Updated version to 0.2.1.

[tclrobot.git] / dcdot.tcl

diff --git a/dcdot.tcl b/dcdot.tcl
index aeb4f91..8e15fea 100755 (executable)
--- a/dcdot.tcl
+++ b/dcdot.tcl
@@ -1,5 +1,5 @@
 #!/usr/bin/tclsh 
-# $Id: dcdot.tcl,v 1.1 2000/12/07 20:16:11 adam Exp $
+# $Id: dcdot.tcl,v 1.5 2003/01/13 13:59:07 adam Exp $
 #
 
 proc RobotRestart {} {
@@ -10,29 +10,23 @@ proc RobotRestart {} {
 
 proc RobotTextHtml {url} {
     global URL
-    
-    set head 0
-    htmlSwitch $URL($url,buf) \
-        title {
-           set URL($url,title) $body
-        } -nonest meta {
-           set scheme {}
-           if {[info exist parm(scheme)]} {
-               set scheme $parm(scheme)
-               unset parm(scheme)
-           }
-           if {[info exist parm(name)]} {
-               if {[info exist parm(content)]} {
-                   set URL($url,meta,$parm(name),$scheme) $parm(content)
-                   unset parm(content)
-               }
-               unset parm(name)
-           }
-        } a {
-            if {[info exists parm(href)]} {
-               lappend URL($url,links) $parm(href)
-           }
-       }
+   
+    set b $URL($url,buf)
+    set e {<meta[^>]*>}
+    catch {unset $URL($url,meta)}
+    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} {
@@ -76,23 +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)]} {
@@ -105,6 +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
@@ -113,7 +112,6 @@ proc RobotReadHeader {url sock} {
                    }
                }
                default {
-                   Robot404 $url
                    close $sock
                    RobotRestart
                }
@@ -125,7 +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)"
@@ -157,14 +155,7 @@ proc RobotGetUrl {url phost} {
     return 0
 }
 
-if {![llength [info commands htmlSwitch]]} {
-    set e [info sharedlibextension]
-    if {[catch {load ./tclrobot$e}]} {
-       load tclrobot$e
-    }
-}
-
-set agent "zmbot/0.0"
+set agent "dcdot.tcl/0.0"
 if {![catch {set os [exec uname -s -r]}]} {
     set agent "$agent ($os)"
 }
@@ -185,8 +176,22 @@ proc RobotGetDCDOT {url} {
 if {$argc == 1} {
     set url [lindex $argv 0]
     RobotGetDCDOT $url
-    set mask {,meta,[Dd][Cc]\.*}
-    foreach a [array names URL $url$mask] {
-       puts "URL($a) = $URL($a)"
+    set mask {,meta}
+    if {[info exist URL($url,meta)]} {
+        foreach m $URL($url,meta) {
+           puts $m
+        }
+    }
+    if {[info exist URL($url,title)]} {
+        foreach m $URL($url,title) {
+           puts $m
+        }
     }
-}
\ No newline at end of file
+    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
+}