clothing-search.py

import sys
import requests
from pyquery import PyQuery

# Clothing items holds all the info needed to display a clothing item from 
# any source. 
class ClothingItem(object):

	def __init__(self, name, price, imageURL, itemURL):
		super(ClothingItem, self).__init__()
		self.name = name
		self.price = price
		self.imageURL = imageURL
		self.itemURL = itemURL

	# When a list containing ClothingItem objects is printed, this method is 
	# called to create a string representation of the object. 
	def __str__(self):
		strRepresentation = "\n" # "\n" represents a newline. 
		strRepresentation += "Name: " + self.name + "\n"
		strRepresentation += "Price: " + self.price + "\n"
		strRepresentation += "Image: " + self.imageURL + "\n"
		strRepresentation += "Item: " + self.itemURL + "\n"
		return strRepresentation;
	def __repr__(self):
		return self.__str__()

# Returns a list of clothing items from express based on an input string. 
def listFromSearchString(searchStr):
	r = requests.get(searchURLFromString(searchStr))
	htmlItemList = listFromSearchHTML(r.text);
	clothingItemList = map(parseSearchItem, htmlItemList);
	return clothingItemList;

# Creates the url for the search string provided as a parameter. 
def searchURLFromString(search_str):
	template_url = "http://www.express.com/browse/search.jsp?q="
	encoded_str = "+".join(search_str.split(" "))
	return template_url + encoded_str

# Convert search HTML into a list of html elements that represent the 
# clothing items. 
def listFromSearchHTML(htmlStr):
	d = PyQuery(htmlStr);
	return d.find("#browse-gallery").find("li.photo");

# Converts an HTML element that represents a clothing item into a 
# ClothingItem object. Does HTML parsing using PyQuery
def parseSearchItem(item):
	d = PyQuery(item)
	name = d.find("li.prod-name").text();
	price = d.find("li.price").text().split(" ")[0];
	imageURL = "http:" + d.find("img").attr("src");
	itemURL = "http://www.express.com" + d.find("a").attr("href")

	return ClothingItem(name, price, imageURL, itemURL)

# Start Point. Any lines of code that are not inside a function will run as 
# soon as the progream starts from top to bottom. Lines of code in functions 
# are called when the function is called. 

search_str = sys.argv[1]
print listFromSearchString(search_str);