Allow maximum check time specification.

config/linkcheckerrc

@@ -174,7 +174,9 @@
 # that are already expired.
 # The default number of days is 14.
 #sslcertwarndays=14
-
+# Stop checking new URLs after the given number of seconds. Same as if the
+# user hits Ctrl-C after X seconds.
+#maxrunseconds=600
 
 ##################### filtering options ##########################
 [filtering]

doc/changelog.txt

@@ -1,5 +1,8 @@
 8.1 "" (release xx.xx.2012)
 
+Features:
+- checking: Allow specification of maximum checking time or maximum
+  number of checked URLs.
 
 
 8.0 "Luminaris" (released 2.9.2012)

doc/t.t

@@ -1,3 +0,0 @@
-Features:
-- checking: Allow specification of maximum checking time or maximum
-  number of checked URLs.

linkcheck/configuration/__init__.py

@@ -222,6 +222,7 @@ class Configuration (dict):
         self["debugmemory"] = False
         self["localwebroot"] = None
         self["warnsslcertdaysvalid"] = 14
+        self["maxrunseconds"] = None
         from ..logger import Loggers
         self.loggers = dict(**Loggers)
 

linkcheck/configuration/confparse.py

@@ -153,6 +153,7 @@ class LCConfigParser (ConfigParser.RawConfigParser, object):
         self.read_string_option(section, "cookiefile")
         self.read_string_option(section, "localwebroot")
         self.read_int_option(section, "warnsslcertdaysvalid")
+        self.read_int_option(section, "maxrunseconds")
 
     def read_authentication_config (self):
         """Read configuration options in section "authentication"."""

linkcheck/director/aggregator.py

@@ -47,7 +47,8 @@ class Aggregate (object):
         """Spawn threads for URL checking and status printing."""
         if self.config["status"]:
             t = status.Status(self.urlqueue, self.config.status_logger,
-                self.config["status_wait_seconds"])
+                self.config["status_wait_seconds"],
+                self.config["maxrunseconds"])
             t.start()
             self.threads.append(t)
         t = cleanup.Cleanup(self.connections)

linkcheck/director/status.py

@@ -1,5 +1,5 @@
 # -*- coding: iso-8859-1 -*-
-# Copyright (C) 2006-2011 Bastian Kleineidam
+# Copyright (C) 2006-2012 Bastian Kleineidam
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
@@ -22,11 +22,21 @@ from . import task
 class Status (task.LoggedCheckedTask):
     """Status thread."""
 
-    def __init__ (self, urlqueue, logger, wait_seconds):
-        """Store urlqueue object."""
+    def __init__ (self, urlqueue, logger, wait_seconds, max_duration):
+        """Initialize the status logger task.
+        @param urlqueue: the URL queue
+        @ptype urlqueue: Urlqueue
+        @param logger: the logger object to inform about status
+        @ptype logger: console.StatusLogger
+        @param wait_seconds: interval in seconds to report status
+        @ptype wait_seconds: int
+        @param max_duration: abort checking after given number of seconds
+        @ptype max_duration: int or None
+        """
         super(Status, self).__init__(logger)
         self.urlqueue = urlqueue
         self.wait_seconds = wait_seconds
+        self.max_duration = max_duration
 
     def run_checked (self):
         """Print periodic status messages."""
@@ -38,5 +48,7 @@ class Status (task.LoggedCheckedTask):
     def log_status (self):
         """Log a status message."""
         duration = time.time() - self.start_time
+        if self.max_duration is not None and duration > self.max_duration:
+            raise KeyboardInterrupt()
         checked, in_progress, queue = self.urlqueue.status()
         self.logger.log_status(checked, in_progress, queue, duration)