Use requests lib instead of low-level urllib calls

With requests library users can run stackalytics-processor from
environments located behind HTTP/HTTPS proxy. Proxy address is
configured by setting the environment variables HTTP_PROXY or
HTTPS_PROXY.

Closes-Bug: #1351136

Change-Id: I6a65afb0f99b351dc2183294d9127cbbebc35856

requirements.txt

@@ -16,5 +16,7 @@ psutil<2.0.0,>=1.1.1
 PyGithub
 python-memcached>=1.56
 PyYAML>=3.1.0
+requests>=2.5.2
+requests-file
 sh
 six>=1.9.0

stackalytics/processor/launchpad_utils.py

@@ -15,8 +15,6 @@
 
 from oslo_log import log as logging
 import six
-from six.moves import http_client
-from six.moves.urllib import parse
 
 from stackalytics.processor import utils
 
@@ -54,14 +52,11 @@ def lp_profile_by_email(email):
 
 def lp_module_exists(module):
     uri = LP_URI_DEVEL % module
-    parsed_uri = parse.urlparse(uri)
-    conn = http_client.HTTPConnection(parsed_uri.netloc)
-    conn.request('GET', parsed_uri.path)
-    res = conn.getresponse()
+    request = utils.do_request(uri)
+
     LOG.debug('Checked uri: %(uri)s, status: %(status)s',
-              {'uri': uri, 'status': res.status})
-    conn.close()
-    return res.status != 404
+              {'uri': uri, 'status': request.status_code})
+    return request.status_code != 404
 
 
 def lp_blueprint_generator(module):

stackalytics/processor/mls.py

@@ -18,7 +18,6 @@ import re
 
 from oslo_log import log as logging
 import six
-from six.moves import http_client
 from six.moves.urllib import parse
 
 from stackalytics.processor import utils
@@ -60,31 +59,25 @@ def _get_mail_archive_links(uri):
     return [parse.urljoin(uri, link) for link in links]
 
 
-def _link_content_changed(link, runtime_storage_inst):
-    LOG.debug('Check changes for mail archive located at uri: %s', link)
-    parsed_uri = parse.urlparse(link)
-    conn = http_client.HTTPConnection(parsed_uri.netloc)
-    conn.request('HEAD', parsed_uri.path)
-    res = conn.getresponse()
-    last_modified = res.getheader('last-modified')
-    conn.close()
+def _uri_content_changed(uri, runtime_storage_inst):
+    LOG.debug('Check changes for mail archive located at: %s', uri)
+    last_modified = utils.get_uri_last_modified(uri)
 
-    if last_modified != runtime_storage_inst.get_by_key('mail_link:' + link):
+    if last_modified != runtime_storage_inst.get_by_key('mail_link:' + uri):
         LOG.debug('Mail archive changed, last modified at: %s', last_modified)
-        runtime_storage_inst.set_by_key('mail_link:' + link, last_modified)
+        runtime_storage_inst.set_by_key('mail_link:' + uri, last_modified)
         return True
 
     return False
 
 
 def _retrieve_mails(uri):
-    LOG.debug('Retrieving mail archive from uri: %s', uri)
-    content = utils.read_uri(uri)
+    LOG.debug('Retrieving mail archive from: %s', uri)
+    content = utils.read_gzip_from_uri(uri)
     if not content:
-        LOG.error('Error reading mail archive from uri: %s', uri)
+        LOG.error('Error reading mail archive from: %s', uri)
         return
 
-    content = utils.gzip_decompress(content)
     LOG.debug('Mail archive is loaded, start processing')
 
     content += TRAILING_RECORD
@@ -116,7 +109,7 @@ def log(uri, runtime_storage_inst):
 
     links = _get_mail_archive_links(uri)
     for link in links:
-        if _link_content_changed(link, runtime_storage_inst):
+        if _uri_content_changed(link, runtime_storage_inst):
             for mail in _retrieve_mails(link):
                 LOG.debug('New mail: %s', mail['message_id'])
                 yield mail

stackalytics/processor/utils.py

@@ -17,8 +17,6 @@ import calendar
 import cgi
 import datetime
 import gzip
-import io
-import json
 import random
 import re
 import time
@@ -26,6 +24,8 @@ import time
 import iso8601
 from oslo_config import cfg
 from oslo_log import log as logging
+import requests
+import requests_file
 import six
 
 
@@ -117,30 +117,31 @@ user_agents = [
 ]
 
 
+def do_request(uri, method='get'):
+    with requests.Session() as session:
+        session.mount('file://', requests_file.FileAdapter())
+        user_agent = random.choice(user_agents)
+
+        return session.request(method, uri, headers={'User-Agent': user_agent})
+
+
 def read_uri(uri):
     try:
-        req = six.moves.urllib.request.Request(
-            url=uri, headers={'User-Agent': random.choice(user_agents)})
-        fd = six.moves.urllib.request.urlopen(req)
-        if six.PY3:
-            fd = io.TextIOWrapper(fd)
-        raw = fd.read()
-        fd.close()
-        return raw
+        return do_request(uri).text
     except Exception as e:
-        LOG.warn('Error "%(error)s" while reading uri %(uri)s',
+        LOG.warn('Error "%(error)s" retrieving uri %(uri)s',
                  {'error': e, 'uri': uri})
 
 
 def read_json_from_uri(uri):
     try:
-        return json.loads(read_uri(uri))
+        return do_request(uri).json()
     except Exception as e:
         LOG.warn('Error "%(error)s" parsing json from uri %(uri)s',
                  {'error': e, 'uri': uri})
 
 
-def gzip_decompress(content):
+def _gzip_decompress(content):
     if six.PY3:
         return gzip.decompress(content).decode('utf8')
     else:
@@ -148,6 +149,22 @@ def gzip_decompress(content):
         return gzip_fd.read()
 
 
+def read_gzip_from_uri(uri):
+    try:
+        return _gzip_decompress(do_request(uri).content)
+    except Exception as e:
+        LOG.warn('Error "%(error)s" retrieving uri %(uri)s',
+                 {'error': e, 'uri': uri})
+
+
+def get_uri_last_modified(uri):
+    try:
+        return do_request(uri, method='head').headers['last-modified']
+    except Exception as e:
+        LOG.warn('Error "%(error)s" retrieving uri %(uri)s',
+                 {'error': e, 'uri': uri})
+
+
 def cmp_to_key(mycmp):  # ported from python 3
     """Convert a cmp= function into a key= function."""
     class K(object):