Fix(Robots/bkmk_rrequests): No need to re-check error 404 via proxy

[bookmarks_db.git] / Robots / bkmk_rrequests.py

diff --git a/Robots/bkmk_rrequests.py b/Robots/bkmk_rrequests.py
index 114d67094586ed97d928768803aed3704f075790..4612a8d5384dd2b03a31145132066be3ce33bf77 100644 (file)
--- a/Robots/bkmk_rrequests.py
+++ b/Robots/bkmk_rrequests.py
@@ -16,7 +16,7 @@ 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()
 
@@ -36,22 +36,29 @@ class robot_requests(robot_base):
         url_host = split_results.hostname
 
         if url_host in self.proxy_error:
-            return 'proxy error', None, None, None, None
+            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, self.timeout, self.proxy)
+            error, r = request_get(url, headers, self.timeout, self.proxy)
         else:
-            error, r = request_get(url, self.timeout, None)
+            error, r = request_get(url, headers, self.timeout, None)
             if error is not None:
                 self.log('   Error: %s' % error)
-                if self.proxy:
+                if self.proxy and error != '404 not_found':
                     self.log('   Retrying with the proxy...')
-                    error, r = request_get(url, self.timeout, self.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:
+            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)
@@ -64,7 +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:
@@ -72,8 +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