X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=Robots%2Fbkmk_rurllib_py3.py;h=268dad43cbb3444b910c8428d436603282503b49;hb=fbb34ad74643bbc69d141e26e105bcef3cdbc62c;hp=796a6fca3425bfb8d91d035968d02be246605bb3;hpb=5e3685956769a46ded587d33d5cb69ca4d0bbe7a;p=bookmarks_db.git

diff --git a/Robots/bkmk_rurllib_py3.py b/Robots/bkmk_rurllib_py3.py
index 796a6fc..268dad4 100644
--- a/Robots/bkmk_rurllib_py3.py
+++ b/Robots/bkmk_rurllib_py3.py
@@ -5,17 +5,25 @@ This file is a part of Bookmarks database and Internet robot.
 """
 
 __author__ = "Oleg Broytman <phd@phdru.name>"
-__copyright__ = "Copyright (C) 2000-2023 PhiloSoft Design"
+__copyright__ = "Copyright (C) 2000-2024 PhiloSoft Design"
 __license__ = "GNU GPL"
 
 __all__ = ['robot_urllib_py3']
 
 
+import http.client
+import socket
 import sys
+import urllib
 import urllib.request
 
 from Robots.bkmk_robot_base import robot_base, get_error
 
+# Fake to import 'add_headers'
+urllib.URLopener = urllib.request.URLopener
+urllib.ftpwrapper = urllib.request.ftpwrapper
+from Robots.bkmk_rurllib import add_headers  # noqa: E402 import not at top
+
 
 class RedirectException(Exception):
     def __init__(self, errcode, newurl):
@@ -41,6 +49,8 @@ class MyURLopener(urllib.request.URLopener):
     http_error_303 = http_error_301
     # Error 307 -- relocated (temporarily)
     http_error_307 = http_error_301
+    # Error 308 -- relocated (permanently)
+    http_error_308 = http_error_301
 
     # Error 401 -- authentication required
     def http_error_401(self, url, fp, errcode, errmsg, headers, data=None):
@@ -53,25 +63,12 @@ class MyURLopener(urllib.request.URLopener):
             fp.close()
         raise IOError(('http error', errcode, errmsg, headers))
 
+    def open(self, fullurl, data=None, timeout=socket._GLOBAL_DEFAULT_TIMEOUT):
+        return urllib.request.URLopener.open(self, fullurl, data)
 
-urllib.request._opener = opener = MyURLopener()
-
-# Fake headers to pretend this is a real browser
-_user_agent = "Mozilla/5.0 (X11; U; Linux 2.6 i686; en)"
-" Gecko/20001221 Firefox/2.0.0"
-opener.addheaders[0] = ('User-Agent', _user_agent)
-_x_user_agent = "bookmarks_db (Python %d.%d.%d; urllib/%s)" % (
-   sys.version_info[0], sys.version_info[1],
-   sys.version_info[2], urllib.request.__version__
-)
-opener.addheader('X-User-Agent', _x_user_agent)
-opener.addheader('Referer', '')
-
-opener.addheader('Accept', '*/*')
-opener.addheader('Accept-Language', 'ru,en')
-opener.addheader('Cache-Control', 'max-age=300')
-opener.addheader('Connection', 'close')
 
+urllib.request._opener = opener = MyURLopener()
+add_headers(opener)
 
 urllib_ftpwrapper = urllib.request.ftpwrapper
 ftpcache_key = None
@@ -102,15 +99,39 @@ class robot_urllib_py3(robot_base):
                     # Remove Accept-Charset
                     del opener.addheaders[-1]
 
-            infile = open(fname, 'rt')
-            content = infile.read()
-            infile.close()
-
+            possible_encodings = []
+            for encoding in (
+                    bookmark.charset,
+                    sys.getfilesystemencoding(),
+                    'utf-8',
+            ):
+                if encoding and encoding not in possible_encodings:
+                    possible_encodings.append(encoding)
+            content = e = None
+            infile = open(fname, 'rb')
+            try:
+                content = infile.read()
+            except Exception:
+                content = None
+            finally:
+                infile.close()
+
+            if content is None:
+                e = str(e)
+                return (
+                    'ERROR: ' + e,
+                    None, None, None, None
+                )
             return None, None, None, headers, content
 
         except RedirectException as e:
             return None, e.errcode, e.newurl, None, None
 
+        except (OSError, http.client.IncompleteRead) as e:
+            error = str(e)
+            self.log('   Error: %s' % error)
+            return error, None, None, None, None
+
         except IOError as e:
             if (e[0] == "http error") and (e[1] == -1):
                 error = None