def get_url_of_image(query_string): '''gets an image from a yahoo image search of "query string", returns a url''' srch = ImageSearch(app_id="YahooDemo", query='%s'%query_string) srch.results = 1 for each in srch.parse_results(): url = str(each.Url) return url
def __init__(self,query="Rocket",results=50, with_cache=None,rescale=None):
done=False
self.rescale=rescale
if (with_cache):
try:
os.stat(with_cache+"/"+query.lower())
f=file(with_cache+"/"+query+"/reader.pcl","rb")
self.query=pickle.load(f)
self.urls=pickle.load(f)
self.destdir=pickle.load(f)
self.lq=pickle.load(f)
done=True
except:
pass
if not done:
from pycvf.core import settings
from yahoo.search.image import ImageSearch
yahoo_image_search=ImageSearch(settings.YAHOOAPPID)
yahoo_image_search.query = query
yahoo_image_search.results = results
r=yahoo_image_search.get_results()
self.query=query
self.urls=[ ]
for i in range(40):
try:
self.urls.append(r.childNodes[0].getElementsByTagName('Result')[i].getElementsByTagName('Url')[0].childNodes[0].nodeValue)
except:
pass
if not (with_cache):
self.destdir="/tmp/"+yahoo_image_search.query.lower()
else:
self.destdir=with_cache+yahoo_image_search.query.lower()
try:
os.mkdir(self.destdir)
except:
pass
self.lq=len(self.urls)
dl=map(lambda i:(self.urls[i],(self.destdir+"/f-"+str(i)+"."+self.urls[i].split(".")[-1]).lower()) ,range(len(self.urls)))
if (self.lq):
md=MultiDownloader(dl)
md.run()
if (with_cache):
try:
os.mkdir(with_cache+"/"+query.lower())
except:
pass
f=file(with_cache+"/"+query+"/reader.pcl","wb")
pickle.dump(self.query,f,protocol=2)
pickle.dump(self.urls,f,protocol=2)
pickle.dump(self.destdir,f,protocol=2)
pickle.dump(self.lq,f,protocol=2)
f.close()
def _run_query_fired(self):
""" Queries Yahoo! image search for with the query string """
# This is where we run the actual query once the user tells the GUI to
# do so. We're using the pYsearch module to do a Yahoo image search,
# which basically exposes a list of image urls.
srch = ImageSearch(app_id="YahooDemo", query=self.query_string)
# Run the search and return the top 10 results
srch.results = 10
# We make a list of the top 10 image urls.
imagrul_list = []
for res in srch.parse_results():
imagrul_list.append(res.Url)
# We will try each image url in succession to download the image. We
# need to do this in a complicated control loop because not all image
# urls returned from the search still work. We need to try/except
# the urllib2 connection errors and check that the image url linked to
# a valid image we can download. If not, move on to the next url. If
# no image urls were returned in the search (for example, if the search
# was too specific and returned no results), print an explanation
# in the GUI by writing it into the image_url string.
final_imgurl = False
for imgurl in imagrul_list:
print "Connecting to", imgurl
error_encountered = False
try:
response = urllib2.urlopen(imgurl)
except:
error_encountered = True
print "Oh no, HTTP error encountered on " + imgurl
if not error_encountered:
final_imgurl = imgurl
break
if final_imgurl:
print "Successfully accessed", final_imgurl
self.image_url = final_imgurl
image_string = response.read()
response.close()
image_file = file("image.jpg", "w")
image_file.write(image_string)
image_file.close()
image_file = file("image_original.jpg", "w")
image_file.write(image_string)
image_file.close()
self.display_image()
print "Retrieved image and saved it as image.jpg"
else:
print ("Failed to retrieve any image, no " +
"results found.")
self.image_url = ("Failed to retrieve any image, " +
"no results found.")
def getImage(q) :
srch = ImageSearch(app_id="YahooDemo",query=q)
# get one result in png format
srch.results = 1
srch.format = "png"
for res in srch.parse_results():
# the file-like object returned by
# urllib doesn't implement some of the methods needed by imread. i
# tried to read() the urllib object and convert that to a string
# stream, but imread didn't like that, either. i knew creating
# a file and reading it would work, so i did it.
commands.getoutput("wget '"+res.Url+"' -O tmpfile")
x = plt.imread("tmpfile")
commands.getoutput("rm tmpfile")
# images get reversed for some reason, reverse it back
x = x[::-1]
return (x, res.Url)