With the current Nutch trunk, the scenario you described gets reproduced with multiple agents but not with single agent:
$ cat robots.txt User-agent: * Disallow: / User-agent: nutch Allow: / $ bin/nutch org.apache.nutch.protocol.RobotRulesParser robots.txt urls.txt nutch,* not allowed: http://temp.com/index.html $ bin/nutch org.apache.nutch.protocol.RobotRulesParser robots.txt urls.txt nutch allowed: http://temp.com/index.html You need not add additional "*" in the http.robots.agents string as CC would internally add it for you and smartly do a match with (User-agent: *) when no agent names are matching with the rules from robots.txt Multiple agents without any "*" -------------------------------------------- As CC is greedy, it matched "nutch" and allowed in the first two examples below. In the last case, it correctly did mismatch. $ bin/nutch org.apache.nutch.protocol.RobotRulesParser robots.txt urls.txt foobar,nutch allowed: http://temp.com/index.html $ bin/nutch org.apache.nutch.protocol.RobotRulesParser robots.txt urls.txt nutch,foobar allowed: http://temp.com/index.html $ bin/nutch org.apache.nutch.protocol.RobotRulesParser robots.txt urls.txt foobar not allowed: http://temp.com/index.html Multiple agents with "*" ----------------------------------- Bad result as it would match with "*" as soon as it scans the first rule in the robots file. $ bin/nutch org.apache.nutch.protocol.RobotRulesParser robots.txt urls.txt foobar,nutch,* not allowed: http://temp.com/index.html $ bin/nutch org.apache.nutch.protocol.RobotRulesParser robots.txt urls.txt nutch,foobar,* not allowed: http://temp.com/index.html Thanks, Tejas On Fri, Jan 24, 2014 at 8:27 PM, Markus Jelsma <[email protected]>wrote: > > -----Original message----- > > From:Tejas Patil <[email protected]> > > Sent: Friday 24th January 2014 15:49 > > To: [email protected] > > Subject: Re: Order of robots file > > > > I am working on the scenario you just pointed out. By Apache Nutch, you > > mean the current codebase with CC or version before that ? > > No, this is current trunk. > > > > > CC differs from original nutch code as CC has kinda a greedy approach > > wherein it tries to get a match / mismatch after every line is sees from > > the robots file. While the time I was working on delegation of robots > > parsing to Crawler commons (CC), I remember that there was difference in > > the semantics of original parsing code and CC's implementation for > multiple > > robots agents. > > Here was my observation at that time: > > > https://issues.apache.org/jira/browse/NUTCH-1031?focusedCommentId=13558217&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13558217 > > The problem i describe persists with one and multiple robot agent names. > So it is also a problem with: > <property> > <name>http.robots.agents</name> > <value>nutch</value> > </property> > > > > > > ~tejas > > > > > > On Fri, Jan 24, 2014 at 8:11 PM, Markus Jelsma > > <[email protected]>wrote: > > > > > Tejas, the problem exists in Apache Nutch as well. We'll take > localhost as > > > example and the following config and robots.txt > > > > > > > > > # cat /var/www/robots.txt > > > User-agent: * > > > Disallow: / > > > > > > User-agent: nutch > > > Allow: / > > > > > > > > > > > > config: > > > <property> > > > <name>http.agent.name</name> > > > <value>Mozilla</value> > > > </property> > > > <property> > > > <name>http.agent.version</name> > > > <value>5.0</value> > > > </property> > > > <property> > > > <name>http.robots.agents</name> > > > <value>nutch,*</value> > > > </property> > > > <property> > > > <name>http.agent.description</name> > > > <value>compatible; NutchCrawler</value> > > > </property> > > > <property> > > > <name>http.agent.url</name> > > > <value>+http://example.org/</value> > > > </property> > > > > > > > > > URL: http://localhost/ > > > Version: 7 > > > Status: 3 (db_gone) > > > Fetch time: Mon Mar 10 15:36:48 CET 2014 > > > Modified time: Thu Jan 01 01:00:00 CET 1970 > > > Retries since fetch: 0 > > > Retry interval: 3888000 seconds (45 days) > > > Score: 0.0 > > > Signature: null > > > Metadata: > > > _pst_=robots_denied(18), lastModified=0 > > > > > > > > > > > > Can you confirm? > > > > > > > > > > > > > > > > > > -----Original message----- > > > > From:Markus Jelsma <[email protected]> > > > > Sent: Friday 24th January 2014 15:29 > > > > To: [email protected] > > > > Subject: RE: Order of robots file > > > > > > > > Hi, sorry for being unclear. You understand correctly, i had to > change > > > the robots.txt order and put our crawler ABOVE User-Agent: *. > > > > > > > > I have tried a unit test for lib-http to demonstrate the problem but > i > > > does not fail. I think i did something wrong in merging the code base. > I'll > > > look further. > > > > > > > > markus@midas:~/projects/apache/nutch/trunk$ svn diff > > > > src/plugin/lib-http/src/test/org/apache/nutch/protocol/http/api/TestRobotRulesParser.java > > > > Index: > > > > src/plugin/lib-http/src/test/org/apache/nutch/protocol/http/api/TestRobotRulesParser.java > > > > =================================================================== > > > > --- > > > > src/plugin/lib-http/src/test/org/apache/nutch/protocol/http/api/TestRobotRulesParser.java > > > (revision 1560984) > > > > +++ > > > > src/plugin/lib-http/src/test/org/apache/nutch/protocol/http/api/TestRobotRulesParser.java > > > (working copy) > > > > @@ -50,6 +50,23 @@ > > > > + "" + CR > > > > + "User-Agent: *" + CR > > > > + "Disallow: /foo/bar/" + CR; // no crawl delay for other > agents > > > > + > > > > + private static final String ROBOTS_STRING_REVERSE = > > > > + "User-Agent: *" + CR > > > > + + "Disallow: /foo/bar/" + CR // no crawl delay for other > agents > > > > + + "" + CR > > > > + + "User-Agent: Agent1 #foo" + CR > > > > + + "Disallow: /a" + CR > > > > + + "Disallow: /b/a" + CR > > > > + + "#Disallow: /c" + CR > > > > + + "Crawl-delay: 10" + CR // set crawl delay for Agent1 as 10 > sec > > > > + + "" + CR > > > > + + "" + CR > > > > + + "User-Agent: Agent2" + CR > > > > + + "Disallow: /a/bloh" + CR > > > > + + "Disallow: /c" + CR > > > > + + "Disallow: /foo" + CR > > > > + + "Crawl-delay: 20" + CR; > > > > > > > > private static final String[] TEST_PATHS = new String[] { > > > > "http://example.com/a";, > > > > @@ -80,6 +97,29 @@ > > > > /** > > > > * Test that the robots rules are interpreted correctly by the > robots > > > rules parser. > > > > */ > > > > + public void testRobotsAgentReverse() { > > > > + rules = parser.parseRules("testRobotsAgent", > > > ROBOTS_STRING_REVERSE.getBytes(), CONTENT_TYPE, SINGLE_AGENT); > > > > + > > > > + for(int counter = 0; counter < TEST_PATHS.length; counter++) { > > > > + assertTrue("testing on agent (" + SINGLE_AGENT + "), and " > > > > + + "path " + TEST_PATHS[counter] > > > > + + " got " + rules.isAllowed(TEST_PATHS[counter]), > > > > + rules.isAllowed(TEST_PATHS[counter]) == > RESULTS[counter]); > > > > + } > > > > + > > > > + rules = parser.parseRules("testRobotsAgent", > > > ROBOTS_STRING_REVERSE.getBytes(), CONTENT_TYPE, MULTIPLE_AGENTS); > > > > + > > > > + for(int counter = 0; counter < TEST_PATHS.length; counter++) { > > > > + assertTrue("testing on agents (" + MULTIPLE_AGENTS + "), and " > > > > + + "path " + TEST_PATHS[counter] > > > > + + " got " + rules.isAllowed(TEST_PATHS[counter]), > > > > + rules.isAllowed(TEST_PATHS[counter]) == > RESULTS[counter]); > > > > + } > > > > + } > > > > + > > > > + /** > > > > + * Test that the robots rules are interpreted correctly by the > robots > > > rules parser. > > > > + */ > > > > public void testRobotsAgent() { > > > > rules = parser.parseRules("testRobotsAgent", > > > ROBOTS_STRING.getBytes(), CONTENT_TYPE, SINGLE_AGENT); > > > > > > > > > > > > -----Original message----- > > > > > From:Tejas Patil <[email protected]> > > > > > Sent: Friday 24th January 2014 15:24 > > > > > To: [email protected] > > > > > Subject: Re: Order of robots file > > > > > > > > > > Hi Markus, > > > > > I am trying to understand the problem you described. You meant that > > > with > > > > > the original Nutch's robots parsing code, the robots file below > allowed > > > > > your crawler to crawl stuff: > > > > > > > > > > User-agent: * > > > > > Disallow: / > > > > > > > > > > User-agent: our_crawler > > > > > Allow: / > > > > > > > > > > But now that started using the change from NUTCH-1031 [0], (ie. > > > delegation > > > > > of robots parsing to crawler commons), it blocked your crawler. To > make > > > > > things work, you had to change your robots file to this: > > > > > > > > > > User-agent: our_crawler > > > > > Allow: / > > > > > > > > > > User-agent: * > > > > > Disallow: / > > > > > > > > > > Did I understand the problem correctly ? > > > > > > > > > > [0] : https://issues.apache.org/jira/browse/NUTCH-1031 > > > > > > > > > > Thanks, > > > > > Tejas > > > > > > > > > > > > > > > On Fri, Jan 24, 2014 at 7:29 PM, Markus Jelsma > > > > > <[email protected]>wrote: > > > > > > > > > > > Hi, > > > > > > > > > > > > I am attempting to merge some Nutch changes back to our own. We > > > aren't > > > > > > using Nutch' CrawlerCommons impl but the old stuff. But because > of > > > > > > recording of response time and rudimentary SSL support i decided > to > > > move it > > > > > > back to our version. Suddenly i realized a local crawl does not > work > > > > > > anymore, it seems because of the order of the robots definitions. > > > > > > > > > > > > For example: > > > > > > > > > > > > User-agent: * > > > > > > Disallow: / > > > > > > > > > > > > User-agent: our_crawler > > > > > > Allow: / > > > > > > > > > > > > Does not allow our crawler to fetch URL's. But > > > > > > > > > > > > User-agent: our_crawler > > > > > > Allow: / > > > > > > > > > > > > User-agent: * > > > > > > Disallow: / > > > > > > > > > > > > Does! This was not the case before, anyone here aware of this? By > > > design? > > > > > > Or is it a flaw? > > > > > > > > > > > > Thanks > > > > > > Markus > > > > > > > > > > > > > > > > > > > > >
