X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/b7bcf49446150838af3c7cd1dec335ba45a2fc6b..00f6437b30bbbd24904cbecfa1c00ab91f7315e5:/utils/h2ph.PL diff --git a/utils/h2ph.PL b/utils/h2ph.PL index f647831..a2d737b 100644 --- a/utils/h2ph.PL +++ b/utils/h2ph.PL @@ -42,8 +42,13 @@ use Config; use File::Path qw(mkpath); use Getopt::Std; -getopts('Dd:rlhaQ'); -use vars qw($opt_D $opt_d $opt_r $opt_l $opt_h $opt_a $opt_Q); +# Make sure read permissions for all are set: +if (defined umask && (umask() & 0444)) { + umask (umask() & ~0444); +} + +getopts('Dd:rlhaQe'); +use vars qw($opt_D $opt_d $opt_r $opt_l $opt_h $opt_a $opt_Q $opt_e); die "-r and -a options are mutually exclusive\n" if ($opt_r and $opt_a); my @inc_dirs = inc_dirs() if $opt_a; @@ -53,25 +58,34 @@ my $Dest_dir = $opt_d || $Config{installsitearch}; die "Destination directory $Dest_dir doesn't exist or isn't a directory\n" unless -d $Dest_dir; -my @isatype = split(' ',<$Dest_dir/$outfile") || die "Can't create $outfile: $!\n"; } - print OUT "require '_h2ph_pre.ph';\n\n"; - while (defined (local $_ = next_line())) { + print OUT + "require '_h2ph_pre.ph';\n\n", + "no warnings qw(redefine misc);\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 = '() '; + my $proto = '() '; if ($args ne '') { - $proto = ''; + $proto = ''; foreach my $arg (split(/,\s*/,$args)) { $arg =~ s/^\s*([^\s].*[^\s])\s*$/$1/; $curargs{$arg} = 1; } $args =~ s/\b(\w)/\$$1/g; - $args = "local($args) = \@_;\n$t "; + $args = "my($args) = \@_;\n$t "; } s/^\s+//; expr(); $new =~ s/(["\\])/\\$1/g; #"]); - $new = reindent($new); - $args = reindent($args); - if ($t ne '') { - $new =~ s/(['\\])/\\$1/g; #']); - if ($opt_h) { - print OUT $t, - "eval \"\\n#line $eval_index $outfile\\n\" . 'sub $name $proto\{\n$t ${args}eval q($new);\n$t}' unless defined(\&$name);\n"; - $eval_index++; - } else { - print OUT $t, - "eval 'sub $name $proto\{\n$t ${args}eval q($new);\n$t}' unless defined(\&$name);\n"; - } - } else { - print OUT "unless(defined(\&$name)) {\n sub $name $proto\{\n\t${args}eval q($new);\n }\n}\n"; - } - %curargs = (); + EMIT($proto); } else { s/^\s+//; expr(); + $new = 1 if $new eq ''; + + # Shunt around such directives as '#define FOO FOO': + next if $new =~ /^\s*&\Q$name\E\s*\z/; + $new = reindent($new); $args = reindent($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++; - } else { - 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; + $new =~ s/(['\\])/\\$1/g; #']); - print OUT $t,"unless(defined(\&$name)) {\n sub $name () {\t",$new,";}\n}\n"; + print OUT $t, 'eval '; + if ($opt_h) { + print OUT "\"\\n#line $eval_index $outfile\\n\" . "; + $eval_index++; } + print OUT "'sub $name () {$new;}' unless defined(&$name);\n"; } - } elsif (/^(include|import)\s*[<"](.*)[>"]/) { - ($incl = $2) =~ s/\.h$/.ph/; - print OUT $t,"require '$incl';\n"; - } elsif(/^include_next\s*[<"](.*)[>"]/) { - ($incl = $1) =~ s/\.h$/.ph/; + } elsif (/^(include|import|include_next)\s*([<\"])(.*)[>\"]/) { + $incl_type = $1; + $incl_quote = $2; + $incl = $3; + if (($incl_type eq 'include_next') || + ($opt_e && exists($bad_file{$incl}))) { + $incl =~ s/\.h$/.ph/; print OUT ($t, "eval {\n"); $tab += 4; $t = "\t" x ($tab / 8) . ' ' x ($tab % 8); + print OUT ($t, "my(\@REM);\n"); + if ($incl_type eq 'include_next') { print OUT ($t, "my(\%INCD) = map { \$INC{\$_} => 1 } ", - "(grep { \$_ eq \"$incl\" } keys(\%INC));\n"); + "(grep { \$_ eq \"$incl\" } ", + "keys(\%INC));\n"); print OUT ($t, - "my(\@REM) = map { \"\$_/$incl\" } ", + "\@REM = map { \"\$_/$incl\" } ", "(grep { not exists(\$INCD{\"\$_/$incl\"})", - "and -f \"\$_/$incl\" } \@INC);\n"); + " and -f \"\$_/$incl\" } \@INC);\n"); + } else { + print OUT ($t, + "\@REM = map { \"\$_/$incl\" } ", + "(grep {-r \"\$_/$incl\" } \@INC);\n"); + } print OUT ($t, "require \"\$REM[0]\" if \@REM;\n"); $tab -= 4; @@ -191,6 +202,14 @@ while (defined (my $file = next_file())) { "};\n"); print OUT ($t, "warn(\$\@) if \$\@;\n"); + } else { + $incl =~ s/\.h$/.ph/; + # copy the prefix in the quote syntax (#include "x.h") case + if ($incl !~ m|/| && $incl_quote eq q{"} && $file =~ m|^(.*)/|) { + $incl = "$1/$incl"; + } + print OUT $t,"require '$incl';\n"; + } } elsif (/^ifdef\s+(\w+)/) { print OUT $t,"if(defined(&$1)) {\n"; $tab += 4; @@ -238,20 +257,27 @@ while (defined (my $file = next_file())) { } elsif(/^ident\s+(.*)/) { print OUT $t, "# $1\n"; } - } elsif(/^\s*(typedef\s*)?enum\s*(\s+[a-zA-Z_]\w*\s*)?\{/) { - until(/\}.*?;/) { - chomp($next = ); + } elsif (/^\s*(typedef\s*)?enum\s*(\s+[a-zA-Z_]\w*\s*)?/) { # { for vi + until(/\{[^}]*\}.*;/ || /;/) { + last unless defined ($next = next_line($file)); + chomp $next; + # drop "#define FOO FOO" in enums + $next =~ s/^\s*#\s*define\s+(\w+)\s+\1\s*$//; + # #defines in enums (aliases) + $next =~ s/^\s*#\s*define\s+(\w+)\s+(\w+)\s*$/$1 = $2,/; $_ .= $next; print OUT "# $next\n" if $opt_D; } + s/#\s*if.*?#\s*endif//g; # drop #ifdefs s@/\*.*?\*/@@g; s/\s+/ /g; - /^\s?(typedef\s?)?enum\s?([a-zA-Z_]\w*)?\s?\{(.*)\}\s?([a-zA-Z_]\w*)?\s?;/; + 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_name or next; $enum_value =~ s/^=//; $enum_val = (length($enum_value) ? $enum_value : $enum_val + 1); if ($opt_h) { @@ -266,26 +292,123 @@ while (defined (my $file = next_file())) { "unless defined(\&$enum_name);\n"); } } + } elsif (/^(?:__extension__\s+)?(?:extern|static)\s+(?:__)?inline(?:__)?\s+/ + and !/;\s*$/ and !/{\s*}\s*$/) + { # { for vi + # This is a hack to parse the inline functions in the glibc headers. + # Warning: massive kludge ahead. We suppose inline functions + # are mainly constructed like macros. + while (1) { + last unless defined ($next = next_line($file)); + chomp $next; + undef $_, last if $next =~ /__THROW\s*;/ + or $next =~ /^(__extension__|extern|static)\b/; + $_ .= " $next"; + print OUT "# $next\n" if $opt_D; + last if $next =~ /^}|^{.*}\s*$/; + } + next if not defined; # because it's only a prototype + s/\b(__extension__|extern|static|(?:__)?inline(?:__)?)\b//g; + # violently drop #ifdefs + s/#\s*if.*?#\s*endif//g + and print OUT "# some #ifdef were dropped here -- fill in the blanks\n"; + if (s/^(?:\w|\s|\*)*\s(\w+)\s*//) { + $name = $1; + } else { + warn "name not found"; next; # shouldn't occur... + } + my @args; + if (s/^\(([^()]*)\)\s*(\w+\s*)*//) { + for my $arg (split /,/, $1) { + if ($arg =~ /(\w+)\s*$/) { + $curargs{$1} = 1; + push @args, $1; + } + } + } + $args = ( + @args + ? "my(" . (join ',', map "\$$_", @args) . ") = \@_;\n$t " + : "" + ); + my $proto = @args ? '' : '() '; + $new = ''; + s/\breturn\b//g; # "return" doesn't occur in macros usually... + expr(); + # try to find and perlify local C variables + our @local_variables = (); # needs to be a our(): (?{...}) bug workaround + { + use re "eval"; + my $typelist = join '|', keys %isatype; + $new =~ s[' + (?:(?:__)?const(?:__)?\s+)? + (?:(?:un)?signed\s+)? + (?:long\s+)? + (?:$typelist)\s+ + (\w+) + (?{ push @local_variables, $1 }) + '] + [my \$$1]gx; + $new =~ s[' + (?:(?:__)?const(?:__)?\s+)? + (?:(?:un)?signed\s+)? + (?:long\s+)? + (?:$typelist)\s+ + ' \s+ &(\w+) \s* ; + (?{ push @local_variables, $1 }) + ] + [my \$$1;]gx; + } + $new =~ s/&$_\b/\$$_/g for @local_variables; + $new =~ s/(["\\])/\\$1/g; #"]); + # now that's almost like a macro (we hope) + EMIT($proto); } } - print OUT "1;\n"; - $Is_converted{$file} = 1; - queue_includes_from($file) if ($opt_a); + if ($opt_e && exists($bad_file{$file})) { + unlink($Dest_dir . '/' . $outfile); + $next = ''; + } else { + print OUT "1;\n"; + queue_includes_from($file) if $opt_a; + } } -exit $Exit; +if ($opt_e && (scalar(keys %bad_file) > 0)) { + warn "Was unable to convert the following files:\n"; + warn "\t" . join("\n\t",sort(keys %bad_file)) . "\n"; +} +exit $Exit; -sub reindent($) { - my($text) = shift; - $text =~ s/\n/\n /g; - $text =~ s/ /\t/g; - $text; +sub EMIT { + my $proto = shift; + + $new = reindent($new); + $args = reindent($args); + if ($t ne '') { + $new =~ s/(['\\])/\\$1/g; #']); + if ($opt_h) { + print OUT $t, + "eval \"\\n#line $eval_index $outfile\\n\" . 'sub $name $proto\{\n$t ${args}eval q($new);\n$t}' unless defined(\&$name);\n"; + $eval_index++; + } else { + print OUT $t, + "eval 'sub $name $proto\{\n$t ${args}eval q($new);\n$t}' unless defined(\&$name);\n"; + } + } else { + print OUT "unless(defined(\&$name)) {\n sub $name $proto\{\n\t${args}eval q($new);\n }\n}\n"; + } + %curargs = (); + return; } - sub expr { + if (/\b__asm__\b/) { # freak out + $new = '"(assembly code)"'; + return + } my $joined_args; if(keys(%curargs)) { $joined_args = join('|', keys(%curargs)); @@ -294,8 +417,21 @@ sub expr { 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 "erroneous" 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 { @@ -333,10 +469,16 @@ sub expr { }; # Eliminate typedefs /\(([\w\s]+)[\*\s]*\)\s*[\w\(]/ && do { + my $doit = 1; foreach (split /\s+/, $1) { # Make sure all the words are types, - last unless ($isatype{$_} or $_ eq 'struct'); + unless($isatype{$_} or $_ eq 'struct' or $_ eq 'union'){ + $doit = 0; + last; + } + } + if( $doit ){ + s/\([\w\s]+[\*\s]*\)// && next; # then eliminate them. } - s/\([\w\s]+[\*\s]*\)// && next; # then eliminate them. }; # struct/union member, including arrays: s/^([_A-Z]\w*(\[[^\]]+\])?((\.|->)[_A-Z]\w*(\[[^\]]+\])?)+)//i && do { @@ -357,7 +499,7 @@ sub expr { }; s/^([_a-zA-Z]\w*)// && do { my $id = $1; - if ($id eq 'struct') { + if ($id eq 'struct' || $id eq 'union') { s/^\s+(\w+)//; $id .= ' ' . $1; $isatype{$id} = 1; @@ -374,7 +516,7 @@ sub expr { s/^\s*\((\w),/("$1",/ if $id =~ /^_IO[WR]*$/i; # cheat $new .= " &$id"; } elsif ($isatype{$id}) { - if ($new =~ /{\s*$/) { + if ($new =~ /\{\s*$/) { $new .= "'$id'"; } elsif ($new =~ /\(\s*$/ && /^[\s*]*\)/) { $new =~ s/\(\s*$//; @@ -383,8 +525,14 @@ sub expr { $new .= q(').$id.q('); } } else { - if ($inif && $new !~ /defined\s*\($/) { - $new .= '(defined(&' . $id . ') ? &' . $id . ' : 0)'; + if ($inif) { + if ($new =~ /defined\s*$/) { + $new .= '(&' . $id . ')'; + } elsif ($new =~ /defined\s*\($/) { + $new .= '&' . $id; + } else { + $new .= '(defined(&' . $id . ') ? &' . $id . ' : undef)'; + } } elsif (/^\[/) { $new .= " \$$id"; } else { @@ -400,6 +548,7 @@ sub expr { sub next_line { + my $file = shift; my ($in, $out); my $pre_sub_tri_graphs = 1; @@ -410,7 +559,7 @@ sub next_line while (length $in) { if ($pre_sub_tri_graphs) { - # Preprocess all tri-graphs + # Preprocess all tri-graphs # including things stuck in quoted string constants. $in =~ s/\?\?=/#/g; # | ??=| #| $in =~ s/\?\?\!/|/g; # | ??!| || @@ -422,6 +571,22 @@ sub next_line $in =~ s/\?\?/}/g; # | ??>| }| } + if ($in =~ /^\#ifdef __LANGUAGE_PASCAL__/) { + # Tru64 disassembler.h evilness: mixed C and Pascal. + while () { + last if /^\#endif/; + } + $in = ""; + next READ; + } + if ($in =~ /^extern inline / && # Inlined assembler. + $^O eq 'linux' && $file =~ m!(?:^|/)asm/[^/]+\.h$!) { + while () { + last if /^}/; + } + $in = ""; + next READ; + } if ($in =~ s/\\$//) { # \-newline $out .= ' '; next READ; @@ -429,12 +594,20 @@ sub next_line $out .= $1; } elsif ($in =~ s/^(\\.)//) { # \... $out .= $1; - } elsif ($in =~ s/^('(\\.|[^'\\])*')//) { # '... - $out .= $1; - } elsif ($in =~ s/^("(\\.|[^"\\])*")//) { # "... - $out .= $1; + } elsif ($in =~ /^'/) { # '... + if ($in =~ s/^('(\\.|[^'\\])*')//) { + $out .= $1; + } else { + next READ; + } + } elsif ($in =~ /^"/) { # "... + if ($in =~ s/^("(\\.|[^"\\])*")//) { + $out .= $1; + } else { + next READ; + } } elsif ($in =~ s/^\/\/.*//) { # //... - last READ; + # fall through } elsif ($in =~ m/^\/\*/) { # /*... # C comment removal adapted from perlfaq6: if ($in =~ s/^\/\*[^*]*\*+([^\/*][^*]*\*+)*\///) { @@ -446,12 +619,24 @@ sub next_line $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$in\n"; + if ($opt_e) { + warn "Cannot parse $file:\n$in\n"; + $bad_file{$file} = 1; + $in = ''; + $out = undef; + last READ; + } else { + die "Cannot parse:\n$in\n"; + } } } - last READ; + last READ if $out =~ /\S/; } return $out; @@ -473,12 +658,12 @@ sub next_file if ($opt_r) { expand_glob($file); } else { - print STDERR "Skipping directory `$file'\n"; + print STDERR "Skipping directory '$file'\n"; } } elsif ($opt_a) { return $file; } else { - print STDERR "Skipping `$file': not a file or directory\n"; + print STDERR "Skipping '$file': not a file or directory\n"; } } @@ -557,8 +742,13 @@ sub queue_includes_from $line .=
; } - if ($line =~ /^#\s*include\s+<(.*?)>/) { - push(@ARGV, $1) unless $Is_converted{$1}; + if ($line =~ /^#\s*include\s+([<"])(.*?)[>"]/) { + my ($delimiter, $new_file) = ($1, $2); + # copy the prefix in the quote syntax (#include "x.h") case + if ($delimiter eq q{"} && $file =~ m|^(.*)/|) { + $new_file = "$1/$new_file"; + } + push(@ARGV, $new_file) unless $Is_converted{$new_file}; } } close HEADER; @@ -566,13 +756,11 @@ sub queue_includes_from # Determine include directories; $Config{usrinc} should be enough for (all -# non-GCC?) C compilers, but gcc uses an additional include directory. +# non-GCC?) C compilers, but gcc uses additional include directories. 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}); + my $from_gcc = `LC_ALL=C $Config{cc} -v -E - < /dev/null 2>&1 | awk '/^#include/, /^End of search list/' | grep '^ '`; + length($from_gcc) ? (split(' ', $from_gcc), $Config{usrinc}) : ($Config{usrinc}); } @@ -581,7 +769,7 @@ sub inc_dirs sub build_preamble_if_necessary { # Increment $VERSION every time this function is modified: - my $VERSION = 2; + my $VERSION = 3; my $preamble = "$Dest_dir/_h2ph_pre.ph"; # Can we skip building the preamble file? @@ -599,25 +787,49 @@ sub build_preamble_if_necessary 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"; - } + print PREAMBLE "# This file was created by h2ph version $VERSION\n"; - if ($define{$_} =~ /^\d+$/) { - print PREAMBLE - "unless (defined &$_) { sub $_() { $define{$_} } }\n\n"; - } elsif ($define{$_} =~ /^\w+$/) { - print PREAMBLE - "unless (defined &$_) { sub $_() { &$define{$_} } }\n\n"; - } else { - print PREAMBLE - "unless (defined &$_) { sub $_() { \"", - quotemeta($define{$_}), "\" } }\n\n"; - } - } + foreach (sort keys %define) { + if ($opt_D) { + print PREAMBLE "# $_=$define{$_}\n"; + } + if ($define{$_} =~ /^\((.*)\)$/) { + # parenthesized value: d=(v) + $define{$_} = $1; + } + if (/^(\w+)\((\w)\)$/) { + my($macro, $arg) = ($1, $2); + my $def = $define{$_}; + $def =~ s/$arg/\$\{$arg\}/g; + print PREAMBLE < $2\n"; - } + # Split compiler pre-definitions into 'key=value' pairs: + while ($allsymbols =~ /([^\s]+)=((\\\s|[^\s])+)/g) { + $define{$1} = $2; + if ($opt_D) { + print STDERR "$_: $1 -> $2\n"; + } } return %define; @@ -668,6 +879,10 @@ It is most easily run while in /usr/include: or + cd /usr/include; h2ph * sys/* arpa/* netinet/* + +or + cd /usr/include; h2ph -r -l . The output files are placed in the hierarchy rooted at Perl's @@ -683,7 +898,7 @@ If run with no arguments, filters standard input to standard output. =item -d destination_dir Put the resulting B<.ph> files beneath B, instead of -beneath the default Perl library location (C<$Config{'installsitsearch'}>). +beneath the default Perl library location (C<$Config{'installsitearch'}>). =item -r @@ -705,7 +920,7 @@ is not specified, then links are skipped over. =item -h -Put ``hints'' in the .ph files which will help in locating problems with +Put 'hints' in the .ph files which will help in locating problems with I. In those cases when you B a B<.ph> file containing syntax errors, instead of the cryptic @@ -724,7 +939,7 @@ This is primarily used for debugging I. =item -Q -``Quiet'' mode; don't print out the names of the files being converted. +'Quiet' mode; don't print out the names of the files being converted. =back @@ -768,10 +983,10 @@ installation. Doesn't handle complicated expressions built piecemeal, a la: enum { - FIRST_VALUE, - SECOND_VALUE, + FIRST_VALUE, + SECOND_VALUE, #ifdef ABC - THIRD_VALUE + THIRD_VALUE #endif };