From: Adam Dickmeiss <adam@indexdata.dk>
Date: Thu, 7 Jun 2001 08:17:00 +0000 (+0000)
Subject: Fixes for robots.txt handling (bug introduced by previous commit).
X-Git-Tag: ZMBOT.0.1~20
X-Git-Url: http://git.indexdata.com/?p=tclrobot.git;a=commitdiff_plain;h=9b7fbb227055fc0ad4eeb10d482fd9e8ada7ce4e

Fixes for robots.txt handling (bug introduced by previous commit).
---

diff --git a/robot.tcl b/robot.tcl
index fa3c595..b8db6c4 100755
--- a/robot.tcl
+++ b/robot.tcl
@@ -1,5 +1,5 @@
 #!/usr/bin/tclsh 
-# $Id: robot.tcl,v 1.17 2001/06/07 08:10:10 adam Exp $
+# $Id: robot.tcl,v 1.18 2001/06/07 08:17:00 adam Exp $
 #
 proc RobotFileNext1 {area lead} {
     # puts "RobotFileNext1 area=$area lead=$lead"
@@ -550,7 +550,7 @@ proc RobotsTxt0 {v buf} {
     global URL agent
     set section 0
     foreach l [split $buf \n] {
-	if {[regexp {([-A-Za-z]+):[ \t]*([^\#\t ]+)} $l match cmd arg]} {
+	if {[regexp {([-A-Za-z]+):[ ]*([^\# ]+)} $l match cmd arg]} {
 	    puts "cmd=$cmd arg=$arg"
 	    switch -- [string tolower $cmd] {
 		user-agent {
@@ -579,7 +579,8 @@ proc RobotTextPlain {url out} {
     global URL
 
     puts $out "<documentcontent>"
-    puts $out $URL($url,buf)
+    regsub -all {<} $URL($url,buf) {\&lt;} content
+    puts $out $content
     puts $out "</documentcontent>"
 
     if {![string compare $URL($url,path) /robots.txt]} {
@@ -616,7 +617,7 @@ proc Robot200 {url} {
 	    }
 	}
 	text/plain {
-	    RobotTextPlain $url $out $outr
+	    RobotTextPlain $url $out
 	}
 	application/pdf {
 	    set pdff [open test.pdf w]