Merge pull request #14 from mixmaxhq/return-character-offsets

Return the character offsets of parsed search parts. The use case is …

README.md

@@ -20,7 +20,13 @@ And turns it into an object like this:
     from: '1/10/2013',
     to: '15/04/2014'
     },
-  text: 'photos'
+  text: 'photos',
+  offsets: 
+   [ { keyword: 'from', value: 'hi@retrace.io,foo@gmail.com', offsetStart: 0, offsetEnd: 32 },
+     { keyword: 'to', value: 'me', offsetStart: 33, offsetEnd: 38 },
+     { keyword: 'subject', value: 'vacations', offsetStart: 39, offsetEnd: 56 },
+     { keyword: 'date', value: '1/10/2013-15/04/2014', offsetStart: 57, offsetEnd: 82 },
+     { text: 'photos', offsetStart: 83, offsetEnd: 89 } ]
 }
 ```
 

lib/search-query-parser.js

@@ -15,10 +15,6 @@ exports.parse = function (string, options) {
     string = '';
   }
 
-  // Regularize white spacing
-  // Make in-between white spaces a unique space
-  string = string.trim().replace(/\s+/g, ' ');
-
   // When a simple string, return it
   if (-1 === string.indexOf(':')) {
     return string;
@@ -30,16 +26,19 @@ exports.parse = function (string, options) {
   // Otherwise parse the advanced query syntax
   else {
     // Our object to store the query object
-    var query = {text: []};
+    var query = {text: [], offsets: []};
     var exclusion = {};
+    var terms = [];
     // Get a list of search terms respecting single and double quotes
-    var terms = string.match(/(\S+:'(?:[^'\\]|\\.)*')|(\S+:"(?:[^"\\]|\\.)*")|\S+|\S+:\S+/g);
-    for (var i = 0; i < terms.length; i++) {
-      var sepIndex = terms[i].indexOf(':');
-      if(sepIndex !== -1) {
-        var split = terms[i].split(':'),
-            key = terms[i].slice(0, sepIndex),
-            val = terms[i].slice(sepIndex + 1);
+    var regex = /(\S+:'(?:[^'\\]|\\.)*')|(\S+:"(?:[^"\\]|\\.)*")|\S+|\S+:\S+/g;
+    var match;
+    while ((match = regex.exec(string)) !== null) {
+      var term = match[0];
+      var sepIndex = term.indexOf(':');
+      if (sepIndex !== -1) {
+        var split = term.split(':'),
+            key = term.slice(0, sepIndex),
+            val = term.slice(sepIndex + 1);
         // Strip surrounding quotes
         val = val.replace(/^\"|\"$|^\'|\'$/g, '');
         // Strip backslashes respecting escapes
@@ -55,7 +54,18 @@ exports.parse = function (string, options) {
             return n1;
           }
         });
-        terms[i] = key + ':' + val;
+        terms.push({
+          keyword: key,
+          value: val,
+          offsetStart: match.index,
+          offsetEnd: match.index + term.length
+        });
+      } else {
+        terms.push({
+          text: term,
+          offsetStart: match.index,
+          offsetEnd: match.index + term.length
+        });
       }
     }
     // Reverse to ensure proper order when pop()'ing.
@@ -63,17 +73,15 @@ exports.parse = function (string, options) {
     // For each search term
     var term;
     while (term = terms.pop()) {
-      // Advanced search terms syntax has key and value
-      // separated with a colon
-      var sepIdx = term.indexOf(':');
       // When just a simple term
-      if (-1 === sepIdx) {
+      if (term.text) {
         // We add it as pure text
-        query.text.push(term);
+        query.text.push(term.text);
+        query.offsets.push(term);
       }
       // We got an advanced search syntax
       else {
-        var key = term.slice(0, sepIdx);
+        var key = term.keyword;
         // Check if the key is a registered keyword
         options.keywords = options.keywords || [];
         var isKeyword = false;
@@ -88,12 +96,20 @@ exports.parse = function (string, options) {
                 isExclusion = true;
             }
         }
+
         // Check if the key is a registered range
         options.ranges = options.ranges || [];
         var isRange = !(-1 === options.ranges.indexOf(key));
         // When the key matches a keyword
         if (isKeyword) {
-          var value = term.slice(sepIdx + 1);
+          query.offsets.push({
+            keyword: key,
+            value: term.value,
+            offsetStart: isExclusion ? term.offsetStart + 1 : term.offsetStart,
+            offsetEnd: term.offsetEnd
+          });
+
+          var value = term.value;
           // When value is a thing
           if (value.length) {
             // Get an array of values when several are there
@@ -174,7 +190,9 @@ exports.parse = function (string, options) {
         }
         // The key allows a range
         else if (isRange) {
-          var value = term.slice(sepIdx + 1);
+          query.offsets.push(term);
+
+          var value = term.value;
           // Range are separated with a dash
           var rangeValues = value.split('-');
           // When both end of the range are specified
@@ -196,7 +214,14 @@ exports.parse = function (string, options) {
         }
         else {
           // We add it as pure text
-          query.text.push(term);
+          var text = term.keyword + ':' + term.value;
+          query.text.push(text);
+
+          query.offsets.push({
+            text: text,
+            offsetStart: term.offsetStart,
+            offsetEnd: term.offsetEnd
+          });
         }
       }
     }

test/test.js

@@ -23,6 +23,12 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.should.be.an.Object;
     parsedSearchQuery.should.have.property('from', 'jul@foo.com');
     parsedSearchQuery.should.not.have.property('text');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'from',
+      value: 'jul@foo.com',
+      offsetStart: 0,
+      offsetEnd: 16
+    }]);
   });
 
 
@@ -34,10 +40,50 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.should.be.an.Object;
     parsedSearchQuery.should.have.property('from', 'jul@foo.com');
     parsedSearchQuery.should.have.property('text', 'hey buddy!');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'from',
+      value: 'jul@foo.com',
+      offsetStart: 0,
+      offsetEnd: 16
+    }, {
+      text: 'hey',
+      offsetStart: 17,
+      offsetEnd: 20
+    }, {
+      text: 'buddy!',
+      offsetStart: 21,
+      offsetEnd: 27
+    }]);
   });
 
 
-  it('should parse a single keyword with free text before it', function () {
+  it('should ignore keywords that are not specified', function() {
+    var searchQuery = 'test another other:jul@foo.com';
+    var options = {
+      keywords: ['from']
+    };
+    var parsedSearchQuery = searchquery.parse(searchQuery, options);
+
+    parsedSearchQuery.should.be.an.Object;
+    parsedSearchQuery.should.have.not.have.property('other');
+    parsedSearchQuery.should.have.property('text', 'test another other:jul@foo.com');
+    parsedSearchQuery.should.have.property('offsets', [{
+      text: 'test',
+      offsetStart: 0,
+      offsetEnd: 4
+    }, {
+      text: 'another',
+      offsetStart: 5,
+      offsetEnd: 12
+    }, {
+      text: 'other:jul@foo.com',
+      offsetStart: 13,
+      offsetEnd: 30
+    }]);
+  });
+
+
+  it('should parse a single keyword with free text before it', function() {
     var searchQuery = 'hey you! from:jul@foo.com';
     var options = {keywords: ['from']};
     var parsedSearchQuery = searchquery.parse(searchQuery, options);
@@ -45,6 +91,20 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.should.be.an.Object;
     parsedSearchQuery.should.have.property('from', 'jul@foo.com');
     parsedSearchQuery.should.have.property('text', 'hey you!');
+    parsedSearchQuery.should.have.property('offsets', [{
+      text: 'hey',
+      offsetStart: 0,
+      offsetEnd: 3
+    }, {
+      text: 'you!',
+      offsetStart: 4,
+      offsetEnd: 8
+    }, {
+      keyword: 'from',
+      value: 'jul@foo.com',
+      offsetStart: 9,
+      offsetEnd: 25
+    }]);
   });
 
 
@@ -56,6 +116,24 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.should.be.an.Object;
     parsedSearchQuery.should.have.property('from', 'jul@foo.com');
     parsedSearchQuery.should.have.property('text', 'hey you! pouet');
+    parsedSearchQuery.should.have.property('offsets', [{
+      text: 'hey',
+      offsetStart: 0,
+      offsetEnd: 3
+    }, {
+      text: 'you!',
+      offsetStart: 4,
+      offsetEnd: 8
+    }, {
+      keyword: 'from',
+      value: 'jul@foo.com',
+      offsetStart: 9,
+      offsetEnd: 25
+    }, {
+      text: 'pouet',
+      offsetStart: 26,
+      offsetEnd: 31
+    }]);
   });
 
 
@@ -68,6 +146,24 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.should.be.an.Object;
     parsedSearchQuery.should.have.property('from', 'jul@foo.com');
     parsedSearchQuery.should.have.property('text', 'hey you! pouet');
+    parsedSearchQuery.should.have.property('offsets', [{
+      text: 'hey',
+      offsetStart: 3,
+      offsetEnd: 6
+    }, {
+      text: 'you!',
+      offsetStart: 11,
+      offsetEnd: 15
+    }, {
+      keyword: 'from',
+      value: 'jul@foo.com',
+      offsetStart: 16,
+      offsetEnd: 32
+    }, {
+      text: 'pouet',
+      offsetStart: 35,
+      offsetEnd: 40
+    }]);
   });
 
 
@@ -80,6 +176,37 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.should.have.property('from', 'jul@foo.com');
     parsedSearchQuery.should.have.property('to', 'bar@hey.ya');
     parsedSearchQuery.should.have.property('text', 'hey, so what\'s up gents');
+    parsedSearchQuery.should.have.property('offsets', [{
+      text: 'hey,',
+      offsetStart: 0,
+      offsetEnd: 4
+    }, {
+      keyword: 'from',
+      value: 'jul@foo.com',
+      offsetStart: 5,
+      offsetEnd: 21
+    }, {
+      keyword: 'to',
+      value: 'bar@hey.ya',
+      offsetStart: 22,
+      offsetEnd: 35
+    }, {
+      text: 'so',
+      offsetStart: 36,
+      offsetEnd: 38
+    }, {
+      text: 'what\'s',
+      offsetStart: 39,
+      offsetEnd: 45
+    }, {
+      text: 'up',
+      offsetStart: 46,
+      offsetEnd: 48
+    }, {
+      text: 'gents',
+      offsetStart: 49,
+      offsetEnd: 54
+    }]);
   });
 
 
@@ -95,6 +222,21 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.from.length.should.equal(2);
     parsedSearchQuery.from.should.containEql('jul@foo.com');
     parsedSearchQuery.from.should.containEql('bar@hey.ya');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'from',
+      value: 'jul@foo.com',
+      offsetStart: 0,
+      offsetEnd: 16
+    }, {
+      keyword: 'from',
+      value: 'bar@hey.ya',
+      offsetStart: 17,
+      offsetEnd: 32
+    }, {
+      text: 'vaccationessss',
+      offsetStart: 33,
+      offsetEnd: 47
+    }]);
   });
 
 
@@ -109,6 +251,12 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.from.length.should.equal(2);
     parsedSearchQuery.from.should.containEql('jul@foo.com');
     parsedSearchQuery.from.should.containEql('bar@hey.ya');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'from',
+      value: 'jul@foo.com,bar@hey.ya',
+      offsetStart: 0,
+      offsetEnd: 27
+    }]);
   });
 
 
@@ -126,6 +274,21 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.from.should.containEql('bar@hey.ya');
     parsedSearchQuery.from.should.containEql('a@b.c');
     parsedSearchQuery.from.should.containEql('d@e.f');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'from',
+      value: 'jul@foo.com,bar@hey.ya',
+      offsetStart: 0,
+      offsetEnd: 27
+    }, {
+      keyword: 'from',
+      value: 'a@b.c,d@e.f',
+      offsetStart: 28,
+      offsetEnd: 44
+    }, {
+      text: 'ouch!#',
+      offsetStart: 45,
+      offsetEnd: 51
+    }]);
   });
 
 
@@ -139,6 +302,16 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.should.have.property('date');
     parsedSearchQuery.date.should.be.an.Object;
     parsedSearchQuery.date.from.should.containEql('12/12/2012');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'date',
+      value: '12/12/2012',
+      offsetStart: 0,
+      offsetEnd: 15
+    }, {
+      text: 'ahaha',
+      offsetStart: 16,
+      offsetEnd: 21
+    }]);
   });
 
   it('should parse range with 2 ends and free text', function () {
@@ -152,6 +325,16 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.date.should.be.an.Object;
     parsedSearchQuery.date.from.should.containEql('12/12/2012');
     parsedSearchQuery.date.to.should.containEql('01/01/2014');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'date',
+      value: '12/12/2012-01/01/2014',
+      offsetStart: 0,
+      offsetEnd: 26
+    }, {
+      text: 'ahaha',
+      offsetStart: 27,
+      offsetEnd: 32
+    }]);
   });
 
 
@@ -198,6 +381,48 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.to.length.should.equal(2);
     parsedSearchQuery.to.should.containEql('me@me.com');
     parsedSearchQuery.to.should.containEql('toto@hey.co');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'date',
+      value: '12/12/2012-01/01/2014',
+      offsetStart: 3,
+      offsetEnd: 29
+    }, {
+      text: 'ahaha',
+      offsetStart: 30,
+      offsetEnd: 35
+    }, {
+      keyword: 'from',
+      value: 'jul@foo.com,bar@hey.ya',
+      offsetStart: 36,
+      offsetEnd: 63
+    }, {
+      keyword: 'from',
+      value: 'a@b.c,d@e.f',
+      offsetStart: 64,
+      offsetEnd: 80
+    }, {
+      text: 'ouch!#',
+      offsetStart: 81,
+      offsetEnd: 87
+    }, {
+      keyword: 'to',
+      value: 'me@me.com',
+      offsetStart: 90,
+      offsetEnd: 102
+    }, {
+      keyword: 'to',
+      value: 'toto@hey.co',
+      offsetStart: 103,
+      offsetEnd: 117
+    }, {
+      text: 'about',
+      offsetStart: 118,
+      offsetEnd: 123
+    }, {
+      text: 'that',
+      offsetStart: 124,
+      offsetEnd: 128
+    }]);
   });
 
 
@@ -209,6 +434,17 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.should.be.an.Object;
     parsedSearchQuery.should.have.property('name', 'Bob Saget');
     parsedSearchQuery.should.have.property('description', 'Banana Sandwiche');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'name',
+      value: 'Bob Saget',
+      offsetStart: 0,
+      offsetEnd: 16
+    }, {
+      keyword: 'description',
+      value: 'Banana Sandwiche',
+      offsetStart: 17,
+      offsetEnd: 47
+    }]);
   });
 
 
@@ -220,6 +456,17 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.should.be.an.Object;
     parsedSearchQuery.should.have.property('case1', 'This "is" \'a\' test');
     parsedSearchQuery.should.have.property('case2', 'This "is" \'a\' test');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'case1',
+      value: 'This "is" \'a\' test',
+      offsetStart: 0,
+      offsetEnd: 28
+    }, {
+      keyword: 'case2',
+      value: 'This "is" \'a\' test',
+      offsetStart: 29,
+      offsetEnd: 57
+    }]);
   });
 
 
@@ -232,6 +479,12 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.exclude.should.be.an.Object;
     parsedSearchQuery.exclude.should.have.property('from', 'jul@foo.com');
     parsedSearchQuery.should.not.have.property('text');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'from',
+      value: 'jul@foo.com',
+      offsetStart: 1,
+      offsetEnd: 17
+    }]);
   });
 
   it('should concatenate a keyword multiple values in exclusion syntax', function() {
@@ -244,11 +497,19 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.exclude.from.should.containEql('jul@foo.com');
     parsedSearchQuery.exclude.from.should.containEql('mar@foo.com');
     parsedSearchQuery.should.not.have.property('text');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'from',
+      value: 'jul@foo.com,mar@foo.com',
+      offsetStart: 1,
+      offsetEnd: 29
+    }]);
   });
 
   it('should support keywords which appear multiple times with exclusion syntax', function() {
     var searchQuery = '-from:jul@foo.com,mar@foo.com -from:jan@foo.com';
-    var options = {keywords: ['from']};
+    var options = {
+      keywords: ['from']
+    };
     var parsedSearchQuery = searchquery.parse(searchQuery, options);
 
     parsedSearchQuery.should.be.an.Object;
@@ -257,5 +518,16 @@ describe('Search query syntax parser', function () {
     parsedSearchQuery.exclude.from.should.containEql('mar@foo.com');
     parsedSearchQuery.exclude.from.should.containEql('jan@foo.com');
     parsedSearchQuery.should.not.have.property('text');
+    parsedSearchQuery.should.have.property('offsets', [{
+      keyword: 'from',
+      value: 'jul@foo.com,mar@foo.com',
+      offsetStart: 1,
+      offsetEnd: 29
+    }, {
+      keyword: 'from',
+      value: 'jan@foo.com',
+      offsetStart: 31,
+      offsetEnd: 47
+    }]);
   });
 });