var SparqlParser = Editor.Parser = (function() {
  function wordRegexp(words) {
    return new RegExp("^(?:" + words.join("|") + ")$", "i");
  }
  var ops = wordRegexp(["str", "lang", "langmatches", "datatype", "bound", "sameterm", "isiri", "isuri",
                        "isblank", "isliteral", "union", "a",
                        // SPARQL 1.1
                        "abs", "ceil", "floor", "round", "concat", "substring", "strlen", "ucase",
                        "lcase", "encode", "contains", "starts", "ends", "rand", "strdt", "strlang",
                        "bnode", "iri", "uri", "datatype", "regex", "isnumeric", "if", "coalesce",
                        "count", "sum", "min", "max", "avg", "group_concat", "sample"
                        ]);
  var keywords = wordRegexp(["base", "prefix", "select", "distinct", "reduced", "construct", "describe",
                             "ask", "from", "named", "where", "order", "limit", "offset", "filter", "optional",
                             "graph", "by", "asc", "desc",
                             // SPARQL 1.1
                             "group",
                             "insert", "delete", "data", "load", "clear", "drop", "graph", "default",
                             "using", "with"]);
  var operatorChars = /[*+\-<>=&|]/;

  var tokenizeSparql = (function() {
    function normal(source, setState) {
      var ch = source.next();
      if (ch == "$" || ch == "?") {
        source.nextWhileMatches(/[\w\d]/);
        return "sp-var";
      }
      else if (ch == "<" && !source.matches(/[\s\u00a0=]/)) {
        source.nextWhileMatches(/[^\s\u00a0>]/);
        if (source.equals(">")) source.next();
        return "sp-uri";
      }
      else if (ch == "\"" || ch == "'") {
        setState(inLiteral(ch));
        return null;
      }
      else if (/[{}\(\),\.;\[\]]/.test(ch)) {
        return "sp-punc";
      }
      else if (ch == "#") {
        while (!source.endOfLine()) source.next();
        return "sp-comment";
      }
      else if (operatorChars.test(ch)) {
        source.nextWhileMatches(operatorChars);
        return "sp-operator";
      }
      else if (ch == ":") {
        source.nextWhileMatches(/[\w\d\._\-]/);
        return "sp-prefixed";
      }
      else {
        source.nextWhileMatches(/[_\w\d]/);
        if (source.equals(":")) {
          source.next();
          source.nextWhileMatches(/[\w\d_\-]/);
          return "sp-prefixed";
        }
        var word = source.get(), type;
        if (ops.test(word))
          type = "sp-operator";
        else if (keywords.test(word))
          type = "sp-keyword";
        else
          type = "sp-word";
        return {style: type, content: word};
      }
    }

    function inLiteral(quote) {
      return function(source, setState) {
        var escaped = false;
        while (!source.endOfLine()) {
          var ch = source.next();
          if (ch == quote && !escaped) {
            setState(normal);
            break;
          }
          escaped = !escaped && ch == "\\";
        }
        return "sp-literal";
      };
    }

    return function(source, startState) {
      return tokenizer(source, startState || normal);
    };
  })();

  function indentSparql(context) {
    return function(nextChars) {
      var firstChar = nextChars && nextChars.charAt(0);
      if (/[\]\}]/.test(firstChar))
        while (context && context.type == "pattern") context = context.prev;

      var closing = context && firstChar == matching[context.type];
      if (!context)
        return 0;
      else if (context.type == "pattern")
        return context.col;
      else if (context.align)
        return context.col - (closing ? context.width : 0);
      else
        return context.indent + (closing ? 0 : indentUnit);
    }
  }

  function parseSparql(source) {
    var tokens = tokenizeSparql(source);
    var context = null, indent = 0, col = 0;
    function pushContext(type, width) {
      context = {prev: context, indent: indent, col: col, type: type, width: width};
    }
    function popContext() {
      context = context.prev;
    }

    var iter = {
      next: function() {
        var token = tokens.next(), type = token.style, content = token.content, width = token.value.length;

        if (content == "\n") {
          token.indentation = indentSparql(context);
          indent = col = 0;
          if (context && context.align == null) context.align = false;
        }
        else if (type == "whitespace" && col == 0) {
          indent = width;
        }
        else if (type != "sp-comment" && context && context.align == null) {
          context.align = true;
        }

        if (content != "\n") col += width;

        if (/[\[\{\(]/.test(content)) {
          pushContext(content, width);
        }
        else if (/[\]\}\)]/.test(content)) {
          while (context && context.type == "pattern")
            popContext();
          if (context && content == matching[context.type])
            popContext();
        }
        else if (content == "." && context && context.type == "pattern") {
          popContext();
        }
        else if ((type == "sp-word" || type == "sp-prefixed" || type == "sp-uri" || type == "sp-var" || type == "sp-literal") &&
                 context && /[\{\[]/.test(context.type)) {
          pushContext("pattern", width);
        }

        return token;
      },

      copy: function() {
        var _context = context, _indent = indent, _col = col, _tokenState = tokens.state;
        return function(source) {
          tokens = tokenizeSparql(source, _tokenState);
          context = _context;
          indent = _indent;
          col = _col;
          return iter;
        };
      }
    };
    return iter;
  }

  return {make: parseSparql, electricChars: "}]"};
})();