Added support to replay update to crawl again all the pages downloaded in the replay file.

--HG--
extra : convert_revision : svn%3Ab85faa78-f9eb-468e-a121-7cced6da292c%40169

scrapy/trunk/scrapy/command/cmdline.py

@@ -42,14 +42,14 @@ def custom_commands_dict():
                 print 'WARNING: Custom command module %s exists but Command class not found' % modname
     return d
 
-def getcmdname():
-    for arg in sys.argv[1:]:
+def getcmdname(argv):
+    for arg in argv[1:]:
         if not arg.startswith('-'):
             return arg
 
-def usage():
-    s  = "usage: %s <subcommand> [options] [args]\n" % sys.argv[0]
-    s += "       %s <subcommand> -h\n\n" % sys.argv[0]
+def usage(argv):
+    s  = "usage: %s <subcommand> [options] [args]\n" % argv[0]
+    s += "       %s <subcommand> -h\n\n" % argv[0]
     s += "Built-in subcommands:\n"
 
     builtin_cmds = builtin_commands_dict()
@@ -96,16 +96,19 @@ def command_settings(cmdname):
 command_executed = {}
 
 def execute():
+    execute_with_args(sys.argv)
+
+def execute_with_args(argv):
     spiders.load()
     cmds = builtin_commands_dict()
     cmds.update(custom_commands_dict())
 
-    cmdname = getcmdname()
+    cmdname = getcmdname(argv)
     command_settings(cmdname)
 
     if not cmdname:
         print "Scrapy %s\n" % scrapy.__version__
-        print usage()
+        print usage(argv)
         sys.exit()
 
     parser = optparse.OptionParser()
@@ -118,10 +121,10 @@ def execute():
     else:
         print "Scrapy %s\n" % scrapy.__version__
         print "Unknown command: %s\n" % cmdname
-        print 'Type "%s -h" for help' % sys.argv[0]
+        print 'Type "%s -h" for help' % argv[0]
         sys.exit()
 
-    (opts, args) = parser.parse_args()
+    (opts, args) = parser.parse_args(args=argv[1:])
     del args[0]  # args[0] is cmdname
 
     # storing command executed info for later reference

scrapy/trunk/scrapy/command/commands/replay.py

@@ -4,6 +4,7 @@ from scrapy.command import ScrapyCommand
 from scrapy.replay import Replay
 from scrapy.utils import display
 from scrapy.conf import settings
+from scrapy.command import cmdline
 
 class Command(ScrapyCommand):
     def syntax(self):
@@ -31,6 +32,8 @@ class Command(ScrapyCommand):
         parser.add_option("--nocolour", dest="nocolour", action="store_true", help="disable colorized output (for console only)")
         parser.add_option("-i", "--ignore", dest="ignores", action="append", help="item attribute to ignore. can be passed multiple times", metavar="ATTR")
         parser.add_option("--target", dest="targets", action="append", help="crawl TARGET instead of recorded urls/domains. can be passed multiple times")
+        # adding option to update 
+        parser.add_option("--pages", dest="pages", action="store_true", help="update all the pages in the replay file, recording it again.")
 
     def process_options(self, args, opts):
         ScrapyCommand.process_options(self, args, opts)
@@ -75,7 +78,16 @@ class Command(ScrapyCommand):
         self.replay.play(args=opts.targets)
 
     def action_update(self, opts):
-        self.action_crawl(opts)
+        self.replay.update(args=opts.targets, opts=opts.__dict__)
+        if (opts.pages):
+            args = ['decobot-ctl.py', 'crawl']
+            args.extend(self.replay.options['args'])
+            for k in self.replay.options['opts']:                
+                if self.replay.options['opts'][k]:                   
+                    args.append("--%s" % k)
+                    if self.replay.options['opts'][k] != True:
+                        args.append(self.replay.options['opts'][k])
+            cmdline.execute_with_args(args)
 
     def action_showitems(self, opts):
         s = ""

scrapy/trunk/scrapy/replay/__init__.py

@@ -68,6 +68,7 @@ class Replay(object):
         scrapymanager.runonce(*args, **opts)
 
     def record(self, args=None, opts=None):
+        self.recording = True
         self.options.clear()
         self.options['args'] = args or []
         self.options['opts'] = opts or {}
@@ -79,8 +80,13 @@ class Replay(object):
 
     def update(self, args=None, opts=None):
         self.updating = True
-
-        self.play(args, opts)
+        if (opts['pages']):
+            args = self.options['args']
+            opts = self.options['opts']
+            settings.overrides['CACHE2_EXPIRATION_SECS'] = 0            
+            settings.overrides['CACHE2_IGNORE_MISSING'] = False
+        else:            
+            self.play(args, opts)
 
     def engine_started(self):
         log.msg("Replay: recording session in %s" % self.repfile)