[projects/distfiles.git] / specparser.pl

#!/usr/bin/perl
#
# USAGE: specparser.pl file.spec
#
# Output:
#   stdout:
#     list of:
#       <md5><space><url>\n
#     for given specfile.
#     except for lines prefixed with: ERROR:
#   exit status:
#     0 - normal
#     2 - cannot open spec file
#
use strict;
use warnings;

my %no_source;
my $spec;
my $base_spec;
my @spec;
my @sources;
my $sources_file;
my %patchset;

sub print_source($$$);

sub next_spec($)
{
	$spec = shift;
	@spec = ();
	$base_spec = $spec;
	$base_spec =~ s|.*/||;
}

sub error($)
{
	print_once( "ERROR: $base_spec: $_[0]" );
}

sub warning($)
{
	print "ERROR: $base_spec: $_[0]\n";
}

sub trim_spaces($)
{
	my $v = shift;

	$v =~ s/\s+$//;
	$v =~ s/^\s+//;

	return $v;
}

# expand macros in string
sub expand($$) # {{{
{
	my $v = trim_spaces(shift);
	my $macrotree = shift;
	my $cnt = 20;

	while ($v =~ /\%\{([^\}]+)\}/) {
		my $value;
		if (defined $macrotree->{$1}) {
			$value = $macrotree->{$1};
		} else {
			error("undefined macro $1");
			$value = "UNDEFINED";
			return undef;
		}
		$v =~ s/\%\{([^\}]+)\}/$value/;

		return $v if (length $v > 1000 or $cnt-- <= 0)
	}

	while ($v =~ s/\%\(\s*echo\s+([^\|]+?)\s*\|\s*tr\s*(-d|)\s+([^\)]+?)\s*\)/\@\@tr-me\@\@/) {
		my ($what, $d_opt, $how) = ($1, $2, $3);
		my ($from, $to) = ($how, "");
		($from, $to) = ($1, $2)
			if $how =~ /^([^\s]+)\s+([^\s]+)$/;
		if ($d_opt and $to ne "") {
			error("tr -d with second string)");
		} elsif (($from . $to) =~ /^[a-zA-Z0-9\+_\-\.]+$/) {
			if ($d_opt) {
				eval "\$what =~ tr/$from//d;";
			} else {
				eval "\$what =~ tr/$from/$to/;";
			}
		} else {
			error("illegal characters in tr string(s) '$from' '$to'");
		}
		$v =~ s/\@\@tr-me\@\@/$what/;

		return $v if (length $v > 1000 or $cnt-- <= 0)
	}

	error("unexpanded macros in $v")
		if ($v =~ /\%[^0-9]/);

	return $v;
} # }}}

sub preparse_spec($) # {{{
{
	@spec = ("");

	open(F, "< $_[0]") or die("failed opening: " . $_[0]);
	while (<F>) {
		chomp;
		if (/^\s*(\%(description|package|prep|install|pre|post|files)|BuildRoot|URL)/) {
			last;
		} elsif (/^\s*(\%if.*|\%else|\%endif|\%define\s+.*|Version.*|Name.*)\s*$/) {
			$_ = $1;
			if ($spec[$#spec] =~ /\%if/) {
				if (/\%else/) {
					next; # don't include empty %if-%else
				} elsif (/\%endif/) {
					# remove empty %if-%endif
					pop @spec;
					next;
				}
			}
			push @spec, $_;
		} elsif (/^NoSource\s*:\s*(\d+)\s*$/i) {
			$no_source{ "source" . $1 } = 1;
		} elsif (my ($patchset) = /^%patchset_source\s+(.+)$/) {
			use Getopt::Long qw(GetOptionsFromString);
			my ($f, $s);
			my ($ret, $args) = GetOptionsFromString($patchset, 's=s' => \$s, 'f=s' => \$f);
			%patchset = (
				pattern => $f,
				filelist => $s,
				start => $args->[0],
				end => $args->[1],
			);
		}
	}
	close(F);

	shift @spec;
} # }}}

# read in 'sources' file
sub read_sources_file {
	my $filename = $_[0] || $sources_file;
	return () unless $filename and -e $filename;

	my %files;

	open(my $fh, '<', $filename) or die $!;
	while (<$fh>) {
		chomp;
		next unless my ($hash, $filename) = /^([a-f0-9]{32})\s+\*?(.+)$/;
		$files{$filename} = $hash;
	}
	return \%files;
}

sub process_patchset($) {
	my $macros = shift;
	my $checksums;

	return unless %patchset;

	# print all files from sources file
	if ($patchset{filelist}) {
		my $prefix = expand($patchset{pattern}, $macros);
		my $filelist = expand($patchset{filelist}, $macros);
		$checksums = read_sources_file($filelist);
		while (my($file, $hash) = each %$checksums) {
			my $url = $prefix . $file;
			print_source "patchset $file", $hash, $url;
		}
		return;
	}

	# parse sources file sequences
	$checksums = read_sources_file() or return;

	# print out patchset entries which source md5 is present in source file
	my $start = expand($patchset{start}, $macros);
	my $end = expand($patchset{end}, $macros);
	my $pattern = expand($patchset{pattern}, $macros);
	for (my $i = $start; $i <= $end; $i++) {
		my $url = sprintf($pattern, $i);
		my ($basename) = $url =~ m{/([^/]+$)};
		my $hash = $checksums->{$basename} or next;
		print_source "patchset $i", $hash, $url;
	}
}


my $total = 0;

sub cont($$);
sub cont($$) # {{{
{
	my ($spec, $macros) = @_;
	local $_;
	while ($_ = shift @{$spec}) {
		if (0 <= index $_, '%if') { # if, ifarch, ifos

			# split spec parsing
			my @speccopy = @{$spec};
			my %macroscopy = %{$macros};
			cont(\@speccopy, \%macroscopy);

			my $level = 0;
			while ($_ = shift @{$spec}) {
				if ($level <= 0 and ($_ eq '%else' or $_ eq '%endif')) {
					last;
				} elsif (0 <= index $_, '%if') {
					$level++;
				} elsif ($_ eq '%endif') {
					$level--;
				}
			}

			# continue parsing
		} elsif ($_ eq '%else') {

			# %else happens only when %if was interpreted
			# so skip until %endif

			my $level = 0;
			while ($_ = shift @{$spec}) {
				if ($level <= 0 and $_ eq '%endif') {
					last;
				} elsif (0 <= index $_, '%if') {
					$level++;
				} elsif ($_ eq '%endif') {
					$level--;
				}
			}

		} elsif (/^\s*\%(define|global)\s+([^\s]+)\s+([^\s].*?)\s*$/) {
			$macros->{$2} = $3;

		} elsif (/^Version\s*:\s*(.*?)\s*$/i) {
			$macros->{"version"} = $1;
		} elsif (/^Name\s*:\s*(.*?)\s*$/i) {
			$macros->{"name"} = $1;
		} elsif (!/\%endif/) {
			warn "unrecognised line: $_\n";
		}
	}

	# so we have macros now, can parse patchset source
	process_patchset($macros);

	# the end, yuppie !
	foreach my $s (@sources) {
		my $src = expand( $s->[2], $macros );
		if (defined $src) {
			our %tried;
			unless (exists $tried{$src}) {
				print_source( $s->[0], $s->[1], $src );
				$tried{$src} = 1;
			}
		}
	}

	if (++$total > 10000) {
		error("maximum number of bcond posibilities exceeded");
		exit 0;
	}
} # }}}

sub print_once($)
{
	our %printed;
	my $l = shift;
	unless (exists $printed{$l}) {
		print $l . "\n";
		$printed{$l} = 1;
	}
}

sub print_source($$$) # {{{
{
	my ($no, $md5, $s) = @_;

	if ($s =~ /^([a-z0-9A-Z;:\=\?&\@\+\~\.,\-\/_]|\%[0-9])+(#\/[a-zA-Z0-9\._-]+)?$/) {
		if ($s =~ /^(ftp|http|https):\/\//) {
			if ($s =~ /\/$/) {
				error("source $no ($s) is directory");
			} else {
				if ($s =~ /:\/\/distfiles\.pld-linux\.org\/src/) {
					$s =~ s|.*/||;
					print_once( "$md5 no-url-copy://$s" );
				} else {
					print_once( "$md5 $s" );
				}
			}
		} else {
			$s =~ s|.*/||;
			print_once( "$md5 no-url://$s");
		}
	} else {
		error("source $no url $s is ill-formatted");
	}
} # }}}

sub add_md5_to_print($) # {{{
{
	open(F, "< $_[0]") or die("failed opening: " . $_[0]);
	my %sourcemap = ();
	while (<F>) {
		chomp;
		if (/^((?:Source|Patch)\d+)\s*:\s*(.*)/i) {
			my $sourceno = lc $1;
			my $source = $2;
			# master.dl is outdated currently
			# $source =~ s/dl.sourceforge.net/master.dl.sourceforge.net/;
			$sourcemap{ $sourceno } = $source;
		} elsif (/^\s*#\s*((?:source|patch)\d+)-md5\s*:\s*([a-f0-9]{32})/i) {
			my $no = lc $1;
			my $md5 = $2;
			if (defined $no_source{$no}) {
				error("both NoSource: $no and md5 given");
			} elsif (defined $sourcemap{ $no} ) {
				my $source = $sourcemap{ $no };
				push @sources, [$no, $md5, $source];
			} else {
				error("source $no not defined (# SourceN-md5: has to be placed just after SourceN:)");
			}
		} elsif (/^\s*BuildRequires:\s*digest\(%((?:SOURCE|PATCH)\d+)\)\s*=\s*([a-f0-9]{32})/i) {
			my $no = lc $1;
			my $md5 = $2;
			if (defined $no_source{ $no }) {
				error("both NoSource: $no and md5 given");
			} elsif (defined $sourcemap{ $no }) {
				my $source = $sourcemap{ $no };
				push @sources, [$no, $md5, $source];
			} else {
				error("source $no not defined (# Source digest has to be placed after SourceN:)");
			}
		}
	}
	close(F);

	# find extra sources from detached checksum file
	my $checksums = read_sources_file();
	my @mapped = map { $_->[0] } @sources;
	delete @sourcemap{@mapped};
	while (my($source, $url) = each %sourcemap) {
		my ($basename) = $url =~ m{/([^/]+$)};
		next unless defined $basename;
		my $hash = $checksums->{$basename} or next;
		push @sources, [$source, $hash, $url];
	}
} # }}}

next_spec(shift);
$sources_file = shift if @ARGV;
preparse_spec($spec);
add_md5_to_print($spec);
cont( \@spec, { "nil" => "" } );

exit(0);

# vim: ts=4:sw=4:fdm=marker
Commit	Line	Data
0db58532 MM	1	#!/usr/bin/perl
	2	#
	3	# USAGE: specparser.pl file.spec
	4	#
	5	# Output:
	6	# stdout:
	7	# list of:
	8	# <md5><space><url>\n
	9	# for given specfile.
1b5ce8c2	10	# except for lines prefixed with: ERROR:
0db58532 MM	11	# exit status:
	12	# 0 - normal
	13	# 2 - cannot open spec file
	14	#
8079fcc5	15	use strict;
	16	use warnings;
	17
	18	my %no_source;
	19	my $spec;
	20	my $base_spec;
	21	my @spec;
	22	my @sources;
1db641df	23	my $sources_file;
8b8c50a0	24	my %patchset;
0db58532	25
ca6092e6 ER	26	sub print_source($$$);
ca6092e6 ER	27
0dd6294b MM	28	sub next_spec($)
0dd6294b MM	29	{
e3413dab ER	30	$spec = shift;
	31	@spec = ();
	32	$base_spec = $spec;
	33	$base_spec =~ s\|.*/\|\|;
0dd6294b	34	}
0db58532	35
0dd6294b	36	sub error($)
0db58532	37	{
e3413dab	38	print_once( "ERROR: $base_spec: $_[0]" );
0dd6294b MM	39	}
	40
	41	sub warning($)
	42	{
e3413dab	43	print "ERROR: $base_spec: $_[0]\n";
0dd6294b	44	}
51a652ef	45
0dd6294b MM	46	sub trim_spaces($)
0dd6294b MM	47	{
e3413dab	48	my $v = shift;
62c5617b	49
e3413dab ER	50	$v =~ s/\s+$//;
e3413dab ER	51	$v =~ s/^\s+//;
62c5617b	52
e3413dab	53	return $v;
0dd6294b MM	54	}
	55
	56	# expand macros in string
8079fcc5	57	sub expand($$) # {{{
0dd6294b	58	{
e3413dab ER	59	my $v = trim_spaces(shift);
	60	my $macrotree = shift;
	61	my $cnt = 20;
	62
	63	while ($v =~ /\%\{([^\}]+)\}/) {
	64	my $value;
	65	if (defined $macrotree->{$1}) {
	66	$value = $macrotree->{$1};
	67	} else {
	68	error("undefined macro $1");
	69	$value = "UNDEFINED";
	70	return undef;
	71	}
	72	$v =~ s/\%\{([^\}]+)\}/$value/;
	73
	74	return $v if (length $v > 1000 or $cnt-- <= 0)
	75	}
	76
	77	while ($v =~ s/\%\(\secho\s+([^\\|]+?)\s\\|\str\s(-d\|)\s+([^\)]+?)\s*\)/\@\@tr-me\@\@/) {
	78	my ($what, $d_opt, $how) = ($1, $2, $3);
	79	my ($from, $to) = ($how, "");
	80	($from, $to) = ($1, $2)
	81	if $how =~ /^([^\s]+)\s+([^\s]+)$/;
	82	if ($d_opt and $to ne "") {
	83	error("tr -d with second string)");
	84	} elsif (($from . $to) =~ /^[a-zA-Z0-9\+_\-\.]+$/) {
	85	if ($d_opt) {
	86	eval "\$what =~ tr/$from//d;";
	87	} else {
	88	eval "\$what =~ tr/$from/$to/;";
	89	}
	90	} else {
	91	error("illegal characters in tr string(s) '$from' '$to'");
	92	}
	93	$v =~ s/\@\@tr-me\@\@/$what/;
	94
	95	return $v if (length $v > 1000 or $cnt-- <= 0)
	96	}
	97
	98	error("unexpanded macros in $v")
	99	if ($v =~ /\%[^0-9]/);
	100
	101	return $v;
8079fcc5	102	} # }}}
	103
	104	sub preparse_spec($) # {{{
0db58532	105	{
e3413dab ER	106	@spec = ("");
	107
	108	open(F, "< $_[0]") or die("failed opening: " . $_[0]);
	109	while (<F>) {
	110	chomp;
	111	if (/^\s*(\%(description\|package\|prep\|install\|pre\|post\|files)\|BuildRoot\|URL)/) {
	112	last;
	113	} elsif (/^\s(\%if.\|\%else\|\%endif\|\%define\s+.\|Version.\|Name.)\s$/) {
	114	$_ = $1;
	115	if ($spec[$#spec] =~ /\%if/) {
	116	if (/\%else/) {
	117	next; # don't include empty %if-%else
	118	} elsif (/\%endif/) {
	119	# remove empty %if-%endif
	120	pop @spec;
	121	next;
	122	}
	123	}
	124	push @spec, $_;
	125	} elsif (/^NoSource\s:\s(\d+)\s*$/i) {
	126	$no_source{ "source" . $1 } = 1;
736b2879 ER	127	} elsif (my ($patchset) = /^%patchset_source\s+(.+)$/) {
	128	use Getopt::Long qw(GetOptionsFromString);
	129	my ($f, $s);
	130	my ($ret, $args) = GetOptionsFromString($patchset, 's=s' => \$s, 'f=s' => \$f);
e3413dab	131	%patchset = (
736b2879 ER	132	pattern => $f,
	133	filelist => $s,
	134	start => $args->[0],
	135	end => $args->[1],
e3413dab ER	136	);
	137	}
	138	}
	139	close(F);
	140
	141	shift @spec;
8079fcc5	142	} # }}}
8079fcc5	143
1db641df	144	# read in 'sources' file
736b2879 ER	145	sub read_sources_file {
	146	my $filename = $_[0] \|\| $sources_file;
	147	return () unless $filename and -e $filename;
e3413dab	148
736b2879	149	my %files;
1db641df	150
736b2879	151	open(my $fh, '<', $filename) or die $!;
e3413dab ER	152	while (<$fh>) {
	153	chomp;
	154	next unless my ($hash, $filename) = /^([a-f0-9]{32})\s+\*?(.+)$/;
	155	$files{$filename} = $hash;
	156	}
1db641df ER	157	return \%files;
	158	}
	159
	160	sub process_patchset($) {
	161	my $macros = shift;
736b2879 ER	162	my $checksums;
736b2879 ER	163
1db641df ER	164	return unless %patchset;
1db641df ER	165
736b2879 ER	166	# print all files from sources file
	167	if ($patchset{filelist}) {
	168	my $prefix = expand($patchset{pattern}, $macros);
	169	my $filelist = expand($patchset{filelist}, $macros);
	170	$checksums = read_sources_file($filelist);
	171	while (my($file, $hash) = each %$checksums) {
	172	my $url = $prefix . $file;
	173	print_source "patchset $file", $hash, $url;
	174	}
	175	return;
	176	}
	177
	178	# parse sources file sequences
	179	$checksums = read_sources_file() or return;
e3413dab ER	180
	181	# print out patchset entries which source md5 is present in source file
	182	my $start = expand($patchset{start}, $macros);
	183	my $end = expand($patchset{end}, $macros);
	184	my $pattern = expand($patchset{pattern}, $macros);
	185	for (my $i = $start; $i <= $end; $i++) {
	186	my $url = sprintf($pattern, $i);
	187	my ($basename) = $url =~ m{/([^/]+$)};
1db641df	188	my $hash = $checksums->{$basename} or next;
e3413dab ER	189	print_source "patchset $i", $hash, $url;
e3413dab ER	190	}
8b8c50a0 ER	191	}
8b8c50a0 ER	192
8079fcc5	193
	194	my $total = 0;
	195
	196	sub cont($$);
	197	sub cont($$) # {{{
	198	{
e3413dab ER	199	my ($spec, $macros) = @_;
	200	local $_;
	201	while ($_ = shift @{$spec}) {
	202	if (0 <= index $_, '%if') { # if, ifarch, ifos
	203
	204	# split spec parsing
	205	my @speccopy = @{$spec};
	206	my %macroscopy = %{$macros};
	207	cont(\@speccopy, \%macroscopy);
	208
	209	my $level = 0;
	210	while ($_ = shift @{$spec}) {
	211	if ($level <= 0 and ($_ eq '%else' or $_ eq '%endif')) {
	212	last;
	213	} elsif (0 <= index $_, '%if') {
	214	$level++;
	215	} elsif ($_ eq '%endif') {
	216	$level--;
	217	}
	218	}
	219
	220	# continue parsing
	221	} elsif ($_ eq '%else') {
	222
	223	# %else happens only when %if was interpreted
	224	# so skip until %endif
	225
	226	my $level = 0;
	227	while ($_ = shift @{$spec}) {
	228	if ($level <= 0 and $_ eq '%endif') {
	229	last;
	230	} elsif (0 <= index $_, '%if') {
	231	$level++;
	232	} elsif ($_ eq '%endif') {
	233	$level--;
	234	}
	235	}
	236
	237	} elsif (/^\s\%(define\|global)\s+([^\s]+)\s+([^\s].?)\s*$/) {
	238	$macros->{$2} = $3;
	239
	240	} elsif (/^Version\s:\s(.?)\s$/i) {
	241	$macros->{"version"} = $1;
	242	} elsif (/^Name\s:\s(.?)\s$/i) {
	243	$macros->{"name"} = $1;
	244	} elsif (!/\%endif/) {
	245	warn "unrecognised line: $_\n";
	246	}
	247	}
	248
	249	# so we have macros now, can parse patchset source
	250	process_patchset($macros);
	251
	252	# the end, yuppie !
	253	foreach my $s (@sources) {
	254	my $src = expand( $s->[2], $macros );
	255	if (defined $src) {
	256	our %tried;
	257	unless (exists $tried{$src}) {
	258	print_source( $s->[0], $s->[1], $src );
	259	$tried{$src} = 1;
	260	}
	261	}
	262	}
263
264	if (++$total > 10000) {
265	error("maximum number of bcond posibilities exceeded");
266	exit 0;
267	}
8079fcc5	268	} # }}}
	269
	270	sub print_once($)
0db58532	271	{
e3413dab ER	272	our %printed;
	273	my $l = shift;
	274	unless (exists $printed{$l}) {
	275	print $l . "\n";
	276	$printed{$l} = 1;
	277	}
0db58532 MM	278	}
0db58532 MM	279
8079fcc5	280	sub print_source($$$) # {{{
0db58532	281	{
e3413dab ER	282	my ($no, $md5, $s) = @_;
	283
	284	if ($s =~ /^([a-z0-9A-Z;:\=\?&\@\+\~\.,\-\/_]\|\%[0-9])+(#\/[a-zA-Z0-9\._-]+)?$/) {
	285	if ($s =~ /^(ftp\|http\|https):\/\//) {
	286	if ($s =~ /\/$/) {
	287	error("source $no ($s) is directory");
	288	} else {
	289	if ($s =~ /:\/\/distfiles\.pld-linux\.org\/src/) {
	290	$s =~ s\|.*/\|\|;
	291	print_once( "$md5 no-url-copy://$s" );
	292	} else {
	293	print_once( "$md5 $s" );
	294	}
	295	}
	296	} else {
	297	$s =~ s\|.*/\|\|;
	298	print_once( "$md5 no-url://$s");
	299	}
	300	} else {
	301	error("source $no url $s is ill-formatted");
	302	}
8079fcc5	303	} # }}}
	304
	305	sub add_md5_to_print($) # {{{
	306	{
e3413dab ER	307	open(F, "< $_[0]") or die("failed opening: " . $_[0]);
	308	my %sourcemap = ();
	309	while (<F>) {
	310	chomp;
	311	if (/^((?:Source\|Patch)\d+)\s:\s(.*)/i) {
	312	my $sourceno = lc $1;
	313	my $source = $2;
	314	# master.dl is outdated currently
	315	# $source =~ s/dl.sourceforge.net/master.dl.sourceforge.net/;
	316	$sourcemap{ $sourceno } = $source;
	317	} elsif (/^\s#\s((?:source\|patch)\d+)-md5\s:\s([a-f0-9]{32})/i) {
	318	my $no = lc $1;
	319	my $md5 = $2;
	320	if (defined $no_source{$no}) {
	321	error("both NoSource: $no and md5 given");
	322	} elsif (defined $sourcemap{ $no} ) {
	323	my $source = $sourcemap{ $no };
	324	push @sources, [$no, $md5, $source];
	325	} else {
	326	error("source $no not defined (# SourceN-md5: has to be placed just after SourceN:)");
	327	}
	328	} elsif (/^\sBuildRequires:\sdigest\(%((?:SOURCE\|PATCH)\d+)\)\s=\s([a-f0-9]{32})/i) {
	329	my $no = lc $1;
	330	my $md5 = $2;
	331	if (defined $no_source{ $no }) {
	332	error("both NoSource: $no and md5 given");
	333	} elsif (defined $sourcemap{ $no }) {
	334	my $source = $sourcemap{ $no };
	335	push @sources, [$no, $md5, $source];
	336	} else {
	337	error("source $no not defined (# Source digest has to be placed after SourceN:)");
	338	}
	339	}
	340	}
	341	close(F);
0de41088 ER	342
	343	# find extra sources from detached checksum file
	344	my $checksums = read_sources_file();
	345	my @mapped = map { $_->[0] } @sources;
	346	delete @sourcemap{@mapped};
	347	while (my($source, $url) = each %sourcemap) {
	348	my ($basename) = $url =~ m{/([^/]+$)};
164cf5c1	349	next unless defined $basename;
0de41088 ER	350	my $hash = $checksums->{$basename} or next;
	351	push @sources, [$source, $hash, $url];
	352	}
8079fcc5	353	} # }}}
0db58532	354
0dd6294b	355	next_spec(shift);
1db641df	356	$sources_file = shift if @ARGV;
8079fcc5	357	preparse_spec($spec);
	358	add_md5_to_print($spec);
	359	cont( \@spec, { "nil" => "" } );
0db58532 MM	360
0db58532 MM	361	exit(0);
8079fcc5	362
8079fcc5	363	# vim: ts=4:sw=4:fdm=marker