XML headers with character encoding as specified by HTTP server

diff --git a/robot.tcl b/robot.tcl
index a90d6e8..3ab1d81 100755 (executable)
--- a/robot.tcl
+++ b/robot.tcl
@@ -1,5 +1,5 @@
 #!/usr/bin/tclsh 
-# $Id: robot.tcl,v 1.43 2003/06/11 09:40:22 adam Exp $
+# $Id: robot.tcl,v 1.44 2003/06/11 10:11:39 adam Exp $
 #
 proc RobotFileNext1 {area lead} {
     # puts "RobotFileNext1 area=$area lead=$lead"
@@ -32,6 +32,7 @@ proc RobotFileNext1 {area lead} {
 }
 
 proc RobotWriteRecord {outf fromurl distance} {
+    puts $outf {<?xml version="1.0" encoding="ISO-8859-1" standalone="yes"?>}
     puts $outf "<zmbot>"
     puts $outf "<distance>"
     puts $outf $distance
@@ -47,6 +48,7 @@ proc RobotReadRecord {inf fromurlx distancex} {
     upvar $distancex distance
     gets $inf
     gets $inf
+    gets $inf
     set distance [string trim [gets $inf]]
     # puts "got distance = $distance"
     gets $inf
@@ -801,6 +803,8 @@ proc RobotTextPlain {task url out} {
 proc RobotWriteMetadata {task url out} {
     global URL
 
+    set charset $URL($task,$url,charset)
+    puts $out "<?xml version=\"1.0\" encoding=\"$charset\" standalone=\"yes\"?>"
     puts $out "<zmbot>"
 
     set distance 1000
@@ -886,6 +890,7 @@ proc RobotReadHeader {task url sock} {
            set version {}
            set headbuf [string range $URL($task,$url,buf) 0 $n]
            incr n 4
+           set URL($task,$url,charset) ISO-8859-1
            set URL($task,$url,buf) [string range $URL($task,$url,buf) $n end]
            
            regexp {^HTTP/([0-9.]+)[ ]+([0-9]+)} $headbuf x version code
@@ -894,6 +899,7 @@ proc RobotReadHeader {task url sock} {
                if {[regexp {^([^:]+):[ ]+([^;]*)} $line x name value]} {
                    set URL($task,$url,head,[string tolower $name]) [string trim $value]
                }
+               regexp {^Content-Type:.*charset=([A-Za-z0-9_-]*)} $line x URL($task,$url,charset)
            }
            puts "HTTP CODE $code"
            set URL($task,$url,state) skip