linux/scripts/get_abi.pl
Finn Behrens c25ce589dc tweewide: Fix most Shebang lines
Change every shebang which does not need an argument to use /usr/bin/env.
This is needed as not every distro has everything under /usr/bin,
sometimes not even bash.

Signed-off-by: Finn Behrens <me@kloenk.de>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2020-12-08 23:30:04 +09:00

628 lines
14 KiB
Perl
Executable file

#!/usr/bin/env perl
# SPDX-License-Identifier: GPL-2.0
use strict;
use warnings;
use utf8;
use Pod::Usage;
use Getopt::Long;
use File::Find;
use Fcntl ':mode';
my $help = 0;
my $man = 0;
my $debug = 0;
my $enable_lineno = 0;
my $prefix="Documentation/ABI";
#
# If true, assumes that the description is formatted with ReST
#
my $description_is_rst = 1;
GetOptions(
"debug|d+" => \$debug,
"enable-lineno" => \$enable_lineno,
"rst-source!" => \$description_is_rst,
"dir=s" => \$prefix,
'help|?' => \$help,
man => \$man
) or pod2usage(2);
pod2usage(1) if $help;
pod2usage(-exitstatus => 0, -verbose => 2) if $man;
pod2usage(2) if (scalar @ARGV < 1 || @ARGV > 2);
my ($cmd, $arg) = @ARGV;
pod2usage(2) if ($cmd ne "search" && $cmd ne "rest" && $cmd ne "validate");
pod2usage(2) if ($cmd eq "search" && !$arg);
require Data::Dumper if ($debug);
my %data;
my %symbols;
#
# Displays an error message, printing file name and line
#
sub parse_error($$$$) {
my ($file, $ln, $msg, $data) = @_;
$data =~ s/\s+$/\n/;
print STDERR "Warning: file $file#$ln:\n\t$msg";
if ($data ne "") {
print STDERR ". Line\n\t\t$data";
} else {
print STDERR "\n";
}
}
#
# Parse an ABI file, storing its contents at %data
#
sub parse_abi {
my $file = $File::Find::name;
my $mode = (stat($file))[2];
return if ($mode & S_IFDIR);
return if ($file =~ m,/README,);
my $name = $file;
$name =~ s,.*/,,;
my $fn = $file;
$fn =~ s,Documentation/ABI/,,;
my $nametag = "File $fn";
$data{$nametag}->{what} = "File $name";
$data{$nametag}->{type} = "File";
$data{$nametag}->{file} = $name;
$data{$nametag}->{filepath} = $file;
$data{$nametag}->{is_file} = 1;
$data{$nametag}->{line_no} = 1;
my $type = $file;
$type =~ s,.*/(.*)/.*,$1,;
my $what;
my $new_what;
my $tag = "";
my $ln;
my $xrefs;
my $space;
my @labels;
my $label = "";
print STDERR "Opening $file\n" if ($debug > 1);
open IN, $file;
while(<IN>) {
$ln++;
if (m/^(\S+)(:\s*)(.*)/i) {
my $new_tag = lc($1);
my $sep = $2;
my $content = $3;
if (!($new_tag =~ m/(what|where|date|kernelversion|contact|description|users)/)) {
if ($tag eq "description") {
# New "tag" is actually part of
# description. Don't consider it a tag
$new_tag = "";
} elsif ($tag ne "") {
parse_error($file, $ln, "tag '$tag' is invalid", $_);
}
}
# Invalid, but it is a common mistake
if ($new_tag eq "where") {
parse_error($file, $ln, "tag 'Where' is invalid. Should be 'What:' instead", "");
$new_tag = "what";
}
if ($new_tag =~ m/what/) {
$space = "";
$content =~ s/[,.;]$//;
push @{$symbols{$content}->{file}}, " $file:" . ($ln - 1);
if ($tag =~ m/what/) {
$what .= ", " . $content;
} else {
if ($what) {
parse_error($file, $ln, "What '$what' doesn't have a description", "") if (!$data{$what}->{description});
foreach my $w(split /, /, $what) {
$symbols{$w}->{xref} = $what;
};
}
$what = $content;
$label = $content;
$new_what = 1;
}
push @labels, [($content, $label)];
$tag = $new_tag;
push @{$data{$nametag}->{symbols}}, $content if ($data{$nametag}->{what});
next;
}
if ($tag ne "" && $new_tag) {
$tag = $new_tag;
if ($new_what) {
@{$data{$what}->{label_list}} = @labels if ($data{$nametag}->{what});
@labels = ();
$label = "";
$new_what = 0;
$data{$what}->{type} = $type;
if (!defined($data{$what}->{file})) {
$data{$what}->{file} = $name;
$data{$what}->{filepath} = $file;
} else {
if ($name ne $data{$what}->{file}) {
$data{$what}->{file} .= " " . $name;
$data{$what}->{filepath} .= " " . $file;
}
}
print STDERR "\twhat: $what\n" if ($debug > 1);
$data{$what}->{line_no} = $ln;
} else {
$data{$what}->{line_no} = $ln if (!defined($data{$what}->{line_no}));
}
if (!$what) {
parse_error($file, $ln, "'What:' should come first:", $_);
next;
}
if ($new_tag eq "description") {
$sep =~ s,:, ,;
$content = ' ' x length($new_tag) . $sep . $content;
while ($content =~ s/\t+/' ' x (length($&) * 8 - length($`) % 8)/e) {}
if ($content =~ m/^(\s*)(\S.*)$/) {
# Preserve initial spaces for the first line
$space = $1;
$content = "$2\n";
$data{$what}->{$tag} .= $content;
} else {
undef($space);
}
} else {
$data{$what}->{$tag} = $content;
}
next;
}
}
# Store any contents before tags at the database
if (!$tag && $data{$nametag}->{what}) {
$data{$nametag}->{description} .= $_;
next;
}
if ($tag eq "description") {
my $content = $_;
while ($content =~ s/\t+/' ' x (length($&) * 8 - length($`) % 8)/e) {}
if (m/^\s*\n/) {
$data{$what}->{$tag} .= "\n";
next;
}
if (!defined($space)) {
# Preserve initial spaces for the first line
if ($content =~ m/^(\s*)(\S.*)$/) {
$space = $1;
$content = "$2\n";
}
} else {
$space = "" if (!($content =~ s/^($space)//));
}
$data{$what}->{$tag} .= $content;
next;
}
if (m/^\s*(.*)/) {
$data{$what}->{$tag} .= "\n$1";
$data{$what}->{$tag} =~ s/\n+$//;
next;
}
# Everything else is error
parse_error($file, $ln, "Unexpected content", $_);
}
$data{$nametag}->{description} =~ s/^\n+// if ($data{$nametag}->{description});
if ($what) {
parse_error($file, $ln, "What '$what' doesn't have a description", "") if (!$data{$what}->{description});
foreach my $w(split /, /,$what) {
$symbols{$w}->{xref} = $what;
};
}
close IN;
}
sub create_labels {
my %labels;
foreach my $what (keys %data) {
next if ($data{$what}->{file} eq "File");
foreach my $p (@{$data{$what}->{label_list}}) {
my ($content, $label) = @{$p};
$label = "abi_" . $label . " ";
$label =~ tr/A-Z/a-z/;
# Convert special chars to "_"
$label =~s/([\x00-\x2f\x3a-\x40\x5b-\x60\x7b-\xff])/_/g;
$label =~ s,_+,_,g;
$label =~ s,_$,,;
# Avoid duplicated labels
while (defined($labels{$label})) {
my @chars = ("A".."Z", "a".."z");
$label .= $chars[rand @chars];
}
$labels{$label} = 1;
$data{$what}->{label} = $label;
# only one label is enough
last;
}
}
}
#
# Outputs the book on ReST format
#
# \b doesn't work well with paths. So, we need to define something else
my $bondary = qr { (?<![\w\/\`\{])(?=[\w\/\`\{])|(?<=[\w\/\`\{])(?![\w\/\`\{]) }x;
sub output_rest {
create_labels();
my $part = "";
foreach my $what (sort {
($data{$a}->{type} eq "File") cmp ($data{$b}->{type} eq "File") ||
$a cmp $b
} keys %data) {
my $type = $data{$what}->{type};
my @file = split / /, $data{$what}->{file};
my @filepath = split / /, $data{$what}->{filepath};
if ($enable_lineno) {
printf "#define LINENO %s%s#%s\n\n",
$prefix, $file[0],
$data{$what}->{line_no};
}
my $w = $what;
$w =~ s/([\(\)\_\-\*\=\^\~\\])/\\$1/g;
if ($type ne "File") {
my $cur_part = $what;
if ($what =~ '/') {
if ($what =~ m#^(\/?(?:[\w\-]+\/?){1,2})#) {
$cur_part = "Symbols under $1";
$cur_part =~ s,/$,,;
}
}
if ($cur_part ne "" && $part ne $cur_part) {
$part = $cur_part;
my $bar = $part;
$bar =~ s/./-/g;
print "$part\n$bar\n\n";
}
printf ".. _%s:\n\n", $data{$what}->{label};
my @names = split /, /,$w;
my $len = 0;
foreach my $name (@names) {
$name = "**$name**";
$len = length($name) if (length($name) > $len);
}
print "+-" . "-" x $len . "-+\n";
foreach my $name (@names) {
printf "| %s", $name . " " x ($len - length($name)) . " |\n";
print "+-" . "-" x $len . "-+\n";
}
print "\n";
}
for (my $i = 0; $i < scalar(@filepath); $i++) {
my $path = $filepath[$i];
my $f = $file[$i];
$path =~ s,.*/(.*/.*),$1,;;
$path =~ s,[/\-],_,g;;
my $fileref = "abi_file_".$path;
if ($type eq "File") {
print ".. _$fileref:\n\n";
} else {
print "Defined on file :ref:`$f <$fileref>`\n\n";
}
}
if ($type eq "File") {
my $bar = $w;
$bar =~ s/./-/g;
print "$w\n$bar\n\n";
}
my $desc = "";
$desc = $data{$what}->{description} if (defined($data{$what}->{description}));
$desc =~ s/\s+$/\n/;
if (!($desc =~ /^\s*$/)) {
if ($description_is_rst) {
# Remove title markups from the description
# Having titles inside ABI files will only work if extra
# care would be taken in order to strictly follow the same
# level order for each markup.
$desc =~ s/\n[\-\*\=\^\~]+\n/\n\n/g;
# Enrich text by creating cross-references
$desc =~ s,Documentation/(?!devicetree)(\S+)\.rst,:doc:`/$1`,g;
my @matches = $desc =~ m,Documentation/ABI/([\w\/\-]+),;
foreach my $f (@matches) {
my $xref = $f;
my $path = $f;
$path =~ s,.*/(.*/.*),$1,;;
$path =~ s,[/\-],_,g;;
$xref .= " <abi_file_" . $path . ">";
$desc =~ s,\bDocumentation/ABI/$f\b,:ref:`$xref`,g;
}
@matches = $desc =~ m,$bondary(/sys/[^\s\.\,\;\:\*\s\`\'\(\)]+)$bondary,;
foreach my $s (@matches) {
if (defined($data{$s}) && defined($data{$s}->{label})) {
my $xref = $s;
$xref =~ s/([\x00-\x1f\x21-\x2f\x3a-\x40\x7b-\xff])/\\$1/g;
$xref = ":ref:`$xref <" . $data{$s}->{label} . ">`";
$desc =~ s,$bondary$s$bondary,$xref,g;
}
}
print "$desc\n\n";
} else {
$desc =~ s/^\s+//;
# Remove title markups from the description, as they won't work
$desc =~ s/\n[\-\*\=\^\~]+\n/\n\n/g;
if ($desc =~ m/\:\n/ || $desc =~ m/\n[\t ]+/ || $desc =~ m/[\x00-\x08\x0b-\x1f\x7b-\xff]/) {
# put everything inside a code block
$desc =~ s/\n/\n /g;
print "::\n\n";
print " $desc\n\n";
} else {
# Escape any special chars from description
$desc =~s/([\x00-\x08\x0b-\x1f\x21-\x2a\x2d\x2f\x3c-\x40\x5c\x5e-\x60\x7b-\xff])/\\$1/g;
print "$desc\n\n";
}
}
} else {
print "DESCRIPTION MISSING for $what\n\n" if (!$data{$what}->{is_file});
}
if ($data{$what}->{symbols}) {
printf "Has the following ABI:\n\n";
foreach my $content(@{$data{$what}->{symbols}}) {
my $label = $data{$symbols{$content}->{xref}}->{label};
# Escape special chars from content
$content =~s/([\x00-\x1f\x21-\x2f\x3a-\x40\x7b-\xff])/\\$1/g;
print "- :ref:`$content <$label>`\n\n";
}
}
if (defined($data{$what}->{users})) {
my $users = $data{$what}->{users};
$users =~ s/\n/\n\t/g;
printf "Users:\n\t%s\n\n", $users if ($users ne "");
}
}
}
#
# Searches for ABI symbols
#
sub search_symbols {
foreach my $what (sort keys %data) {
next if (!($what =~ m/($arg)/));
my $type = $data{$what}->{type};
next if ($type eq "File");
my $file = $data{$what}->{filepath};
my $bar = $what;
$bar =~ s/./-/g;
print "\n$what\n$bar\n\n";
my $kernelversion = $data{$what}->{kernelversion} if (defined($data{$what}->{kernelversion}));
my $contact = $data{$what}->{contact} if (defined($data{$what}->{contact}));
my $users = $data{$what}->{users} if (defined($data{$what}->{users}));
my $date = $data{$what}->{date} if (defined($data{$what}->{date}));
my $desc = $data{$what}->{description} if (defined($data{$what}->{description}));
$kernelversion =~ s/^\s+// if ($kernelversion);
$contact =~ s/^\s+// if ($contact);
if ($users) {
$users =~ s/^\s+//;
$users =~ s/\n//g;
}
$date =~ s/^\s+// if ($date);
$desc =~ s/^\s+// if ($desc);
printf "Kernel version:\t\t%s\n", $kernelversion if ($kernelversion);
printf "Date:\t\t\t%s\n", $date if ($date);
printf "Contact:\t\t%s\n", $contact if ($contact);
printf "Users:\t\t\t%s\n", $users if ($users);
print "Defined on file(s):\t$file\n\n";
print "Description:\n\n$desc";
}
}
# Ensure that the prefix will always end with a slash
# While this is not needed for find, it makes the patch nicer
# with --enable-lineno
$prefix =~ s,/?$,/,;
#
# Parses all ABI files located at $prefix dir
#
find({wanted =>\&parse_abi, no_chdir => 1}, $prefix);
print STDERR Data::Dumper->Dump([\%data], [qw(*data)]) if ($debug);
#
# Handles the command
#
if ($cmd eq "search") {
search_symbols;
} else {
if ($cmd eq "rest") {
output_rest;
}
# Warn about duplicated ABI entries
foreach my $what(sort keys %symbols) {
my @files = @{$symbols{$what}->{file}};
next if (scalar(@files) == 1);
printf STDERR "Warning: $what is defined %d times: @files\n",
scalar(@files);
}
}
__END__
=head1 NAME
abi_book.pl - parse the Linux ABI files and produce a ReST book.
=head1 SYNOPSIS
B<abi_book.pl> [--debug] [--enable-lineno] [--man] [--help]
[--(no-)rst-source] [--dir=<dir>] <COMAND> [<ARGUMENT>]
Where <COMMAND> can be:
=over 8
B<search> [SEARCH_REGEX] - search for [SEARCH_REGEX] inside ABI
B<rest> - output the ABI in ReST markup language
B<validate> - validate the ABI contents
=back
=head1 OPTIONS
=over 8
=item B<--dir>
Changes the location of the ABI search. By default, it uses
the Documentation/ABI directory.
=item B<--rst-source> and B<--no-rst-source>
The input file may be using ReST syntax or not. Those two options allow
selecting between a rst-compliant source ABI (--rst-source), or a
plain text that may be violating ReST spec, so it requres some escaping
logic (--no-rst-source).
=item B<--enable-lineno>
Enable output of #define LINENO lines.
=item B<--debug>
Put the script in verbose mode, useful for debugging. Can be called multiple
times, to increase verbosity.
=item B<--help>
Prints a brief help message and exits.
=item B<--man>
Prints the manual page and exits.
=back
=head1 DESCRIPTION
Parse the Linux ABI files from ABI DIR (usually located at Documentation/ABI),
allowing to search for ABI symbols or to produce a ReST book containing
the Linux ABI documentation.
=head1 EXAMPLES
Search for all stable symbols with the word "usb":
=over 8
$ scripts/get_abi.pl search usb --dir Documentation/ABI/stable
=back
Search for all symbols that match the regex expression "usb.*cap":
=over 8
$ scripts/get_abi.pl search usb.*cap
=back
Output all obsoleted symbols in ReST format
=over 8
$ scripts/get_abi.pl rest --dir Documentation/ABI/obsolete
=back
=head1 BUGS
Report bugs to Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
=head1 COPYRIGHT
Copyright (c) 2016-2019 by Mauro Carvalho Chehab <mchehab+samsung@kernel.org>.
License GPLv2: GNU GPL version 2 <http://gnu.org/licenses/gpl.html>.
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.
=cut