Ejemplo n.º 1
0
ace.define("ace/theme/sense-dark", ['require', 'exports', 'module'],
  function (require, exports) {
    exports.isDark = true;
    exports.cssClass = "ace-sense-dark";
    exports.cssText = ".ace-sense-dark .ace_gutter {\
background: #2e3236;\
color: #bbbfc2;\
}\
.ace-sense-dark .ace_print-margin {\
width: 1px;\
background: #555651\
}\
.ace-sense-dark .ace_scroller {\
background-color: #202328;\
}\
.ace-sense-dark .ace_content {\
}\
.ace-sense-dark .ace_text-layer {\
color: #F8F8F2\
}\
.ace-sense-dark .ace_cursor {\
border-left: 2px solid #F8F8F0\
}\
.ace-sense-dark .ace_overwrite-cursors .ace_cursor {\
border-left: 0px;\
border-bottom: 1px solid #F8F8F0\
}\
.ace-sense-dark .ace_marker-layer .ace_selection {\
background: #222\
}\
.ace-sense-dark.ace_multiselect .ace_selection.ace_start {\
box-shadow: 0 0 3px 0px #272822;\
border-radius: 2px\
}\
.ace-sense-dark .ace_marker-layer .ace_step {\
background: rgb(102, 82, 0)\
}\
.ace-sense-dark .ace_marker-layer .ace_bracket {\
margin: -1px 0 0 -1px;\
border: 1px solid #49483E\
}\
.ace-sense-dark .ace_marker-layer .ace_active-line {\
background: #202020\
}\
.ace-sense-dark .ace_gutter-active-line {\
background-color: #272727\
}\
.ace-sense-dark .ace_marker-layer .ace_selected-word {\
border: 1px solid #49483E\
}\
.ace-sense-dark .ace_invisible {\
color: #49483E\
}\
.ace-sense-dark .ace_entity.ace_name.ace_tag,\
.ace-sense-dark .ace_keyword,\
.ace-sense-dark .ace_meta,\
.ace-sense-dark .ace_storage {\
color: #F92672\
}\
.ace-sense-dark .ace_constant.ace_character,\
.ace-sense-dark .ace_constant.ace_language,\
.ace-sense-dark .ace_constant.ace_numeric,\
.ace-sense-dark .ace_constant.ace_other {\
color: #AE81FF\
}\
.ace-sense-dark .ace_invalid {\
color: #F8F8F0;\
background-color: #F92672\
}\
.ace-sense-dark .ace_invalid.ace_deprecated {\
color: #F8F8F0;\
background-color: #AE81FF\
}\
.ace-sense-dark .ace_support.ace_constant,\
.ace-sense-dark .ace_support.ace_function {\
color: #66D9EF\
}\
.ace-sense-dark .ace_fold {\
background-color: #A6E22E;\
border-color: #F8F8F2\
}\
.ace-sense-dark .ace_storage.ace_type,\
.ace-sense-dark .ace_support.ace_class,\
.ace-sense-dark .ace_support.ace_type {\
font-style: italic;\
color: #66D9EF\
}\
.ace-sense-dark .ace_entity.ace_name.ace_function,\
.ace-sense-dark .ace_entity.ace_other.ace_attribute-name,\
.ace-sense-dark .ace_variable {\
color: #A6E22E\
}\
.ace-sense-dark .ace_variable.ace_parameter {\
font-style: italic;\
color: #FD971F\
}\
.ace-sense-dark .ace_string {\
color: #E6DB74\
}\
.ace-sense-dark .ace_comment {\
color: #629755\
}\
.ace-sense-dark .ace_markup.ace_underline {\
text-decoration: underline\
}\
.ace-sense-dark .ace_indent-guide {\
background: url(data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAACCAYAAACZgbYnAAAAEklEQVQImWNQ11D6z7Bq1ar/ABCKBG6g04U2AAAAAElFTkSuQmCC) right repeat-y\
}";

    var dom = require("ace/lib/dom");
    dom.importCssString(exports.cssText, exports.cssClass);
  });
Ejemplo n.º 2
0
module.exports = function (editor) {

  function isSeparatorToken(token) {
    switch ((token || {}).type) {
      case "url.slash":
      case "url.comma":
      case "url.questionmark":
      case "paren.lparen":
      case "paren.rparen":
      case "punctuation.colon":
      case "punctuation.comma":
      case "whitespace":
        return true;
      default:
        // standing on white space, quotes or another punctuation - no replacing
        return false;
    }
  }

  function isUrlPathToken(token) {
    switch ((token || {}).type) {
      case "url.slash":
      case "url.comma":
      case "url.part":
        return true;
      default:
        return false;
    }
  }

  function isUrlParamsToken(token) {
    switch ((token || {}).type) {
      case "url.param":
      case "url.equal":
      case "url.value":
      case "url.questionmark":
      case "url.amp":
        return true;
      default:
        return false;
    }
  }

  function getAutoCompleteValueFromToken(token) {
    switch ((token || {}).type) {
      case "variable":
      case "string":
      case "text":
      case "constant.numeric":
      case "constant.language.boolean":
        return token.value.replace(/"/g, '');
      case "method":
      case "url.part":
        return token.value;
      default:
        // standing on white space, quotes or another punctuation - no replacing
        return "";
    }
  }

  function addMetaToTermsList(list, meta, template) {
    return _.map(list, function (t) {
      if (typeof t !== "object") {
        t = { name: t };
      }
      return _.defaults(t, { meta: meta, template: template });
    });
  }

  function termToFilterRegex(term, prefix, suffix) {
    if (!prefix) {
      prefix = "";
    }
    if (!suffix) {
      suffix = "";
    }

    return new RegExp(prefix + term.replace(/[-\/\\^$*+?.()|[\]{}]/g, '\\$&') + suffix, 'i');
  }

  function applyTerm(term) {
    var session = editor.getSession();

    var context = term.context;

    // make sure we get up to date replacement info.
    addReplacementInfoToContext(context, editor.getCursorPosition(), term.insert_value);

    var termAsString;
    if (context.autoCompleteType == "body") {
      termAsString = typeof term.insert_value == "string" ? '"' + term.insert_value + '"' : term.insert_value + "";
      if (term.insert_value === "[" || term.insert_value === "{") {
        termAsString = "";
      }
    }
    else {
      termAsString = term.insert_value + "";
    }

    var valueToInsert = termAsString;
    var templateInserted = false;
    if (context.addTemplate && !_.isUndefined(term.template) && !_.isNull(term.template)) {
      var indentedTemplateLines = utils.jsonToString(term.template, true).split("\n");
      var currentIndentation = session.getLine(context.rangeToReplace.start.row);
      currentIndentation = currentIndentation.match(/^\s*/)[0];
      for (var i = 1; i < indentedTemplateLines.length; i++) // skip first line
        indentedTemplateLines[i] = currentIndentation + indentedTemplateLines[i];

      valueToInsert += ": " + indentedTemplateLines.join("\n");
      templateInserted = true;
    }
    else {
      templateInserted = true;
      if (term.value === "[") {
        valueToInsert += "[]";
      }
      else if (term.value == "{") {
        valueToInsert += "{}";
      }
      else {
        templateInserted = false;
      }
    }

    valueToInsert = context.prefixToAdd + valueToInsert + context.suffixToAdd;

    // disable listening to the changes we are making.
    removeChangeListener();

    if (context.rangeToReplace.start.column != context.rangeToReplace.end.column) {
      session.replace(context.rangeToReplace, valueToInsert);
    }
    else {
      editor.insert(valueToInsert);
    }

    editor.clearSelection(); // for some reason the above changes selection

    // go back to see whether we have one of ( : { & [ do not require a comma. All the rest do.
    var newPos = {
      row: context.rangeToReplace.start.row,
      column: context.rangeToReplace.start.column + termAsString.length + context.prefixToAdd.length
      + (templateInserted ? 0 : context.suffixToAdd.length)
    };

    var tokenIter = editor.iterForPosition(newPos.row, newPos.column);

    if (context.autoCompleteType === "body") {
      // look for the next place stand, just after a comma, {
      var nonEmptyToken = editor.parser.nextNonEmptyToken(tokenIter);
      switch (nonEmptyToken ? nonEmptyToken.type : "NOTOKEN") {
        case "paren.rparen":
          newPos = { row: tokenIter.getCurrentTokenRow(), column: tokenIter.getCurrentTokenColumn() };
          break;
        case "punctuation.colon":
          nonEmptyToken = editor.parser.nextNonEmptyToken(tokenIter);
          if ((nonEmptyToken || {}).type == "paren.lparen") {
            nonEmptyToken = editor.parser.nextNonEmptyToken(tokenIter);
            newPos = { row: tokenIter.getCurrentTokenRow(), column: tokenIter.getCurrentTokenColumn() };
            if (nonEmptyToken && nonEmptyToken.value.indexOf('"') === 0) {
              newPos.column++;
            } // don't stand on "
          }
          break;
        case "paren.lparen":
        case "punctuation.comma":
          tokenIter.stepForward();
          newPos = { row: tokenIter.getCurrentTokenRow(), column: tokenIter.getCurrentTokenColumn() };
          break;
      }
      editor.moveCursorToPosition(newPos);
    }

    // re-enable listening to typing
    addChangeListener();
  }

  function getAutoCompleteContext(editor, session, pos) {
    // deduces all the parameters need to position and insert the auto complete
    var context = {
      autoCompleteSet: null, // instructions for what can be here
      endpoint: null,
      urlPath: null,
      method: null,
      activeScheme: null,
      editor: editor
    };

    //  context.updatedForToken = session.getTokenAt(pos.row, pos.column);
    //
    //  if (!context.updatedForToken)
    //    context.updatedForToken = { value: "", start: pos.column }; // empty line
    //
    //  context.updatedForToken.row = pos.row; // extend

    context.autoCompleteType = getAutoCompleteType(pos);
    switch (context.autoCompleteType) {
      case "path":
        addPathAutoCompleteSetToContext(context, pos);
        break;
      case "url_params":
        addUrlParamsAutoCompleteSetToContext(context, pos);
        break;
      case "method":
        addMethodAutoCompleteSetToContext(context, pos);
        break;
      case "body":
        addBodyAutoCompleteSetToContext(context, pos);
        break;
      default:
        return null;
    }

    if (!context.autoCompleteSet) {
      return null; // nothing to do..
    }

    addReplacementInfoToContext(context, pos);

    context.createdWithToken = _.clone(context.updatedForToken);

    return context;
  }

  function getAutoCompleteType(pos) {
    // return "method", "path" or "body" to determine auto complete type.

    var rowMode = editor.parser.getRowParseMode(pos.row);

    //noinspection JSBitwiseOperatorUsage
    if (rowMode & editor.parser.MODE.IN_REQUEST) {
      return "body";
    }
    //noinspection JSBitwiseOperatorUsage
    if (rowMode & editor.parser.MODE.REQUEST_START) {
      // on url path, url params or method.
      var tokenIter = editor.iterForPosition(pos.row, pos.column);
      var t = tokenIter.getCurrentToken();

      while (t.type == "url.comma") {
        t = tokenIter.stepBackward();
      }
      switch (t.type) {
        case "method":
          return "method";
        case "whitespace":
          t = editor.parser.prevNonEmptyToken(tokenIter);

          switch ((t || {}).type) {
            case "method":
              // we moved one back
              return "path";
              break;
            default:
              if (isUrlPathToken(t)) {
                return "path";
              }
              if (isUrlParamsToken(t)) {
                return "url_params";
              }
              return null;
          }
          break;
        default:
          if (isUrlPathToken(t)) {
            return "path";
          }
          if (isUrlParamsToken(t)) {
            return "url_params";
          }
          return null;
      }
    }

    // after start to avoid single line url only requests
    //noinspection JSBitwiseOperatorUsage
    if (rowMode & editor.parser.MODE.REQUEST_END) {
      return "body"
    }

    // in between request on an empty
    if ((editor.getSession().getLine(pos.row) || "").trim() === "") {
      // check if the previous line is a single line begging of a new request
      rowMode = editor.parser.getRowParseMode(pos.row - 1);
      //noinspection JSBitwiseOperatorUsage
      if ((rowMode & editor.parser.MODE.REQUEST_START) && (rowMode & editor.parser.MODE.REQUEST_END)) {
        return "body";
      }
      //o.w suggest a method
      return "method";
    }

    return null;
  }

  function addReplacementInfoToContext(context, pos, replacingTerm) {
    // extract the initial value, rangeToReplace & textBoxPosition

    // Scenarios for current token:
    //   -  Nice token { "bla|"
    //   -  Broken text token {   bla|
    //   -  No token : { |
    //   - Broken scenario { , bla|
    //   - Nice token, broken before: {, "bla"

    var session = editor.getSession();
    var insertingRelativeToToken;

    context.updatedForToken = _.clone(session.getTokenAt(pos.row, pos.column));
    if (!context.updatedForToken) {
      context.updatedForToken = { value: "", start: pos.column };
    } // empty line

    var anchorToken = context.createdWithToken;
    if (!anchorToken) {
      anchorToken = context.updatedForToken;
    }

    switch (context.updatedForToken.type) {
      case "variable":
      case "string":
      case "text":
      case "constant.numeric":
      case "constant.language.boolean":
      case "method":
      case "url.index":
      case "url.type":
      case "url.id":
      case "url.method":
      case "url.endpoint":
      case "url.part":
      case "url.param":
      case "url.value":
        insertingRelativeToToken = 0;
        context.rangeToReplace = new AceRange(
          pos.row, anchorToken.start, pos.row,
          context.updatedForToken.start + context.updatedForToken.value.length
        );
        context.replacingToken = true;
        break;
      default:
        if (replacingTerm && context.updatedForToken.value == replacingTerm) {
          insertingRelativeToToken = 0;
          context.rangeToReplace = new AceRange(
            pos.row, anchorToken.start, pos.row,
            context.updatedForToken.start + context.updatedForToken.value.length
          );
          context.replacingToken = true;
        }
        else {
          // standing on white space, quotes or another punctuation - no replacing
          context.rangeToReplace = new AceRange(
            pos.row, pos.column, pos.row, pos.column
          );
          context.replacingToken = false;
        }
        break;
    }

    context.textBoxPosition = { row: context.rangeToReplace.start.row, column: context.rangeToReplace.start.column };

    switch (context.autoCompleteType) {
      case "path":
        addPathPrefixSuffixToContext(context);
        break;
      case "url_params":
        addUrlParamsPrefixSuffixToContext(context);
        break;
      case "method":
        addMethodPrefixSuffixToContext(context);
        break;
      case "body":
        addBodyPrefixSuffixToContext(context);
        break;
    }
  }

  function addBodyPrefixSuffixToContext(context) {
    // Figure out what happens next to the token to see whether it needs trailing commas etc.

    // Templates will be used if not destroying existing structure.
    // -> token : {} or token ]/} or token , but not token : SOMETHING ELSE

    context.prefixToAdd = "";
    context.suffixToAdd = "";

    var tokenIter = editor.iterForCurrentLoc();
    var nonEmptyToken = editor.parser.nextNonEmptyToken(tokenIter);
    switch (nonEmptyToken ? nonEmptyToken.type : "NOTOKEN") {
      case "NOTOKEN":
      case "paren.lparen":
      case "paren.rparen":
      case "punctuation.comma":
        context.addTemplate = true;
        break;
      case "punctuation.colon":
        // test if there is an empty object - if so we replace it
        context.addTemplate = false;

        nonEmptyToken = editor.parser.nextNonEmptyToken(tokenIter);
        if (!(nonEmptyToken && nonEmptyToken.value == "{")) {
          break;
        }
        nonEmptyToken = editor.parser.nextNonEmptyToken(tokenIter);
        if (!(nonEmptyToken && nonEmptyToken.value == "}")) {
          break;
        }
        context.addTemplate = true;
        // extend range to replace to include all up to token
        context.rangeToReplace.end.row = tokenIter.getCurrentTokenRow();
        context.rangeToReplace.end.column = tokenIter.getCurrentTokenColumn() + nonEmptyToken.value.length;

        // move one more time to check if we need a trailing comma
        nonEmptyToken = editor.parser.nextNonEmptyToken(tokenIter);
        switch (nonEmptyToken ? nonEmptyToken.type : "NOTOKEN") {
          case "NOTOKEN":
          case "paren.rparen":
          case "punctuation.comma":
          case "punctuation.colon":
            break;
          default:
            context.suffixToAdd = ", "
        }

        break;
      default:
        context.addTemplate = true;
        context.suffixToAdd = ", ";
        break; // for now play safe and do nothing. May be made smarter.
    }

    // go back to see whether we have one of ( : { & [ do not require a comma. All the rest do.
    tokenIter = editor.iterForCurrentLoc();
    nonEmptyToken = tokenIter.getCurrentToken();
    var insertingRelativeToToken; // -1 is before token, 0 middle, +1 after token
    if (context.replacingToken) {
      insertingRelativeToToken = 0;
    }
    else {
      var pos = editor.getCursorPosition();
      if (pos.column == context.updatedForToken.start) {
        insertingRelativeToToken = -1;
      }
      else if (pos.column < context.updatedForToken.start + context.updatedForToken.value.length) {
        insertingRelativeToToken = 0;
      }
      else {
        insertingRelativeToToken = 1;
      }

    }
    // we should actually look at what's happening before this token
    if (editor.parser.isEmptyToken(nonEmptyToken) || insertingRelativeToToken <= 0) {
      nonEmptyToken = editor.parser.prevNonEmptyToken(tokenIter);
    }

    switch (nonEmptyToken ? nonEmptyToken.type : "NOTOKEN") {
      case "NOTOKEN":
      case "paren.lparen":
      case "punctuation.comma":
      case "punctuation.colon":
      case "method":
        break;
      default:
        if (nonEmptyToken && nonEmptyToken.type.indexOf("url") < 0) {
          context.prefixToAdd = ", "
        }
    }

    return context;
  }

  function addUrlParamsPrefixSuffixToContext(context) {
    context.prefixToAdd = "";
    context.suffixToAdd = "";
  }

  function addMethodPrefixSuffixToContext(context) {
    context.prefixToAdd = "";
    context.suffixToAdd = "";
    var tokenIter = editor.iterForCurrentLoc();
    var row = tokenIter.getCurrentTokenRow();
    var t = editor.parser.nextNonEmptyToken(tokenIter);

    if (tokenIter.getCurrentTokenRow() != row || !t) {
      // we still have nothing next to the method, add a space..
      context.suffixToAdd = " ";
    }
  }

  function addPathPrefixSuffixToContext(context) {
    context.prefixToAdd = "";
    context.suffixToAdd = "";
  }

  function addMethodAutoCompleteSetToContext(context, pos) {
    context.autoCompleteSet = _.map(["GET", "PUT", "POST", "DELETE", "HEAD"], function (m, i) {
      return { name: m, score: -i, meta: "method" }
    })
  }

  function addPathAutoCompleteSetToContext(context, pos) {
    var ret = getCurrentMethodAndTokenPaths(pos);
    context.method = ret.method;
    context.token = ret.token;
    context.otherTokenValues = ret.otherTokenValues;
    context.urlTokenPath = ret.urlTokenPath;
    autocomplete_engine.populateContext(ret.urlTokenPath, context, editor, true, kb.getTopLevelUrlCompleteComponents());
    context.autoCompleteSet = addMetaToTermsList(context.autoCompleteSet, "endpoint");
  }

  function addUrlParamsAutoCompleteSetToContext(context, pos) {
    var ret = getCurrentMethodAndTokenPaths(pos);
    context.method = ret.method;
    context.otherTokenValues = ret.otherTokenValues;
    context.urlTokenPath = ret.urlTokenPath;
    if (!ret.urlTokenPath) { // zero length tokenPath is true

      console.log("Can't extract a valid url token path.");
      return context;
    }

    autocomplete_engine.populateContext(ret.urlTokenPath, context, editor, false, kb.getTopLevelUrlCompleteComponents());

    if (!context.endpoint) {
      console.log("couldn't resolve an endpoint.");
      return context;
    }

    if (!ret.urlParamsTokenPath) { // zero length tokenPath is true
      console.log("Can't extract a valid urlParams token path.");
      return context;
    }
    var tokenPath = [], currentParam = ret.urlParamsTokenPath.pop();
    if (currentParam) {
      tokenPath = Object.keys(currentParam); // single key object
      context.otherTokenValues = currentParam[tokenPath[0]];
    }

    autocomplete_engine.populateContext(tokenPath, context, editor, true,
      context.endpoint.paramsAutocomplete.getTopLevelComponents());
    return context;
  }

  function addBodyAutoCompleteSetToContext(context, pos) {

    var ret = getCurrentMethodAndTokenPaths(pos);
    context.method = ret.method;
    context.otherTokenValues = ret.otherTokenValues;
    context.urlTokenPath = ret.urlTokenPath;
    context.requestStartRow = ret.requestStartRow;
    if (!ret.urlTokenPath) { // zero length tokenPath is true
      console.log("Can't extract a valid url token path.");
      return context;
    }

    autocomplete_engine.populateContext(ret.urlTokenPath, context, editor, false, kb.getTopLevelUrlCompleteComponents());

    context.bodyTokenPath = ret.bodyTokenPath;
    if (!ret.bodyTokenPath) { // zero length tokenPath is true

      console.log("Can't extract a valid body token path.");
      return context;
    }

    // needed for scope linking + global term resolving
    context.endpointComponentResolver = kb.getEndpointBodyCompleteComponents;
    context.globalComponentResolver = kb.getGlobalAutocompleteComponents;
    var components;
    if (context.endpoint) {
      components = context.endpoint.bodyAutocompleteRootComponents;
    }
    else {
      components = kb.getUnmatchedEndpointComponents();
    }
    autocomplete_engine.populateContext(ret.bodyTokenPath, context, editor, true, components);

    return context;
  }

  function getCurrentMethodAndTokenPaths(pos) {
    var tokenIter = editor.iterForPosition(pos.row, pos.column);
    var startPos = pos;
    var bodyTokenPath = [], last_var = "", first_scope = true, ret = {};

    var STATES = {
      looking_for_key: 0, // looking for a key but without jumping over anything but white space and colon.
      looking_for_scope_start: 1, // skip everything until scope start
      start: 3
    };
    var state = STATES.start;

    // initialization problems -
    var t = tokenIter.getCurrentToken();
    if (t) {
      if (startPos.column == 0) {
        // if we are at the beginning of the line, the current token is the one after cursor, not before which
        // deviates from the standard.
        t = tokenIter.stepBackward();
        state = STATES.looking_for_scope_start;
      }
    }
    else {
      if (startPos.column == 0) {
        // empty lines do no have tokens, move one back
        t = tokenIter.stepBackward();
        state = STATES.start;
      }
    }

    var walkedSomeBody = false;

    // climb one scope at a time and get the scope key
    for (; t && t.type.indexOf("url") == -1 && t.type != "method"; t = tokenIter.stepBackward()) {

      if (t.type != "whitespace") {
        walkedSomeBody = true;
      } // marks we saw something

      switch (t.type) {
        case "variable":
          if (state == STATES.looking_for_key) {
            bodyTokenPath.unshift(t.value.trim().replace(/"/g, ''));
          }
          state = STATES.looking_for_scope_start; // skip everything until the beginning of this scope
          break;

        case "paren.lparen":
          bodyTokenPath.unshift(t.value);
          if (state == STATES.looking_for_scope_start) {
            // found it. go look for the relevant key
            state = STATES.looking_for_key;
          }
          break;
        case "paren.rparen":
          // reset he search for key
          state = STATES.looking_for_scope_start;
          // and ignore this sub scope..
          var parenCount = 1;
          t = tokenIter.stepBackward();
          while (t && parenCount > 0) {
            switch (t.type) {
              case "paren.lparen":
                parenCount--;
                break;
              case "paren.rparen":
                parenCount++;
                break;
            }
            if (parenCount > 0) {
              t = tokenIter.stepBackward();
            }
          }
          if (!t) // oops we run out.. we don't know what's up return null;
          {
            return {};
          }
          continue;
        case "punctuation.end_triple_quote":
          // reset the search for key
          state = STATES.looking_for_scope_start;
          for (t = tokenIter.stepBackward(); t; t = tokenIter.stepBackward()) {
            if (t.type === "punctuation.start_tripple_qoute") {
                t = tokenIter.stepBackward();
                break;
            }
          }
          if (!t) // oops we run out.. we don't know what's up return null;
          {
            return {};
          }
          continue;
        case "punctuation.start_triple_quote":
          if (state == STATES.start) {
            state = STATES.looking_for_key;
          }
          else if (state == STATES.looking_for_key) {
            state = STATES.looking_for_scope_start;
          }
          bodyTokenPath.unshift('"""');
          continue;
        case "string":
        case "constant.numeric":
        case "constant.language.boolean":
        case "text":
          if (state == STATES.start) {
            state = STATES.looking_for_key;
          }
          else if (state == STATES.looking_for_key) {
            state = STATES.looking_for_scope_start;
          }

          break;
        case "punctuation.comma":
          if (state == STATES.start) {
            state = STATES.looking_for_scope_start;
          }
          break;
        case "punctuation.colon":
        case "whitespace":
          if (state == STATES.start) {
            state = STATES.looking_for_key;
          }
          break; // skip white space

      }
    }

    if (walkedSomeBody && (!bodyTokenPath || bodyTokenPath.length == 0)) {
      // we had some content and still no path -> the cursor is position after a closed body -> no auto complete
      return {};
    }

    if (tokenIter.getCurrentTokenRow() == startPos.row) {
      if (t.type === "url.part" || t.type === "url.param" || t.type === "url.value") {
        // we are on the same line as cursor and dealing with a url. Current token is not part of the context
        t = tokenIter.stepBackward();
      }
      bodyTokenPath = null; // no not on a body line.
    }

    ret.bodyTokenPath = bodyTokenPath;
    ret.urlTokenPath = [];
    ret.urlParamsTokenPath = null;
    ret.requestStartRow = tokenIter.getCurrentTokenRow();
    var curUrlPart;

    while (t && isUrlParamsToken(t)) {
      switch (t.type) {
        case "url.value":
          if (_.isArray(curUrlPart)) {
            curUrlPart.unshift(t.value);
          }
          else if (curUrlPart) {
            curUrlPart = [t.value, curUrlPart];
          }
          else {
            curUrlPart = t.value;
          }
          break;
        case "url.comma":
          if (!curUrlPart) {
            curUrlPart = [];
          }
          else if (!_.isArray(curUrlPart)) {
            curUrlPart = [curUrlPart];
          }
          break;
        case "url.param":
          var v = curUrlPart;
          curUrlPart = {};
          curUrlPart[t.value] = v;
          break;
        case "url.amp":
        case "url.questionmark":
          if (!ret.urlParamsTokenPath) {
            ret.urlParamsTokenPath = [];
          }
          ret.urlParamsTokenPath.unshift(curUrlPart || {});
          curUrlPart = null;
          break;
      }
      t = tokenIter.stepBackward();
    }

    curUrlPart = null;
    while (t && t.type.indexOf("url") != -1) {
      switch (t.type) {
        case "url.part":
          if (_.isArray(curUrlPart)) {
            curUrlPart.unshift(t.value);
          }
          else if (curUrlPart) {
            curUrlPart = [t.value, curUrlPart];
          }
          else {
            curUrlPart = t.value;
          }
          break;
        case "url.comma":
          if (!curUrlPart) {
            curUrlPart = [];
          }
          else if (!_.isArray(curUrlPart)) {
            curUrlPart = [curUrlPart];
          }
          break;
        case "url.slash":
          ret.urlTokenPath.unshift(curUrlPart);
          curUrlPart = null;
          break;
      }
      t = editor.parser.prevNonEmptyToken(tokenIter);
    }

    if (curUrlPart) {
      ret.urlTokenPath.unshift(curUrlPart);
    }

    if (!ret.bodyTokenPath && !ret.urlParamsTokenPath) {

      if (ret.urlTokenPath.length > 0) {
        // started on the url, first token is current token
        ret.otherTokenValues = ret.urlTokenPath.splice(-1)[0];
      }
    }
    else {
      // mark the url as completed.
      ret.urlTokenPath.push(url_pattern_matcher.URL_PATH_END_MARKER);
    }

    if (t && t.type == "method") {
      ret.method = t.value;
    }
    return ret;
  }

  var evaluateCurrentTokenAfterAChange = _.debounce(function evaluateCurrentTokenAfterAChange(pos) {
    var session = editor.getSession();
    var currentToken = session.getTokenAt(pos.row, pos.column);

    if (!currentToken) {
      if (pos.row == 0) {
        LAST_EVALUATED_TOKEN = null;
        return;
      }
      currentToken = { start: 0, value: "" }; // empty row
    }

    currentToken.row = pos.row; // extend token with row. Ace doesn't supply it by default
    if (editor.parser.isEmptyToken(currentToken)) {
      // empty token. check what's coming next
      var nextToken = session.getTokenAt(pos.row, pos.column + 1);
      if (editor.parser.isEmptyToken(nextToken)) {
        // Empty line, or we're not on the edge of current token. Save the current position as base
        currentToken.start = pos.column;
        LAST_EVALUATED_TOKEN = currentToken;
      }
      else {
        nextToken.row = pos.row;
        LAST_EVALUATED_TOKEN = nextToken;
      }
      return;
    }

    if (!LAST_EVALUATED_TOKEN) {
      LAST_EVALUATED_TOKEN = currentToken;
      return; // wait for the next typing.
    }

    if (LAST_EVALUATED_TOKEN.start != currentToken.start || LAST_EVALUATED_TOKEN.row != currentToken.row
      || LAST_EVALUATED_TOKEN.value === currentToken.value) {
      // not on the same place or nothing changed, cache and wait for the next time
      LAST_EVALUATED_TOKEN = currentToken;
      return;
    }

    // don't automatically open the auto complete if some just hit enter (new line) or open a parentheses
    switch (currentToken.type || "UNKNOWN") {
      case "paren.lparen":
      case "paren.rparen":
      case "punctuation.colon":
      case "punctuation.comma":
      case "UNKOWN":
        return;
    }

    LAST_EVALUATED_TOKEN = currentToken;
    editor.execCommand("startAutocomplete");
  }, 100);

  function editorChangeListener(e) {
    var cursor = editor.selection.lead;
    if (editor.__ace.completer && editor.__ace.completer.activated) {
      return;
    }
    evaluateCurrentTokenAfterAChange(cursor);
  }

  function addChangeListener() {
    editor.on("changeSelection", editorChangeListener);
  }

  function removeChangeListener() {
    editor.off("changeSelection", editorChangeListener)
  }

  function getCompletions(aceEditor, session, pos, prefix, callback) {
    try {

      var context = getAutoCompleteContext(editor, session, pos);
      if (!context) {
        callback(null, []);
      }
      else {
        var terms = _.map(context.autoCompleteSet, function (term) {
          if (typeof term !== "object") {
            term = {
              name: term
            }
          } else {
            term = _.clone(term);
          }

          return _.defaults(term, {
            value: term.name,
            meta: "API",
            score: 0,
            context: context,
            completer: {
              insertMatch: function () {
                applyTerm(term);
              }
            }
          });
        });

        terms.sort(function (t1, t2) {
          /* score sorts from high to low */
          if (t1.score > t2.score) {
            return -1;
          }
          if (t1.score < t2.score) {
            return 1;
          }
          /* names sort from low to high */
          if (t1.name < t2.name) {
            return -1;
          }
          if (t1.name === t2.name) {
            return 0;
          }
          return 1;
        });

        callback(null, _.map(terms, function (t, i) {
          t.insert_value = t.insert_value || t.value;
          t.value = '' + t.value; // normalize to strings
          t.score = -i;
          return t;
        }));
      }
    }
    catch (e) {
      console.log("error while getting completion terms", e);
      callback(e, null);
    }
  }

  addChangeListener();

  // Hook into Ace

  // disable standard context based autocompletion.
  ace.define('ace/autocomplete/text_completer', ['require', 'exports', 'module'], function (require, exports,
                                                                                            module) {
    exports.getCompletions = function (editor, session, pos, prefix, callback) {
      callback(null, []);
    }
  });

  var langTools = ace.require('ace/ext/language_tools');
  var aceUtils = ace.require('ace/autocomplete/util');
  var aceAutoComplete = ace.require('ace/autocomplete');

  langTools.addCompleter({
    getCompletions: getCompletions
  });

  editor.setOptions({
    enableBasicAutocompletion: true
  });

  // Ace doesn't care about tokenization when calculating prefix. It will thus stop on . in keys names.
  // we patch this behavior.
  // CHECK ON ACE UPDATE
  var aceAutoCompleteInstance = new aceAutoComplete.Autocomplete();
  aceAutoCompleteInstance.autoInsert = false;
  aceAutoCompleteInstance.gatherCompletions = function (ace_editor, callback) {
    var session = ace_editor.getSession();
    var pos = ace_editor.getCursorPosition();
    var prefix = "";
    // change starts here
    var token = session.getTokenAt(pos.row, pos.column);
    this.base = _.clone(pos);
    if (!editor.parser.isEmptyToken(token) && !isSeparatorToken(token)) {
      if (token.value.indexOf('"') == 0) {
        this.base.column = token.start + 1;
      }
      else {
        this.base.column = token.start;
      }

      prefix = getAutoCompleteValueFromToken(token);
    }

    var matches = [];
    aceUtils.parForEach(ace_editor.completers, function (completer, next) {
      completer.getCompletions(ace_editor, session, pos, prefix, function (err, results) {
        if (!err) {
          matches = matches.concat(results);
        }
        next();
      });
    }, function () {
      callback(null, {
        prefix: prefix,
        matches: matches
      });
    });
    return true;
  };

  editor.__ace.completer = aceAutoCompleteInstance;

  return {
    _test: {
      getCompletions: getCompletions,
      addReplacementInfoToContext: addReplacementInfoToContext,
      addChangeListener: addChangeListener,
      removeChangeListener: removeChangeListener
    }
  }
};