Fix check for content-type

diff --git a/dcdot.tcl b/dcdot.tcl
index 3b2edcb..8e15fea 100755 (executable)
--- a/dcdot.tcl
+++ b/dcdot.tcl
@@ -1,5 +1,5 @@
 #!/usr/bin/tclsh 
-# $Id: dcdot.tcl,v 1.4 2000/12/11 17:11:03 adam Exp $
+# $Id: dcdot.tcl,v 1.5 2003/01/13 13:59:07 adam Exp $
 #
 
 proc RobotRestart {} {
@@ -10,15 +10,23 @@ proc RobotRestart {} {
 
 proc RobotTextHtml {url} {
     global URL
-    
+   
     set b $URL($url,buf)
-    set e {<[mM][eE][tT][aA][^>]*>}
+    set e {<meta[^>]*>}
     catch {unset $URL($url,meta)}
-    while {[regexp -indices $e $b i]} {
+    while {[regexp -nocase -indices $e $b i]} {
         set meta [string range $b [lindex $i 0] [lindex $i 1]]
        lappend URL($url,meta) $meta
         set b [string range $b [lindex $i 1] end]
     }
+    set b $URL($url,buf)
+    set e {<title>[^>]*>}
+    catch {unset $URL($url,meta)}
+    while {[regexp -nocase -indices $e $b i]} {
+        set title [string range $b [lindex $i 0] [lindex $i 1]]
+       lappend URL($url,title) $title
+        set b [string range $b [lindex $i 1] end]
+    }
 }
 
 proc Robot200 {url} {
@@ -74,11 +82,12 @@ proc RobotReadHeader {url sock} {
            regexp {^HTTP/([0-9.]+)[ ]+([0-9]+)} $headbuf x version code
            set lines [split $headbuf \n]
            foreach line $lines {
-               if {[regexp {^([^:]+):[ ]+(.*)} $line x name value]} {
+               if {[regexp {^([^:]+):[ ]+([^;]*)} $line x name value]} {
                    set URL($url,head,[string tolower $name]) [string trim $value]
                }
            }
            set URL($url,state) skip
+            puts "code=$code"  
            switch $code {
                200 {
                    if {![info exists URL($url,head,content-type)]} {
@@ -173,6 +182,11 @@ if {$argc == 1} {
            puts $m
         }
     }
+    if {[info exist URL($url,title)]} {
+        foreach m $URL($url,title) {
+           puts $m
+        }
+    }
     foreach v [array names URL $url,head,*] {
        puts "$v = $URL($v)"
     }