[Richard Suchenwirth] -- Here's a tiny script (executable for Unix) that takes 1..3 Tcl scripts (not files - literal code!) as arguments and evaluates
   * the first (if more than one - may be an empty string) in the beginning
   * the second (or first if only one) for each line from stdin
   * the third (if present) in the end: after eof on stdin
This is of course modeled after the 'BEGIN{this}{that}END{finally}' pattern of awk scripts, but I didn't want to introduce special keywords. To make it more awk-like, I however introduced special variables FS (field separator), OFS (output field separator), NF (number of fields in current line), NR (number of records - lines - so far), 0 (the whole input line), 1..$NF (fields as split by FS; you get the last field with [[set $NF]]).

I use this script for short Tcl tasks that I don't want to write a program file for, e.g.
  owh '' 'lappend t $0' 'puts [join [lsort -index end -integer -decreasing $t] \n]'
to sort output of a pipe the way I want it, but it can be used simply like
 owh 'puts [string toupper $0]'
 owh 'set n 0' 'incr n' 'puts $n' <infile ;# substitutes 'wc -l'
 owh '' '' 'set NR' <infile               ;# ditto, just more compact
 owh 'if $NR>10 break; puts $0' <infile   ;# substitutes 'head'
As for awk, the scripts should be single-quoted so the shell doesn't see all those dollars, brackets etc. These single quotes are not seen by tclsh.
 #!/bin/sh
 # use -*-Tcl-*- \
 exec tclsh "$0" "$@"

 if {[llength $argv]<1} {
	puts "usage: owh ?init? body ?exit?
	performs body (in Tcl) for each line (\$0) from stdin
	owh: Ousterhout - Welch - Hobbs, to name a few"
	exit -1
 }
 proc awksplit {text {split default}} {
    set no 0
    if {$split!="default"} {
	set t [split $text $split]
    } else {
	eval set t [list $text]
    }
    uplevel 1 [list set NF [llength $t]]
    foreach i $t {uplevel 1 [list set [incr no] $i]}
    uplevel 1 {set 0 "";trace variable 0 ru 0}
 }
 proc 0 {_name index op} {
    switch $op {
	r {
	    uplevel {
		set 0 ""
		for {set i 1} {$i<=$NF} {incr i} {lappend 0 [set $i]}
		set 0 [join $0 $OFS]
	    }
	}
	u {rename 0 {} ;# leave no traces of the trace..}
    }
 } 

 set FS default
 set OFS " "

 if {[llength $argv]>1} {
    eval [lindex $argv 0]
    set _body [lindex $argv 1] ;# strip outer braces
    set _exit [lindex $argv 2]
 } else {
    set _body [lindex $argv 0] ;# strip outer braces
    set _exit {}
 }

 set NR 1
 while 1 {
    gets stdin line
    if [eof stdin] break
    awksplit $line $FS
    eval $_body
    incr NR
 }
 set res [eval $_exit]
 if [string length $res] {puts $res}
----
Okay Richard, just to prove that I really am on it, here's
the current state of perlytcl (gimme a coupla days to finish!):

 #!/bin/sh
 # use -*-Tcl-*- \
 exec tclsh "$0" "$@"
 set awk   0
 set print 1
 set bak   {}
 foreach arg $args {
   # if we are doing in-place edits, get the bak pattern
   regexp {^(-[an]*i)\.?(.+)?} $arg -> arg bak 
   switch -exact -- $arg {
     -an -
     -na -
      -a -
      -n { # default will be to print, so we won't support p
            if { [ regexp {a} $arg ] } {
               set awk 1
            }
            if { [ regexp {n} $arg ] } {
               set print 0
            }
         }
    -ani -
    -nai -
      -i {
         }
    -ane -
    -nae -
      -e { # script follows! then filenames
         }
 default {
            if { [ string match $switch -e ] } {
               set cmd $arg
               set switch {}   
            # sorry, we silently lose non-existent files
            } elseif { [ file exists $arg ] } {
               lappend files $arg
            }   
         }
   } # end of switch   
 }

 foreach file $files {
 }