gistfile1.diff

4a5,7
> import Queue
> import threading
> import heapq
13,16c16,29
< class WebSpider(object):
<     def __init__(self):
<         self.results = {}
< 
---
> class WebSpider(threading.Thread):
>     def __init__(self, urls, results):
>         threading.Thread.__init__(self)
>         self.results = results
>         self.urls = urls
>         
> 
>     def run(self):
>             while True:
>                 try:
>                     self.fetch(self.urls.get(False))
>                 except Queue.Empty:
>                     break
>     
26c39,40
<             self.results[url] = resp.read()
---
>             self.results.put((url, resp.read()), True)
>             #self.results[url] = resp.read()
30,36c44,48
<     def fetch_all(self, iterable):
<         for url in iterable:
<             self.fetch(url)
< 
<     def most_vowels(self, num=10):
<         """
<         Return a list of the URLs which had the most vowels.
---
>     #def fetch_all(self, iterable):
>     #    for url in iterable:
>     #        self.fetch(url)
> 
> 
38c50,55
<         Each item in the list is:
---
> 
> def most_vowels(self, num=10):
>     """
>     Return a list of the URLs which had the most vowels.
> 
>     Each item in the list is:
41,43c58,66
<         """
<         scored_urls = [(u, b, count_vowels(b)) for u, b in self.results.items()]
<         return sorted(scored_urls, key=lambda x: x[-1], reverse=True)[:num]
---
>     """
>     #scored_urls = [(u, b, count_vowels(b)) for u, b in self.results.items()]
>     #return sorted(scored_urls, key=lambda x: x[-1], reverse=True)[:num]
>     h = heapq.heapq()
>     # *****set max heap size to num
>     for u,b in results:
>         h.heappush((u, b))
> 
>      # **** return heap *******
47c70
<     urls = []
---
>     urls = Queue.Queue()
51c74
<             urls.append(line.strip())
---
>             urls.put(line.strip())
53,54c76,86
<     spider = WebSpider()
<     spider.fetch_all(urls)
---
>     results = Queue.Queue()
>     spiders = set([WebSpider(urls, results) for _ in range(100)])
>     for s in spiders:
>         s.run()
>     while len(spiders) > 0:
>         if s.is_alive():
>             del spiders[s]
>             try:
>                 s.join()
>             except RuntimeError: pass
>         
59,60c91,92
<     for url, content, score in spider.most_vowels(10):
<         print '%-35s %-50s %d' % (url, content, score)
---
>     #for url, content, score in spider.most_vowels(10):
>      #   print '%-35s %-50s %d' % (url, content, score)