X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=check_urls.py;h=d6d038d2c624a116c22b60f772260095661909a1;hb=a04eaa0346e8aa5ad86a195f8f4d36487ebfe09c;hp=ab1985ba01b95d4fa45c5b84e484ceed61dfd3c9;hpb=a99768db3986f2b125c5b47928d999b38bd22d62;p=bookmarks_db.git

diff --git a/check_urls.py b/check_urls.py
index ab1985b..d6d038d 100755
--- a/check_urls.py
+++ b/check_urls.py
@@ -1,13 +1,16 @@
 #! /usr/bin/env python
-"""
-   Robot interface - check URLs from bookmarks database
+"""Robot interface - check URLs from bookmarks database
 
-   Written by BroytMann. Copyright (C) 2000-2007 PhiloSoft Design.
+This file is a part of Bookmarks database and Internet robot.
 """
 
-
+from __future__ import print_function
 import sys, os
-tempfname = "check_urls.tmp"
+
+
+__author__ = "Oleg Broytman <phd@phdru.name>"
+__copyright__ = "Copyright (C) 2000-2017 PhiloSoft Design"
+__license__ = "GNU GPL"
 
 
 def run():
@@ -31,7 +34,7 @@ def run():
       pass
 
    if report_stats:
-      print "BroytMann check_urls, Copyright (C) 1997-2007 PhiloSoft Design"
+      print("Broytman check_urls, Copyright (C) 1997-2017 PhiloSoft Design")
 
    if args:
       sys.stderr.write("check_urls: too many arguments\n")
@@ -52,18 +55,18 @@ def run():
       log = openlog("check.log")
       log("chk_urls restarted for errors")
       if report_stats:
-         print "chk_urls restarted for errors"
+         print("chk_urls restarted for errors")
    else:
       log = makelog("check.log")
       log("check_urls started")
       if report_stats:
-         print "   check_urls: normal start"
+         print("   check_urls: normal start")
 
    from storage import storage
    storage = storage()
 
    from robots import robot
-   robot = robot(tempfname, log)
+   robot = robot(log)
 
    if report_stats:
       sys.stdout.write("Loading %s: " % storage.filename)
@@ -75,7 +78,7 @@ def run():
    objects = len(root_folder.linear)
 
    if report_stats:
-      print "Ok"
+      print("Ok")
 
    if report_stats:
       if only_errors:
@@ -103,26 +106,32 @@ def run():
       object_count = object_count + 1
 
       if object.isBookmark:
+         href = object.href
+         if (href.startswith('place:') # Firefox SmartBookmarks
+               or '%s' in href): # Bookmark with keyword
+            log("Skipped %s" % href)
+            continue
+
          if only_errors:
             if hasattr(object, "error"):
                delattr(object, "error")
             else:
                continue
 
-         if checked.has_key(object.href):
-            log("Already checked %s" % object.href)
-            old_object = root_folder.linear[checked[object.href]]
+         if checked.has_key(href):
+            log("Already checked %s" % href)
+            old_object = root_folder.linear[checked[href]]
             for attr_name in ("last_visit", "last_modified",
                   "error", "no_error", "moved", "size", "md5", "real_title",
                   "last_tested", "test_time", "icon", "charset"):
                if hasattr(old_object, attr_name):
                   setattr(object, attr_name, getattr(old_object, attr_name))
          else:
-            log("Checking %s" % object.href)
+            log("Checking %s" % href)
             rcode = robot.check_url(object)
 
             if rcode:
-               checked[object.href] = object_no
+               checked[href] = object_no
                urls_no = urls_no + 1
                try:
                   size = size + int(object.size)
@@ -137,10 +146,10 @@ def run():
       del pbar
 
    if report_stats:
-      print "Ok"
-      print object_count, "objects passed"
-      print urls_no, "URLs checked"
-      print size, "bytes eaten"
+      print("Ok")
+      print(object_count, "objects passed")
+      print(urls_no, "URLs checked")
+      print(size, "bytes eaten")
 
    break_tree(root_folder.linear)
    storage.store(root_folder)
@@ -149,11 +158,6 @@ def run():
       log("check_urls finished ok")
    log.close()
 
-   try:
-      os.unlink(tempfname)
-   except os.error:
-      pass
-
 
 if __name__ == '__main__':
    run()