metasqlqueryparser-basic.peg

/**
 * MetaSQL PEG.js Grammar
 * ======================
 *
 * This file is a PEG.js [1] Grammar for xTuple's MetaSQL language. It is used
 * to generate a Javascipt parser that can be passed a MetaSQL query string and
 * an options object containing a params object. The parser returns a SQl query
 * string with all MetaSQL parsed and replaced with the appropriate params
 * values.
 *
 * MetaSQL[2] is a scripting language developed by xTuple for use by the report
 * writer. The language is designed to handle dynamic database queries. MetaSQL
 * statements are embedded within standard SQL—for example, within the Query
 * Source of a report definition. When a report is run, a parsing engine
 * interprets the MetaSQL using a list of named parameters. The parsed result
 * is standard SQL, which in turn is sent to the target database.
 *
 * [1] http://pegjs.org/
 * [2] http://www.xtuple.org/node/270
 *
 * Example generated parser usage:
 *  var options = {
 *        params: {
 *          foo: 'bar'
 *        }
 *      },
 *      metaSQL: 'SELECT <? value("foo") ?> AS baz;';
 *
 *  parser.parse(metaSQL, options); // returns "SELECT 'bar' AS baz;"
 *
 * TODO:
 * - Add legacy mode flag
 * - break - currently fails like the old C++ parser.
 * - continue - currently fails like the old C++ parser.
 * - Support parameterized query output.
 */

/**
 * PEG.js grammar initializer
 */
{
  // The default params below are for the MetaSQL test here: https://github.com/xtuple/openrpt/blob/master/MetaSQL/test.mql
  params = options.params ? options.params : {
      start: true
      ,test_0_1_0: 'test_0_1_0'
      ,test_0_1_2: ['test_0_1_2']
      ,test_0_1_3: ['test_0_1_3', 'test_0_1_3_2', 'test_0_1_3_3']
      ,test_0_1_4: 'true THEN true END AS test_0_1_4) AS tests; SELECT current_user--'
      ,test_0_2_0: 'test_0_2_0'
      ,test_0_2_2: ['test_0_2_2']
      ,test_0_2_3: ['test_0_2_3','test_0_2_3_2', 'test_0_2_3_3']
      ,test_0_2_4: "(SELECT usr_username FROM usr WHERE usr_username = 'admin')"
      ,test_0_3_0: true
      ,test_0_3_2: null
      ,test_0_3_3: ''
      ,test_0_3_4: 'true THEN true END AS test_0_3_4) AS test; SELECT current_user--'
      ,test_0_4_0a: 'test_0_4_0'
      ,test_0_4_2a: null
      ,test_0_4_3a: ''
      ,test_0_4_4a: 'true THEN true END AS test_0_4_4) AS test; SELECT current_user--'
      ,test_0_5_0: ['test_0_5_0']
      ,test_0_5_1: ['test_0_5_1','test_0_5_1_2', 'test_0_5_1_3']
      ,test_0_5_2: 'test_0_5_2'
      ,test_0_5_4: ['true THEN true END AS test_0_5_4) AS test; SELECT current_user--']
      ,test_0_6_0: ['test_0_6_0']
      ,test_0_6_1: ['test_0_6_1','test_0_6_1_2', 'test_0_6_1_3']
      ,test_0_6_2: 'test_0_6_2'
      ,test_0_6_4: ['true THEN true END AS test_0_6_4) AS test; SELECT current_user--']
      ,test_1_3_0: 'test_1_3_0'
      ,test_1_3_1: 'test_1_3_1'
      ,test_1_3_1_2: 'test_1_3_1_2'
      ,test_2_3_0: 'test_2_3_0'
      ,test_2_3_1: 'test_2_3_1'
      ,test_3_3_0: 'test_3_3_0'
      ,test_3_3_1: 'test_3_3_1'
      ,test_3_3_2: 'test_3_3_2'
      ,test_4_0_0: ['test_4_0_0', 'test_4_0_0_2', 'test_4_0_0_3']
      ,test_4_0_1: ['true THEN true END AS test_4_0_1) AS test; SELECT current_user--']
      ,test_41_0_0: ['test_41_0_0', 'test_41_0_0_2', 'test_41_0_0_3']
      ,test_42_0_0: ['test_42_0_0', 'test_42_0_0_2', 'test_42_0_0_3']
      ,test_43_0_0: ['test_43_0_0', 'test_43_0_0_2', 'test_43_0_0_3']
      ,test_4_1_0: 'test_4_1_0'
      ,test_4_1_2: ['test_4_1_2']
      ,test_4_1_3: ['test_4_1_3', 'test_4_1_3_2', 'test_4_1_3_3']
      ,test_4_1_4: ['test_4_1_4', 'test_4_1_4_2', 'test_4_1_4_3']
      ,test_4_1_42: 'test_4_1_42'
      ,test_4_1_5: ['test_4_1_5', 'test_4_1_5_2', 'test_4_1_5_3']
      ,test_4_1_52: ['test_4_1_52']
      ,test_4_1_6: ['test_4_1_6', 'test_4_1_6_2', 'test_4_1_6_3']
      ,test_4_1_62: ['test_4_1_62', 'test_4_1_62_2', 'test_4_1_62_3']
      ,test_4_1_7: [1]
      ,test_4_1_8: [true]
      ,test_4_2_0: 'test_4_2_0'
      ,test_4_2_2: ['test_4_2_2']
      ,test_4_2_3: ['test_4_2_3', 'test_4_2_3_2', 'test_4_2_3_3']
      ,test_4_2_4: ['test_4_2_4', 'test_4_2_4_2', 'test_4_2_4_3']
      ,test_4_2_42: 'test_4_2_42'
      ,test_4_2_5: ['test_4_2_5', 'test_4_2_5_2', 'test_4_2_5_3']
      ,test_4_2_52: ['test_4_2_52']
      ,test_4_2_6: ['test_4_2_6', 'test_4_2_6_2', 'test_4_2_6_3']
      ,test_4_2_62: ['test_4_2_62', 'test_4_2_62_2', 'test_4_2_62_3']
      ,test_4_3_0: ['test_4_3_0']
      ,test_4_3_01: 'test_4_3_01'
      ,test_4_3_1: ['test_4_3_1']
      ,test_4_3_2: ['test_4_3_2']
      ,test_4_3_21: null
      ,test_4_3_3: ['test_4_3_3']
      ,test_4_3_31: ''
      ,test_4_4_0: ['test_4_4_0']
      ,test_4_4_01: 'test_4_4_01'
      ,test_4_4_1: ['test_4_4_1']
      ,test_4_4_2: ['test_4_4_2']
      ,test_4_4_21: null
      ,test_4_4_3: ['test_4_4_3']
      ,test_4_4_31: ''
      ,test_4_5_0: ['test_4_5_0']
      ,test_4_5_1: ['test_4_5_1', 'test_4_5_1_2']
      ,test_4_5_2: ['test_4_5_2', 'test_4_5_2_2']
      ,test_4_5_3: ['test_4_5_3']
      ,test_4_5_32: 'test_4_5_32'
      ,test_4_5_4: ['test_4_5_4']
      ,test_4_6_0: ['test_4_6_0']
      ,test_4_6_1: ['test_4_6_1', 'test_4_6_1_2']
      ,test_4_6_2: ['test_4_6_2', 'test_4_6_2_2']
      ,test_4_6_3: ['test_4_6_3']
      ,test_4_6_32: 'test_4_6_32'
      ,test_4_6_4: ['test_4_6_4']
      ,test_4_7_0: ['test_4_7_0', 'test_4_7_0_2', 'test_4_7_0_3']
      ,test_4_8_0: ['test_4_8_0', 'test_4_8_0_2', 'test_4_8_0_3']
      ,test_44_0_0: ['test_44_0_0', 'test_44_0_0_2']
      ,test_44_0_02: ['test_44_0_02', 'test_44_0_02_2']
      ,test_44_0_1: ['test_44_0_1', 'test_44_0_1_2']
      ,test_441_0_0: ['test_441_0_0', 'test_441_0_0_2']
      ,test_441_0_02: ['test_441_0_02', 'test_441_0_02_2']
      ,test_442_0_0: ['test_442_0_0', 'test_442_0_0_2']
      ,test_442_0_02: ['test_442_0_02', 'test_442_0_02_2']
      ,test_443_0_0: ['test_443_0_0', 'test_443_0_0_2', 'test_443_0_0_3']
      ,test_443_0_02: ['test_443_0_02', 'test_443_0_02_2', 'test_443_0_02_3']
      ,test_444_0_0: ['test_444_0_0', 'test_444_0_0_2']
      ,test_444_0_02: ['test_444_0_02', 'test_444_0_01_2']
      ,test_444_0_03: ['test_444_0_03', 'test_444_0_01_3']
      ,end: true
    }
  ,loopstack = options.loopstack ? options.loopstack : []
  ,stackIdx = function (id) {
      var idx = 1;

      if (loopstack.length > 0) {
        for (idx = 0; idx < loopstack.length; idx++) {
          if (loopstack[idx].id === id) {
            break;
          }
        }
      }

      return idx;
    }
  ,extractList = function (list, index) {
      var result = new Array(list.length),
          i;

      for (i = 0; i < list.length; i++) {
        result[i] = list[i][index];
      }

      return result;
    }
  ,buildList = function (first, rest, index) {
      return [first].concat(extractList(rest, index));
    }
  ,optionalList = function (value) {
      return value !== null ? value : [];
    }
}

/**
 * Grammar startRule
 */
Start
  = start:__ metasql:MetaSQL end:__ {
      return start.join('') + metasql + end.join('');
    }

/**
 * MetaSQL Grammer
 */
MetaSQL
  = sql:SourceElements? {
      return sql;
    }

SourceElements
  = first:(SourceElement) rest:(SourceElement)* {
      return first + rest.join('');
    }

SourceElement
  = quoted:QuotedString {
      //return "quoted: " + quoted;
      return quoted;
    }
  / comment:Comment {
      //return "comment: " + comment;
      return comment;
    }
  / state:Statement {
      //return "statement: " + state;
      return state;
    }
  / sql:SQLText {
      //return "sql: " + sql;
      return sql;
    }

/**
 * MetaSQL Statements
 */
Statement
  = value:MQLValue {
      //return "MQLValue: " + value;
      return value;
    }
  / literal:MQLLiteral {
      //return "MQLLiteral: " + literal;
      return literal;
    }
  / loopcontrol:MQLLoopControl {
      //return "MQLLoopControl: " + loopcontrol;
      return loopcontrol;
    }
  / rawcond:MQLRawMQLCondition {
      //return "MQLRawMQLCondition: " + rawcond;
      return rawcond;
    }
  / controlstate:MQLControlStatement {
      //return "MQLControlStatement: " + controlstate;
      return controlstate;
    }
  / generic:MQLGeneric {
      //return "MQLGeneric: " + generic;
      return generic;
    }

MQLValue
  = MQLOpen WhiteSpace? Value WhiteSpace? '(' WhiteSpace? id:QuotedIdentifier WhiteSpace? ')' WhiteSpace? MQLClose {
      var stackidx,
          listidx;

      if (Object.prototype.toString.call(params[id]) === '[object Array]') {
        stackidx = stackIdx(id);

        if (stackidx < loopstack.length) {
          listidx = loopstack[stackidx].count;

          if (listidx >= params[id].length) {
            listidx = params[id].length - 1;
          }

          if (typeof params[id][0] == 'string') {
            return ("'" + params[id][listidx] + "'");
          } else {
            return params[id][listidx];
          }
        }
        else {
          // Not in a foreach loop
          if (typeof params[id][0] == 'string') {
            return "'" + params[id][0] + "'";
          } else {
            return params[id][0];
          }
        }
      }

      if (typeof params[id] == 'string') {
        return ("'" + (params[id] ? params[id] : 'NULL ') + "'");
      } else {
        return (params[id] ? params[id] : 'NULL ');
      }
    }

MQLLiteral
  = MQLOpen WhiteSpace? Literal WhiteSpace? '(' WhiteSpace? id:QuotedIdentifier WhiteSpace? ')' WhiteSpace? MQLClose {
      var stackidx,
          listidx;

      if (Object.prototype.toString.call(params[id]) === '[object Array]') {
        stackidx = stackIdx(id);

        if (stackidx < loopstack.length) {
          listidx = loopstack[stackidx].count;

          if (listidx >= params[id].length) {
            listidx = params[id].length - 1;
          }

          return params[id][listidx];
        }
        else {
          // Not in a foreach loop
          return params[id][0];
        }
      }

      return (params[id] ? params[id] : '');
    }

MQLLoopControl
  = MQLOpen WhiteSpace? loop:MQLLoopFunction WhiteSpace? MQLClose {
      return loop;
    }

MQLRawMQLCondition
  = MQLOpen WhiteSpace? cond:MQLCondition WhiteSpace? MQLClose {
      return cond;
    }

MQLControlStatement
  = ifcond:MQLIf ifbody:(!(MQLElse / MQLElseIf / MQLEndIf) SourceElement)* MQLEndIf {
      if (ifcond) {
        return (ifbody ? extractList(ifbody, 1).join('') : '');
      } else {
        return '';
      }
    }
  / ifcond:MQLIf ifbody:(!(MQLElse / MQLElseIf / MQLEndIf) SourceElement)*
    MQLElse elsebody:(!(MQLElseIf / MQLEndIf) SourceElement)* MQLEndIf {
      if (ifcond) {
        return (ifbody ? extractList(ifbody, 1).join('') : '');
      } else {
        return (elsebody ? extractList(elsebody, 1).join('') : '');
      }
    }
  / ifcond:MQLIf ifbody:(!(MQLElse / MQLElseIf / MQLEndIf) SourceElement)*
    elseifwraper:(MQLElseIf (!(MQLElseIf / MQLElse / MQLEndIf) SourceElement)*)*
    MQLElse elsebody:(!(MQLEndIf) SourceElement)* MQLEndIf {
      if (ifcond) {
        return (ifbody ? extractList(ifbody, 1).join('') : '');
      }

      if (elseifwraper !== null) {
        for (var i = 0; i < elseifwraper.length; i++) {
          // Find a valid elseif if any exists.
          if (elseifwraper[i][0]) {
            return (elseifwraper[i][1] ? extractList(elseifwraper[i][1], 1).join('') : '');
          }
        }
      }

      // The ifcond and elseifwraper conditions did not match.
      return (elsebody ? extractList(elsebody, 1).join('') : '');
    }
  / wrapper:ForEachWrapper {
      var body = wrapper.body,
          foreachcond = wrapper.foreachstart.id,
          foreachtext = wrapper.foreachstart.text,
          extractedbody = '',
          result = '',
          stackidx = stackIdx(foreachcond),
          extractForeach = function (foreachbody) {
            var foreachraw = '';

            for (var k = 0; k < foreachbody.length; k++) {
              // Index [k][1] is a nested foreach, build it recursively.
              if (foreachbody[k][1]) {
                foreachraw += foreachbody[k][1].foreachstart.text;
                foreachraw += extractForeach(foreachbody[k][1].body);
                foreachraw += foreachbody[k][1].foreachend;
              }

              // Index [k][2] is this foreach's body text.
              foreachraw += foreachbody[k][2];
            }

            return foreachraw;
          };

      extractedbody += extractForeach(body);

      if (params[foreachcond]) {
        if (Object.prototype.toString.call(params[foreachcond]) === '[object Array]') {
          for (var i = 0; i < params[foreachcond].length; i++) {
            var recurseOptions = {
                  loopstack: loopstack
                };

            // Call the parser recursively on the body.
            result += parser.parse(extractedbody, recurseOptions);

            loopstack[stackidx].count++;
          }

          loopstack.shift().id;
          return result;
        } else {
          return '';
        }
      }
    }

MQLGeneric
  = start:MQLOpen body:(!MQLClose SourceCharacter)* end:MQLClose {
      return start + (body ? extractList(body, 1).join('') : '') + end;
    }

/**
 * MetaSQL Helpers
 */
MQLCondition
  = Not WhiteSpace+ Exists WhiteSpace? '(' WhiteSpace? id:QuotedIdentifier WhiteSpace? ')' {
      return ! (id in params);
    }
  / Exists WhiteSpace? '(' WhiteSpace? id:QuotedIdentifier WhiteSpace? ')' {
      return (id in params);
    }
  / ReExists WhiteSpace? '(' WhiteSpace? re:RegExp WhiteSpace? ')' {
      for (var prop in params) {
        if (prop.match(re)) {
          return true;
        }
      }

      return '';
    }
  / IsFirst WhiteSpace? '(' WhiteSpace? id:QuotedIdentifier WhiteSpace? ')' {
      var stackidx = stackIdx(id);

      if (Object.prototype.toString.call(params[id]) === '[object Array]') {
        if (stackidx < loopstack.length) {
          listidx = loopstack[stackidx].count;

          // Figure out if this is the first item in the array.
          if (listidx === 0) {
            return true;
          } else {
            return false;
          }
        }
        else {
          // Not in a foreach loop
          return (id in params);
        }
      } else {
        // Not a list parameter or does not exist.
        return true;
      }
    }
  / IsLast  WhiteSpace? '(' WhiteSpace? id:QuotedIdentifier WhiteSpace? ')' {
      var stackidx = stackIdx(id);

      if (Object.prototype.toString.call(params[id]) === '[object Array]') {
        if (stackidx < loopstack.length) {
          listidx = loopstack[stackidx].count;

          // Figure out if this is the last item in the array.
          if (listidx === (params[id].length - 1)) {
            return true;
          } else {
            return false;
          }
        }
        else {
          // Not in a foreach loop
          if ((id in params) && params[id].length === 1) {
            return true;
          } else {
            return false;
          }
        }
      } else {
        // Not a list parameter or does not exist.
        return false;
      }
    }

MQLIf
  = MQLOpen WhiteSpace? If WhiteSpace? cond:MQLCondition WhiteSpace? MQLClose {
      return cond;
    }

MQLElseIf
  = MQLOpen WhiteSpace? ElseIf WhiteSpace? cond:MQLCondition WhiteSpace? MQLClose {
      return cond;
    }

MQLElse
  = MQLOpen WhiteSpace? Else WhiteSpace? MQLClose

MQLEndIf
  = MQLOpen WhiteSpace? EndIf WhiteSpace? MQLClose

MQLForeach
  = MQLOpen WhiteSpace? Foreach WhiteSpace? '(' WhiteSpace? id:QuotedIdentifier WhiteSpace? ')' WhiteSpace? MQLClose {
      loopstack.unshift({ id: id, count: 0 });

      return {
        text: text(),
        id: id
      };
    }

MQLEndForeach
  = MQLOpen WhiteSpace? EndForeach WhiteSpace? MQLClose {
      return text();
    }

ForEachWrapper
  = foreachstart:MQLForeach body:(!(MQLEndForeach) ForEachWrapper? SourceCharacter)* foreachend:MQLEndForeach {
      var wrapper = {
        foreachstart: foreachstart,
        body: body,
        foreachend: foreachend
      };

      return wrapper;
    }

MQLLoopFunction
  = Continue WhiteSpace? loop:(MQLLoopCount)? {
      // TODO: Figure out what the passed loop count is.
      // TODO: Figure out where we are in the loop.
      // TODO: Continue logic.
      loop = loop || 1;
      return '';
    }
  / Break WhiteSpace? loop:(MQLLoopCount)? {
      // TODO: Figure out what the passed loop count is.
      // TODO: Figure out where we are in the loop.
      // TODO: Break logic.
      loop = loop || 1;
      return '';
    }

MQLLoopCount
  = '(' WhiteSpace? loop:DecimalDigit* WhiteSpace? ')' {
      return loop;
    }

MQLOpen    = '<?'
MQLClose   = '?>'
Value      = [Vv] [Aa] [Ll] [Uu] [Ee]
Literal    = [Ll] [Ii] [Tt] [Ee] [Rr] [Aa] [Ll]
If         = [Ii] [Ff]
ElseIf     = [Ee] [Ll] [Ss] [Ee] [Ii] [Ff]
Else       = [Ee] [Ll] [Ss] [Ee]
EndIf      = [Ee] [Nn] [Dd] [Ii] [Ff]
Not        = [Nn] [Oo] [Tt]
Exists     = [Ee] [Xx] [Ii] [Ss] [Tt] [Ss]
ReExists   = [Rr] [Ee] [Ee] [Xx] [Ii] [Ss] [Tt] [Ss]
IsFirst    = [Ii] [Ss] [Ff] [Ii] [Rr] [Ss] [Tt]
IsLast     = [Ii] [Ss] [Ll] [Aa] [Ss] [Tt]
Foreach    = [Ff] [Oo] [Rr] [Ee] [Aa] [Cc] [Hh]
EndForeach = [Ee] [Nn] [Dd] [Ff] [Oo] [Rr] [Ee] [Aa] [Cc] [Hh]
Continue   = [Cc] [Oo] [Nn] [Tt] [Ii] [Nn] [Uu] [Ee]
Break      = [Bb] [Rr] [Ee] [Aa] [Kk]

/**
 * Simple SQL Grammer
 */
SQLText
  = text:(SQLPattern) {
      return (text ? text : '');
    }

SQLPattern
  = body:SourceCharacter {
      return (body ? body : '');
    }

/**
 * Generic Grammar
 */
SourceCharacter
  = .

WhiteSpace "whitespace"
  = "\t"
  / "\v"
  / "\f"
  / " "
  / "\u00A0"
  / "\uFEFF"
  / Zs

LineTerminator
  = line:[\n\r\u2028\u2029] {
      return line + '&';
    }

LineTerminatorSequence "end of line"
  = "\n"
  / "\r\n"
  / "\r"
  / "\u2028"
  / "\u2029"

DecimalDigit
  = [0-9]

HexDigit
  = [0-9a-f]i

RegExp
  = "'" str:([^']*) "'" {
      return str.join("");
    }
  / '"' str:([^"]*) '"' {
      return str.join("");
    }

Comment "comment"
  = MultiLineComment
  / SingleLineComment

MultiLineComment
  = start:"/*" body:(!"*/" SourceCharacter)* end:"*/" {
      return start + (body ? extractList(body, 1).join('') : '') + end;
    }

MultiLineCommentNoLineTerminator
  = start:"/*" body:(!("*/" / LineTerminator) SourceCharacter)* end:"*/" {
      return start + (body ? extractList(body, 1).join('') : '') + end;
    }

SingleLineComment
  = start:"--" body:(!LineTerminator SourceCharacter)* {
      return start + (body ? extractList(body, 1).join('') : '');
    }

QuotedIdentifier "quoted identifier"
  = SingleQuotedIdentifier
  / DoubleQuotedIdentifier

SingleQuotedIdentifier
  = start:"'" body:SingleStringCharacter* end:"'" {
      return (body ? body.join('') : '');
    }

DoubleQuotedIdentifier
  = start:'"' body:DoubleStringCharacter* end:'"' {
      return (body ? body.join('') : '');
    }

QuotedString "quoted string"
  = SingleQuotedString
  / DoubleQuotedString

SingleQuotedString
  = start:"'" body:SingleStringCharacter* end:"'" {
      return start + (body ? body.join('') : '') + end;
    }

DoubleQuotedString
  = start:'"' body:DoubleStringCharacter* end:'"' {
      return start + (body ? body.join('') : '') + end;
    }

DoubleStringCharacter
  = !('"' / "\\" / LineTerminator) SourceCharacter {
      return text();
    }
  / "\\" sequence:EscapeSequence {
      return sequence;
    }
  / LineContinuation

SingleStringCharacter
  = !("'" / "\\" / LineTerminator) SourceCharacter {
      return text();
    }
  / "\\" sequence:EscapeSequence {
      return sequence;
    }
  / LineContinuation

LineContinuation
  = "\\" LineTerminatorSequence {
      return "";
    }

EscapeSequence
  = CharacterEscapeSequence
  / "0" !DecimalDigit {
      return "\0";
    }
  / HexEscapeSequence
  / UnicodeEscapeSequence

CharacterEscapeSequence
  = SingleEscapeCharacter
  / NonEscapeCharacter

SingleEscapeCharacter
  = "'"
  / '"'
  / "\\"
  / "b"  {
      return "\b";
    }
  / "f"  {
      return "\f";
    }
  / "n"  {
      return "\n";
    }
  / "r"  {
      return "\r";
    }
  / "t"  {
      return "\t";
    }
  / "v"  {
      return "\x0B"; // IE does not recognize "\v".
    }

NonEscapeCharacter
  = !(EscapeCharacter / LineTerminator) SourceCharacter {
      return text();
    }

EscapeCharacter
  = SingleEscapeCharacter
  / DecimalDigit
  / "x"
  / "u"

HexEscapeSequence
  = "x" digits:$(HexDigit HexDigit) {
      return String.fromCharCode(parseInt(digits, 16));
    }

UnicodeEscapeSequence
  = "u" digits:$(HexDigit HexDigit HexDigit HexDigit) {
      return String.fromCharCode(parseInt(digits, 16));
    }
__
  = (WhiteSpace / LineTerminatorSequence / Comment)*

_
  = (WhiteSpace / MultiLineCommentNoLineTerminator)*

// Separator, Space
Zs = [\u0020\u00A0\u1680\u2000-\u200A\u202F\u205F\u3000]