First draft of theaters scraping

lib/projection.js

@@ -4,15 +4,15 @@ var request = require('request'),
   cheerio = require('cheerio');
 
 var Projection = function() {
-  this.GOOGLE_ENDPOINT = "http://www.google.com/movies";
+  this.GOOGLE_ENDPOINT = 'http://www.google.com/movies';
 };
 
 // Find showtimes of nearby theaters
 Projection.prototype.findTheaters = function(near, callback) {
   var err,
   	result = {};
 
-  var url = this.GOOGLE_ENDPOINT + "?near=" + near;
+  var url = this.GOOGLE_ENDPOINT + '?near=' + near;
 
   request(url, function (error, response, body) {
 	  if (error) {
@@ -26,27 +26,73 @@ Projection.prototype.findTheaters = function(near, callback) {
 	  }
 
 		var $ = cheerio.load(body);
-
-		var theaters = $(".theater").text();
-
-		//result = $(".theater").find(".desc .name").text();
-
 		var theaters = [];
 
-		$(".theater").each(function(i, value) {
+		$('.theater').each(function(i, t) {
+			t = $(t);
+
 			var theater = {
-				name: $(this).find(".desc .name").text(),
-				address: $(this).find(".desc .info").text().trim('-')[0],
-				phone: $(this).find(".desc .info").text().trim(' - ')[1]
+				name: t.find('.desc .name').text(),
+				address: t.find('.desc .info').text().split(' - ')[0].trim(),
+				phone: t.find('.desc .info').text().split(' - ')[1].trim(),
+				note: t.find('.desc .closure').text(), //status
+				movies: []
 			};
 
+			// Google movie info format : Duration - Rating - Genre - Trailer - IMDB
+			t.find('.showtimes .movie').each(function(j, m) {
+				m = $(m);
+				var infos = m.find('.info').text().split(' - ');
+
+				var cursor = 0,
+					duration = null,
+					rating 	 = null,
+					genre 	 = null,
+					trailer  = null;
+
+				if (infos[cursor].match(/(1|2|hr|min)/)){
+					duration = infos[cursor].trim();
+					cursor++;
+				}
+
+				if (infos[cursor].match(/(G|\+|13|16|18)/)){
+					rating = infos[cursor].trim();
+					cursor++;
+				}
+
+				if (infos[cursor].match(/(^\D*$)/)){
+					genre = infos[cursor].trim();
+					cursor++;
+				}
+
+				if (m.find('.info a').attr('href') && (m.find('.info a').attr('href').match(/(youtube|vimeo|daily)/))) {
+					trailer = m.find('.info a').attr('href').replace('/url?q=', '').trim();
+				}
+
+
+				var showtimes = m.find('.times').text().split(' ');
+				showtimes = showtimes.map(function(s) {
+					return s.trim(); // TODO new Date (w/ am/pm)
+					//return new Date(s);
+				});
+
+				//console.log(infos);
+				var movie = {
+					title: m.find('.name').text(),
+					duration: duration,
+					rating: rating,
+					genre: genre,
+					trailer: trailer,
+					showtimes: showtimes
+				};
+
+				theater.movies.push(movie);
+			});			
+
 			theaters.push(theater);
-		  //theaters.push(value.text());
 		});
 
-		result = theaters;
-
-		callback(err, result);
+		callback(err, theaters);
 	})
 
 };
@@ -57,4 +103,9 @@ Projection.prototype.findMovie = function(callback) {
   // TODO
 };
 
+// var p = new Projection();
+// p.findTheaters('Montreal', function(err, theaters) {
+// 	console.log(theaters[2].movies[0	]);
+// });
+
 module.exports = Projection;

package.json

@@ -1,10 +1,10 @@
 {
   "name": "projection",
-  "version": "1.0.0",
-  "description": "A nodejs package for movie showtimes around the globe",
+  "version": "0.1.0",
+  "description": "Fast acquisition of movie showtimes around the globe",
   "main": "index.js",
   "scripts": {
-    "test": "mocha ./test/**.js",
+    "test": "mocha ./test/*.js",
     "test-travis": "istanbul cover ./node_modules/mocha/bin/_mocha -- -R spec ./test/*.js"
   },
   "repository": {

test/tests.js

@@ -1,6 +1,6 @@
 'use strict';
 
-var assert = require("assert")
+var assert = require('assert')
 var Projection = require('../lib/projection');
 
 var p; // Projection
@@ -19,13 +19,13 @@ describe('Projection', function(){
 	describe('new Projection()', function(){
     it('should have the corresponding endpoint', function(){
     	p = new Projection();
-    	assert.equal(p.GOOGLE_ENDPOINT, "http://www.google.com/movies");
+    	assert.equal(p.GOOGLE_ENDPOINT, 'http://www.google.com/movies');
     })
   })
 
   describe('#findTheaters()', function(){
     it('should find theaters by town', function(done){
-			p.findTheaters("Montreal", function(err, theaters) {
+			p.findTheaters('Montreal', function(err, theaters) {
 				//console.log(theaters);
 
 				assert.equal(err, null);
@@ -37,30 +37,30 @@ describe('Projection', function(){
     })
 
     it('should find theaters by zipcode', function(done){
-    	p.findTheaters("Montreal", function(err, theaters) {
+    	p.findTheaters('Montreal', function(err, theaters) {
 				assert.equal(err, null);
 				done();
 			});
     })
 
     it('should find theaters by lat/long', function(done){
-    	p.findTheaters("Montreal", function(err, theaters) {
+    	p.findTheaters('Montreal', function(err, theaters) {
 				assert.equal(err, null);
 				done();
 			});
     })
 
     it('should return requestjs error', function(done){
-    	p.GOOGLE_ENDPOINT = "abc" // Override endpoint
-    	p.findTheaters("Montreal", function(err, theaters) {
-				assert.equal(err, "Error: Invalid URI \"abc?near=Montreal\"");
+    	p.GOOGLE_ENDPOINT = 'abc'; // Override endpoint
+    	p.findTheaters('Montreal', function(err, theaters) {
+				assert.equal(err, 'Error: Invalid URI "abc?near=Montreal"');
 				done();
 			});
     })
 
     it('should return a 404 error', function(done){
-    	p.GOOGLE_ENDPOINT = "http://httpstat.us/404" // Override endpoint
-    	p.findTheaters("Montreal", function(err, theaters) {
+    	p.GOOGLE_ENDPOINT = 'http://httpstat.us/404' // Override endpoint
+    	p.findTheaters('Montreal', function(err, theaters) {
 				assert.equal(err, 404);
 				done();
 			});