tartrazine/lexers/html.xml

<lexer>
  <config>
    <name>HTML</name>
    <alias>html</alias>
    <filename>*.html</filename>
    <filename>*.htm</filename>
    <filename>*.xhtml</filename>
    <filename>*.xslt</filename>
    <mime_type>text/html</mime_type>
    <mime_type>application/xhtml+xml</mime_type>
    <case_insensitive>true</case_insensitive>
    <dot_all>true</dot_all>
    <not_multiline>true</not_multiline>
  </config>
  <rules>
    <state name="script-content">
      <rule pattern="(&lt;)(\s*)(/)(\s*)(script)(\s*)(&gt;)">
        <bygroups>
          <token type="Punctuation"/>
          <token type="Text"/>
          <token type="Punctuation"/>
          <token type="Text"/>
          <token type="NameTag"/>
          <token type="Text"/>
          <token type="Punctuation"/>
        </bygroups>
        <pop depth="1"/>
      </rule>
      <rule pattern=".+?(?=&lt;\s*/\s*script\s*&gt;)">
        <using lexer="Javascript"/>
      </rule>
    </state>
    <state name="style-content">
      <rule pattern="(&lt;)(\s*)(/)(\s*)(style)(\s*)(&gt;)">
        <bygroups>
          <token type="Punctuation"/>
          <token type="Text"/>
          <token type="Punctuation"/>
          <token type="Text"/>
          <token type="NameTag"/>
          <token type="Text"/>
          <token type="Punctuation"/>
        </bygroups>
        <pop depth="1"/>
      </rule>
      <rule pattern=".+?(?=&lt;\s*/\s*style\s*&gt;)">
        <using lexer="CSS"/>
      </rule>
    </state>
    <state name="attr">
      <rule pattern="&#34;.*?&#34;">
        <token type="LiteralString"/>
        <pop depth="1"/>
      </rule>
      <rule pattern="&#39;.*?&#39;">
        <token type="LiteralString"/>
        <pop depth="1"/>
      </rule>
      <rule pattern="[^\s&gt;]+">
        <token type="LiteralString"/>
        <pop depth="1"/>
      </rule>
    </state>
    <state name="root">
      <rule pattern="[^&lt;&amp;]+">
        <token type="Text"/>
      </rule>
      <rule pattern="&amp;\S*?;">
        <token type="NameEntity"/>
      </rule>
      <rule pattern="\&lt;\!\[CDATA\[.*?\]\]\&gt;">
        <token type="CommentPreproc"/>
      </rule>
      <rule pattern="&lt;!--">
        <token type="Comment"/>
        <push state="comment"/>
      </rule>
      <rule pattern="&lt;\?.*?\?&gt;">
        <token type="CommentPreproc"/>
      </rule>
      <rule pattern="&lt;![^&gt;]*&gt;">
        <token type="CommentPreproc"/>
      </rule>
      <rule pattern="(&lt;)(\s*)(script)(\s*)">
        <bygroups>
          <token type="Punctuation"/>
          <token type="Text"/>
          <token type="NameTag"/>
          <token type="Text"/>
        </bygroups>
        <push state="script-content" state="tag"/>
      </rule>
      <rule pattern="(&lt;)(\s*)(style)(\s*)">
        <bygroups>
          <token type="Punctuation"/>
          <token type="Text"/>
          <token type="NameTag"/>
          <token type="Text"/>
        </bygroups>
        <push state="style-content" state="tag"/>
      </rule>
      <rule pattern="(&lt;)(\s*)([\w:.-]+)">
        <bygroups>
          <token type="Punctuation"/>
          <token type="Text"/>
          <token type="NameTag"/>
        </bygroups>
        <push state="tag"/>
      </rule>
      <rule pattern="(&lt;)(\s*)(/)(\s*)([\w:.-]+)(\s*)(&gt;)">
        <bygroups>
          <token type="Punctuation"/>
          <token type="Text"/>
          <token type="Punctuation"/>
          <token type="Text"/>
          <token type="NameTag"/>
          <token type="Text"/>
          <token type="Punctuation"/>
        </bygroups>
      </rule>
    </state>
    <state name="comment">
      <rule pattern="[^-]+">
        <token type="Comment"/>
      </rule>
      <rule pattern="--&gt;">
        <token type="Comment"/>
        <pop depth="1"/>
      </rule>
      <rule pattern="-">
        <token type="Comment"/>
      </rule>
    </state>
    <state name="tag">
      <rule pattern="\s+">
        <token type="Text"/>
      </rule>
      <rule pattern="([\w:-]+\s*)(=)(\s*)">
        <bygroups>
          <token type="NameAttribute"/>
          <token type="Operator"/>
          <token type="Text"/>
        </bygroups>
        <push state="attr"/>
      </rule>
      <rule pattern="[\w:-]+">
        <token type="NameAttribute"/>
      </rule>
      <rule pattern="(/?)(\s*)(&gt;)">
        <bygroups>
          <token type="Punctuation"/>
          <token type="Text"/>
          <token type="Punctuation"/>
        </bygroups>
        <pop depth="1"/>
      </rule>
    </state>
  </rules>
</lexer>