warcserver: define default HTTPAdapter in warcserver.http.default_adapter, for use with both index sources and responseloader

responseloader uses existing pool from shared HTTPAdapter
fix tests: call_release_conn() checks if release_conn() exists before calling, else default to close()

pywb/utils/io.py

@@ -27,7 +27,10 @@ def call_release_conn(stream):
     try:
         yield stream
     finally:
-        stream.release_conn()
+        if hasattr(stream, 'release_conn'):
+            stream.release_conn()
+        else:
+            stream.close()
 
 
 #=============================================================================

pywb/warcserver/http.py

@@ -0,0 +1,5 @@
+from requests.adapters import HTTPAdapter
+
+default_adapter = HTTPAdapter(max_retries=3)
+
+

pywb/warcserver/index/indexsource.py

@@ -5,6 +5,7 @@ from pywb.utils.wbexception import NotFoundException
 from warcio.timeutils import timestamp_to_http_date, http_date_to_timestamp
 from warcio.timeutils import timestamp_now, pad_timestamp, PAD_14_DOWN
 
+from pywb.warcserver.http import default_adapter
 from pywb.warcserver.index.cdxobject import CDXObject
 
 from pywb.utils.format import ParamFormatter, res_template
@@ -31,6 +32,11 @@ class BaseIndexSource(object):
         else:
             return None
 
+    def _init_sesh(self):
+        self.sesh = requests.Session()
+        self.sesh.mount('http://', default_adapter)
+        self.sesh.mount('https://', default_adapter)
+
 
 #=============================================================================
 class FileIndexSource(BaseIndexSource):
@@ -96,6 +102,7 @@ class RemoteIndexSource(BaseIndexSource):
         self.replay_url = replay_url
         self.url_field = url_field
         self.closest_limit = closest_limit
+        self._init_sesh()
 
     def _get_api_url(self, params):
         api_url = res_template(self.api_url, params)
@@ -106,8 +113,11 @@ class RemoteIndexSource(BaseIndexSource):
 
     def load_index(self, params):
         api_url = self._get_api_url(params)
-        r = requests.get(api_url, timeout=params.get('_timeout'))
-        if r.status_code >= 400:
+        try:
+            r = self.sesh.get(api_url, timeout=params.get('_timeout'))
+            r.raise_for_status()
+        except Exception as e:
+            print('FAILED: ' + str(e))
             raise NotFoundException(api_url)
 
         lines = r.content.strip().split(b'\n')
@@ -317,12 +327,6 @@ class MementoIndexSource(BaseIndexSource):
         self.replay_url = replay_url
         self._init_sesh()
 
-    def _init_sesh(self):
-        self.sesh = requests.Session()
-        adapt = requests.adapters.HTTPAdapter(max_retries=3)
-        self.sesh.mount('http://', adapt)
-        self.sesh.mount('https://', adapt)
-
     def links_to_cdxobject(self, link_header, def_name):
         results = MementoUtils.parse_links(link_header, def_name)
 
@@ -360,10 +364,9 @@ class MementoIndexSource(BaseIndexSource):
             headers = self._get_headers(params)
             headers['Accept-Datetime'] = accept_dt
             res = self.sesh.head(url, headers=headers)
-            if res.status_code >= 400:
-                raise NotFoundException(url)
+            res.raise_for_status()
         except Exception as e:
-            print('FAILED:', e)
+            print('FAILED: ' + str(e))
             raise NotFoundException(url)
 
         links = res.headers.get('Link')

pywb/warcserver/resource/responseloader.py

@@ -14,6 +14,8 @@ from pywb.utils.format import ParamFormatter
 from pywb.warcserver.resource.resolvingloader import ResolvingLoader
 from pywb.warcserver.resource.pathresolvers import DefaultResolverMixin
 
+from pywb.warcserver.http import default_adapter
+
 from six.moves.urllib.parse import urlsplit, quote, unquote
 
 from io import BytesIO
@@ -26,7 +28,6 @@ import glob
 import datetime
 
 from requests.models import PreparedRequest
-from requests.packages import urllib3
 
 import six.moves.http_client
 six.moves.http_client._MAXHEADERS = 10000
@@ -233,14 +234,14 @@ class LiveWebLoader(BaseLoader):
 
     UNREWRITE_HEADERS = ('location', 'content-location')
 
-    def __init__(self, forward_proxy_prefix=None):
-        self.num_retries = 3
-        self.num_pools = 10
-        self.num_conn_per_pool = 10
+    def __init__(self, forward_proxy_prefix=None, adapter=None):
         self.forward_proxy_prefix = forward_proxy_prefix
 
-        self.pool = urllib3.PoolManager(num_pools=self.num_pools,
-                                        maxsize=self.num_conn_per_pool)
+        if not adapter:
+            adapter = default_adapter
+
+        self.pool = adapter.poolmanager
+        self.max_retries = adapter.max_retries
 
     def load_resource(self, cdx, params):
         load_url = cdx.get('load_url')
@@ -442,7 +443,7 @@ class LiveWebLoader(BaseLoader):
                                              assert_same_host=False,
                                              preload_content=False,
                                              decode_content=False,
-                                             retries=self.num_retries,
+                                             retries=self.max_retries,
                                              timeout=params.get('_timeout'))
 
             return upstream_res