xml/impl/src/com/intellij/lexer/_HtmlLexer.flex

   1 /* It's an automatically generated code. Do not modify it. */
   2 package com.intellij.lexer;
   3
   4 import com.intellij.psi.tree.IElementType;
   5 import com.intellij.psi.*;
   6 import com.intellij.psi.xml.*;
   7
   8 %%
   9
  10 %unicode
  11
  12 %{
  13   private IElementType elTokenType = XmlTokenType.XML_DATA_CHARACTERS;
  14   private IElementType elTokenType2 = XmlTokenType.XML_ATTRIBUTE_VALUE_TOKEN;
  15
  16   public void setElTypes(IElementType _elTokenType,IElementType _elTokenType2) {
  17     elTokenType = _elTokenType;
  18     elTokenType2 = _elTokenType2;
  19   }
  20
  21   public _HtmlLexer() {
  22     this((java.io.Reader)null);
  23   }
  24 %}
  25
  26 %class _HtmlLexer
  27 %public
  28 %implements FlexLexer,ELHostLexer
  29 %function advance
  30 %type IElementType
  31 %eof{ return;
  32 %eof}
  33
  34 %state DOC_TYPE
  35 %state COMMENT
  36 %state START_TAG_NAME
  37 %state END_TAG_NAME
  38 %state TAG_ATTRIBUTES
  39 %state ATTRIBUTE_VALUE_START
  40 %state ATTRIBUTE_VALUE_DQ
  41 %state ATTRIBUTE_VALUE_SQ
  42 %state PROCESSING_INSTRUCTION
  43 %state START_TAG_NAME2
  44 %state END_TAG_NAME2
  45 %state TAG_CHARACTERS
  46 %state C_COMMENT_START
  47 %state C_COMMENT_END
  48 /* IMPORTANT! number of states should not exceed 16. See JspHighlightingLexer. */
  49
  50 ALPHA=[:letter:]
  51 DIGIT=[0-9]
  52 WHITE_SPACE_CHARS=[ \n\r\t\f]+
  53
  54 TAG_NAME=({ALPHA}|"_"|":")({ALPHA}|{DIGIT}|"_"|":"|"."|"-")*
  55 TAG_NAME_FWT=("#")({ALPHA}|{DIGIT}|"_"|":"|"."|"-")*
  56 ATTRIBUTE_NAME=({ALPHA}|"_"|":")({ALPHA}|{DIGIT}|"_"|":"|"."|"-")*
  57
  58 DTD_REF= "\"" [^\"]* "\"" | "'" [^']* "'"
  59 DOCTYPE= "<!" (D|d)(O|o)(C|c)(T|t)(Y|y)(P|p)(E|e)
  60 HTML= (H|h)(T|t)(M|m)(L|l)
  61 PUBLIC= (P|p)(U|u)(B|b)(L|l)(I|i)(C|c)
  62 EL_EMBEDDMENT="${" [^\}]* "}"
  63
  64 END_COMMENT="--"[ \n\r\t\f]*">"
  65
  66 CONDITIONAL_COMMENT_CONDITION=({ALPHA})({ALPHA}|{WHITE_SPACE_CHARS}|{DIGIT}|"."|"("|")"|"|"|"!"|"&")*
  67 %%
  68
  69 <YYINITIAL> "<?" { yybegin(PROCESSING_INSTRUCTION); return XmlTokenType.XML_PI_START; }
  70 <PROCESSING_INSTRUCTION> "?"? ">" { yybegin(YYINITIAL); return XmlTokenType.XML_PI_END; }
  71 <PROCESSING_INSTRUCTION> ([^\?\>] | (\?[^\>]))* { return XmlTokenType.XML_PI_TARGET; }
  72
  73 <YYINITIAL> {DOCTYPE} { yybegin(DOC_TYPE); return XmlTokenType.XML_DOCTYPE_START; }
  74 <DOC_TYPE> {HTML} { return XmlTokenType.XML_NAME; }
  75 <DOC_TYPE> {PUBLIC} { return XmlTokenType.XML_DOCTYPE_PUBLIC; }
  76 <DOC_TYPE> {DTD_REF} { return XmlTokenType.XML_ATTRIBUTE_VALUE_TOKEN;}
  77 <DOC_TYPE> ">" { yybegin(YYINITIAL); return XmlTokenType.XML_DOCTYPE_END; }
  78 <YYINITIAL> {WHITE_SPACE_CHARS} { return XmlTokenType.XML_REAL_WHITE_SPACE; }
  79 <DOC_TYPE,TAG_ATTRIBUTES,ATTRIBUTE_VALUE_START,PROCESSING_INSTRUCTION, START_TAG_NAME, END_TAG_NAME, END_TAG_NAME2, TAG_CHARACTERS> {WHITE_SPACE_CHARS} { return XmlTokenType.XML_WHITE_SPACE; }
  80 <YYINITIAL> "<" {TAG_NAME} { yybegin(START_TAG_NAME); yypushback(yylength()); }
  81 <YYINITIAL> "<" {TAG_NAME_FWT} { yybegin(START_TAG_NAME2); yypushback(yylength()); }
  82 <START_TAG_NAME, START_TAG_NAME2, TAG_CHARACTERS> "<" { return XmlTokenType.XML_START_TAG_START; }
  83
  84 <YYINITIAL> "</" {TAG_NAME} { yybegin(END_TAG_NAME); yypushback(yylength()); }
  85 <YYINITIAL> "</" {TAG_NAME_FWT} { yybegin(END_TAG_NAME2); yypushback(yylength()); }
  86 <YYINITIAL, END_TAG_NAME, END_TAG_NAME2> "</" { return XmlTokenType.XML_END_TAG_START; }
  87
  88 <YYINITIAL> "<!--" { yybegin(COMMENT); return XmlTokenType.XML_COMMENT_START; }
  89 <COMMENT> "[" { yybegin(C_COMMENT_START); return XmlTokenType.XML_CONDITIONAL_COMMENT_START; }
  90 <COMMENT> "<![" { yybegin(C_COMMENT_END); return XmlTokenType.XML_CONDITIONAL_COMMENT_END_START; }
  91 <COMMENT> {END_COMMENT} { yybegin(YYINITIAL); return XmlTokenType.XML_COMMENT_END; }
  92 <COMMENT> [^] { return XmlTokenType.XML_COMMENT_CHARACTERS; }
  93
  94 <C_COMMENT_START,C_COMMENT_END> {CONDITIONAL_COMMENT_CONDITION} { return XmlTokenType.XML_COMMENT_CHARACTERS; }
  95 <C_COMMENT_START> [^] { yybegin(COMMENT); return XmlTokenType.XML_COMMENT_CHARACTERS; }
  96 <C_COMMENT_START> "]>" { yybegin(COMMENT); return XmlTokenType.XML_CONDITIONAL_COMMENT_START_END; }
  97 <C_COMMENT_START,C_COMMENT_END> {END_COMMENT} { yybegin(YYINITIAL); return XmlTokenType.XML_COMMENT_END; }
  98 <C_COMMENT_END> "]" { yybegin(COMMENT); return XmlTokenType.XML_CONDITIONAL_COMMENT_END; }
  99 <C_COMMENT_END> [^] { yybegin(COMMENT); return XmlTokenType.XML_COMMENT_CHARACTERS; }
 100
 101 <YYINITIAL> \\\$ {
 102   return XmlTokenType.XML_DATA_CHARACTERS;
 103 }
 104
 105 <YYINITIAL> {EL_EMBEDDMENT} {
 106   return elTokenType;
 107 }
 108
 109 <START_TAG_NAME, END_TAG_NAME> {TAG_NAME} { yybegin(TAG_ATTRIBUTES); return XmlTokenType.XML_NAME; }
 110 <END_TAG_NAME2> {TAG_NAME_FWT} { return XmlTokenType.XML_NAME; }
 111 <START_TAG_NAME2> {TAG_NAME_FWT} { yybegin(TAG_CHARACTERS); return XmlTokenType.XML_NAME; }
 112
 113 <TAG_ATTRIBUTES, END_TAG_NAME2, TAG_CHARACTERS> ">" { yybegin(YYINITIAL); return XmlTokenType.XML_TAG_END; }
 114 <TAG_ATTRIBUTES, TAG_CHARACTERS> "/>" { yybegin(YYINITIAL); return XmlTokenType.XML_EMPTY_ELEMENT_END; }
 115 <TAG_ATTRIBUTES> {ATTRIBUTE_NAME} { return XmlTokenType.XML_NAME; }
 116 <TAG_ATTRIBUTES> "=" { yybegin(ATTRIBUTE_VALUE_START); return XmlTokenType.XML_EQ; }
 117 <TAG_ATTRIBUTES,START_TAG_NAME, END_TAG_NAME, END_TAG_NAME2> [^] { yybegin(YYINITIAL); yypushback(1); break; }
 118
 119 <TAG_CHARACTERS> [^] { return XmlTokenType.XML_TAG_CHARACTERS; }
 120
 121 <ATTRIBUTE_VALUE_START> ">" { yybegin(YYINITIAL); return XmlTokenType.XML_TAG_END; }
 122 <ATTRIBUTE_VALUE_START> "/>" { yybegin(YYINITIAL); return XmlTokenType.XML_EMPTY_ELEMENT_END; }
 123  <ATTRIBUTE_VALUE_START> {EL_EMBEDDMENT} {
 124   return elTokenType2;
 125 }
 126
 127 <ATTRIBUTE_VALUE_START> ([^ \n\r\t\f'\"\>]|(\/[^\>]))* { yybegin(TAG_ATTRIBUTES); return XmlTokenType.XML_ATTRIBUTE_VALUE_TOKEN; }
 128 <ATTRIBUTE_VALUE_START> "\"" { yybegin(ATTRIBUTE_VALUE_DQ); return XmlTokenType.XML_ATTRIBUTE_VALUE_START_DELIMITER; }
 129 <ATTRIBUTE_VALUE_START> "'" { yybegin(ATTRIBUTE_VALUE_SQ); return XmlTokenType.XML_ATTRIBUTE_VALUE_START_DELIMITER; }
 130
 131 <ATTRIBUTE_VALUE_DQ> {
 132   "\"" { yybegin(TAG_ATTRIBUTES); return XmlTokenType.XML_ATTRIBUTE_VALUE_END_DELIMITER; }
 133   \\\$ { return XmlTokenType.XML_ATTRIBUTE_VALUE_TOKEN; }
 134   "${" [^\}\"]* "}" { return elTokenType2; }
 135   [^] { return XmlTokenType.XML_ATTRIBUTE_VALUE_TOKEN;}
 136 }
 137
 138 <ATTRIBUTE_VALUE_SQ> {
 139   "'" { yybegin(TAG_ATTRIBUTES); return XmlTokenType.XML_ATTRIBUTE_VALUE_END_DELIMITER; }
 140   \\\$ { return XmlTokenType.XML_ATTRIBUTE_VALUE_TOKEN; }
 141   "${" [^\}\']* "}" { return elTokenType2; }
 142   [^] { return XmlTokenType.XML_ATTRIBUTE_VALUE_TOKEN;}
 143 }
 144
 145 "&lt;" |
 146 "&gt;" |
 147 "&apos;" |
 148 "&quot;" |
 149 "&nbsp;" |
 150 "&amp;" |
 151 "&#"{DIGIT}+";" |
 152 "&#x"({DIGIT}|[a-fA-F])+";" { return XmlTokenType.XML_CHAR_ENTITY_REF; }
 153 "&"{TAG_NAME}";" { return XmlTokenType.XML_ENTITY_REF_TOKEN; }
 154
 155 <YYINITIAL> ([^<&\$# \n\r\t\f]|(\\\$)|(\\#))* { return XmlTokenType.XML_DATA_CHARACTERS; }
 156 <YYINITIAL> [^] { return XmlTokenType.XML_DATA_CHARACTERS; }
 157 [^] { return XmlTokenType.XML_BAD_CHARACTER; }