Use simpler regular expression to avoid Tcl regsub error (Tcl8.0.4-5).

[tclrobot.git] / robot.tcl

diff --git a/robot.tcl b/robot.tcl
index 5210452..2e175d6 100755 (executable)
--- a/robot.tcl
+++ b/robot.tcl
@@ -1,5 +1,5 @@
 #!/usr/bin/tclsh 
-# $Id: robot.tcl,v 1.23 2001/10/31 08:51:49 adam Exp $
+# $Id: robot.tcl,v 1.25 2001/11/07 11:50:07 adam Exp $
 #
 proc RobotFileNext1 {area lead} {
     # puts "RobotFileNext1 area=$area lead=$lead"
@@ -441,8 +441,9 @@ proc RobotTextHtml {url out} {
             }
            puts $out {></meta>}
        } body {
-           regsub -all -nocase {<script([^<]|(<!.*>))*</script>} $body {} abody
-           regsub -all {<[^\>]+>} $abody {} nbody
+           regsub -all {<!--[^-]*->} $body { } abody
+           regsub -all -nocase {<script[^<]*</script>} $abody {} bbody
+           regsub -all {<[^\>]+>} $bbody {} nbody
            puts $out "<documentcontent>"
             puts $out $nbody
             puts $out "</documentcontent>"
@@ -647,12 +648,6 @@ proc Robot200 {url} {
     RobotWriteMetadata $url $out
     RobotFileClose $out
 
-    if {[file isdirectory flat]} {
-        regsub -all {/} $URL($url,hostport).$URL($url,path) {.} fname
-        set out [open "flat/$fname" w]
-        RobotWriteMetadata $url $out
-        close $out
-    }
     RobotFileUnlink unvisited $URL($url,hostport) $URL($url,path)
 }
 
@@ -873,17 +868,28 @@ proc checkrule {type this} {
             # consider type
             if {[lindex $l 1] != $type} continue
             # consider mask (! negates)
-            set mask [lindex $l 2]
-            if {[string index $mask 0] == "!"} {
-                set mask [string range $mask 1 end]
-                if {[string match $mask $this]} continue
-            } else {
-                if {![string match $mask $this]} continue
+            set masks [lindex $l 2]
+           set ok 0
+           foreach mask $masks {       
+                if {$debuglevel > 4} {
+                    puts "consider single mask $mask"
+                }
+                if {[string index $mask 0] == "!"} {
+                    set mask [string range $mask 1 end]
+                    if {[string match $mask $this]}  continue
+                } else {
+                    if {![string match $mask $this]} continue
+                }
+                set ok 1
+            }
+            if {$debuglevel > 4} {
+                puts "ok = $ok"
             }
+            if {!$ok} continue
             # OK, we have a match
             if {[lindex $l 0] == "allow"} {
                 if {$debuglevel > 3} {
-                    puts "CHECKRULE MATH OK"
+                    puts "CHECKRULE MATCH OK"
                 }
                 return 1
             } else {
@@ -895,7 +901,7 @@ proc checkrule {type this} {
         }
     }
     if {$debuglevel > 3} {
-        puts "CHECKRULE MATH OK"
+        puts "CHECKRULE MATCH OK"
     }
     return 1
 }