Fix(Robots/bkmk_rrequests): Add forgotten spaces in log

[bookmarks_db.git] / Robots / bkmk_rrequests.py

diff --git a/Robots/bkmk_rrequests.py b/Robots/bkmk_rrequests.py
index e1f00198dcda56a7f4d54835ce322d29ba51bc69..c6a91af3a962a3da3057c022b08edaaf96542ee3 100644 (file)
--- a/Robots/bkmk_rrequests.py
+++ b/Robots/bkmk_rrequests.py
@@ -11,10 +11,12 @@ __license__ = "GNU GPL"
 __all__ = ['robot_requests']
 
 
+from urllib.parse import urlsplit
+
 import requests
 import requests_ftp
 
-from Robots.bkmk_robot_base import robot_base
+from Robots.bkmk_robot_base import robot_base, request_headers
 
 requests_ftp.monkeypatch_session()
 
@@ -24,16 +26,42 @@ class robot_requests(robot_base):
     # BKMK_ROBOT=requests:proxy=socks5h%3a//localhost%3a1080
     proxy = None
 
+    # Store hosts for which we already know they require proxy...
+    proxy_ok = set()
+    # ...but aren't accessible even through proxy
+    proxy_error = set()
+
     def get(self, bookmark, url, accept_charset=False):
-        error, r = request_get(url, self.timeout, None)
-        if error is not None:
-            self.log('   Error: %s' % error)
-            if self.proxy:
-                self.log('   Retrying with the proxy...')
-                error, r = request_get(url, self.timeout, self.proxy)
+        split_results = urlsplit(url)
+        url_host = split_results.hostname
+
+        if url_host in self.proxy_error:
+            return 'see prev. error', None, None, None, None
+
+        if accept_charset and bookmark.charset:
+            headers = request_headers.copy()
+            headers['Accept-Charset'] = bookmark.charset
+        else:
+            headers = request_headers
+
+        if url_host in self.proxy_ok:
+            self.log('   Immediately trying with the proxy')
+            error, r = request_get(url, headers, self.timeout, self.proxy)
+        else:
+            error, r = request_get(url, headers, self.timeout, None)
+            if error is not None:
+                self.log('   Error          : %s' % error)
+                if self.proxy and error != '404 not_found':
+                    self.log('   Retrying with the proxy...')
+                    error, r = request_get(url, headers,
+                                           self.timeout, self.proxy)
+                    if error is None:
+                        self.proxy_ok.add(url_host)
         if error is not None:
-            if self.proxy:
-                self.log('   Proxy error: %s' % error)
+            if self.proxy and error != '404 not_found':
+                self.log('   Proxy error    : %s' % error)
+                if url_host not in self.proxy_ok:
+                    self.proxy_error.add(url_host)
             return error, None, None, None, None
         if r.is_redirect:
             return None, r.status_code, r.next.url, None, None
@@ -43,7 +71,7 @@ class robot_requests(robot_base):
         return ''  # Alas, requests_ftp doesn't store welcome message
 
 
-def request_get(url, timeout, proxy):
+def request_get(url, headers, timeout, proxy):
     if proxy:
         proxies = {'http': proxy, 'https': proxy}
     else:
@@ -51,8 +79,12 @@ def request_get(url, timeout, proxy):
 
     try:
         r = requests.Session().get(
-            url, timeout=timeout, allow_redirects=False, proxies=proxies)
+            url, headers=headers, timeout=timeout,
+            allow_redirects=False, proxies=proxies)
     except requests.RequestException as e:
         return str(e), None
     else:
+        if r.status_code >= 400:
+            error = requests.status_codes._codes[r.status_code][0]
+            return '%d %s' % (r.status_code, error), None
         return None, r