I came across Scrapy a few days back and have grown to really love it. This tutorial will illustrate how you can write a simple spider using Scrapy to scrape data off Paul Smith. All this in 10 minutes.

Lets begin

1. Download and install scrapy and its dependencies.
2. This done, open up your terminal and type python scrapy-ctl.py startproject paul_smith. A scrapy project will be created.

3. Navigate to ~/paul_smith/paul_smith/spiders and create the file paul_smith.py with the following contents:

   paul_smith.py

   1 2 3 4 5 6 7 8 9 10

   from scrapy.spider import BaseSpider   class PaulSmithSpider(BaseSpider): domain_name = "paulsmith.co.uk" start_urls = ["http://www.paulsmith.co.uk/paul-smith-jeans-253/category.html"]   def parse(self, response): open('paulsmith.html', 'wb').write(response.body)   SPIDER = PaulSmithSpider()

4. To run the spider, go to ~/paul_smith type python scrapy-ctl.py crawl paulsmith.co.uk on the command line. This will fetch the page and save it to paulsmith.html.

5. The next step is to parse the contents of the page. Open the page in your favourite editor and try to understand the pattern of the items we want to capture. You can see that <div class="yui-u"> contains the required information. We are going to modify out code like so:

   paul_smith.py

   1 2 3 4 5 6 7 8 9 10 11 12 13 14

   from scrapy.spider import BaseSpider from scrapy.selector import HtmlXPathSelector   class PaulSmithSpider(BaseSpider): domain_name = "paulsmith.co.uk" start_urls = ["http://www.paulsmith.co.uk/paul-smith-jeans-253/category.html"]   def parse(self, response): hxs = HtmlXPathSelector(response) sites = hxs.select('//div[@class="yui-u"]') for site in sites: print site.extract()   SPIDER = PaulSmithSpider()

   You can read more on XPath Selectors here.

6. Finally, looking at the HTML again, we can extract title, link, img-src & sale-price like so:

   paul_smith.py

   1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

   from scrapy.spider import BaseSpider from scrapy.selector import HtmlXPathSelector import random   class PaulSmithSpider(BaseSpider): domain_name = "paulsmith.co.uk" start_urls = ["http://www.paulsmith.co.uk/paul-smith-jeans-253/category.html"]   def parse(self, response): hxs = HtmlXPathSelector(response) sites = hxs.select('//div[@class="yui-u"]') random.shuffle(sites) for site in sites: title = site.select('a/strong[@class="thumbnail-text"]/text()').extract() hlink = site.select('a/@href').extract() price = site.select('a/strong[@class="sale"]/text()').extract() image = site.select('a/img/@src').extract()   print title, hlink, image, price   SPIDER = PaulSmithSpider()

   You can save this data to your datastore in whatever way you wish.

7. The output of 3 random items scraped using the above code can be seen below.

Output

Shawl Collar Block Stripe Jumper
Sale: £ 74.00

Crew Neck Placement Stripe Jumper
Sale: £ 67.00

Tailored Fit, Organic Cotton Cravat Print Shirt
Sale: £ 74.00

A person’s URS is a phrase that could be used instead of his/her usual name in all circumstances, which makes it absolutely clear who he/she is. A good URS for a person should meet the following criteria:

• Everyone familiar with the person will confidently recognise him/her from the URS.
• There is no possibility that the URS could also describe anyone other than the person.
• Even someone who isn’t familiar with the person will have some understanding of who he/she is from the URS.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106

#!/usr/bin/python
""" 
Script to generate URS from the starting paragraph of Wikipedia 
articles about persons.
 
by Pravin Paratey (pravinp -at- gmail.com)
 
Current Implementation:
----------------------
1. Extract first sentence
2. Clean wiki markup
3. Observing given data, and the data on wikipedia, shows that there 
   is a pattern that is followed while writing wikipedia entries for
   persons. Replacing (was/is)(an/a/the/) with (/the) does the trick
4. Output sentence formed
 
Ideally:
--------
Ideally, the piece of code should identify the following concepts:
1. Name of person
2. Time period
3. Son/Daughter/Father/Mother of (in case of famous personality)
4. Renowned for
 
How do we go about it?
1 and 2 - straight forward. Wikipedia gives cues through its markup
3 - straight forward. String matching using "son of", "daughter of", etc
4 - will need to match against a database.
 
For 3, we only keep the "son of", "daughter of", "X of Y" if Y is a prominent
person. An easy way of doing this is using incoming links on wikipedia OR
to search for X and Y individually on google and noting the number of results.
"""
 
import re, sys, codecs
 
def cleanUri(m):
    """ Cleans Uri wiki markup """
    word = m.group(1)
    if '|' in word: word = word.split('|')[1]
    return word.strip()
 
def dotRemove(m):
    """ Replaces . by # inside tags """
    return m.group(0).replace('.', '#')
 
def cleanMarkup(text):
    """ Removes
    1. wiki markup
    2. sanitize html entities 
    3. comments """
    #text = re.sub(r"\[\[[\w\s\-,]+\|(\w+)\]\]", r"\1", text)
    text = re.sub(r"\[\[(.*?)\]\]", cleanUri, text)
    text = re.sub(r"\{\{.*?\}\}", r"", text)
    text = re.sub(r"<ref>.*?<\/ref>", r"", text)
    text = re.sub(r"<!--.*?-->", r"", text)
    text = re.sub(r"\[.*?\]", r"", text)
    text = text.replace("'''", "").replace("''", "'")
    text = text.replace("[[", "").replace("]]", "")
    text = text.replace("&ndash;", "-").replace("&amp;", "&")
    return text
 
def getFirstSentence(text):
    """ Returns the text until first instance of '.'
    It also makes sure that the '.' isn't part of a wiki link
    or name"""
    tmp = re.sub(r"\[\[.*?\]\]", dotRemove, text)
    tmp = re.sub(r"\[.*?\]", dotRemove, tmp)
    tmp = re.sub(r"<ref>.*?<\/ref>", dotRemove, tmp)
    tmp = re.sub(r"<!--.*?-->", dotRemove, tmp)
    tmp = re.sub(r"'''.*?'''", dotRemove, tmp)
    tmp = re.sub(r"''.*?''", dotRemove, tmp)
    index = tmp.find('.')
 
    if index == -1: 
        return text
    else:
        return text[:index]
 
def makeArticle(m):
    """ Changes a, an to the when appropriate """
    retval = ', the'
    if len(m.group(2)) == 0:
        retval = ' '
    return retval
 
def extractURS(text):
    """ The function to call. Returns the URS """
    text = getFirstSentence(text)
    text = cleanMarkup(text)
    text = re.sub(r",?\s+(was|is)\s+(an|the|a|)", makeArticle, text)
    return text
 
if __name__ == '__main__':
    #fp = open(sys.argv[1])
    fp = codecs.open("input.txt", "r", "utf-8")
    fp2 = codecs.open("output.txt", "w", "utf-8")
    fp2.write(codecs.BOM_UTF8.decode("utf-8")), # Add BOM for UTF-8
    for line in fp:
        line = line.rstrip()
        if len(line) == 0 or line.startswith("#"): # For debugging
            continue
        urs = extractURS(line)
        fp2.write(urs + '\r\n')
    fp.close()
    fp2.close()

Example Inputs and Outputs

These are inputs from Wikipedia (Click on the article and then Edit). Ex Lala Lajpat Rai. The above script outputs the URS.

Example Input: ”’B. S. Johnson ”’ (Bryan Stanley Johnson) ([[5 February]],[[1933]] – [[13 November]],[[1973]]) was an English experimental novelist, poet, literary critic and film-maker.

Script Output: B. S. Johnson (Bryan Stanley Johnson) (5 February,1933 – 13 November,1973), the English experimental novelist, poet, literary critic and film-maker

How are URS used?

URS can be directly substituted in a sentence containing that persons’ name. (Hover over Bhagat Singh to see this URS.

ex. Bhagat Singh was executed by the British in 1931.

This way, a person who had no idea who Bhagat Singh was, now has more context about the person.