use Config;
use File::Basename qw(basename dirname);
+use Cwd;
# List explicitly here the variables you want Configure to
# generate. Metaconfig only looks for shell variables, so you
# This forces PL files to create target in same directory as PL file.
# This is so that make depend always knows where to find PL derivatives.
+$origdir = cwd;
chdir dirname($0);
$file = basename($0, '.PL');
$file .= '.com' if $^O eq 'VMS';
print OUT <<'!NO!SUBS!';
+use strict;
+
use Config;
use File::Path qw(mkpath);
use Getopt::Std;
-getopts('Dd:rlh');
+getopts('Dd:rlhaQ');
+use vars qw($opt_D $opt_d $opt_r $opt_l $opt_h $opt_a $opt_Q);
+die "-r and -a options are mutually exclusive\n" if ($opt_r and $opt_a);
+my @inc_dirs = inc_dirs() if $opt_a;
my $Exit = 0;
die "Destination directory $Dest_dir doesn't exist or isn't a directory\n"
unless -d $Dest_dir;
-@isatype = split(' ',<<END);
+my @isatype = split(' ',<<END);
char uchar u_char
short ushort u_short
int uint u_int
FILE key_t caddr_t
END
+my %isatype;
@isatype{@isatype} = (1) x @isatype;
-$inif = 0;
+my $inif = 0;
+my %Is_converted;
@ARGV = ('-') unless @ARGV;
-while (defined ($file = next_file())) {
+build_preamble_if_necessary();
+
+my ($t, $tab, %curargs, $new, $eval_index, $dir, $name, $args, $outfile);
+my ($incl, $next);
+while (defined (my $file = next_file())) {
if (-l $file and -d $file) {
link_if_possible($file) if ($opt_l);
next;
open(OUT, ">-");
} else {
($outfile = $file) =~ s/\.h$/.ph/ || next;
- print "$file -> $outfile\n";
+ print "$file -> $outfile\n" unless $opt_Q;
if ($file =~ m|^(.*)/|) {
$dir = $1;
mkpath "$Dest_dir/$dir";
}
+
+ if ($opt_a) { # automagic mode: locate header file in @inc_dirs
+ foreach (@inc_dirs) {
+ chdir $_;
+ last if -f $file;
+ }
+ }
+
open(IN,"$file") || (($Exit = 1),(warn "Can't open $file: $!\n"),next);
open(OUT,">$Dest_dir/$outfile") || die "Can't create $outfile: $!\n";
}
- while (<IN>) {
- chop;
- while (/\\$/) {
- chop;
- $_ .= <IN>;
- chop;
- }
- print OUT "# $_\n" if $opt_D;
- if (s:/\*:\200:g) {
- s:\*/:\201:g;
- s/\200[^\201]*\201//g; # delete single line comments
- if (s/\200.*//) { # begin multi-line comment?
- $_ .= '/*';
- $_ .= <IN>;
- redo;
- }
- }
- if (s/^\s*#\s*//) {
+
+ print OUT "require '_h2ph_pre.ph';\n\n";
+
+ while (defined (local $_ = next_line($file))) {
+ if (s/^\s*\#\s*//) {
if (s/^define\s+(\w+)//) {
$name = $1;
$new = '';
s/\s+$//;
+ s/\(\w+\s*\(\*\)\s*\(\w*\)\)\s*(-?\d+)/$1/; # (int (*)(foo_t))0
if (s/^\(([\w,\s]*)\)//) {
$args = $1;
my $proto = '() ';
if ($args ne '') {
$proto = '';
- foreach $arg (split(/,\s*/,$args)) {
+ foreach my $arg (split(/,\s*/,$args)) {
$arg =~ s/^\s*([^\s].*[^\s])\s*$/$1/;
$curargs{$arg} = 1;
}
$args = reindent($args);
if ($t ne '') {
$new =~ s/(['\\])/\\$1/g; #']);
+
if ($opt_h) {
print OUT $t,"eval \"\\n#line $eval_index $outfile\\n\" . 'sub $name () {",$new,";}' unless defined(\&$name);\n";
$eval_index++;
print OUT $t,"eval 'sub $name () {",$new,";}' unless defined(\&$name);\n";
}
} else {
+ # Shunt around such directives as `#define FOO FOO':
+ next if " \&$name" eq $new;
+
print OUT $t,"unless(defined(\&$name)) {\n sub $name () {\t",$new,";}\n}\n";
}
}
print OUT $t,"require '$incl';\n";
} elsif(/^include_next\s*[<"](.*)[>"]/) {
($incl = $1) =~ s/\.h$/.ph/;
- # should've read up on #include_next properly before attempting
- # to implement it...
- #
- #print OUT $t, "{\n";
- #$tab += 4;
- #$t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
- #print OUT $t, "my(\$INC) = shift(\@INC);\n";
- #print OUT $t, "require '$incl';\n";
- #print OUT $t, "unshift(\@INC, \$INC);}\n";
- #$tab -= 4;
- #$t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
- #print OUT $t, "}\n";
- #
- # try this instead:
- print OUT ($t, "my(\$i) = 0;\n");
- print OUT ($t, "if(exists(\$INC{$incl})) {\n");
- $tab += 4;
- $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
- print OUT ($t, "++\$i while (\$i <= \$#INC",
- " and \$INC[\$i].'/$incl' ne \$INC{'$incl'});\n");
- print OUT ($t, "\$i = 0 if \$INC[\$i].'/$incl' ne",
- " \$INC{'$incl'};\n");
- $tab -= 4;
- $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
- print OUT ($t, "}\n");
print OUT ($t,
- "eval(\"require '\" . ",
- "(\$i ? \$INC[\$i].'/' : '') . \"\$incl';\");");
- # any better? require is smart enough not to try and include a
- # file twice, i believe, so require-ing the same actual file
- # should end up just being a null operation...
+ "eval {\n");
+ $tab += 4;
+ $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
+ print OUT ($t,
+ "my(\%INCD) = map { \$INC{\$_} => 1 } ",
+ "(grep { \$_ eq \"$incl\" } keys(\%INC));\n");
+ print OUT ($t,
+ "my(\@REM) = map { \"\$_/$incl\" } ",
+ "(grep { not exists(\$INCD{\"\$_/$incl\"})",
+ "and -f \"\$_/$incl\" } \@INC);\n");
+ print OUT ($t,
+ "require \"\$REM[0]\" if \@REM;\n");
+ $tab -= 4;
+ $t = "\t" x ($tab / 8) . ' ' x ($tab % 8);
+ print OUT ($t,
+ "};\n");
+ print OUT ($t,
+ "warn(\$\@) if \$\@;\n");
} elsif (/^ifdef\s+(\w+)/) {
print OUT $t,"if(defined(&$1)) {\n";
$tab += 4;
print OUT $t,"}\n";
} elsif(/^undef\s+(\w+)/) {
print OUT $t, "undef(&$1) if defined(&$1);\n";
+ } elsif(/^error\s+(".*")/) {
+ print OUT $t, "die($1);\n";
} elsif(/^error\s+(.*)/) {
- print OUT $t, "die(\"$1\");\n";
+ print OUT $t, "die(\"", quotemeta($1), "\");\n";
} elsif(/^warning\s+(.*)/) {
- print OUT $t, "warn(\"$1\");\n";
+ print OUT $t, "warn(\"", quotemeta($1), "\");\n";
} elsif(/^ident\s+(.*)/) {
print OUT $t, "# $1\n";
}
+ } elsif(/^\s*(typedef\s*)?enum\s*(\s+[a-zA-Z_]\w*\s*)?/) {
+ until(/\{[^}]*\}.*;/ || /;/) {
+ last unless defined ($next = next_line($file));
+ chomp $next;
+ # drop "#define FOO FOO" in enums
+ $next =~ s/^\s*#\s*define\s+(\w+)\s+\1\s*$//;
+ $_ .= $next;
+ print OUT "# $next\n" if $opt_D;
+ }
+ s/#\s*if.*?#\s*endif//g; # drop #ifdefs
+ s@/\*.*?\*/@@g;
+ s/\s+/ /g;
+ next unless /^\s?(typedef\s?)?enum\s?([a-zA-Z_]\w*)?\s?\{(.*)\}\s?([a-zA-Z_]\w*)?\s?;/;
+ (my $enum_subs = $3) =~ s/\s//g;
+ my @enum_subs = split(/,/, $enum_subs);
+ my $enum_val = -1;
+ foreach my $enum (@enum_subs) {
+ my ($enum_name, $enum_value) = $enum =~ /^([a-zA-Z_]\w*)(=.+)?$/;
+ $enum_value =~ s/^=//;
+ $enum_val = (length($enum_value) ? $enum_value : $enum_val + 1);
+ if ($opt_h) {
+ print OUT ($t,
+ "eval(\"\\n#line $eval_index $outfile\\n",
+ "sub $enum_name () \{ $enum_val; \}\") ",
+ "unless defined(\&$enum_name);\n");
+ ++ $eval_index;
+ } else {
+ print OUT ($t,
+ "eval(\"sub $enum_name () \{ $enum_val; \}\") ",
+ "unless defined(\&$enum_name);\n");
+ }
+ }
}
}
print OUT "1;\n";
+
+ $Is_converted{$file} = 1;
+ queue_includes_from($file) if ($opt_a);
}
exit $Exit;
+
sub reindent($) {
my($text) = shift;
$text =~ s/\n/\n /g;
$text;
}
+
sub expr {
+ my $joined_args;
if(keys(%curargs)) {
- my($joined_args) = join('|', keys(%curargs));
+ $joined_args = join('|', keys(%curargs));
}
while ($_ ne '') {
s/^\&\&// && do { $new .= " &&"; next;}; # handle && operator
s/^\&([\(a-z\)]+)/$1/i; # hack for things that take the address of
s/^(\s+)// && do {$new .= ' '; next;};
- s/^(0X[0-9A-F]+)[UL]*//i && do {$new .= lc($1); next;};
- s/^(-?\d+\.\d+E[-+]\d+)F?//i && do {$new .= $1; next;};
+ s/^0X([0-9A-F]+)[UL]*//i
+ && do {my $hex = $1;
+ $hex =~ s/^0+//;
+ if (length $hex > 8 && !$Config{use64bitint}) {
+ # Croak if nv_preserves_uv_bits < 64 ?
+ $new .= hex(substr($hex, -8)) +
+ 2**32 * hex(substr($hex, 0, -8));
+ # The above will produce "errorneus" code
+ # if the hex constant was e.g. inside UINT64_C
+ # macro, but then again, h2ph is an approximation.
+ } else {
+ $new .= lc("0x$hex");
+ }
+ next;};
+ s/^(-?\d+\.\d+E[-+]?\d+)[FL]?//i && do {$new .= $1; next;};
s/^(\d+)\s*[LU]*//i && do {$new .= $1; next;};
s/^("(\\"|[^"])*")// && do {$new .= $1; next;};
s/^'((\\"|[^"])*)'// && do {
# Eliminate typedefs
/\(([\w\s]+)[\*\s]*\)\s*[\w\(]/ && do {
foreach (split /\s+/, $1) { # Make sure all the words are types,
- last unless ($isatype{$_} or $_ eq 'struct');
+ last unless ($isatype{$_} or $_ eq 'struct' or $_ eq 'union');
}
s/\([\w\s]+[\*\s]*\)// && next; # then eliminate them.
};
# struct/union member, including arrays:
s/^([_A-Z]\w*(\[[^\]]+\])?((\.|->)[_A-Z]\w*(\[[^\]]+\])?)+)//i && do {
- $id = $1;
+ my $id = $1;
$id =~ s/(\.|(->))([^\.\-]*)/->\{$3\}/g;
$id =~ s/\b([^\$])($joined_args)/$1\$$2/g if length($joined_args);
while($id =~ /\[\s*([^\$\&\d\]]+)\]/) {
$new .= " (\$$id)";
};
s/^([_a-zA-Z]\w*)// && do {
- $id = $1;
- if ($id eq 'struct') {
+ my $id = $1;
+ if ($id eq 'struct' || $id eq 'union') {
s/^\s+(\w+)//;
$id .= ' ' . $1;
$isatype{$id} = 1;
$new .= '->' if /^[\[\{]/;
} elsif ($id eq 'defined') {
$new .= 'defined';
- } elsif (/^\(/) {
- s/^\((\w),/("$1",/ if $id =~ /^_IO[WR]*$/i; # cheat
+ } elsif (/^\s*\(/) {
+ s/^\s*\((\w),/("$1",/ if $id =~ /^_IO[WR]*$/i; # cheat
$new .= " &$id";
} elsif ($isatype{$id}) {
if ($new =~ /{\s*$/) {
}
+sub next_line
+{
+ my $file = shift;
+ my ($in, $out);
+ my $pre_sub_tri_graphs = 1;
+
+ READ: while (not eof IN) {
+ $in .= <IN>;
+ chomp $in;
+ next unless length $in;
+
+ while (length $in) {
+ if ($pre_sub_tri_graphs) {
+ # Preprocess all tri-graphs
+ # including things stuck in quoted string constants.
+ $in =~ s/\?\?=/#/g; # | ??=| #|
+ $in =~ s/\?\?\!/|/g; # | ??!| ||
+ $in =~ s/\?\?'/^/g; # | ??'| ^|
+ $in =~ s/\?\?\(/[/g; # | ??(| [|
+ $in =~ s/\?\?\)/]/g; # | ??)| ]|
+ $in =~ s/\?\?\-/~/g; # | ??-| ~|
+ $in =~ s/\?\?\//\\/g; # | ??/| \|
+ $in =~ s/\?\?</{/g; # | ??<| {|
+ $in =~ s/\?\?>/}/g; # | ??>| }|
+ }
+ if ($in =~ /^\#ifdef __LANGUAGE_PASCAL__/) {
+ # Tru64 disassembler.h evilness: mixed C and Pascal.
+ while (<IN>) {
+ last if /^\#endif/;
+ }
+ next READ;
+ }
+ if ($in =~ /^extern inline / && # Inlined assembler.
+ $^O eq 'linux' && $file =~ m!(?:^|/)asm/[^/]+\.h$!) {
+ while (<IN>) {
+ last if /^}/;
+ }
+ next READ;
+ }
+ if ($in =~ s/\\$//) { # \-newline
+ $out .= ' ';
+ next READ;
+ } elsif ($in =~ s/^([^"'\\\/]+)//) { # Passthrough
+ $out .= $1;
+ } elsif ($in =~ s/^(\\.)//) { # \...
+ $out .= $1;
+ } elsif ($in =~ s/^('(\\.|[^'\\])*')//) { # '...
+ $out .= $1;
+ } elsif ($in =~ s/^("(\\.|[^"\\])*")//) { # "...
+ $out .= $1;
+ } elsif ($in =~ s/^\/\/.*//) { # //...
+ # fall through
+ } elsif ($in =~ m/^\/\*/) { # /*...
+ # C comment removal adapted from perlfaq6:
+ if ($in =~ s/^\/\*[^*]*\*+([^\/*][^*]*\*+)*\///) {
+ $out .= ' ';
+ } else { # Incomplete /* */
+ next READ;
+ }
+ } elsif ($in =~ s/^(\/)//) { # /...
+ $out .= $1;
+ } elsif ($in =~ s/^([^\'\"\\\/]+)//) {
+ $out .= $1;
+ } elsif ($^O eq 'linux' &&
+ $file =~ m!(?:^|/)linux/byteorder/pdp_endian\.h$! &&
+ $in =~ s!\'T KNOW!!) {
+ $out =~ s!I DON$!I_DO_NOT_KNOW!;
+ } else {
+ die "Cannot parse:\n$in\n";
+ }
+ }
+
+ last READ if $out =~ /\S/;
+ }
+
+ return $out;
+}
+
+
# Handle recursive subdirectories without getting a grotesquely big stack.
# Could this be implemented using File::Find?
sub next_file
} else {
print STDERR "Skipping directory `$file'\n";
}
- } else {
+ } elsif ($opt_a) {
+ return $file;
+ } else {
print STDERR "Skipping `$file': not a file or directory\n";
}
}
# expand_glob() is going to be called until $ARGV[0] isn't a
# directory; so push directories, and unshift everything else.
- if (-d "$directory/$_") {
- push @ARGV, "$directory/$_";
- } else {
- unshift @ARGV, "$directory/$_";
- }
+ if (-d "$directory/$_") { push @ARGV, "$directory/$_" }
+ else { unshift @ARGV, "$directory/$_" }
}
closedir DIR;
}
unlink "$Dest_dir/$dirlink" or
print STDERR "Could not remove link $Dest_dir/$dirlink: $!\n";
}
+
if (eval 'symlink($target, "$Dest_dir/$dirlink")') {
print "Linking $target -> $Dest_dir/$dirlink\n";
# Make sure that the link _links_ to something:
if (! -e "$Dest_dir/$target") {
- mkdir("$Dest_dir/$target", 0755) or
+ mkpath("$Dest_dir/$target", 0755) or
print STDERR "Could not create $Dest_dir/$target/\n";
}
} else {
}
+# Push all #included files in $file onto our stack, except for STDIN
+# and files we've already processed.
+sub queue_includes_from
+{
+ my ($file) = @_;
+ my $line;
+
+ return if ($file eq "-");
+
+ open HEADER, $file or return;
+ while (defined($line = <HEADER>)) {
+ while (/\\$/) { # Handle continuation lines
+ chop $line;
+ $line .= <HEADER>;
+ }
+
+ if ($line =~ /^#\s*include\s+<(.*?)>/) {
+ push(@ARGV, $1) unless $Is_converted{$1};
+ }
+ }
+ close HEADER;
+}
+
+
+# Determine include directories; $Config{usrinc} should be enough for (all
+# non-GCC?) C compilers, but gcc uses an additional include directory.
+sub inc_dirs
+{
+ my $from_gcc = `$Config{cc} -v 2>&1`;
+ $from_gcc =~ s:^Reading specs from (.*?)/specs\b.*:$1/include:s;
+
+ length($from_gcc) ? ($from_gcc, $Config{usrinc}) : ($Config{usrinc});
+}
+
+
+# Create "_h2ph_pre.ph", if it doesn't exist or was built by a different
+# version of h2ph.
+sub build_preamble_if_necessary
+{
+ # Increment $VERSION every time this function is modified:
+ my $VERSION = 2;
+ my $preamble = "$Dest_dir/_h2ph_pre.ph";
+
+ # Can we skip building the preamble file?
+ if (-r $preamble) {
+ # Extract version number from first line of preamble:
+ open PREAMBLE, $preamble or die "Cannot open $preamble: $!";
+ my $line = <PREAMBLE>;
+ $line =~ /(\b\d+\b)/;
+ close PREAMBLE or die "Cannot close $preamble: $!";
+
+ # Don't build preamble if a compatible preamble exists:
+ return if $1 == $VERSION;
+ }
+
+ my (%define) = _extract_cc_defines();
+
+ open PREAMBLE, ">$preamble" or die "Cannot open $preamble: $!";
+ print PREAMBLE "# This file was created by h2ph version $VERSION\n";
+
+ foreach (sort keys %define) {
+ if ($opt_D) {
+ print PREAMBLE "# $_=$define{$_}\n";
+ }
+
+ if ($define{$_} =~ /^(\d+)U?L{0,2}$/i) {
+ print PREAMBLE
+ "unless (defined &$_) { sub $_() { $1 } }\n\n";
+ } elsif ($define{$_} =~ /^\w+$/) {
+ print PREAMBLE
+ "unless (defined &$_) { sub $_() { &$define{$_} } }\n\n";
+ } else {
+ print PREAMBLE
+ "unless (defined &$_) { sub $_() { \"",
+ quotemeta($define{$_}), "\" } }\n\n";
+ }
+ }
+ close PREAMBLE or die "Cannot close $preamble: $!";
+}
+
+
+# %Config contains information on macros that are pre-defined by the
+# system's compiler. We need this information to make the .ph files
+# function with perl as the .h files do with cc.
+sub _extract_cc_defines
+{
+ my %define;
+ my $allsymbols = join " ",
+ @Config{'ccsymbols', 'cppsymbols', 'cppccsymbols'};
+
+ # Split compiler pre-definitions into `key=value' pairs:
+ foreach (split /\s+/, $allsymbols) {
+ /(.+?)=(.+)/ and $define{$1} = $2;
+
+ if ($opt_D) {
+ print STDERR "$_: $1 -> $2\n";
+ }
+ }
+
+ return %define;
+}
+
+
1;
##############################################################################
=head1 SYNOPSIS
-B<h2ph [-d destination directory] [-r] [-l] [headerfiles]>
+B<h2ph [-d destination directory] [-r | -a] [-l] [headerfiles]>
=head1 DESCRIPTION
or
+ cd /usr/include; h2ph * sys/* arpa/* netinet/*
+
+or
+
cd /usr/include; h2ph -r -l .
The output files are placed in the hierarchy rooted at Perl's
=item -r
Run recursively; if any of B<headerfiles> are directories, then run I<h2ph>
-on all files in those directories (and their subdirectories, etc.).
+on all files in those directories (and their subdirectories, etc.). B<-r>
+and B<-a> are mutually exclusive.
+
+=item -a
+
+Run automagically; convert B<headerfiles>, as well as any B<.h> files
+which they include. This option will search for B<.h> files in all
+directories which your C compiler ordinarily uses. B<-a> and B<-r> are
+mutually exclusive.
=item -l
However, the B<.ph> files almost double in size when built using B<-h>.
+=item -D
+
+Include the code from the B<.h> file as a comment in the B<.ph> file.
+This is primarily used for debugging I<h2ph>.
+
+=item -Q
+
+``Quiet'' mode; don't print out the names of the files being converted.
+
=back
=head1 ENVIRONMENT
It's only intended as a rough tool.
You may need to dicker with the files produced.
+You have to run this program by hand; it's not run as part of the Perl
+installation.
+
+Doesn't handle complicated expressions built piecemeal, a la:
+
+ enum {
+ FIRST_VALUE,
+ SECOND_VALUE,
+ #ifdef ABC
+ THIRD_VALUE
+ #endif
+ };
+
+Doesn't necessarily locate all of your C compiler's internally-defined
+symbols.
+
=cut
!NO!SUBS!
close OUT or die "Can't close $file: $!";
chmod 0755, $file or die "Can't reset permissions for $file: $!\n";
exec("$Config{'eunicefix'} $file") if $Config{'eunicefix'} ne ':';
+chdir $origdir;