Merge pull request #30 from kwhitley/feature/v2.0.1

v2.0.1

Author: K. R. Whitley

README.md

@@ -1,6 +1,6 @@
 # Treeize.js
 
-[![Build Status via Travis CI](https://travis-ci.org/kwhitley/treeize.svg?branch=feature%2Fmulti-format)](https://travis-ci.org/kwhitley/treeize)
+[![Build Status via Travis CI](https://travis-ci.org/kwhitley/treeize.svg)](https://travis-ci.org/kwhitley/treeize)
 
 Converts row data (in JSON/associative array format or flat array format) to object/tree structure based on simple column naming conventions.
 
@@ -108,12 +108,14 @@ people.getData() == [
 ##### 1. get/set options (optional)
 
 - [`options([options])`](#options) - getter/setter for options
+- [`getOptions()`](#getOptions) - returns options
 - [`setOptions(options)`](#setOptions) - merges new `[options]` with existing
 - [`resetOptions()`](#resetOptions) - resets options to defaults
 
 ##### 2a. set data signature manually if needed (optional)
 
 - [`signature([row], [options])`](#signature) - getter/setter for signature definitions
+- [`getSignature()`](#getSignature) - returns currently defined signature
 - [`setSignature(row, [options])`](#setSignature) - sets signature using a specific row of data/headers (preserves signature between data sets if uniformity option is enabled)
 - [`clearSignature()`](#clearSignature) - clear signature between data sets (only needed when previously defined a uniform signature via `setSignature`)
 
@@ -127,8 +129,6 @@ people.getData() == [
 
 ##### * misc/internal methods
 
-- [`getOptions()`](#getOptions) - returns options
-- [`getSignature()`](#getSignature) - returns currently defined signature
 - [`getStats()`](#getStats) - returns object with growth statistics
 - [`toString()`](#toString) - uses `util` to return data in visually formatted object graph
 - [`log(arg1, arg2, arg3)`](#log) - console.log output of `arg1..n` when `log` option is set to `true`
@@ -461,7 +461,7 @@ In this example, we'll take our dump (as if from a CSV or SQL result) - and name
 group by movies (as if for an `/api/movies`).
 
 ```js
-var movieDump = [
+var movieData = [
   {
     'title':             'The Prestige',
     'director':          'Christopher Nolan',
@@ -497,7 +497,7 @@ var movieDump = [
 var Treeize = require('treeize');
 var movies  = new Treeize();
 
-movies.grow(movieDump);
+movies.grow(movieData);
 
 /*
 
@@ -595,7 +595,7 @@ var moviesDump = [
 var Treeize = require('treeize');
 var actors  = new Treeize();
 
-actors.grow(movieDump);
+actors.grow(moviesData);
 
 /*
 
@@ -651,3 +651,10 @@ actors.grow(movieDump);
 
 */
 ```
+
+# Changelog
+
+### 2.0.1
+
+- minor README modifications
+- performance tuning... ~400% performance boost over 2.0.0

lib/treeize.js

@@ -47,7 +47,7 @@ function Treeize(options) {
 }
 
 Treeize.prototype.log = function() {
-  if (this.options().log) {
+  if (this._options.log) {
     console.log.apply(this, arguments);
   }
 
@@ -79,21 +79,21 @@ Treeize.prototype.signature = function(row, options, auto) {
 
   var nodes         = this.data.signature.nodes = [];
   var isArray       = _.isArray(row);
-  var opt           = merge(this.options(), options || {});
+  var opt           = merge(this._options, options || {});
 
   this.data.signature.type = isArray ? 'array' : 'object';
 
   _.each(row, function(value, key) {
     // set up attribute
     var attr        = {}
 
-    attr.key        = typeof key === 'number' ? key : key.replace(/^[\*\-\+]|[\*\-\+]$/g,'');
+    attr.key        = typeof key === 'number' ? key : key;//.replace(/^[\*\-\+]|[\*\-\+]$/g,'');
     attr.fullPath   = isArray ? value : key;
     attr.split      = attr.fullPath.split(opt.input.delimiter);
-    attr.path       = _.initial(attr.split).join(opt.input.delimiter);//.replace(/^[\*\-\+]|[\*\-\+]$/g,'');
-    attr.parent     = _.initial(attr.split, 2).join(opt.input.delimiter).replace(/^[\*\-\+]|[\*\-\+]$/g,'');
+    attr.path       = attr.split.slice(0,attr.split.length-1).join(opt.input.delimiter);
+    attr.parent     = attr.split.slice(0,attr.split.length-2).join(opt.input.delimiter);//.replace(/^[\*\-\+]|[\*\-\+]$/g,'');
     attr.node       = attr.split[attr.split.length - 2];
-    attr.attr       = _.last(attr.split);
+    attr.attr       = attr.split[attr.split.length - 1];
 
     if (attr.attr.match(/\*/gi)) {
       attr.attr = attr.attr.replace(/[\*]/gi,'');
@@ -123,7 +123,7 @@ Treeize.prototype.signature = function(row, options, auto) {
 
     node.name         = attr.node;
     node.depth        = attr.split.length - 1;
-    node.parent       = _.initial(attr.split, 2).join(opt.input.delimiter);
+    node.parent       = attr.split.slice(0, attr.split.length - 2).join(opt.input.delimiter);
     node.attributes.push({ name: attr.attr, key: attr.key });
     if (attr.pk) {
       this.log('adding node to blueprint');
@@ -133,7 +133,7 @@ Treeize.prototype.signature = function(row, options, auto) {
   }, this);
 
   // backfill blueprint when not specifically defined
-  _.each(nodes, function(node) {
+  nodes.forEach(function(node) {
     if (!node.blueprint.length) {
       node.blueprint = node.attributes;
     }
@@ -170,7 +170,7 @@ Treeize.prototype.clearSignature = function() {
 
 
 Treeize.prototype.grow = function(data, options) {
-  var opt = merge(this.options(), options || {});
+  var opt = merge(this._options, options || {});
 
   // chain past if no data to grow
   if (!data || !_.isArray(data) || !data.length) {
@@ -198,7 +198,7 @@ Treeize.prototype.grow = function(data, options) {
       data = [];
 
       // copy data without original signature row before processing
-      _.each(originalData, function(row, index) {
+      originalData.forEach(function(row, index) {
         if (index > 0) {
           data.push(row);
         }
@@ -215,7 +215,7 @@ Treeize.prototype.grow = function(data, options) {
   this.stats.sources++;
   var t1 = (new Date()).getTime();
 
-  _.each(data, function(row) {
+  data.forEach(function(row) {
     var trails = {}; // LUT for trails (find parent of new node in trails path)
     var trail = root = this.data.tree; // OPTIMIZATION: do we need to reset this trail for each row?
     this.log('CURRENT TRAIL STATUS>', trail);
@@ -248,20 +248,20 @@ Treeize.prototype.grow = function(data, options) {
       }, this);
     }
 
-    _.each(this.signature().nodes, function(node) {
+    this.signature().nodes.forEach(function(node) {
       this.log('PROCESSING NODE>', node);
       var blueprint = {};
       var blueprintExtended = {};
 
       // create blueprint for locating existing nodes
-      _.each(node.blueprint, function(attribute) {
+      node.blueprint.forEach(function(attribute) {
         var key = (node.path ? (node.path + ':') : '') + attribute.name;
         blueprint[attribute.name] = row[attribute.key];
         this.log('creating attribute "' + attribute.name + '" within blueprint', row[attribute.key]);
       }, this);
 
       // create full node signature for insertion/updating
-      _.each(node.attributes, function(attribute) {
+      node.attributes.forEach(function(attribute) {
         var key = (node.path ? (node.path + ':') : '') + attribute.name;
         var value = row[attribute.key];
 
@@ -338,7 +338,7 @@ Treeize.prototype.grow = function(data, options) {
                 this.log('..attempting path location for "' + pathAttempt + '"');
 
                 //infinite loop kickout
-                if (segmentsStripped > 5) break;
+                if (segmentsStripped > 15) break;
               }
               this.log('path FOUND for location for "' + pathAttempt + '" after removing ' + segmentsStripped + ' segments');
 
@@ -369,7 +369,7 @@ Treeize.prototype.grow = function(data, options) {
               // node attribute exists, set path for next pass
               // TODO: extend trail??
               this.log('object at node "' + node.name + '" exists as "' + trail[node.name] + '", skipping insertion and adding trail');
-              if (_.isObject(trail[node.name])) {
+              if (typeof trail[node.name] === 'object') {
                 trail[node.name] = merge(trail[node.name], blueprintExtended);
               }
               this.log('trail[node.name] updated to "' + trail[node.name]);
@@ -410,7 +410,7 @@ Treeize.prototype.options = function(options) {
 };
 
 Treeize.prototype.getOptions = function() {
-  return this.options();
+  return this._options;
 };
 
 Treeize.prototype.setOptions = function(options) {

package.json

@@ -1,6 +1,6 @@
 {
   "name": "treeize",
-  "version": "2.0.0",
+  "version": "2.0.1",
   "description": "Converts tabular row data (as from SQL joins, flat JSON, etc) to deep object graphs based on simple column naming conventions - without the use of an ORM or models.",
   "main": "./lib/treeize.js",
   "repository": {
@@ -10,6 +10,8 @@
   "keywords": [
     "JSON",
     "SQL",
+    "CSV",
+    "excel",
     "tree",
     "object",
     "graph",