Updated version to 0.2.1.

[tclrobot.git] / dcdot.tcl

diff --git a/dcdot.tcl b/dcdot.tcl
index af0bef7..8e15fea 100755 (executable)
--- a/dcdot.tcl
+++ b/dcdot.tcl
@@ -1,5 +1,5 @@
 #!/usr/bin/tclsh 
-# $Id: dcdot.tcl,v 1.2 2000/12/08 08:55:35 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} {
@@ -62,23 +70,24 @@ proc RobotReadHeader {url sock} {
        # puts "Got $readCount bytes"
        set URL($url,buf) $URL($url,buf)$buffer
        
-       set n [string first \n\n $URL($url,buf)]
+       set n [string first \r\n\r\n $URL($url,buf)]
        if {$n > 1} {
+           puts "string first match n = $n"
            set code 0
            set version {}
            set headbuf [string range $URL($url,buf) 0 $n]
-           incr n
-           incr n
+           incr n 4
            set URL($url,buf) [string range $URL($url,buf) $n end]
            
            regexp {^HTTP/([0-9.]+)[ ]+([0-9]+)} $headbuf x version code
            set lines [split $headbuf \n]
            foreach line $lines {
-               if {[regexp {^([^:]+):[ ]+(.*)} $line x name value]} {
-                   set URL($url,head,[string tolower $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)]} {
@@ -91,6 +100,10 @@ proc RobotReadHeader {url sock} {
                        text/plain {
                            fileevent $sock readable [list RobotReadContent $url $sock]
                        }
+                       application/pdf {
+                           puts "ok preceeed with this thingy"
+                           fileevent $sock readable [list RobotReadContent $url $sock]
+                       }
                        default {
                            close $sock
                            Robot200 $url
@@ -99,7 +112,6 @@ proc RobotReadHeader {url sock} {
                    }
                }
                default {
-                   Robot404 $url
                    close $sock
                    RobotRestart
                }
@@ -111,7 +123,7 @@ proc RobotReadHeader {url sock} {
 proc RobotConnect {url sock} {
     global URL agent
 
-    fconfigure $sock -translation {auto crlf} -blocking 0
+    fconfigure $sock -translation {lf crlf} -blocking 0
     fileevent $sock readable [list RobotReadHeader $url $sock]
     puts $sock "GET $URL($url,path) HTTP/1.0"
     puts $sock "Host: $URL($url,host)"
@@ -170,4 +182,16 @@ 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)"
+    }
+    puts "Buffer length is [string length $URL($url,buf)]"
+    set f [open out.pdf w]
+    puts -nonewline $f $URL($url,buf)
+    close $f
 }