[p5sagit/p5-mst-13.2.git] / x2p / s2p.SH

: This forces SH files to create target in same directory as SH file.
: This is so that make depend always knows where to find SH derivatives.
case "$0" in
*/*) cd `expr X$0 : 'X\(.*\)/'` ;;
esac
case $CONFIG in
'')
    if test ! -f config.sh; then
	ln ../config.sh . || \
	ln -s ../config.sh . || \
	ln ../../config.sh . || \
	ln ../../../config.sh . || \
	(echo "Can't find config.sh."; exit 1)
    fi 2>/dev/null
    . ./config.sh
    ;;
esac
echo "Extracting s2p (with variable substitutions)"
: This section of the file will have variable substitutions done on it.
: Move anything that needs config subs from !NO!SUBS! section to !GROK!THIS!.
: Protect any dollar signs and backticks that you do not want interpreted
: by putting a backslash in front.  You may delete these comments.
$spitshell >s2p <<!GROK!THIS!
#!$bin/perl

\$bin = '$bin';
!GROK!THIS!

: In the following dollars and backticks do not need the extra backslash.
$spitshell >>s2p <<'!NO!SUBS!'

# $RCSfile: s2p.SH,v $$Revision: 4.0.1.1 $$Date: 91/06/07 12:19:18 $
#
# $Log:	s2p.SH,v $
# Revision 4.0.1.1  91/06/07  12:19:18  lwall
# patch4: s2p now handles embedded newlines better and optimizes common idioms
# 
# Revision 4.0  91/03/20  01:57:59  lwall
# 4.0 baseline.
# 
#

$indent = 4;
$shiftwidth = 4;
$l = '{'; $r = '}';

while ($ARGV[0] =~ /^-/) {
    $_ = shift;
  last if /^--/;
    if (/^-D/) {
	$debug++;
	open(BODY,'>-');
	next;
    }
    if (/^-n/) {
	$assumen++;
	next;
    }
    if (/^-p/) {
	$assumep++;
	next;
    }
    die "I don't recognize this switch: $_\n";
}

unless ($debug) {
    open(BODY,">/tmp/sperl$$") ||
      &Die("Can't open temp file: $!\n");
}

if (!$assumen && !$assumep) {
    print BODY &q(<<'EOT');
:	while ($ARGV[0] =~ /^-/) {
:	    $_ = shift;
:	  last if /^--/;
:	    if (/^-n/) {
:		$nflag++;
:		next;
:	    }
:	    die "I don't recognize this switch: $_\\n";
:	}
:	
EOT
}

print BODY &q(<<'EOT');
:	#ifdef PRINTIT
:	#ifdef ASSUMEP
:	$printit++;
:	#else
:	$printit++ unless $nflag;
:	#endif
:	#endif
:	<><>
:	$\ = "\n";		# automatically add newline on print
:	<><>
:	#ifdef TOPLABEL
:	LINE:
:	while (chop($_ = <>)) {
:	#else
:	LINE:
:	while (<>) {
:	    chop;
:	#endif
EOT

LINE:
while (<>) {

    # Wipe out surrounding whitespace.

    s/[ \t]*(.*)\n$/$1/;

    # Perhaps it's a label/comment.

    if (/^:/) {
	s/^:[ \t]*//;
	$label = &make_label($_);
	if ($. == 1) {
	    $toplabel = $label;
	    if (/^(top|(re)?start|redo|begin(ning)|again|input)$/i) {
		$_ = <>;
		redo LINE; # Never referenced, so delete it if not a comment.
	    }
	}
	$_ = "$label:";
	if ($lastlinewaslabel++) {
	    $indent += 4;
	    print BODY &tab, ";\n";
	    $indent -= 4;
	}
	if ($indent >= 2) {
	    $indent -= 2;
	    $indmod = 2;
	}
	next;
    } else {
	$lastlinewaslabel = '';
    }

    # Look for one or two address clauses

    $addr1 = '';
    $addr2 = '';
    if (s/^([0-9]+)//) {
	$addr1 = "$1";
	$addr1 = "\$. == $addr1" unless /^,/;
    }
    elsif (s/^\$//) {
	$addr1 = 'eof()';
    }
    elsif (s|^/||) {
	$addr1 = &fetchpat('/');
    }
    if (s/^,//) {
	if (s/^([0-9]+)//) {
	    $addr2 = "$1";
	} elsif (s/^\$//) {
	    $addr2 = "eof()";
	} elsif (s|^/||) {
	    $addr2 = &fetchpat('/');
	} else {
	    &Die("Invalid second address at line $.\n");
	}
	$addr1 .= " .. $addr2";
    }

    # Now we check for metacommands {, }, and ! and worry
    # about indentation.

    s/^[ \t]+//;
    # a { to keep vi happy
    if ($_ eq '}') {
	$indent -= 4;
	next;
    }
    if (s/^!//) {
	$if = 'unless';
	$else = "$r else $l\n";
    } else {
	$if = 'if';
	$else = '';
    }
    if (s/^{//) {	# a } to keep vi happy
	$indmod = 4;
	$redo = $_;
	$_ = '';
	$rmaybe = '';
    } else {
	$rmaybe = "\n$r";
	if ($addr2 || $addr1) {
	    $space = ' ' x $shiftwidth;
	} else {
	    $space = '';
	}
	$_ = &transmogrify();
    }

    # See if we can optimize to modifier form.

    if ($addr1) {
	if ($_ !~ /[\n{}]/ && $rmaybe && !$change &&
	  $_ !~ / if / && $_ !~ / unless /) {
	    s/;$/ $if $addr1;/;
	    $_ = substr($_,$shiftwidth,1000);
	} else {
	    $_ = "$if ($addr1) $l\n$change$_$rmaybe";
	}
	$change = '';
	next LINE;
    }
} continue {
    @lines = split(/\n/,$_);
    for (@lines) {
	unless (s/^ *<<--//) {
	    print BODY &tab;
	}
	print BODY $_, "\n";
    }
    $indent += $indmod;
    $indmod = 0;
    if ($redo) {
	$_ = $redo;
	$redo = '';
	redo LINE;
    }
}
if ($lastlinewaslabel++) {
    $indent += 4;
    print BODY &tab, ";\n";
    $indent -= 4;
}

if ($appendseen || $tseen || !$assumen) {
    $printit++ if $dseen || (!$assumen && !$assumep);
    print BODY &q(<<'EOT');
:	#ifdef SAWNEXT
:	}
:	continue {
:	#endif
:	#ifdef PRINTIT
:	#ifdef DSEEN
:	#ifdef ASSUMEP
:	    print if $printit++;
:	#else
:	    if ($printit)
:		{ print; }
:	    else
:		{ $printit++ unless $nflag; }
:	#endif
:	#else
:	    print if $printit;
:	#endif
:	#else
:	    print;
:	#endif
:	#ifdef TSEEN
:	    $tflag = 0;
:	#endif
:	#ifdef APPENDSEEN
:	    if ($atext) { chop $atext; print $atext; $atext = ''; }
:	#endif
EOT

print BODY &q(<<'EOT');
:	}
EOT
}

close BODY;

unless ($debug) {
    open(HEAD,">/tmp/sperl2$$.c")
      || &Die("Can't open temp file 2: $!\n");
    print HEAD "#define PRINTIT\n"	if $printit;
    print HEAD "#define APPENDSEEN\n"	if $appendseen;
    print HEAD "#define TSEEN\n"	if $tseen;
    print HEAD "#define DSEEN\n"	if $dseen;
    print HEAD "#define ASSUMEN\n"	if $assumen;
    print HEAD "#define ASSUMEP\n"	if $assumep;
    print HEAD "#define TOPLABEL\n"	if $toplabel;
    print HEAD "#define SAWNEXT\n"	if $sawnext;
    if ($opens) {print HEAD "$opens\n";}
    open(BODY,"/tmp/sperl$$")
      || &Die("Can't reopen temp file: $!\n");
    while (<BODY>) {
	print HEAD $_;
    }
    close HEAD;

    print &q(<<"EOT");
:	#!$bin/perl
:	eval 'exec $bin/perl -S \$0 \${1+"\$@"}'
:		if \$running_under_some_shell;
:	
EOT
    open(BODY,"cc -E /tmp/sperl2$$.c |") ||
	&Die("Can't reopen temp file: $!\n");
    while (<BODY>) {
	/^# [0-9]/ && next;
	/^[ \t]*$/ && next;
	s/^<><>//;
	print;
    }
}

&Cleanup;
exit;

sub Cleanup {
    chdir "/tmp";
    unlink "sperl$$", "sperl2$$", "sperl2$$.c";
}
sub Die {
    &Cleanup;
    die $_[0];
}
sub tab {
    "\t" x ($indent / 8) . ' ' x ($indent % 8);
}
sub make_filehandle {
    local($_) = $_[0];
    local($fname) = $_;
    if (!$seen{$fname}) {
	$_ = "FH_" . $_ if /^\d/;
	s/[^a-zA-Z0-9]/_/g;
	s/^_*//;
	$_ = "\U$_";
	if ($fhseen{$_}) {
	    for ($tmp = "a"; $fhseen{"$_$tmp"}; $a++) {}
	    $_ .= $tmp;
	}
	$fhseen{$_} = 1;
	$opens .= &q(<<"EOT");
:	open($_, '>$fname') || die "Can't create $fname: \$!";
EOT
	$seen{$fname} = $_;
    }
    $seen{$fname};
}

sub make_label {
    local($label) = @_;
    $label =~ s/[^a-zA-Z0-9]/_/g;
    if ($label =~ /^[0-9_]/) { $label = 'L' . $label; }
    $label = substr($label,0,8);

    # Could be a reserved word, so capitalize it.
    substr($label,0,1) =~ y/a-z/A-Z/
      if $label =~ /^[a-z]/;

    $label;
}

sub transmogrify {
    {	# case
	if (/^d/) {
	    $dseen++;
	    chop($_ = &q(<<'EOT'));
:	<<--#ifdef PRINTIT
:	$printit = 0;
:	<<--#endif
:	next LINE;
EOT
	    $sawnext++;
	    next;
	}

	if (/^n/) {
	    chop($_ = &q(<<'EOT'));
:	<<--#ifdef PRINTIT
:	<<--#ifdef DSEEN
:	<<--#ifdef ASSUMEP
:	print if $printit++;
:	<<--#else
:	if ($printit)
:	    { print; }
:	else
:	    { $printit++ unless $nflag; }
:	<<--#endif
:	<<--#else
:	print if $printit;
:	<<--#endif
:	<<--#else
:	print;
:	<<--#endif
:	<<--#ifdef APPENDSEEN
:	if ($atext) {chop $atext; print $atext; $atext = '';}
:	<<--#endif
:	$_ = <>;
:	chop;
:	<<--#ifdef TSEEN
:	$tflag = 0;
:	<<--#endif
EOT
	    next;
	}

	if (/^a/) {
	    $appendseen++;
	    $command = $space . "\$atext .= <<'End_Of_Text';\n<<--";
	    $lastline = 0;
	    while (<>) {
		s/^[ \t]*//;
		s/^[\\]//;
		unless (s|\\$||) { $lastline = 1;}
		s/^([ \t]*\n)/<><>$1/;
		$command .= $_;
		$command .= '<<--';
		last if $lastline;
	    }
	    $_ = $command . "End_Of_Text";
	    last;
	}

	if (/^[ic]/) {
	    if (/^c/) { $change = 1; }
	    $addr1 = 1 if $addr1 eq '';
	    $addr1 = '$iter = (' . $addr1 . ')';
	    $command = $space .
	      "    if (\$iter == 1) { print <<'End_Of_Text'; }\n<<--";
	    $lastline = 0;
	    while (<>) {
		s/^[ \t]*//;
		s/^[\\]//;
		unless (s/\\$//) { $lastline = 1;}
		s/'/\\'/g;
		s/^([ \t]*\n)/<><>$1/;
		$command .= $_;
		$command .= '<<--';
		last if $lastline;
	    }
	    $_ = $command . "End_Of_Text";
	    if ($change) {
		$dseen++;
		$change = "$_\n";
		chop($_ = &q(<<"EOT"));
:	<<--#ifdef PRINTIT
:	$space\$printit = 0;
:	<<--#endif
:	${space}next LINE;
EOT
		$sawnext++;
	    }
	    last;
	}

	if (/^s/) {
	    $delim = substr($_,1,1);
	    $len = length($_);
	    $repl = $end = 0;
	    $inbracket = 0;
	    for ($i = 2; $i < $len; $i++) {
		$c = substr($_,$i,1);
		if ($c eq $delim) {
		    if ($inbracket) {
			substr($_, $i, 0) = '\\';
			$i++;
			$len++;
		    }
		    else {
			if ($repl) {
			    $end = $i;
			    last;
			} else {
			    $repl = $i;
			}
		    }
		}
		elsif ($c eq '\\') {
		    $i++;
		    if ($i >= $len) {
			$_ .= 'n';
			$_ .= <>;
			$len = length($_);
			$_ = substr($_,0,--$len);
		    }
		    elsif (substr($_,$i,1) =~ /^[n]$/) {
			;
		    }
		    elsif (!$repl &&
		      substr($_,$i,1) =~ /^[(){}\w]$/) {
			$i--;
			$len--;
			substr($_, $i, 1) = '';
		    }
		    elsif (!$repl &&
		      substr($_,$i,1) =~ /^[<>]$/) {
			substr($_,$i,1) = 'b';
		    }
		}
		elsif ($c eq '[' && !$repl) {
		    $i++ if substr($_,$i,1) eq '^';
		    $i++ if substr($_,$i,1) eq ']';
		    $inbracket = 1;
		}
		elsif ($c eq ']') {
		    $inbracket = 0;
		}
		elsif ($c eq "\t") {
		    substr($_, $i, 1) = '\\t';
		    $i++;
		    $len++;
		}
		elsif (!$repl && index("()+",$c) >= 0) {
		    substr($_, $i, 0) = '\\';
		    $i++;
		    $len++;
		}
	    }
	    &Die("Malformed substitution at line $.\n")
	      unless $end;
	    $pat = substr($_, 0, $repl + 1);
	    $repl = substr($_, $repl+1, $end-$repl-1);
	    $end = substr($_, $end + 1, 1000);
	    &simplify($pat);
	    $dol = '$';
	    $repl =~ s/\$/\\$/;
	    $repl =~ s'&'$&'g;
	    $repl =~ s/[\\]([0-9])/$dol$1/g;
	    $subst = "$pat$repl$delim";
	    $cmd = '';
	    while ($end) {
		if ($end =~ s/^g//) {
		    $subst .= 'g';
		    next;
		}
		if ($end =~ s/^p//) {
		    $cmd .= ' && (print)';
		    next;
		}
		if ($end =~ s/^w[ \t]*//) {
		    $fh = &make_filehandle($end);
		    $cmd .= " && (print $fh \$_)";
		    $end = '';
		    next;
		}
		&Die("Unrecognized substitution command".
		  "($end) at line $.\n");
	    }
	    chop ($_ = &q(<<"EOT"));
:	<<--#ifdef TSEEN
:	$subst && \$tflag++$cmd;
:	<<--#else
:	$subst$cmd;
:	<<--#endif
EOT
	    next;
	}

	if (/^p/) {
	    $_ = 'print;';
	    next;
	}

	if (/^w/) {
	    s/^w[ \t]*//;
	    $fh = &make_filehandle($_);
	    $_ = "print $fh \$_;";
	    next;
	}

	if (/^r/) {
	    $appendseen++;
	    s/^r[ \t]*//;
	    $file = $_;
	    $_ = "\$atext .= `cat $file 2>/dev/null`;";
	    next;
	}

	if (/^P/) {
	    $_ = 'print $1 if /^(.*)/;';
	    next;
	}

	if (/^D/) {
	    chop($_ = &q(<<'EOT'));
:	s/^.*\n?//;
:	redo LINE if $_;
:	next LINE;
EOT
	    $sawnext++;
	    next;
	}

	if (/^N/) {
	    chop($_ = &q(<<'EOT'));
:	$_ .= "\n";
:	$len1 = length;
:	$_ .= <>;
:	chop if $len1 < length;
:	<<--#ifdef TSEEN
:	$tflag = 0;
:	<<--#endif
EOT
	    next;
	}

	if (/^h/) {
	    $_ = '$hold = $_;';
	    next;
	}

	if (/^H/) {
	    $_ = '$hold .= "\n"; $hold .= $_;';
	    next;
	}

	if (/^g/) {
	    $_ = '$_ = $hold;';
	    next;
	}

	if (/^G/) {
	    $_ = '$_ .= "\n"; $_ .= $hold;';
	    next;
	}

	if (/^x/) {
	    $_ = '($_, $hold) = ($hold, $_);';
	    next;
	}

	if (/^b$/) {
	    $_ = 'next LINE;';
	    $sawnext++;
	    next;
	}

	if (/^b/) {
	    s/^b[ \t]*//;
	    $lab = &make_label($_);
	    if ($lab eq $toplabel) {
		$_ = 'redo LINE;';
	    } else {
		$_ = "goto $lab;";
	    }
	    next;
	}

	if (/^t$/) {
	    $_ = 'next LINE if $tflag;';
	    $sawnext++;
	    $tseen++;
	    next;
	}

	if (/^t/) {
	    s/^t[ \t]*//;
	    $lab = &make_label($_);
	    $_ = q/if ($tflag) {$tflag = 0; /;
	    if ($lab eq $toplabel) {
		$_ .= 'redo LINE;}';
	    } else {
		$_ .= "goto $lab;}";
	    }
	    $tseen++;
	    next;
	}

	if (/^y/) {
	    s/abcdefghijklmnopqrstuvwxyz/a-z/g;
	    s/ABCDEFGHIJKLMNOPQRSTUVWXYZ/A-Z/g;
	    s/abcdef/a-f/g;
	    s/ABCDEF/A-F/g;
	    s/0123456789/0-9/g;
	    s/01234567/0-7/g;
	    $_ .= ';';
	}

	if (/^=/) {
	    $_ = 'print $.;';
	    next;
	}

	if (/^q/) {
	    chop($_ = &q(<<'EOT'));
:	close(ARGV);
:	@ARGV = ();
:	next LINE;
EOT
	    $sawnext++;
	    next;
	}
    } continue {
	if ($space) {
	    s/^/$space/;
	    s/(\n)(.)/$1$space$2/g;
	}
	last;
    }
    $_;
}

sub fetchpat {
    local($outer) = @_;
    local($addr) = $outer;
    local($inbracket);
    local($prefix,$delim,$ch);

    # Process pattern one potential delimiter at a time.

    DELIM: while (s#^([^\]+(|)[\\/]*)([]+(|)[\\/])##) {
	$prefix = $1;
	$delim = $2;
	if ($delim eq '\\') {
	    s/(.)//;
	    $ch = $1;
	    $delim = '' if $ch =~ /^[(){}A-Za-mo-z]$/;
	    $ch = 'b' if $ch =~ /^[<>]$/;
	    $delim .= $ch;
	}
	elsif ($delim eq '[') {
	    $inbracket = 1;
	    s/^\^// && ($delim .= '^');
	    s/^]// && ($delim .= ']');
	}
	elsif ($delim eq ']') {
	    $inbracket = 0;
	}
	elsif ($inbracket || $delim ne $outer) {
	    $delim = '\\' . $delim;
	}
	$addr .= $prefix;
	$addr .= $delim;
	if ($delim eq $outer && !$inbracket) {
	    last DELIM;
	}
    }
    $addr =~ s/\t/\\t/g;
    &simplify($addr);
    $addr;
}

sub q {
    local($string) = @_;
    local($*) = 1;
    $string =~ s/^:\t?//g;
    $string;
}

sub simplify {
    $_[0] =~ s/_a-za-z0-9/\\w/ig;
    $_[0] =~ s/a-z_a-z0-9/\\w/ig;
    $_[0] =~ s/a-za-z_0-9/\\w/ig;
    $_[0] =~ s/a-za-z0-9_/\\w/ig;
    $_[0] =~ s/_0-9a-za-z/\\w/ig;
    $_[0] =~ s/0-9_a-za-z/\\w/ig;
    $_[0] =~ s/0-9a-z_a-z/\\w/ig;
    $_[0] =~ s/0-9a-za-z_/\\w/ig;
    $_[0] =~ s/\[\\w\]/\\w/g;
    $_[0] =~ s/\[^\\w\]/\\W/g;
    $_[0] =~ s/\[0-9\]/\\d/g;
    $_[0] =~ s/\[^0-9\]/\\D/g;
    $_[0] =~ s/\\d\\d\*/\\d+/g;
    $_[0] =~ s/\\D\\D\*/\\D+/g;
    $_[0] =~ s/\\w\\w\*/\\w+/g;
    $_[0] =~ s/\\t\\t\*/\\t+/g;
    $_[0] =~ s/(\[.[^]]*\])\1\*/$1+/g;
    $_[0] =~ s/([\w\s!@#%^&-=,:;'"])\1\*/$1+/g;
}

!NO!SUBS!
chmod 755 s2p
$eunicefix s2p
Commit	Line	Data
a687059c	1	: This forces SH files to create target in same directory as SH file.
	2	: This is so that make depend always knows where to find SH derivatives.
	3	case "$0" in
	4	/) cd `expr X$0 : 'X\(.*\)/'` ;;
	5	esac
	6	case $CONFIG in
	7	'')
	8	if test ! -f config.sh; then
	9	ln ../config.sh . \|\| \
27e2fb84	10	ln -s ../config.sh . \|\| \
a687059c	11	ln ../../config.sh . \|\| \
	12	ln ../../../config.sh . \|\| \
	13	(echo "Can't find config.sh."; exit 1)
fe14fcc3	14	fi 2>/dev/null
e5d73d77	15	. ./config.sh
a687059c	16	;;
	17	esac
	18	echo "Extracting s2p (with variable substitutions)"
	19	: This section of the file will have variable substitutions done on it.
	20	: Move anything that needs config subs from !NO!SUBS! section to !GROK!THIS!.
	21	: Protect any dollar signs and backticks that you do not want interpreted
	22	: by putting a backslash in front. You may delete these comments.
	23	$spitshell >s2p <<!GROK!THIS!
	24	#!$bin/perl
	25
	26	\$bin = '$bin';
	27	!GROK!THIS!
	28
	29	: In the following dollars and backticks do not need the extra backslash.
	30	$spitshell >>s2p <<'!NO!SUBS!'
	31
9ef589d8	32	# $RCSfile: s2p.SH,v $$Revision: 4.0.1.1 $$Date: 91/06/07 12:19:18 $
378cc40b	33	#
a687059c	34	# $Log: s2p.SH,v $
9ef589d8	35	# Revision 4.0.1.1 91/06/07 12:19:18 lwall
	36	# patch4: s2p now handles embedded newlines better and optimizes common idioms
	37	#
fe14fcc3	38	# Revision 4.0 91/03/20 01:57:59 lwall
fe14fcc3	39	# 4.0 baseline.
378cc40b	40	#
378cc40b	41	#
8d063cd8	42
	43	$indent = 4;
	44	$shiftwidth = 4;
	45	$l = '{'; $r = '}';
8d063cd8	46
0a12ae7d	47	while ($ARGV[0] =~ /^-/) {
8d063cd8	48	$_ = shift;
	49	last if /^--/;
	50	if (/^-D/) {
	51	$debug++;
0a12ae7d	52	open(BODY,'>-');
8d063cd8	53	next;
	54	}
	55	if (/^-n/) {
	56	$assumen++;
	57	next;
	58	}
	59	if (/^-p/) {
	60	$assumep++;
	61	next;
	62	}
378cc40b	63	die "I don't recognize this switch: $_\n";
8d063cd8	64	}
	65
	66	unless ($debug) {
0a12ae7d	67	open(BODY,">/tmp/sperl$$") \|\|
0a12ae7d	68	&Die("Can't open temp file: $!\n");
8d063cd8	69	}
	70
	71	if (!$assumen && !$assumep) {
9ef589d8	72	print BODY &q(<<'EOT');
	73	: while ($ARGV[0] =~ /^-/) {
	74	: $_ = shift;
	75	: last if /^--/;
	76	: if (/^-n/) {
	77	: $nflag++;
	78	: next;
	79	: }
	80	: die "I don't recognize this switch: $_\\n";
	81	: }
	82	:
0a12ae7d	83	EOT
8d063cd8	84	}
8d063cd8	85
9ef589d8	86	print BODY &q(<<'EOT');
	87	: #ifdef PRINTIT
	88	: #ifdef ASSUMEP
	89	: $printit++;
	90	: #else
	91	: $printit++ unless $nflag;
	92	: #endif
	93	: #endif
	94	: <><>
	95	: $\ = "\n"; # automatically add newline on print
	96	: <><>
	97	: #ifdef TOPLABEL
	98	: LINE:
	99	: while (chop($_ = <>)) {
	100	: #else
	101	: LINE:
	102	: while (<>) {
	103	: chop;
	104	: #endif
0a12ae7d	105	EOT
0a12ae7d	106
9ef589d8	107	LINE:
9ef589d8	108	while (<>) {
0a12ae7d	109
0a12ae7d	110	# Wipe out surrounding whitespace.
8d063cd8	111
8d063cd8	112	s/[ \t](.)\n$/$1/;
0a12ae7d	113
	114	# Perhaps it's a label/comment.
	115
8d063cd8	116	if (/^:/) {
8d063cd8	117	s/^:[ \t]*//;
0a12ae7d	118	$label = &make_label($_);
8d063cd8	119	if ($. == 1) {
8d063cd8	120	$toplabel = $label;
9ef589d8	121	if (/^(top\|(re)?start\|redo\|begin(ning)\|again\|input)$/i) {
	122	$_ = <>;
	123	redo LINE; # Never referenced, so delete it if not a comment.
	124	}
8d063cd8	125	}
8d063cd8	126	$_ = "$label:";
ffed7fef	127	if ($lastlinewaslabel++) {
ffed7fef	128	$indent += 4;
0a12ae7d	129	print BODY &tab, ";\n";
ffed7fef	130	$indent -= 4;
ffed7fef	131	}
8d063cd8	132	if ($indent >= 2) {
	133	$indent -= 2;
	134	$indmod = 2;
	135	}
	136	next;
	137	} else {
	138	$lastlinewaslabel = '';
	139	}
0a12ae7d	140
	141	# Look for one or two address clauses
	142
8d063cd8	143	$addr1 = '';
	144	$addr2 = '';
	145	if (s/^([0-9]+)//) {
	146	$addr1 = "$1";
9ef589d8	147	$addr1 = "\$. == $addr1" unless /^,/;
8d063cd8	148	}
	149	elsif (s/^\$//) {
	150	$addr1 = 'eof()';
	151	}
	152	elsif (s\|^/\|\|) {
0a12ae7d	153	$addr1 = &fetchpat('/');
8d063cd8	154	}
	155	if (s/^,//) {
	156	if (s/^([0-9]+)//) {
	157	$addr2 = "$1";
	158	} elsif (s/^\$//) {
	159	$addr2 = "eof()";
	160	} elsif (s\|^/\|\|) {
0a12ae7d	161	$addr2 = &fetchpat('/');
8d063cd8	162	} else {
0a12ae7d	163	&Die("Invalid second address at line $.\n");
8d063cd8	164	}
	165	$addr1 .= " .. $addr2";
	166	}
0a12ae7d	167
	168	# Now we check for metacommands {, }, and ! and worry
	169	# about indentation.
	170
378cc40b	171	s/^[ \t]+//;
0a12ae7d	172	# a { to keep vi happy
8d063cd8	173	if ($_ eq '}') {
	174	$indent -= 4;
	175	next;
	176	}
	177	if (s/^!//) {
	178	$if = 'unless';
	179	$else = "$r else $l\n";
	180	} else {
	181	$if = 'if';
	182	$else = '';
	183	}
	184	if (s/^{//) { # a } to keep vi happy
	185	$indmod = 4;
	186	$redo = $_;
	187	$_ = '';
	188	$rmaybe = '';
	189	} else {
	190	$rmaybe = "\n$r";
	191	if ($addr2 \|\| $addr1) {
a687059c	192	$space = ' ' x $shiftwidth;
8d063cd8	193	} else {
	194	$space = '';
	195	}
0a12ae7d	196	$_ = &transmogrify();
8d063cd8	197	}
8d063cd8	198
0a12ae7d	199	# See if we can optimize to modifier form.
0a12ae7d	200
8d063cd8	201	if ($addr1) {
	202	if ($_ !~ /[\n{}]/ && $rmaybe && !$change &&
	203	$_ !~ / if / && $_ !~ / unless /) {
	204	s/;$/ $if $addr1;/;
	205	$_ = substr($_,$shiftwidth,1000);
	206	} else {
0a12ae7d	207	$_ = "$if ($addr1) $l\n$change$_$rmaybe";
8d063cd8	208	}
8d063cd8	209	$change = '';
0a12ae7d	210	next LINE;
8d063cd8	211	}
	212	} continue {
	213	@lines = split(/\n/,$_);
0a12ae7d	214	for (@lines) {
8d063cd8	215	unless (s/^ *<<--//) {
0a12ae7d	216	print BODY &tab;
8d063cd8	217	}
0a12ae7d	218	print BODY $_, "\n";
8d063cd8	219	}
	220	$indent += $indmod;
	221	$indmod = 0;
	222	if ($redo) {
	223	$_ = $redo;
	224	$redo = '';
0a12ae7d	225	redo LINE;
8d063cd8	226	}
8d063cd8	227	}
ffed7fef	228	if ($lastlinewaslabel++) {
ffed7fef	229	$indent += 4;
0a12ae7d	230	print BODY &tab, ";\n";
ffed7fef	231	$indent -= 4;
ffed7fef	232	}
8d063cd8	233
8d063cd8	234	if ($appendseen \|\| $tseen \|\| !$assumen) {
8d063cd8	235	$printit++ if $dseen \|\| (!$assumen && !$assumep);
9ef589d8	236	print BODY &q(<<'EOT');
	237	: #ifdef SAWNEXT
	238	: }
	239	: continue {
	240	: #endif
	241	: #ifdef PRINTIT
	242	: #ifdef DSEEN
	243	: #ifdef ASSUMEP
	244	: print if $printit++;
	245	: #else
	246	: if ($printit)
	247	: { print; }
	248	: else
	249	: { $printit++ unless $nflag; }
	250	: #endif
	251	: #else
	252	: print if $printit;
	253	: #endif
	254	: #else
	255	: print;
	256	: #endif
	257	: #ifdef TSEEN
	258	: $tflag = 0;
	259	: #endif
	260	: #ifdef APPENDSEEN
	261	: if ($atext) { chop $atext; print $atext; $atext = ''; }
	262	: #endif
	263	EOT
	264
	265	print BODY &q(<<'EOT');
	266	: }
0a12ae7d	267	EOT
8d063cd8	268	}
8d063cd8	269
0a12ae7d	270	close BODY;
8d063cd8	271
8d063cd8	272	unless ($debug) {
0a12ae7d	273	open(HEAD,">/tmp/sperl2$$.c")
0a12ae7d	274	\|\| &Die("Can't open temp file 2: $!\n");
9ef589d8	275	print HEAD "#define PRINTIT\n" if $printit;
	276	print HEAD "#define APPENDSEEN\n" if $appendseen;
	277	print HEAD "#define TSEEN\n" if $tseen;
	278	print HEAD "#define DSEEN\n" if $dseen;
	279	print HEAD "#define ASSUMEN\n" if $assumen;
	280	print HEAD "#define ASSUMEP\n" if $assumep;
	281	print HEAD "#define TOPLABEL\n" if $toplabel;
	282	print HEAD "#define SAWNEXT\n" if $sawnext;
0a12ae7d	283	if ($opens) {print HEAD "$opens\n";}
	284	open(BODY,"/tmp/sperl$$")
	285	\|\| &Die("Can't reopen temp file: $!\n");
	286	while (<BODY>) {
	287	print HEAD $_;
8d063cd8	288	}
0a12ae7d	289	close HEAD;
8d063cd8	290
9ef589d8	291	print &q(<<"EOT");
	292	: #!$bin/perl
	293	: eval 'exec $bin/perl -S \$0 \${1+"\$@"}'
	294	: if \$running_under_some_shell;
	295	:
0a12ae7d	296	EOT
	297	open(BODY,"cc -E /tmp/sperl2$$.c \|") \|\|
	298	&Die("Can't reopen temp file: $!\n");
	299	while (<BODY>) {
8d063cd8	300	/^# [0-9]/ && next;
	301	/^[ \t]*$/ && next;
	302	s/^<><>//;
	303	print;
	304	}
	305	}
	306
0a12ae7d	307	&Cleanup;
0a12ae7d	308	exit;
8d063cd8	309
0a12ae7d	310	sub Cleanup {
	311	chdir "/tmp";
	312	unlink "sperl$$", "sperl2$$", "sperl2$$.c";
	313	}
8d063cd8	314	sub Die {
0a12ae7d	315	&Cleanup;
8d063cd8	316	die $_[0];
8d063cd8	317	}
0a12ae7d	318	sub tab {
	319	"\t" x ($indent / 8) . ' ' x ($indent % 8);
	320	}
8d063cd8	321	sub make_filehandle {
0a12ae7d	322	local($_) = $_[0];
0a12ae7d	323	local($fname) = $_;
9ef589d8	324	if (!$seen{$fname}) {
	325	$_ = "FH_" . $_ if /^\d/;
	326	s/[^a-zA-Z0-9]/_/g;
	327	s/^_*//;
	328	$_ = "\U$_";
	329	if ($fhseen{$_}) {
	330	for ($tmp = "a"; $fhseen{"$_$tmp"}; $a++) {}
	331	$_ .= $tmp;
	332	}
	333	$fhseen{$_} = 1;
	334	$opens .= &q(<<"EOT");
	335	: open($_, '>$fname') \|\| die "Can't create $fname: \$!";
0a12ae7d	336	EOT
9ef589d8	337	$seen{$fname} = $_;
8d063cd8	338	}
9ef589d8	339	$seen{$fname};
8d063cd8	340	}
	341
	342	sub make_label {
0a12ae7d	343	local($label) = @_;
8d063cd8	344	$label =~ s/[^a-zA-Z0-9]/_/g;
	345	if ($label =~ /^[0-9_]/) { $label = 'L' . $label; }
	346	$label = substr($label,0,8);
0a12ae7d	347
	348	# Could be a reserved word, so capitalize it.
	349	substr($label,0,1) =~ y/a-z/A-Z/
	350	if $label =~ /^[a-z]/;
	351
8d063cd8	352	$label;
	353	}
	354
	355	sub transmogrify {
	356	{ # case
	357	if (/^d/) {
	358	$dseen++;
9ef589d8	359	chop($_ = &q(<<'EOT'));
	360	: <<--#ifdef PRINTIT
	361	: $printit = 0;
	362	: <<--#endif
	363	: next LINE;
0a12ae7d	364	EOT
9ef589d8	365	$sawnext++;
8d063cd8	366	next;
	367	}
	368
	369	if (/^n/) {
9ef589d8	370	chop($_ = &q(<<'EOT'));
	371	: <<--#ifdef PRINTIT
	372	: <<--#ifdef DSEEN
	373	: <<--#ifdef ASSUMEP
	374	: print if $printit++;
	375	: <<--#else
	376	: if ($printit)
	377	: { print; }
	378	: else
	379	: { $printit++ unless $nflag; }
	380	: <<--#endif
	381	: <<--#else
	382	: print if $printit;
	383	: <<--#endif
	384	: <<--#else
	385	: print;
	386	: <<--#endif
	387	: <<--#ifdef APPENDSEEN
	388	: if ($atext) {chop $atext; print $atext; $atext = '';}
	389	: <<--#endif
	390	: $_ = <>;
	391	: chop;
	392	: <<--#ifdef TSEEN
	393	: $tflag = 0;
	394	: <<--#endif
0a12ae7d	395	EOT
8d063cd8	396	next;
	397	}
	398
	399	if (/^a/) {
	400	$appendseen++;
9ef589d8	401	$command = $space . "\$atext .= <<'End_Of_Text';\n<<--";
8d063cd8	402	$lastline = 0;
	403	while (<>) {
	404	s/^[ \t]*//;
	405	s/^[\\]//;
	406	unless (s\|\\$\|\|) { $lastline = 1;}
8d063cd8	407	s/^([ \t]*\n)/<><>$1/;
	408	$command .= $_;
	409	$command .= '<<--';
	410	last if $lastline;
	411	}
9ef589d8	412	$_ = $command . "End_Of_Text";
8d063cd8	413	last;
	414	}
	415
	416	if (/^[ic]/) {
	417	if (/^c/) { $change = 1; }
9ef589d8	418	$addr1 = 1 if $addr1 eq '';
8d063cd8	419	$addr1 = '$iter = (' . $addr1 . ')';
9ef589d8	420	$command = $space .
9ef589d8	421	" if (\$iter == 1) { print <<'End_Of_Text'; }\n<<--";
8d063cd8	422	$lastline = 0;
	423	while (<>) {
	424	s/^[ \t]*//;
	425	s/^[\\]//;
	426	unless (s/\\$//) { $lastline = 1;}
	427	s/'/\\'/g;
	428	s/^([ \t]*\n)/<><>$1/;
	429	$command .= $_;
	430	$command .= '<<--';
	431	last if $lastline;
	432	}
9ef589d8	433	$_ = $command . "End_Of_Text";
8d063cd8	434	if ($change) {
	435	$dseen++;
	436	$change = "$_\n";
9ef589d8	437	chop($_ = &q(<<"EOT"));
	438	: <<--#ifdef PRINTIT
	439	: $space\$printit = 0;
	440	: <<--#endif
	441	: ${space}next LINE;
0a12ae7d	442	EOT
9ef589d8	443	$sawnext++;
8d063cd8	444	}
	445	last;
	446	}
	447
	448	if (/^s/) {
	449	$delim = substr($_,1,1);
	450	$len = length($_);
	451	$repl = $end = 0;
a687059c	452	$inbracket = 0;
8d063cd8	453	for ($i = 2; $i < $len; $i++) {
8d063cd8	454	$c = substr($_,$i,1);
a687059c	455	if ($c eq $delim) {
a687059c	456	if ($inbracket) {
0a12ae7d	457	substr($_, $i, 0) = '\\';
a687059c	458	$i++;
	459	$len++;
	460	}
	461	else {
	462	if ($repl) {
	463	$end = $i;
	464	last;
	465	} else {
	466	$repl = $i;
	467	}
	468	}
	469	}
	470	elsif ($c eq '\\') {
8d063cd8	471	$i++;
	472	if ($i >= $len) {
	473	$_ .= 'n';
	474	$_ .= <>;
	475	$len = length($_);
	476	$_ = substr($_,0,--$len);
	477	}
00bf170e	478	elsif (substr($_,$i,1) =~ /^[n]$/) {
	479	;
	480	}
0a12ae7d	481	elsif (!$repl &&
0a12ae7d	482	substr($_,$i,1) =~ /^[(){}\w]$/) {
8d063cd8	483	$i--;
8d063cd8	484	$len--;
0a12ae7d	485	substr($_, $i, 1) = '';
8d063cd8	486	}
0a12ae7d	487	elsif (!$repl &&
0a12ae7d	488	substr($_,$i,1) =~ /^[<>]$/) {
9f68db38	489	substr($_,$i,1) = 'b';
9f68db38	490	}
8d063cd8	491	}
a687059c	492	elsif ($c eq '[' && !$repl) {
	493	$i++ if substr($_,$i,1) eq '^';
	494	$i++ if substr($_,$i,1) eq ']';
	495	$inbracket = 1;
8d063cd8	496	}
a687059c	497	elsif ($c eq ']') {
	498	$inbracket = 0;
	499	}
9ef589d8	500	elsif ($c eq "\t") {
	501	substr($_, $i, 1) = '\\t';
	502	$i++;
	503	$len++;
	504	}
ae986130	505	elsif (!$repl && index("()+",$c) >= 0) {
0a12ae7d	506	substr($_, $i, 0) = '\\';
8d063cd8	507	$i++;
	508	$len++;
	509	}
	510	}
0a12ae7d	511	&Die("Malformed substitution at line $.\n")
0a12ae7d	512	unless $end;
8d063cd8	513	$pat = substr($_, 0, $repl + 1);
0a12ae7d	514	$repl = substr($_, $repl+1, $end-$repl-1);
8d063cd8	515	$end = substr($_, $end + 1, 1000);
9ef589d8	516	&simplify($pat);
8d063cd8	517	$dol = '$';
378cc40b	518	$repl =~ s/\$/\\$/;
8d063cd8	519	$repl =~ s'&'$&'g;
	520	$repl =~ s/[\\]([0-9])/$dol$1/g;
	521	$subst = "$pat$repl$delim";
	522	$cmd = '';
	523	while ($end) {
0a12ae7d	524	if ($end =~ s/^g//) {
	525	$subst .= 'g';
	526	next;
	527	}
	528	if ($end =~ s/^p//) {
	529	$cmd .= ' && (print)';
	530	next;
	531	}
8d063cd8	532	if ($end =~ s/^w[ \t]*//) {
0a12ae7d	533	$fh = &make_filehandle($end);
8d063cd8	534	$cmd .= " && (print $fh \$_)";
	535	$end = '';
	536	next;
	537	}
0a12ae7d	538	&Die("Unrecognized substitution command".
0a12ae7d	539	"($end) at line $.\n");
8d063cd8	540	}
9ef589d8	541	chop ($_ = &q(<<"EOT"));
	542	: <<--#ifdef TSEEN
	543	: $subst && \$tflag++$cmd;
	544	: <<--#else
	545	: $subst$cmd;
	546	: <<--#endif
0a12ae7d	547	EOT
8d063cd8	548	next;
	549	}
	550
	551	if (/^p/) {
	552	$_ = 'print;';
	553	next;
	554	}
	555
	556	if (/^w/) {
	557	s/^w[ \t]*//;
0a12ae7d	558	$fh = &make_filehandle($_);
8d063cd8	559	$_ = "print $fh \$_;";
	560	next;
	561	}
	562
	563	if (/^r/) {
	564	$appendseen++;
	565	s/^r[ \t]*//;
	566	$file = $_;
	567	$_ = "\$atext .= `cat $file 2>/dev/null`;";
	568	next;
	569	}
	570
	571	if (/^P/) {
9ef589d8	572	$_ = 'print $1 if /^(.*)/;';
8d063cd8	573	next;
	574	}
	575
	576	if (/^D/) {
9ef589d8	577	chop($_ = &q(<<'EOT'));
	578	: s/^.*\n?//;
	579	: redo LINE if $_;
	580	: next LINE;
0a12ae7d	581	EOT
9ef589d8	582	$sawnext++;
8d063cd8	583	next;
	584	}
	585
	586	if (/^N/) {
9ef589d8	587	chop($_ = &q(<<'EOT'));
	588	: $_ .= "\n";
	589	: $len1 = length;
	590	: $_ .= <>;
	591	: chop if $len1 < length;
	592	: <<--#ifdef TSEEN
	593	: $tflag = 0;
	594	: <<--#endif
0a12ae7d	595	EOT
8d063cd8	596	next;
	597	}
	598
	599	if (/^h/) {
	600	$_ = '$hold = $_;';
	601	next;
	602	}
	603
	604	if (/^H/) {
9ef589d8	605	$_ = '$hold .= "\n"; $hold .= $_;';
8d063cd8	606	next;
	607	}
	608
	609	if (/^g/) {
	610	$_ = '$_ = $hold;';
	611	next;
	612	}
	613
	614	if (/^G/) {
9ef589d8	615	$_ = '$_ .= "\n"; $_ .= $hold;';
8d063cd8	616	next;
	617	}
	618
	619	if (/^x/) {
	620	$_ = '($_, $hold) = ($hold, $_);';
	621	next;
	622	}
	623
	624	if (/^b$/) {
0a12ae7d	625	$_ = 'next LINE;';
9ef589d8	626	$sawnext++;
8d063cd8	627	next;
	628	}
	629
	630	if (/^b/) {
	631	s/^b[ \t]*//;
0a12ae7d	632	$lab = &make_label($_);
8d063cd8	633	if ($lab eq $toplabel) {
0a12ae7d	634	$_ = 'redo LINE;';
8d063cd8	635	} else {
	636	$_ = "goto $lab;";
	637	}
	638	next;
	639	}
	640
	641	if (/^t$/) {
0a12ae7d	642	$_ = 'next LINE if $tflag;';
9ef589d8	643	$sawnext++;
8d063cd8	644	$tseen++;
	645	next;
	646	}
	647
	648	if (/^t/) {
	649	s/^t[ \t]*//;
0a12ae7d	650	$lab = &make_label($_);
9ef589d8	651	$_ = q/if ($tflag) {$tflag = 0; /;
8d063cd8	652	if ($lab eq $toplabel) {
0a12ae7d	653	$_ .= 'redo LINE;}';
8d063cd8	654	} else {
0a12ae7d	655	$_ .= "goto $lab;}";
8d063cd8	656	}
	657	$tseen++;
	658	next;
	659	}
	660
9ef589d8	661	if (/^y/) {
	662	s/abcdefghijklmnopqrstuvwxyz/a-z/g;
	663	s/ABCDEFGHIJKLMNOPQRSTUVWXYZ/A-Z/g;
	664	s/abcdef/a-f/g;
	665	s/ABCDEF/A-F/g;
	666	s/0123456789/0-9/g;
	667	s/01234567/0-7/g;
	668	$_ .= ';';
	669	}
	670
8d063cd8	671	if (/^=/) {
9ef589d8	672	$_ = 'print $.;';
8d063cd8	673	next;
	674	}
	675
	676	if (/^q/) {
9ef589d8	677	chop($_ = &q(<<'EOT'));
	678	: close(ARGV);
	679	: @ARGV = ();
	680	: next LINE;
0a12ae7d	681	EOT
9ef589d8	682	$sawnext++;
8d063cd8	683	next;
	684	}
	685	} continue {
	686	if ($space) {
	687	s/^/$space/;
	688	s/(\n)(.)/$1$space$2/g;
	689	}
	690	last;
	691	}
	692	$_;
	693	}
	694
a687059c	695	sub fetchpat {
	696	local($outer) = @_;
	697	local($addr) = $outer;
	698	local($inbracket);
	699	local($prefix,$delim,$ch);
	700
0a12ae7d	701	# Process pattern one potential delimiter at a time.
	702
	703	DELIM: while (s#^([^\]+(\|)[\\/]*)([]+(\|)[\\/])##) {
a687059c	704	$prefix = $1;
a687059c	705	$delim = $2;
a687059c	706	if ($delim eq '\\') {
	707	s/(.)//;
	708	$ch = $1;
00bf170e	709	$delim = '' if $ch =~ /^[(){}A-Za-mo-z]$/;
9f68db38	710	$ch = 'b' if $ch =~ /^[<>]$/;
9f68db38	711	$delim .= $ch;
a687059c	712	}
	713	elsif ($delim eq '[') {
	714	$inbracket = 1;
	715	s/^\^// && ($delim .= '^');
	716	s/^]// && ($delim .= ']');
a687059c	717	}
	718	elsif ($delim eq ']') {
	719	$inbracket = 0;
	720	}
	721	elsif ($inbracket \|\| $delim ne $outer) {
a687059c	722	$delim = '\\' . $delim;
	723	}
	724	$addr .= $prefix;
	725	$addr .= $delim;
	726	if ($delim eq $outer && !$inbracket) {
0a12ae7d	727	last DELIM;
a687059c	728	}
a687059c	729	}
9ef589d8	730	$addr =~ s/\t/\\t/g;
9ef589d8	731	&simplify($addr);
a687059c	732	$addr;
	733	}
	734
9ef589d8	735	sub q {
	736	local($string) = @_;
	737	local($*) = 1;
	738	$string =~ s/^:\t?//g;
	739	$string;
	740	}
	741
	742	sub simplify {
	743	$_[0] =~ s/_a-za-z0-9/\\w/ig;
	744	$_[0] =~ s/a-z_a-z0-9/\\w/ig;
	745	$_[0] =~ s/a-za-z_0-9/\\w/ig;
	746	$_[0] =~ s/a-za-z0-9_/\\w/ig;
	747	$_[0] =~ s/_0-9a-za-z/\\w/ig;
	748	$_[0] =~ s/0-9_a-za-z/\\w/ig;
	749	$_[0] =~ s/0-9a-z_a-z/\\w/ig;
	750	$_[0] =~ s/0-9a-za-z_/\\w/ig;
	751	$_[0] =~ s/\[\\w\]/\\w/g;
	752	$_[0] =~ s/\[^\\w\]/\\W/g;
	753	$_[0] =~ s/\[0-9\]/\\d/g;
	754	$_[0] =~ s/\[^0-9\]/\\D/g;
	755	$_[0] =~ s/\\d\\d\*/\\d+/g;
	756	$_[0] =~ s/\\D\\D\*/\\D+/g;
	757	$_[0] =~ s/\\w\\w\*/\\w+/g;
	758	$_[0] =~ s/\\t\\t\*/\\t+/g;
	759	$_[0] =~ s/(\[.[^]]\])\1\/$1+/g;
	760	$_[0] =~ s/([\w\s!@#%^&-=,:;'"])\1\*/$1+/g;
	761	}
	762
a687059c	763	!NO!SUBS!
	764	chmod 755 s2p
	765	$eunicefix s2p