GROUP"); + TokenSet KEYWORDS = TokenSet.create(DOT, STAR, QUEST, PLUS); TokenSet CHARACTERS = TokenSet.create(CHARACTER, @@ -123,7 +126,8 @@ public interface RegExpTT { POS_LOOKAHEAD, NEG_LOOKAHEAD, POS_LOOKBEHIND, - NEG_LOOKBEHIND); + NEG_LOOKBEHIND, + PYTHON_NAMED_GROUP); TokenSet BOUNDARIES = TokenSet.create(BOUNDARY, CARET, DOLLAR); } diff --git a/RegExpSupport/src/org/intellij/lang/regexp/_RegExLexer.java b/RegExpSupport/src/org/intellij/lang/regexp/_RegExLexer.java index 8de1c26644c5..a4126fd94117 100644 --- a/RegExpSupport/src/org/intellij/lang/regexp/_RegExLexer.java +++ b/RegExpSupport/src/org/intellij/lang/regexp/_RegExLexer.java @@ -1,21 +1,20 @@ -/* The following code was generated by JFlex 1.4.1 on 18.12.09 20:27 */ +/* The following code was generated by JFlex 1.4.3 on 21.02.10 20:59 */ /* It's an automatically generated code. Do not modify it. */ package org.intellij.lang.regexp; import com.intellij.lexer.FlexLexer; -import com.intellij.psi.StringEscapesTokenTypes; import com.intellij.psi.tree.IElementType; - import java.util.LinkedList; +import com.intellij.psi.StringEscapesTokenTypes; // IDEADEV-11055 @SuppressWarnings({ "ALL", "SameParameterValue", "WeakerAccess", "SameReturnValue", "RedundantThrows", "UnusedDeclaration", "UnusedDeclaration" }) /** * This class is a scanner generated by - * JFlex 1.4.1 - * on 18.12.09 20:27 from the specification file + * JFlex 1.4.3 + * on 21.02.10 20:59 from the specification file * C:/JetBrains/IDEA/tools/lexer/../../community/RegExpSupport/src/org/intellij/lang/regexp/regexp-lexer.flex */ class _RegExLexer implements FlexLexer { @@ -23,86 +22,96 @@ class _RegExLexer implements FlexLexer { private static final int ZZ_BUFFERSIZE = 16384; /** lexical states */ - public static final int CLASS1 = 3; - public static final int PROP = 5; - public static final int EMBRACED = 2; - public static final int QUOTED = 1; + public static final int CLASS1 = 6; + public static final int PROP = 10; + public static final int EMBRACED = 4; + public static final int QUOTED = 2; public static final int YYINITIAL = 0; - public static final int COMMENT = 7; - public static final int OPTIONS = 6; - public static final int CLASS2 = 4; + public static final int COMMENT = 14; + public static final int OPTIONS = 12; + public static final int CLASS2 = 8; + + /** + * ZZ_LEXSTATE[l] is the state in the DFA for the lexical state l + * ZZ_LEXSTATE[l+1] is the state in the DFA for the lexical state l + * at the beginning of a line + * l is of the form l = 2*k, k a non negative integer + */ + private static final int ZZ_LEXSTATE[] = { + 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7 + }; /** * Translates characters to character classes */ private static final String ZZ_CMAP_PACKED = - "\10\0\2\51\1\13\1\0\1\51\1\60\22\0\1\44\1\55\1\0"+ - "\1\57\1\15\1\0\1\50\1\0\1\4\1\5\1\17\1\20\1\47"+ - "\1\42\1\3\1\0\1\2\7\41\2\1\1\52\1\0\1\56\1\54"+ - "\1\53\1\16\1\0\1\24\1\24\1\30\1\27\1\36\1\34\1\25"+ - "\1\43\1\32\6\43\1\33\1\35\1\43\1\26\3\43\1\26\1\26"+ - "\1\43\1\25\1\10\1\12\1\11\1\14\1\45\1\0\1\23\1\24"+ - "\1\31\1\27\1\23\1\23\2\43\1\32\4\43\1\22\1\43\1\33"+ - "\1\43\1\22\1\26\1\22\1\40\1\43\1\26\1\37\1\43\1\25"+ - "\1\6\1\21\1\7\54\0\1\43\12\0\1\43\4\0\1\43\5\0"+ - "\27\43\1\0\37\43\1\0\u013f\43\31\0\162\43\4\0\14\43\16\0"+ - "\5\43\11\0\1\43\213\0\1\43\13\0\1\43\1\0\3\43\1\0"+ - "\1\43\1\0\24\43\1\0\54\43\1\0\46\43\1\0\5\43\4\0"+ - "\202\43\10\0\105\43\1\0\46\43\2\0\2\43\6\0\20\43\41\0"+ - "\46\43\2\0\1\43\7\0\47\43\110\0\33\43\5\0\3\43\56\0"+ - "\32\43\5\0\13\43\25\0\12\46\4\0\2\43\1\0\143\43\1\0"+ - "\1\43\17\0\2\43\7\0\2\43\12\46\3\43\2\0\1\43\20\0"+ - "\1\43\1\0\36\43\35\0\3\43\60\0\46\43\13\0\1\43\u0152\0"+ - "\66\43\3\0\1\43\22\0\1\43\7\0\12\43\4\0\12\46\25\0"+ - "\10\43\2\0\2\43\2\0\26\43\1\0\7\43\1\0\1\43\3\0"+ - "\4\43\3\0\1\43\36\0\2\43\1\0\3\43\4\0\12\46\2\43"+ - "\23\0\6\43\4\0\2\43\2\0\26\43\1\0\7\43\1\0\2\43"+ - "\1\0\2\43\1\0\2\43\37\0\4\43\1\0\1\43\7\0\12\46"+ - "\2\0\3\43\20\0\11\43\1\0\3\43\1\0\26\43\1\0\7\43"+ - "\1\0\2\43\1\0\5\43\3\0\1\43\22\0\1\43\17\0\2\43"+ - "\4\0\12\46\25\0\10\43\2\0\2\43\2\0\26\43\1\0\7\43"+ - "\1\0\2\43\1\0\5\43\3\0\1\43\36\0\2\43\1\0\3\43"+ - "\4\0\12\46\1\0\1\43\21\0\1\43\1\0\6\43\3\0\3\43"+ - "\1\0\4\43\3\0\2\43\1\0\1\43\1\0\2\43\3\0\2\43"+ - "\3\0\3\43\3\0\10\43\1\0\3\43\55\0\11\46\25\0\10\43"+ - "\1\0\3\43\1\0\27\43\1\0\12\43\1\0\5\43\46\0\2\43"+ - "\4\0\12\46\25\0\10\43\1\0\3\43\1\0\27\43\1\0\12\43"+ - "\1\0\5\43\3\0\1\43\40\0\1\43\1\0\2\43\4\0\12\46"+ - "\25\0\10\43\1\0\3\43\1\0\27\43\1\0\20\43\46\0\2\43"+ - "\4\0\12\46\25\0\22\43\3\0\30\43\1\0\11\43\1\0\1\43"+ - "\2\0\7\43\72\0\60\43\1\0\2\43\14\0\7\43\11\0\12\46"+ - "\47\0\2\43\1\0\1\43\2\0\2\43\1\0\1\43\2\0\1\43"+ - "\6\0\4\43\1\0\7\43\1\0\3\43\1\0\1\43\1\0\1\43"+ - "\2\0\2\43\1\0\4\43\1\0\2\43\11\0\1\43\2\0\5\43"+ - "\1\0\1\43\11\0\12\46\2\0\2\43\42\0\1\43\37\0\12\46"+ - "\26\0\10\43\1\0\42\43\35\0\4\43\164\0\42\43\1\0\5\43"+ - "\1\0\2\43\25\0\12\46\6\0\6\43\112\0\46\43\12\0\51\43"+ - "\7\0\132\43\5\0\104\43\5\0\122\43\6\0\7\43\1\0\77\43"+ - "\1\0\1\43\1\0\4\43\2\0\7\43\1\0\1\43\1\0\4\43"+ - "\2\0\47\43\1\0\1\43\1\0\4\43\2\0\37\43\1\0\1\43"+ - "\1\0\4\43\2\0\7\43\1\0\1\43\1\0\4\43\2\0\7\43"+ - "\1\0\7\43\1\0\27\43\1\0\37\43\1\0\1\43\1\0\4\43"+ - "\2\0\7\43\1\0\47\43\1\0\23\43\16\0\11\46\56\0\125\43"+ - "\14\0\u026c\43\2\0\10\43\12\0\32\43\5\0\113\43\25\0\15\43"+ - "\1\0\4\43\16\0\22\43\16\0\22\43\16\0\15\43\1\0\3\43"+ - "\17\0\64\43\43\0\1\43\4\0\1\43\3\0\12\46\46\0\12\46"+ - "\6\0\130\43\10\0\51\43\127\0\35\43\51\0\12\46\36\43\2\0"+ - "\5\43\u038b\0\154\43\224\0\234\43\4\0\132\43\6\0\26\43\2\0"+ - "\6\43\2\0\46\43\2\0\6\43\2\0\10\43\1\0\1\43\1\0"+ - "\1\43\1\0\1\43\1\0\37\43\2\0\65\43\1\0\7\43\1\0"+ - "\1\43\3\0\3\43\1\0\7\43\3\0\4\43\2\0\6\43\4\0"+ - "\15\43\5\0\3\43\1\0\7\43\164\0\1\43\15\0\1\43\202\0"+ - "\1\43\4\0\1\43\2\0\12\43\1\0\1\43\3\0\5\43\6\0"+ - "\1\43\1\0\1\43\1\0\1\43\1\0\4\43\1\0\3\43\1\0"+ - "\7\43\3\0\3\43\5\0\5\43\u0ebb\0\2\43\52\0\5\43\5\0"+ - "\2\43\4\0\126\43\6\0\3\43\1\0\132\43\1\0\4\43\5\0"+ - "\50\43\4\0\136\43\21\0\30\43\70\0\20\43\u0200\0\u19b6\43\112\0"+ - "\u51a6\43\132\0\u048d\43\u0773\0\u2ba4\43\u215c\0\u012e\43\2\0\73\43\225\0"+ - "\7\43\14\0\5\43\5\0\1\43\1\0\12\43\1\0\15\43\1\0"+ - "\5\43\1\0\1\43\1\0\2\43\1\0\2\43\1\0\154\43\41\0"+ - "\u016b\43\22\0\100\43\2\0\66\43\50\0\14\43\164\0\5\43\1\0"+ - "\207\43\23\0\12\46\7\0\32\43\6\0\32\43\13\0\131\43\3\0"+ - "\6\43\2\0\6\43\2\0\6\43\2\0\3\43\43\0"; + "\10\0\2\52\1\13\1\0\1\52\1\61\22\0\1\45\1\56\1\0"+ + "\1\60\1\15\1\0\1\51\1\0\1\4\1\5\1\17\1\20\1\50"+ + "\1\43\1\3\1\0\1\2\7\42\2\1\1\53\1\0\1\57\1\55"+ + "\1\54\1\16\1\0\1\24\1\24\1\30\1\27\1\37\1\35\1\25"+ + "\1\44\1\32\6\44\1\34\1\36\1\44\1\26\3\44\1\26\1\26"+ + "\1\44\1\25\1\10\1\12\1\11\1\14\1\46\1\0\1\23\1\24"+ + "\1\31\1\27\1\23\1\23\2\44\1\32\4\44\1\22\1\44\1\33"+ + "\1\44\1\22\1\26\1\22\1\41\1\44\1\26\1\40\1\44\1\25"+ + "\1\6\1\21\1\7\54\0\1\44\12\0\1\44\4\0\1\44\5\0"+ + "\27\44\1\0\37\44\1\0\u013f\44\31\0\162\44\4\0\14\44\16\0"+ + "\5\44\11\0\1\44\213\0\1\44\13\0\1\44\1\0\3\44\1\0"+ + "\1\44\1\0\24\44\1\0\54\44\1\0\46\44\1\0\5\44\4\0"+ + "\202\44\10\0\105\44\1\0\46\44\2\0\2\44\6\0\20\44\41\0"+ + "\46\44\2\0\1\44\7\0\47\44\110\0\33\44\5\0\3\44\56\0"+ + "\32\44\5\0\13\44\25\0\12\47\4\0\2\44\1\0\143\44\1\0"+ + "\1\44\17\0\2\44\7\0\2\44\12\47\3\44\2\0\1\44\20\0"+ + "\1\44\1\0\36\44\35\0\3\44\60\0\46\44\13\0\1\44\u0152\0"+ + "\66\44\3\0\1\44\22\0\1\44\7\0\12\44\4\0\12\47\25\0"+ + "\10\44\2\0\2\44\2\0\26\44\1\0\7\44\1\0\1\44\3\0"+ + "\4\44\3\0\1\44\36\0\2\44\1\0\3\44\4\0\12\47\2\44"+ + "\23\0\6\44\4\0\2\44\2\0\26\44\1\0\7\44\1\0\2\44"+ + "\1\0\2\44\1\0\2\44\37\0\4\44\1\0\1\44\7\0\12\47"+ + "\2\0\3\44\20\0\11\44\1\0\3\44\1\0\26\44\1\0\7\44"+ + "\1\0\2\44\1\0\5\44\3\0\1\44\22\0\1\44\17\0\2\44"+ + "\4\0\12\47\25\0\10\44\2\0\2\44\2\0\26\44\1\0\7\44"+ + "\1\0\2\44\1\0\5\44\3\0\1\44\36\0\2\44\1\0\3\44"+ + "\4\0\12\47\1\0\1\44\21\0\1\44\1\0\6\44\3\0\3\44"+ + "\1\0\4\44\3\0\2\44\1\0\1\44\1\0\2\44\3\0\2\44"+ + "\3\0\3\44\3\0\10\44\1\0\3\44\55\0\11\47\25\0\10\44"+ + "\1\0\3\44\1\0\27\44\1\0\12\44\1\0\5\44\46\0\2\44"+ + "\4\0\12\47\25\0\10\44\1\0\3\44\1\0\27\44\1\0\12\44"+ + "\1\0\5\44\3\0\1\44\40\0\1\44\1\0\2\44\4\0\12\47"+ + "\25\0\10\44\1\0\3\44\1\0\27\44\1\0\20\44\46\0\2\44"+ + "\4\0\12\47\25\0\22\44\3\0\30\44\1\0\11\44\1\0\1\44"+ + "\2\0\7\44\72\0\60\44\1\0\2\44\14\0\7\44\11\0\12\47"+ + "\47\0\2\44\1\0\1\44\2\0\2\44\1\0\1\44\2\0\1\44"+ + "\6\0\4\44\1\0\7\44\1\0\3\44\1\0\1\44\1\0\1\44"+ + "\2\0\2\44\1\0\4\44\1\0\2\44\11\0\1\44\2\0\5\44"+ + "\1\0\1\44\11\0\12\47\2\0\2\44\42\0\1\44\37\0\12\47"+ + "\26\0\10\44\1\0\42\44\35\0\4\44\164\0\42\44\1\0\5\44"+ + "\1\0\2\44\25\0\12\47\6\0\6\44\112\0\46\44\12\0\51\44"+ + "\7\0\132\44\5\0\104\44\5\0\122\44\6\0\7\44\1\0\77\44"+ + "\1\0\1\44\1\0\4\44\2\0\7\44\1\0\1\44\1\0\4\44"+ + "\2\0\47\44\1\0\1\44\1\0\4\44\2\0\37\44\1\0\1\44"+ + "\1\0\4\44\2\0\7\44\1\0\1\44\1\0\4\44\2\0\7\44"+ + "\1\0\7\44\1\0\27\44\1\0\37\44\1\0\1\44\1\0\4\44"+ + "\2\0\7\44\1\0\47\44\1\0\23\44\16\0\11\47\56\0\125\44"+ + "\14\0\u026c\44\2\0\10\44\12\0\32\44\5\0\113\44\25\0\15\44"+ + "\1\0\4\44\16\0\22\44\16\0\22\44\16\0\15\44\1\0\3\44"+ + "\17\0\64\44\43\0\1\44\4\0\1\44\3\0\12\47\46\0\12\47"+ + "\6\0\130\44\10\0\51\44\127\0\35\44\51\0\12\47\36\44\2\0"+ + "\5\44\u038b\0\154\44\224\0\234\44\4\0\132\44\6\0\26\44\2\0"+ + "\6\44\2\0\46\44\2\0\6\44\2\0\10\44\1\0\1\44\1\0"+ + "\1\44\1\0\1\44\1\0\37\44\2\0\65\44\1\0\7\44\1\0"+ + "\1\44\3\0\3\44\1\0\7\44\3\0\4\44\2\0\6\44\4\0"+ + "\15\44\5\0\3\44\1\0\7\44\164\0\1\44\15\0\1\44\202\0"+ + "\1\44\4\0\1\44\2\0\12\44\1\0\1\44\3\0\5\44\6\0"+ + "\1\44\1\0\1\44\1\0\1\44\1\0\4\44\1\0\3\44\1\0"+ + "\7\44\3\0\3\44\5\0\5\44\u0ebb\0\2\44\52\0\5\44\5\0"+ + "\2\44\4\0\126\44\6\0\3\44\1\0\132\44\1\0\4\44\5\0"+ + "\50\44\4\0\136\44\21\0\30\44\70\0\20\44\u0200\0\u19b6\44\112\0"+ + "\u51a6\44\132\0\u048d\44\u0773\0\u2ba4\44\u215c\0\u012e\44\2\0\73\44\225\0"+ + "\7\44\14\0\5\44\5\0\1\44\1\0\12\44\1\0\15\44\1\0"+ + "\5\44\1\0\1\44\1\0\2\44\1\0\2\44\1\0\154\44\41\0"+ + "\u016b\44\22\0\100\44\2\0\66\44\50\0\14\44\164\0\5\44\1\0"+ + "\207\44\23\0\12\47\7\0\32\44\6\0\32\44\13\0\131\44\3\0"+ + "\6\44\2\0\6\44\2\0\6\44\2\0\3\44\43\0"; /** * Translates characters to character classes @@ -122,12 +131,12 @@ class _RegExLexer implements FlexLexer { "\1\36\1\35\1\37\1\40\1\1\1\41\1\42\1\2"+ "\1\43\1\44\1\45\1\46\1\47\1\50\1\51\1\52"+ "\1\53\1\54\2\55\1\56\1\11\1\57\1\60\1\61"+ - "\1\62\1\63\1\64\1\65\1\66\2\0\1\67\1\70"+ - "\2\60\2\61\1\71\1\72\1\0\1\67\1\60\1\73"+ - "\2\61\1\74\1\67\3\61\1\75"; + "\1\62\1\63\1\0\1\64\1\65\1\66\2\0\1\67"+ + "\1\70\2\60\2\61\1\0\1\71\1\72\1\0\1\67"+ + "\1\60\1\73\2\61\1\74\1\75\1\67\3\61\1\76"; private static int [] zzUnpackAction() { - int [] result = new int[89]; + int [] result = new int[92]; int offset = 0; offset = zzUnpackAction(ZZ_ACTION_PACKED_0, offset, result); return result; @@ -152,21 +161,21 @@ class _RegExLexer implements FlexLexer { private static final int [] ZZ_ROWMAP = zzUnpackRowMap(); private static final String ZZ_ROWMAP_PACKED_0 = - "\0\0\0\61\0\142\0\223\0\304\0\365\0\u0126\0\u0157"+ - "\0\u0188\0\u0188\0\u01b9\0\u0188\0\u0188\0\u01ea\0\u021b\0\u0188"+ - "\0\u0188\0\u0188\0\u0188\0\u0188\0\u0188\0\u0188\0\u0188\0\u0188"+ - "\0\u0188\0\u024c\0\u0188\0\u027d\0\u0188\0\u02ae\0\u0188\0\u0188"+ - "\0\u0188\0\u0188\0\u0188\0\u02df\0\u0188\0\u0188\0\u01ea\0\u0188"+ - "\0\u0188\0\u0310\0\u0341\0\u0188\0\u0188\0\u0372\0\u0188\0\u0188"+ - "\0\u03a3\0\u03d4\0\u0188\0\u0188\0\u0188\0\u0188\0\u0188\0\u0188"+ - "\0\u0405\0\u0188\0\u0188\0\u0188\0\u0436\0\u0467\0\u0188\0\u0188"+ - "\0\u0188\0\u0188\0\u0188\0\u0498\0\u04c9\0\u04fa\0\u0188\0\u052b"+ - "\0\u055c\0\u058d\0\u05be\0\u0188\0\u0188\0\u05ef\0\u0620\0\u0188"+ - "\0\u0188\0\u0651\0\u0682\0\u0188\0\u0188\0\u06b3\0\u06e4\0\u0188"+ - "\0\u0188"; + "\0\0\0\62\0\144\0\226\0\310\0\372\0\u012c\0\u015e"+ + "\0\u0190\0\u0190\0\u01c2\0\u0190\0\u0190\0\u01f4\0\u0226\0\u0190"+ + "\0\u0190\0\u0190\0\u0190\0\u0190\0\u0190\0\u0190\0\u0190\0\u0190"+ + "\0\u0190\0\u0258\0\u0190\0\u028a\0\u0190\0\u02bc\0\u0190\0\u0190"+ + "\0\u0190\0\u0190\0\u0190\0\u02ee\0\u0190\0\u0190\0\u01f4\0\u0190"+ + "\0\u0190\0\u0320\0\u0352\0\u0190\0\u0190\0\u0384\0\u0190\0\u0190"+ + "\0\u03b6\0\u03e8\0\u0190\0\u0190\0\u0190\0\u0190\0\u0190\0\u0190"+ + "\0\u041a\0\u0190\0\u0190\0\u0190\0\u044c\0\u047e\0\u0190\0\u0190"+ + "\0\u04b0\0\u0190\0\u0190\0\u0190\0\u04e2\0\u0514\0\u0546\0\u0190"+ + "\0\u0578\0\u05aa\0\u05dc\0\u060e\0\u0640\0\u0190\0\u0190\0\u0672"+ + "\0\u06a4\0\u0190\0\u0190\0\u06d6\0\u0708\0\u0190\0\u0190\0\u0190"+ + "\0\u073a\0\u076c\0\u0190\0\u0190"; private static int [] zzUnpackRowMap() { - int [] result = new int[89]; + int [] result = new int[92]; int offset = 0; offset = zzUnpackRowMap(ZZ_ROWMAP_PACKED_0, offset, result); return result; @@ -191,40 +200,41 @@ class _RegExLexer implements FlexLexer { private static final String ZZ_TRANS_PACKED_0 = "\3\11\1\12\1\13\1\14\1\15\1\11\1\16\1\11"+ "\1\17\1\20\1\21\1\22\1\23\1\24\1\25\1\26"+ - "\20\11\1\27\1\11\1\30\4\11\1\20\5\11\1\31"+ - "\1\20\12\11\1\32\1\0\45\11\1\33\2\34\4\33"+ - "\1\35\12\33\17\36\1\34\1\33\1\36\2\33\1\34"+ - "\1\37\11\33\11\40\1\41\1\40\1\0\45\40\6\11"+ - "\1\15\1\11\1\16\1\42\1\17\1\43\1\21\25\11"+ + "\21\11\1\27\1\11\1\30\4\11\1\20\5\11\1\31"+ + "\1\20\12\11\1\32\1\0\46\11\1\33\2\34\4\33"+ + "\1\35\12\33\20\36\1\34\1\33\1\36\2\33\1\34"+ + "\1\37\11\33\11\40\1\41\1\40\1\0\46\40\6\11"+ + "\1\15\1\11\1\16\1\42\1\17\1\43\1\21\26\11"+ "\1\27\5\11\1\44\1\43\6\11\1\43\6\45\1\46"+ - "\1\45\1\47\1\45\1\17\46\45\5\50\1\51\14\50"+ - "\17\52\1\50\1\53\1\52\6\50\1\54\6\50\13\10"+ - "\1\55\44\10\1\55\77\0\1\56\53\0\1\57\47\0"+ + "\1\45\1\47\1\45\1\17\47\45\5\50\1\51\14\50"+ + "\20\52\1\50\1\53\1\52\6\50\1\54\6\50\13\10"+ + "\1\55\45\10\1\55\100\0\1\56\54\0\1\57\50\0"+ "\1\60\1\61\1\62\4\63\1\60\1\63\1\60\1\63"+ "\1\64\6\63\2\65\2\66\2\67\1\70\1\71\1\70"+ - "\1\72\1\73\1\74\1\73\1\75\1\76\1\61\1\63"+ - "\1\73\1\64\4\60\1\64\6\60\1\64\36\0\1\77"+ - "\23\0\2\34\36\0\1\34\4\0\1\34\13\0\2\36"+ - "\17\0\20\36\1\0\1\36\1\0\2\36\62\0\1\100"+ - "\32\0\17\52\2\0\1\52\37\0\17\53\2\0\1\53"+ - "\67\0\2\101\1\102\1\103\1\104\1\105\2\0\2\61"+ - "\36\0\1\61\21\0\1\106\36\0\1\106\17\0\61\107"+ - "\1\110\2\111\20\110\2\111\2\110\3\111\2\110\1\111"+ - "\1\110\1\111\2\110\1\111\17\110\1\112\2\113\20\112"+ - "\2\113\2\112\3\113\2\112\1\113\1\112\1\113\2\112"+ - "\1\113\17\112\54\0\1\114\1\115\3\0\5\116\1\0"+ - "\53\116\2\0\1\117\36\0\1\117\17\0\62\120\2\121"+ - "\20\120\2\121\2\120\3\121\2\120\1\121\1\120\1\121"+ - "\2\120\1\121\17\120\62\122\2\123\20\122\2\123\2\122"+ - "\3\123\2\122\1\123\1\122\1\123\2\122\1\123\17\122"+ - "\5\116\1\124\53\116\2\0\1\125\36\0\1\125\17\0"+ - "\62\126\2\127\20\126\2\127\2\126\3\127\2\126\1\127"+ - "\1\126\1\127\2\126\1\127\17\126\62\130\2\131\20\130"+ - "\2\131\2\130\3\131\2\130\1\131\1\130\1\131\2\130"+ - "\1\131\17\130"; + "\2\72\1\73\1\74\1\73\1\75\1\76\1\61\1\63"+ + "\1\73\1\64\4\60\1\64\6\60\1\64\37\0\1\77"+ + "\23\0\2\34\37\0\1\34\4\0\1\34\13\0\2\36"+ + "\17\0\21\36\1\0\1\36\1\0\2\36\63\0\1\100"+ + "\32\0\20\52\2\0\1\52\37\0\20\53\2\0\1\53"+ + "\51\0\1\101\16\0\2\102\1\103\1\104\1\105\1\106"+ + "\2\0\2\61\37\0\1\61\21\0\1\107\37\0\1\107"+ + "\17\0\62\110\1\111\2\112\20\111\2\112\2\111\3\112"+ + "\3\111\1\112\1\111\1\112\2\111\1\112\17\111\1\113"+ + "\2\114\20\113\2\114\2\113\3\114\3\113\1\114\1\113"+ + "\1\114\2\113\1\114\17\113\57\0\1\115\57\0\1\116"+ + "\1\117\3\0\5\120\1\0\54\120\2\0\1\121\37\0"+ + "\1\121\17\0\63\122\2\123\20\122\2\123\2\122\3\123"+ + "\3\122\1\123\1\122\1\123\2\122\1\123\17\122\63\124"+ + "\2\125\20\124\2\125\2\124\3\125\3\124\1\125\1\124"+ + "\1\125\2\124\1\125\17\124\22\0\20\115\2\0\1\115"+ + "\7\0\1\126\5\0\5\120\1\127\54\120\2\0\1\130"+ + "\37\0\1\130\17\0\63\131\2\132\20\131\2\132\2\131"+ + "\3\132\3\131\1\132\1\131\1\132\2\131\1\132\17\131"+ + "\63\133\2\134\20\133\2\134\2\133\3\134\3\133\1\134"+ + "\1\133\1\134\2\133\1\134\17\133"; private static int [] zzUnpackTrans() { - int [] result = new int[1813]; + int [] result = new int[1950]; int offset = 0; offset = zzUnpackTrans(ZZ_TRANS_PACKED_0, offset, result); return result; @@ -265,15 +275,15 @@ class _RegExLexer implements FlexLexer { private static final int [] ZZ_ATTRIBUTE = zzUnpackAttribute(); private static final String ZZ_ATTRIBUTE_PACKED_0 = - "\6\0\2\1\2\11\1\1\2\11\1\3\1\1\12\11"+ - "\1\1\1\11\1\1\1\11\1\1\5\11\1\1\2\11"+ - "\1\3\2\11\2\1\2\11\1\1\1\15\1\11\2\1"+ - "\6\11\1\1\3\11\2\1\5\11\2\0\1\1\1\11"+ - "\4\1\2\11\1\0\1\1\2\11\2\1\2\11\2\1"+ - "\2\11"; + "\6\0\2\1\2\11\1\1\2\11\2\1\12\11\1\1"+ + "\1\11\1\1\1\11\1\1\5\11\1\1\2\11\1\1"+ + "\2\11\2\1\2\11\1\1\2\11\2\1\6\11\1\1"+ + "\3\11\2\1\2\11\1\0\3\11\2\0\1\1\1\11"+ + "\4\1\1\0\2\11\1\0\1\1\2\11\2\1\3\11"+ + "\2\1\2\11"; private static int [] zzUnpackAttribute() { - int [] result = new int[89]; + int [] result = new int[92]; int offset = 0; offset = zzUnpackAttribute(ZZ_ATTRIBUTE_PACKED_0, offset, result); return result; @@ -570,8 +580,6 @@ class _RegExLexer implements FlexLexer { int [] zzTransL = ZZ_TRANS; int [] zzRowMapL = ZZ_ROWMAP; int [] zzAttrL = ZZ_ATTRIBUTE; - int zzPushbackPosL = zzPushbackPos = -1; - boolean zzWasPushback; while (true) { zzMarkedPosL = zzMarkedPos; @@ -580,9 +588,8 @@ class _RegExLexer implements FlexLexer { zzCurrentPosL = zzCurrentPos = zzStartRead = zzMarkedPosL; - zzState = zzLexicalState; + zzState = ZZ_LEXSTATE[zzLexicalState]; - zzWasPushback = false; zzForAction: { while (true) { @@ -597,14 +604,12 @@ class _RegExLexer implements FlexLexer { // store back cached positions zzCurrentPos = zzCurrentPosL; zzMarkedPos = zzMarkedPosL; - zzPushbackPos = zzPushbackPosL; boolean eof = zzRefill(); // get translated positions and possibly new buffer zzCurrentPosL = zzCurrentPos; zzMarkedPosL = zzMarkedPos; zzBufferL = zzBuffer; zzEndReadL = zzEndRead; - zzPushbackPosL = zzPushbackPos; if (eof) { zzInput = YYEOF; break zzForAction; @@ -618,11 +623,7 @@ class _RegExLexer implements FlexLexer { zzState = zzNext; int zzAttributes = zzAttrL[zzState]; - if ( (zzAttributes & 2) == 2 ) - zzPushbackPosL = zzCurrentPosL; - if ( (zzAttributes & 1) == 1 ) { - zzWasPushback = (zzAttributes & 4) == 4; zzAction = zzState; zzMarkedPosL = zzCurrentPosL; if ( (zzAttributes & 8) == 8 ) break zzForAction; @@ -633,246 +634,250 @@ class _RegExLexer implements FlexLexer { // store back cached position zzMarkedPos = zzMarkedPosL; - if (zzWasPushback) - zzMarkedPos = zzPushbackPosL; switch (zzAction < 0 ? zzAction : ZZ_ACTION[zzAction]) { case 7: { if (yystate() != CLASS2) yypushstate(EMBRACED); return RegExpTT.LBRACE; } - case 62: break; + case 63: break; case 6: { return RegExpTT.GROUP_END; } - case 63: break; + case 64: break; case 36: + // lookahead expression with fixed base length + zzMarkedPos = zzStartRead + 1; { yypushstate(CLASS1); return RegExpTT.CLASS_BEGIN; } - case 64: break; + case 65: break; case 14: { return RegExpTT.STAR; } - case 65: break; + case 66: break; case 40: { return RegExpTT.ESC_CHARACTER; } - case 66: break; + case 67: break; case 28: { return commentMode ? com.intellij.psi.TokenType.WHITE_SPACE : RegExpTT.ESC_CHARACTER; } - case 67: break; + case 68: break; case 41: { return commentMode ? RegExpTT.CHARACTER : RegExpTT.REDUNDANT_ESCAPE; } - case 68: break; + case 69: break; case 37: { return RegExpTT.REDUNDANT_ESCAPE; } - case 69: break; + case 70: break; case 24: { return RegExpTT.COMMA; } - case 70: break; + case 71: break; case 57: { return RegExpTT.POS_LOOKBEHIND; } - case 71: break; + case 72: break; case 39: { return RegExpTT.BAD_OCT_VALUE; } - case 72: break; + case 73: break; case 10: { return commentMode ? com.intellij.psi.TokenType.WHITE_SPACE : RegExpTT.CTRL_CHARACTER; } - case 73: break; + case 74: break; case 30: { yypopstate(); yypushstate(EMBRACED); return RegExpTT.LBRACE; } - case 74: break; + case 75: break; case 49: { return StringEscapesTokenTypes.INVALID_UNICODE_ESCAPE_TOKEN; } - case 75: break; + case 76: break; case 58: { return RegExpTT.NEG_LOOKBEHIND; } - case 76: break; - case 61: + case 77: break; + case 62: { return RegExpTT.UNICODE_CHAR; } - case 77: break; + case 78: break; case 56: { if (xmlSchemaMode) { yypushback(1); return RegExpTT.CHAR_CLASS; } else return RegExpTT.CTRL; } - case 78: break; + case 79: break; case 35: { yybegin(OPTIONS); return RegExpTT.SET_OPTIONS; } - case 79: break; + case 80: break; case 12: { return RegExpTT.DOLLAR; } - case 80: break; + case 81: break; case 50: { yypopstate(); return RegExpTT.QUOTE_END; } - case 81: break; + case 82: break; case 53: { return RegExpTT.POS_LOOKAHEAD; } - case 82: break; + case 83: break; case 22: { yypopstate(); return RegExpTT.RBRACE; } - case 83: break; + case 84: break; case 3: { return RegExpTT.CHARACTER; } - case 84: break; + case 85: break; case 54: { return RegExpTT.NEG_LOOKAHEAD; } - case 85: break; + case 86: break; case 42: { return RegExpTT.ESC_CTRL_CHARACTER; } - case 86: break; + case 87: break; case 23: { return RegExpTT.NAME; } - case 87: break; + case 88: break; case 51: { return RegExpTT.ANDAND; } - case 88: break; + case 89: break; case 13: { return RegExpTT.QUEST; } - case 89: break; + case 90: break; case 44: { return RegExpTT.CHAR_CLASS; } - case 90: break; + case 91: break; case 17: { return RegExpTT.MINUS; } - case 91: break; - case 60: + case 92: break; + case 61: { return RegExpTT.COMMENT; } - case 92: break; + case 93: break; case 38: { return yystate() != CLASS2 ? RegExpTT.BACKREF : RegExpTT.ESC_CHARACTER; } - case 93: break; + case 94: break; case 16: { return RegExpTT.UNION; } - case 94: break; + case 95: break; + case 60: + { return RegExpTT.PYTHON_NAMED_GROUP; + } + case 96: break; case 45: { if (xmlSchemaMode) return RegExpTT.CHAR_CLASS; else return StringEscapesTokenTypes.INVALID_CHARACTER_ESCAPE_TOKEN; } - case 95: break; + case 97: break; case 55: { return RegExpTT.OCT_CHAR; } - case 96: break; + case 98: break; case 4: { return RegExpTT.DOT; } - case 97: break; + case 99: break; case 32: { yybegin(YYINITIAL); return RegExpTT.GROUP_END; } - case 98: break; + case 100: break; case 21: { return RegExpTT.NUMBER; } - case 99: break; + case 101: break; case 46: { yypushstate(PROP); return RegExpTT.PROPERTY; } - case 100: break; + case 102: break; case 18: { return commentMode ? com.intellij.psi.TokenType.WHITE_SPACE : RegExpTT.CHARACTER; } - case 101: break; + case 103: break; case 33: { handleOptions(); return RegExpTT.OPTIONS_OFF; } - case 102: break; + case 104: break; case 34: { yybegin(YYINITIAL); return RegExpTT.COLON; } - case 103: break; + case 105: break; case 25: { assert false : yytext(); } - case 104: break; + case 106: break; case 27: { yypopstate(); return RegExpTT.CLASS_END; } - case 105: break; + case 107: break; case 11: { return RegExpTT.CARET; } - case 106: break; + case 108: break; case 1: { handleOptions(); return RegExpTT.OPTIONS_ON; } - case 107: break; + case 109: break; case 29: { yypopstate(); yypushback(1); } - case 108: break; + case 110: break; case 43: { return yystate() != CLASS2 ? RegExpTT.BOUNDARY : RegExpTT.ESC_CHARACTER; } - case 109: break; + case 111: break; case 19: { if (commentMode) { yypushstate(COMMENT); return RegExpTT.COMMENT; } else return RegExpTT.CHARACTER; } - case 110: break; + case 112: break; case 59: { return RegExpTT.HEX_CHAR; } - case 111: break; + case 113: break; case 5: { return RegExpTT.GROUP_BEGIN; } - case 112: break; + case 114: break; case 8: { yypushstate(CLASS2); return RegExpTT.CLASS_BEGIN; } - case 113: break; + case 115: break; case 52: { return RegExpTT.NON_CAPT_GROUP; } - case 114: break; + case 116: break; case 9: { return StringEscapesTokenTypes.INVALID_CHARACTER_ESCAPE_TOKEN; } - case 115: break; + case 117: break; case 48: { return RegExpTT.BAD_HEX_VALUE; } - case 116: break; + case 118: break; case 2: { yypopstate(); return RegExpTT.COMMENT; } - case 117: break; + case 119: break; case 47: { yypushstate(QUOTED); return RegExpTT.QUOTE_BEGIN; } - case 118: break; + case 120: break; case 26: { yybegin(CLASS2); return RegExpTT.CHARACTER; } - case 119: break; + case 121: break; case 15: { return RegExpTT.PLUS; } - case 120: break; + case 122: break; case 20: { if (allowDanglingMetacharacters) { yypopstate(); yypushback(1); @@ -880,11 +885,11 @@ class _RegExLexer implements FlexLexer { return RegExpTT.BAD_CHARACTER; } } - case 121: break; + case 123: break; case 31: { yybegin(YYINITIAL); return RegExpTT.BAD_CHARACTER; } - case 122: break; + case 124: break; default: if (zzInput == YYEOF && zzStartRead == zzCurrentPos) { zzAtEOF = true; diff --git a/RegExpSupport/src/org/intellij/lang/regexp/psi/RegExpGroup.java b/RegExpSupport/src/org/intellij/lang/regexp/psi/RegExpGroup.java index 7770f6ad4376..db6e81c0eae5 100644 --- a/RegExpSupport/src/org/intellij/lang/regexp/psi/RegExpGroup.java +++ b/RegExpSupport/src/org/intellij/lang/regexp/psi/RegExpGroup.java @@ -22,4 +22,6 @@ public interface RegExpGroup extends RegExpAtom { @Nullable RegExpPattern getPattern(); + + boolean isPythonNamedGroup(); } diff --git a/RegExpSupport/src/org/intellij/lang/regexp/psi/impl/RegExpGroupImpl.java b/RegExpSupport/src/org/intellij/lang/regexp/psi/impl/RegExpGroupImpl.java index 9b8fa67ece98..976fb2ba8fef 100644 --- a/RegExpSupport/src/org/intellij/lang/regexp/psi/impl/RegExpGroupImpl.java +++ b/RegExpSupport/src/org/intellij/lang/regexp/psi/impl/RegExpGroupImpl.java @@ -41,4 +41,8 @@ public class RegExpGroupImpl extends RegExpElementImpl implements RegExpGroup { final ASTNode node = getNode().findChildByType(RegExpElementTypes.PATTERN); return node != null ? (RegExpPattern)node.getPsi() : null; } + + public boolean isPythonNamedGroup() { + return getNode().findChildByType(RegExpTT.PYTHON_NAMED_GROUP) != null; + } } diff --git a/RegExpSupport/src/org/intellij/lang/regexp/regexp-lexer.flex b/RegExpSupport/src/org/intellij/lang/regexp/regexp-lexer.flex index 621fc89cf6e6..2b0455a747eb 100644 --- a/RegExpSupport/src/org/intellij/lang/regexp/regexp-lexer.flex +++ b/RegExpSupport/src/org/intellij/lang/regexp/regexp-lexer.flex @@ -205,6 +205,7 @@ HEX_CHAR=[0-9a-fA-F] "(?<=" { return RegExpTT.POS_LOOKBEHIND; } "(?" { return RegExpTT.PYTHON_NAMED_GROUP; } "(?" { yybegin(OPTIONS); return RegExpTT.SET_OPTIONS; } } diff --git a/RegExpSupport/src/org/intellij/lang/regexp/validation/RegExpAnnotator.java b/RegExpSupport/src/org/intellij/lang/regexp/validation/RegExpAnnotator.java index 52307d1cd58d..21e50a871722 100644 --- a/RegExpSupport/src/org/intellij/lang/regexp/validation/RegExpAnnotator.java +++ b/RegExpSupport/src/org/intellij/lang/regexp/validation/RegExpAnnotator.java @@ -30,12 +30,13 @@ import org.intellij.lang.regexp.RegExpLanguageHost; import org.intellij.lang.regexp.RegExpTT; import org.intellij.lang.regexp.psi.*; import org.intellij.lang.regexp.psi.impl.RegExpPropertyImpl; +import org.jetbrains.annotations.NotNull; import org.jetbrains.annotations.Nullable; public final class RegExpAnnotator extends RegExpElementVisitor implements Annotator { private AnnotationHolder myHolder; - public void annotate(PsiElement psiElement, AnnotationHolder holder) { + public void annotate(@NotNull PsiElement psiElement, @NotNull AnnotationHolder holder) { assert myHolder == null : "unsupported concurrent annotator invocation"; try { myHolder = holder; @@ -159,6 +160,12 @@ public final class RegExpAnnotator extends RegExpElementVisitor implements Annot } } } + if (group.isPythonNamedGroup()) { + RegExpLanguageHost host = findRegExpHost(group); + if (host == null || !host.supportsPythonNamedGroups()) { + myHolder.createErrorAnnotation(group, "This named group syntax is not supported"); + } + } } @Override