annotate tools/transFIC/transFIC_web.py @ 2:b47ee319385d

opush
author Saket Choudhary <saketkc@gmail.com>
date Wed, 20 Nov 2013 01:52:26 +0530
parents 7404ab872276
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
1 import sys
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
2 sys.path.insert(0, '/home/saket/requests-new-urllib3-api/requests/packages/')
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
3 sys.path.insert(0, '/home/saket/requests-new-urllib3-api')
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
4
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
5
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
6 import requests
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
7 import pycurl
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
8 import os
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
9 from os.path import getsize
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
10 import argparse
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
11 import sys
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
12 import cStringIO
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
13 from functools import wraps
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
14 import tempfile, shutil,time
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
15
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
16 __url__="http://bg.upf.edu/transfic/taskService"
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
17 def stop_err( msg ):
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
18 sys.stderr.write( '%s\n' % msg )
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
19 sys.exit()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
20
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
21 def retry(ExceptionToCheck, tries=14, delay=3, backoff=2, logger=None):
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
22 """Retry calling the decorated function using an exponential backoff.
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
23
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
24 http://www.saltycrane.com/blog/2009/11/trying-out-retry-decorator-python/
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
25 original from: http://wiki.python.org/moin/PythonDecoratorLibrary#Retry
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
26
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
27 :param ExceptionToCheck: the exception to check. may be a tuple of
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
28 exceptions to check
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
29 :type ExceptionToCheck: Exception or tuple
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
30 :param tries: number of times to try (not retry) before giving up
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
31 :type tries: int
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
32 :param delay: initial delay between retries in seconds
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
33 :type delay: int
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
34 :param backoff: backoff multiplier e.g. value of 2 will double the delay
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
35 each retry
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
36 :type backoff: int
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
37 :param logger: logger to use. If None, print
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
38 :type logger: logging.Logger instance
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
39 """
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
40 def deco_retry(f):
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
41
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
42 @wraps(f)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
43 def f_retry(*args, **kwargs):
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
44 mtries, mdelay = tries, delay
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
45 while mtries > 1:
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
46 try:
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
47 return f(*args, **kwargs)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
48 except ExceptionToCheck, e:
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
49 #msg = "%s, Retrying in %d seconds..." % (str(e), mdelay)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
50 msg = "Retrying in %d seconds..." % (mdelay)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
51 if logger:
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
52 logger.warning(msg)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
53 else:
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
54 #print msg
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
55 pass
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
56 time.sleep(mdelay)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
57 mtries -= 1
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
58 mdelay *= backoff
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
59 return f(*args, **kwargs)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
60
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
61 return f_retry # true decorator
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
62
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
63 return deco_retry
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
64
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
65 class TransficUploader:
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
66 def __init__(self):
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
67
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
68 self.c = pycurl.Curl()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
69 self.c.setopt(pycurl.URL, __url__)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
70 self.c.setopt(pycurl.UPLOAD, 1)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
71 #c.setopt(pycurl.USERPWD, 'saket.kumar:whatsinaname.')
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
72 self.c.setopt(pycurl.PROXY, 'http://saket.kumar:whatsinaname.@netmon.iitb.ac.in:80/')
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
73 #c.setopt(pycurl.PROXYPORT, 80)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
74 #c.setopt(pycurl.PROXYTYPE_HTTP,1)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
75 #self.c.setopt(pycurl.VERBOSE, 1)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
76 self.c.setopt(pycurl.HTTPHEADER, ['Expect:'])
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
77 #c.setopt(pycurl.HTTPPROXYTUNNEL, 1)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
78 self.c.setopt(pycurl.UPLOAD, 1)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
79 #c.perform()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
80 self.c.setopt(pycurl.NOPROGRESS, 1);
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
81 #curl_easy_setopt(hnd, CURLOPT_UPLOAD, 1L);
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
82 self.c.setopt(pycurl.USERAGENT, "curl/7.27.0");
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
83 #curl_easy_setopt(hnd, CURLOPT_HTTPHEADER, slist1);
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
84 #curl_easy_setopt(hnd, CURLOPT_MAXREDIRS, 50L);
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
85 self.c.setopt(pycurl.SSL_VERIFYPEER, 1);
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
86 self.c.setopt(pycurl.CUSTOMREQUEST, "PUT")
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
87 #curl_easy_setopt(hnd, CURLOPT_VERBOSE, 1L);
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
88 self.c.setopt(pycurl.TCP_NODELAY, 1);
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
89 self.buf = cStringIO.StringIO()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
90 self.c.setopt(self.c.WRITEFUNCTION, self.buf.write)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
91
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
92
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
93 def upload_file(self,filepath):
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
94 f = open(filepath)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
95 self.c.setopt(pycurl.INFILE, f)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
96 self.c.setopt(pycurl.INFILESIZE, getsize(filepath))
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
97 #print f.read()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
98
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
99 def run(self):
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
100 self.c.perform()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
101
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
102 def get_url(self):
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
103 return self.buf.getvalue().strip()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
104
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
105 @retry(requests.exceptions.HTTPError)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
106 def result_exists(self,url ):
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
107 #url="http://www.cravat.us/results/%s/%s.zip" %(job_id,job_id)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
108 download_request = requests.request("GET", url)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
109 if download_request.status_code==404 or download_request==500:
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
110 raise requests.HTTPError()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
111 elif "Task status is : error" in download_request.text:
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
112 stop_err("No SNVs found!")
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
113 else:
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
114 return url
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
115 @retry(requests.exceptions.HTTPError)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
116 def download_result(self, url,outpath):
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
117 tmp_dir = tempfile.mkdtemp()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
118 r = requests.get( url, stream=True )
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
119 if r.status_code == 500:
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
120 raise requests.HTTPError()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
121 else:
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
122 path = os.path.join( tmp_dir,"results.csv")
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
123 with open(path, 'wb') as f:
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
124 for chunk in r.iter_content(128):
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
125 f.write(chunk)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
126 shutil.move(path,outpath)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
127 shutil.rmtree(tmp_dir)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
128
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
129
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
130
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
131 def main(params):
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
132 parser = argparse.ArgumentParser()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
133 parser.add_argument("--input",type=str,required=True)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
134 parser.add_argument("--output",type=str,required=True)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
135 args = parser.parse_args(params)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
136 uploader = TransficUploader();
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
137 uploader.upload_file(args.input)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
138 uploader.run()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
139 url = uploader.get_url()
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
140 url = uploader.result_exists(url)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
141 download = uploader.download_result(url,args.output)
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
142 if __name__=="__main__":
7404ab872276 First Commit
Saket Choudhary <saketkc@gmail.com>
parents:
diff changeset
143 main(sys.argv[1:])