From 277fb4ebe62242812df47ecc6721fbc7c393ddf8 Mon Sep 17 00:00:00 2001 From: Corentin Chary Date: Sat, 8 Oct 2011 08:33:03 +0200 Subject: [PATCH] euscan: add new robots.txt exceptions Signed-off-by: Corentin Chary --- pym/euscan/__init__.py | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/pym/euscan/__init__.py b/pym/euscan/__init__.py index 987e105..2d23503 100644 --- a/pym/euscan/__init__.py +++ b/pym/euscan/__init__.py @@ -18,7 +18,7 @@ CONFIG = { 'brute-force-recursive': True, 'brute-force-false-watermark': 50, 'scan-dir': True, - 'oneshot': False, + 'oneshot': True, 'user-agent' : 'escan (http://euscan.iksaif.net)', 'skip-robots-txt' : False, 'cache' : False @@ -60,4 +60,6 @@ ROBOTS_TXT_BLACKLIST_DOMAINS = [ '(.*)sourceforge(.*)', '(.*)github.com', '(.*)berlios(.*)', + '(.*)qt.nokia.com(.*)', + '(.*)chromium.org(.*)', ]