cdramey/tidx
Archived
1
0
Fork 0
Code Releases Activity

Added support for untokenized indexes, various speed improvements

Browse Source 
by using strict equality comparisons where possible, more IE compatibility
fixes.
This commit is contained in:
Christopher Ramey 2012-04-01 22:13:09 +00:00 committed by cdramey
parent 1b174aa0c4
commit 662ff40fa2
2 changed files with 46 additions and 31 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
test.html
View File

@ -44,7 +44,11 @@
var s = new Date().getTime(); var s = new Date().getTime();
for(var i = 0; i < json.length; i++){ for(var i = 0; i < json.length; i++){
for(var f in json[i]){ for(var f in json[i]){
if(f != 'id'){ t.index(i, f, json[i][f]); } switch(f){
case 'id': break;
case 'email': t.index(false, i, f, json[i][f]); break;
default: t.index(true, i, f, json[i][f]); break;
}
} }
} }

59
tidx.js
View File

@ -5,14 +5,14 @@ var Tidx = function()
this._index = {}; this._index = {};
// regex used to find terms inside a value // regex used to find terms inside a value
this.v_rx = new RegExp('(:?[a-z_-]+)|(\\d*\\.\\d+)|(\\d+)', 'img'); this.v_rx = new RegExp('(:?[a-z_-]+)|(\\d*\\.\\d+)|(\\d+)', 'mg');
// regex used to break out terms with a search // regex used to break out terms with a search
this.s_rx = new RegExp('(:?([a-z\\d-_]+):){0,1}(:?(:?"(.+?)")|([a-z_-]+)|(\\d+\\.{0,1}\\d*))', 'img') this.s_rx = new RegExp('(:?([a-z\\d-_]+):){0,1}(:?(:?"(.+?)")|([a-z_-]+)|(\\d*\\.\\d+)|(\\d+))', 'img')
// Adds data to index // Adds data to index
this.index = function(id, field, value) this.index = function(tokenize, id, field, value)
{ {
var f; var f;
switch(typeof field){ switch(typeof field){
@ -30,30 +30,34 @@ var Tidx = function()
case 'string': case 'string':
v = value.toLowerCase(); v = value.toLowerCase();
// Don't index empty fields // Don't index empty fields
if(v.length == 0){ return; } if(v.length === 0){ return; }
break; break;
default: v = String(value); default: v = String(value);
} }
// Add field to field list, as needed // Add field to field list, as needed
if(this._index[f] == undefined){ this._index[f] = {}; } if(this._index[f] === undefined){ this._index[f] = {}; }
if(tokenize === true){
// Iterate over discrete searchable terms // Iterate over discrete searchable terms
var re; var re;
while((re = this.v_rx.exec(value)) !== null){ while((re = this.v_rx.exec(v)) !== null){
var v = re[0].toLowerCase(); var t = re[0];
// If the field in question doesn't have this term already, // If the field in question doesn't have this term already,
// add it. // add it.
if(this._index[f][v] == undefined){ this._index[f][v] = {}; } if(this._index[f][t] === undefined){ this._index[f][t] = {}; }
// Add this id to the reverse index under specific field (f) // Add this id to the reverse index under specific field (f)
// and term (v), if it already exists, increment the weight // and term (v), if it already exists, increment the weight
if(this._index[f][v][id] == undefined){ if(this._index[f][t][id] === undefined){ this._index[f][t][id] = 1; }
this._index[f][v][id] = 1; else { this._index[f][t][id]++; }
} else {
this._index[f][v][id]++;
} }
} else {
if(this._index[f][v] === undefined){ this._index[f][v] = {}; }
if(this._index[f][v][id] === undefined){ this._index[f][v][id] = 1; }
else { this._index[f][v][id]++; }
} }
}; };
@ -63,17 +67,17 @@ var Tidx = function()
this.gsearch = function(result, value) this.gsearch = function(result, value)
{ {
// Refuse empty searches // Refuse empty searches
if(value.length == 0){ return []; } if(value.length === 0){ return []; }
var v = value.toLowerCase(); var v = value.toLowerCase();
// Loop over every field // Loop over every field
for(var f in this._index){ for(var f in this._index){
// Look for the specified search term // Look for the specified search term
if(this._index[f][v] != undefined){ if(this._index[f][v] !== undefined){
// If it exists, add the result to r, adding in the weight // If it exists, add the result to r, adding in the weight
for(var i in this._index[f][v]){ for(var i in this._index[f][v]){
if(result[i] == undefined){ result[i] = 0; } if(result[i] === undefined){ result[i] = 0; }
result[i] += this._index[f][v][i]; result[i] += this._index[f][v][i];
} }
} }
@ -92,13 +96,15 @@ var Tidx = function()
default: f = String(field); default: f = String(field);
} }
if(value.length == 0){ return []; } if(value.length === 0){ return []; }
var v = value.toLowerCase(); var v = value.toLowerCase();
if(this._index[f] == undefined || this._index[f][v] == undefined){ return []; } if(this._index[f] === undefined || this._index[f][v] === undefined){
return [];
}
for(var i in this._index[f][v]){ for(var i in this._index[f][v]){
if(result[i] == undefined){ result[i] = 0; } if(result[i] === undefined){ result[i] = 0; }
result[i] += this._index[f][v][i]; result[i] += this._index[f][v][i];
} }
}; };
@ -115,14 +121,19 @@ var Tidx = function()
var field = re[2]; var field = re[2];
var value; var value;
if(re[5] != undefined){ value = re[5]; } for(var i=5; i < 9; i++){
else if(re[6] != undefined){ value = re[6]; } if(re[i] !== undefined && re[i].length !== 0){
else if(re[7] != undefined){ value = re[7]; } value = re[i];
break;
}
}
// Global term // Global term
if(field == undefined || field == ''){ this.gsearch(r, value); } if(field !== undefined && field.length !== 0){
// Field specific term this.fsearch(r, field, value);
else { this.fsearch(r, field, value); } } else {
this.gsearch(r, value);
}
} }
return this.sortresult(r); return this.sortresult(r);