Simplify the parser

Author: eemeli

parser.pegjs

@@ -1,106 +1,64 @@
-start
-  = messageFormatPattern
+start = token* 
 
-messageFormatPattern
-  = st:(messageFormatElement/string/octothorpe)* {
-      return { type: 'messageFormatPattern', statements: st };
-    }
-
-messageFormatElement
-  = '{' _ argIdx:id efmt:(',' elementFormat)? _ '}' {
-      var res = {
-        type: "messageFormatElement",
-        argumentIndex: argIdx
+token
+  = '{' _ arg:id _ '}' {
+      return {
+        type: 'argument',
+        arg: arg
       };
-      if (efmt && efmt.length) {
-        res.elementFormat = efmt[1];
-      } else {
-        res.output = true;
-      }
-      return res;
-    }
-
-elementFormat
-  = _ t:"plural" _ ',' _ s:pluralFormatPattern _ {
-      return { type: "elementFormat", key: t, val: s };
     }
-  / _ t:"selectordinal" _ ',' _ s:pluralFormatPattern _ {
-      return { type: "elementFormat", key: t, val: s };
-    }
-  / _ t:"select" _ ',' _ s:selectFormatPattern _ {
-      return { type: "elementFormat", key: t, val: s };
-    }
-  / _ t:id p:argStylePattern* {
-      return { type: "elementFormat", key: t, val: p };
+  / '{' _ arg:id _ ',' _ 'select' _ ',' _ cases:selectCase+ _ '}' {
+      return {
+        type: 'select',
+        arg: arg,
+        cases: cases
+      };
     }
-
-pluralFormatPattern
-  = op:offsetPattern? pf:(pluralForm)+ {
-      return { type: "pluralFormatPattern", pluralForms: pf, offset: op || 0 };
+  / '{' _ arg:id _ ',' _ type:('plural'/'selectordinal') _ ',' _ offset:offset? cases:pluralCase+ _ '}' {
+      return {
+        type: type,
+        arg: arg,
+        offset: offset || 0,
+        cases: cases
+      };
     }
-
-offsetPattern
-  = _ "offset" _ ":" _ d:digits _ { return d; }
-
-pluralForm
-  = _ k:pluralKey _ "{" _ mfp:messageFormatPattern _ "}" {
-      return { key: k, val: mfp };
+  / '{' _ arg:id _ ',' _ key:id _ params:functionParams* '}' {
+      return {
+        type: 'function',
+        arg: arg,
+        key: key,
+        params: params
+      };
     }
+  / '#' { return { type: 'octothorpe' }; }
+  / str:char+ { return str.join(''); }
 
-pluralKey
-  = i:id { return i; }
-  / "=" d:digits { return d; }
+id = $([0-9a-zA-Z$_][^ \t\n\r,.+={}]*)
 
-selectFormatPattern
-  = pf:selectForm+ { return { type: "selectFormatPattern", pluralForms: pf }; }
+selectCase = _ key:id _ tokens:caseTokens { return { key: key, tokens: tokens }; }
 
-selectForm
-  = _ k:id _ "{" _ mfp:messageFormatPattern _ "}" {
-      return { key: k, val: mfp };
-    }
-
-argStylePattern
-  = _ "," _ p:id _ { return p; }
+pluralCase = _ key:pluralKey _ tokens:caseTokens { return { key: key, tokens: tokens }; }
 
-octothorpe
-  = '#' { return {type: 'octothorpe'}; };
+caseTokens = '{' (_ & '{')? tokens:token* _ '}' { return tokens; }
 
-string
-  = s:(chars/whitespace)+ { return { type: "string", val: s.join('') }; }
+offset = _ 'offset' _ ':' _ d:digits _ { return d; }
 
-// This is a subset to keep code size down
-// More or less, it has to be a single word
-// that doesn't contain punctuation, etc
-id "identifier"
-  = _ s:$([0-9a-zA-Z$_][^ \t\n\r,.+={}]*) _ { return s; }
+pluralKey = id / '=' d:digits { return d; }
 
-chars
-  = chars:char+ { return chars.join(''); }
+functionParams = _ ',' _ p:id _ { return p; }
 
 char
-  = x:[^{}#\\\0-\x1F\x7f \t\n\r] { return x; }
-  / "\\\\" { return "\\"; }
-  / "\\#" { return "#"; }
-  / "\\{" { return "\u007B"; }
-  / "\\}" { return "\u007D"; }
-  / "\\u" h1:hexDigit h2:hexDigit h3:hexDigit h4:hexDigit {
-      return String.fromCharCode(parseInt("0x" + h1 + h2 + h3 + h4));
+  = [^{}#\\\0-\x08\x0e-\x1f\x7f]
+  / '\\\\' { return '\\'; }
+  / '\\#' { return '#'; }
+  / '\\{' { return '\u007B'; }
+  / '\\}' { return '\u007D'; }
+  / '\\u' h1:hexDigit h2:hexDigit h3:hexDigit h4:hexDigit {
+      return String.fromCharCode(parseInt('0x' + h1 + h2 + h3 + h4));
     }
 
-digits
-  = ds:[0-9]+ {
-    //the number might start with 0 but must not be interpreted as an octal number
-    //Hence, the base is passed to parseInt explicitely
-    return parseInt((ds.join('')), 10);
-  }
-
-hexDigit
-  = [0-9a-fA-F]
+digits = $([0-9]+)
 
-_ "whitespace"
-  = w:whitespace* { return w.join(''); }
+hexDigit = [0-9a-fA-F]
 
-// Whitespace is undefined in the original JSON grammar, so I assume a simple
-// conventional definition consistent with ECMA-262, 5th ed.
-whitespace
-  = [ \t\n\r]
+_ = $([ \t\n\r]*)