Add Vala Lexer (#697)

alecthomas · Nov 5, 2022 · be4aebd · be4aebd
1 parent d6ea504
commit be4aebd
Show file tree

Hide file tree

Showing 3 changed files with 328 additions and 0 deletions.
diff --git a/lexers/embedded/vala.xml b/lexers/embedded/vala.xml
@@ -0,0 +1,72 @@
+
+<lexer>
+  <config>
+    <name>Vala</name>
+    <alias>vala</alias>
+    <alias>vapi</alias>
+    <filename>*.vala</filename>
+    <filename>*.vapi</filename>
+    <mime_type>text/x-vala</mime_type>
+  </config>
+  <rules>
+    <state name="whitespace">
+      <rule pattern="^\s*#if\s+0"><token type="CommentPreproc"/><push state="if0"/></rule>
+      <rule pattern="\n"><token type="TextWhitespace"/></rule>
+      <rule pattern="\s+"><token type="TextWhitespace"/></rule>
+      <rule pattern="\\\n"><token type="Text"/></rule>
+      <rule pattern="//(\n|(.|\n)*?[^\\]\n)"><token type="CommentSingle"/></rule>
+      <rule pattern="/(\\\n)?[*](.|\n)*?[*](\\\n)?/"><token type="CommentMultiline"/></rule>
+    </state>
+    <state name="statements">
+      <rule pattern="[L@]?&quot;"><token type="LiteralString"/><push state="string"/></rule>
+      <rule pattern="L?&#x27;(\\.|\\[0-7]{1,3}|\\x[a-fA-F0-9]{1,2}|[^\\\&#x27;\n])&#x27;"><token type="LiteralStringChar"/></rule>
+      <rule pattern="(?s)&quot;&quot;&quot;.*?&quot;&quot;&quot;"><token type="LiteralString"/></rule>
+      <rule pattern="(\d+\.\d*|\.\d+|\d+)[eE][+-]?\d+[lL]?"><token type="LiteralNumberFloat"/></rule>
+      <rule pattern="(\d+\.\d*|\.\d+|\d+[fF])[fF]?"><token type="LiteralNumberFloat"/></rule>
+      <rule pattern="0x[0-9a-fA-F]+[Ll]?"><token type="LiteralNumberHex"/></rule>
+      <rule pattern="0[0-7]+[Ll]?"><token type="LiteralNumberOct"/></rule>
+      <rule pattern="\d+[Ll]?"><token type="LiteralNumberInteger"/></rule>
+      <rule pattern="[~!%^&amp;*+=|?:&lt;&gt;/-]"><token type="Operator"/></rule>
+      <rule pattern="(\[)(Compact|Immutable|(?:Boolean|Simple)Type)(\])"><bygroups><token type="Punctuation"/><token type="NameDecorator"/><token type="Punctuation"/></bygroups></rule>
+      <rule pattern="(\[)(CCode|(?:Integer|Floating)Type)"><bygroups><token type="Punctuation"/><token type="NameDecorator"/></bygroups></rule>
+      <rule pattern="[()\[\],.]"><token type="Punctuation"/></rule>
+      <rule pattern="(as|base|break|case|catch|construct|continue|default|delete|do|else|enum|finally|for|foreach|get|if|in|is|lock|new|out|params|return|set|sizeof|switch|this|throw|try|typeof|while|yield)\b"><token type="Keyword"/></rule>
+      <rule pattern="(abstract|const|delegate|dynamic|ensures|extern|inline|internal|override|owned|private|protected|public|ref|requires|signal|static|throws|unowned|var|virtual|volatile|weak|yields)\b"><token type="KeywordDeclaration"/></rule>
+      <rule pattern="(namespace|using)(\s+)"><bygroups><token type="KeywordNamespace"/><token type="TextWhitespace"/></bygroups><push state="namespace"/></rule>
+      <rule pattern="(class|errordomain|interface|struct)(\s+)"><bygroups><token type="KeywordDeclaration"/><token type="TextWhitespace"/></bygroups><push state="class"/></rule>
+      <rule pattern="(\.)([a-zA-Z_]\w*)"><bygroups><token type="Operator"/><token type="NameAttribute"/></bygroups></rule>
+      <rule pattern="(void|bool|char|double|float|int|int8|int16|int32|int64|long|short|size_t|ssize_t|string|time_t|uchar|uint|uint8|uint16|uint32|uint64|ulong|unichar|ushort)\b"><token type="KeywordType"/></rule>
+      <rule pattern="(true|false|null)\b"><token type="NameBuiltin"/></rule>
+      <rule pattern="[a-zA-Z_]\w*"><token type="Name"/></rule>
+    </state>
+    <state name="root">
+      <rule><include state="whitespace"/></rule>
+      <rule><push state="statement"/></rule>
+    </state>
+    <state name="statement">
+      <rule><include state="whitespace"/></rule>
+      <rule><include state="statements"/></rule>
+      <rule pattern="[{}]"><token type="Punctuation"/></rule>
+      <rule pattern=";"><token type="Punctuation"/><pop depth="1"/></rule>
+    </state>
+    <state name="string">
+      <rule pattern="&quot;"><token type="LiteralString"/><pop depth="1"/></rule>
+      <rule pattern="\\([\\abfnrtv&quot;\&#x27;]|x[a-fA-F0-9]{2,4}|[0-7]{1,3})"><token type="LiteralStringEscape"/></rule>
+      <rule pattern="[^\\&quot;\n]+"><token type="LiteralString"/></rule>
+      <rule pattern="\\\n"><token type="LiteralString"/></rule>
+      <rule pattern="\\"><token type="LiteralString"/></rule>
+    </state>
+    <state name="if0">
+      <rule pattern="^\s*#if.*?(?&lt;!\\)\n"><token type="CommentPreproc"/><push/></rule>
+      <rule pattern="^\s*#el(?:se|if).*\n"><token type="CommentPreproc"/><pop depth="1"/></rule>
+      <rule pattern="^\s*#endif.*?(?&lt;!\\)\n"><token type="CommentPreproc"/><pop depth="1"/></rule>
+      <rule pattern=".*?\n"><token type="Comment"/></rule>
+    </state>
+    <state name="class">
+      <rule pattern="[a-zA-Z_]\w*"><token type="NameClass"/><pop depth="1"/></rule>
+    </state>
+    <state name="namespace">
+      <rule pattern="[a-zA-Z_][\w.]*"><token type="NameNamespace"/><pop depth="1"/></rule>
+    </state>
+  </rules>
+</lexer>
diff --git a/lexers/testdata/vala.actual b/lexers/testdata/vala.actual
@@ -0,0 +1,42 @@
+class HelloWorld: Object {
+
+    public signal void some_event ();
+
+    public void method () {
+        some_event ();
+    }
+
+    struct Vector {
+        public double x;
+        public double y;
+        public double z;
+    }
+
+    enum WindowType {
+        TOPLEVEL,
+        POPUP
+    }
+
+    private uint year = 0;
+
+    public HelloWorld () {
+    }
+
+    public HelloWorld.with_year (int year) {
+        if (year>0)
+            this.year = year;
+    }
+
+    public void greeting () {
+        if (year == 0)
+            print ("Hello World\n");
+        else
+            /* Strings prefixed with '@' are string templates. */
+            print (@"Hello World, $(this.year)\n"); 
+    }
+}
+
+void main (string[] args) {
+    var helloworld = new HelloWorld.with_year (2021);
+    helloworld.greeting ();
+}
diff --git a/lexers/testdata/vala.expected b/lexers/testdata/vala.expected
@@ -0,0 +1,214 @@
+[
+  {"type":"KeywordDeclaration","value":"class"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameClass","value":"HelloWorld"},
+  {"type":"Operator","value":":"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"Object"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"{"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"KeywordDeclaration","value":"public"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordDeclaration","value":"signal"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"void"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"some_event"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"();"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"KeywordDeclaration","value":"public"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"void"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"method"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"()"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"{"},
+  {"type":"TextWhitespace","value":"\n        "},
+  {"type":"Name","value":"some_event"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"();"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Punctuation","value":"}"},
+  {"type":"TextWhitespace","value":"\n    \n    "},
+  {"type":"KeywordDeclaration","value":"struct"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"NameClass","value":"Vector"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"{"},
+  {"type":"TextWhitespace","value":"\n        "},
+  {"type":"KeywordDeclaration","value":"public"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"double"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"x"},
+  {"type":"Punctuation","value":";"},
+  {"type":"TextWhitespace","value":"\n        "},
+  {"type":"KeywordDeclaration","value":"public"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"double"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"y"},
+  {"type":"Punctuation","value":";"},
+  {"type":"TextWhitespace","value":"\n        "},
+  {"type":"KeywordDeclaration","value":"public"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"double"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"z"},
+  {"type":"Punctuation","value":";"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Punctuation","value":"}"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"Keyword","value":"enum"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"WindowType"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"{"},
+  {"type":"TextWhitespace","value":"\n        "},
+  {"type":"Name","value":"TOPLEVEL"},
+  {"type":"Punctuation","value":","},
+  {"type":"TextWhitespace","value":"\n        "},
+  {"type":"Name","value":"POPUP"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Punctuation","value":"}"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"KeywordDeclaration","value":"private"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"uint"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"year"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Operator","value":"="},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralNumberInteger","value":"0"},
+  {"type":"Punctuation","value":";"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"KeywordDeclaration","value":"public"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"HelloWorld"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"()"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"{"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Punctuation","value":"}"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"KeywordDeclaration","value":"public"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"HelloWorld"},
+  {"type":"Punctuation","value":"."},
+  {"type":"Name","value":"with_year"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"("},
+  {"type":"KeywordType","value":"int"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"year"},
+  {"type":"Punctuation","value":")"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"{"},
+  {"type":"TextWhitespace","value":"\n        "},
+  {"type":"Keyword","value":"if"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"("},
+  {"type":"Name","value":"year"},
+  {"type":"Operator","value":"\u003e"},
+  {"type":"LiteralNumberInteger","value":"0"},
+  {"type":"Punctuation","value":")"},
+  {"type":"TextWhitespace","value":"\n            "},
+  {"type":"Keyword","value":"this"},
+  {"type":"Punctuation","value":"."},
+  {"type":"Name","value":"year"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Operator","value":"="},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"year"},
+  {"type":"Punctuation","value":";"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Punctuation","value":"}"},
+  {"type":"TextWhitespace","value":"\n\n    "},
+  {"type":"KeywordDeclaration","value":"public"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"KeywordType","value":"void"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"greeting"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"()"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"{"},
+  {"type":"TextWhitespace","value":"\n        "},
+  {"type":"Keyword","value":"if"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"("},
+  {"type":"Name","value":"year"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Operator","value":"=="},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"LiteralNumberInteger","value":"0"},
+  {"type":"Punctuation","value":")"},
+  {"type":"TextWhitespace","value":"\n            "},
+  {"type":"Name","value":"print"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralString","value":"\"Hello World"},
+  {"type":"LiteralStringEscape","value":"\\n"},
+  {"type":"LiteralString","value":"\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n        "},
+  {"type":"Keyword","value":"else"},
+  {"type":"TextWhitespace","value":"\n            "},
+  {"type":"CommentMultiline","value":"/* Strings prefixed with '@' are string templates. */"},
+  {"type":"TextWhitespace","value":"\n            "},
+  {"type":"Name","value":"print"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralString","value":"@\"Hello World, $(this.year)"},
+  {"type":"LiteralStringEscape","value":"\\n"},
+  {"type":"LiteralString","value":"\""},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":" \n    "},
+  {"type":"Punctuation","value":"}"},
+  {"type":"TextWhitespace","value":"\n"},
+  {"type":"Punctuation","value":"}"},
+  {"type":"TextWhitespace","value":"\n\n"},
+  {"type":"KeywordType","value":"void"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"main"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"("},
+  {"type":"KeywordType","value":"string"},
+  {"type":"Punctuation","value":"[]"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"args"},
+  {"type":"Punctuation","value":")"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"{"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"KeywordDeclaration","value":"var"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"helloworld"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Operator","value":"="},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Keyword","value":"new"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Name","value":"HelloWorld"},
+  {"type":"Punctuation","value":"."},
+  {"type":"Name","value":"with_year"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"("},
+  {"type":"LiteralNumberInteger","value":"2021"},
+  {"type":"Punctuation","value":");"},
+  {"type":"TextWhitespace","value":"\n    "},
+  {"type":"Name","value":"helloworld"},
+  {"type":"Punctuation","value":"."},
+  {"type":"Name","value":"greeting"},
+  {"type":"TextWhitespace","value":" "},
+  {"type":"Punctuation","value":"();"},
+  {"type":"TextWhitespace","value":"\n"},
+  {"type":"Punctuation","value":"}"},
+  {"type":"TextWhitespace","value":"\n"}
+]