admin (DEVEL): distfiles/specparser.pl - annother trial - robust b...

sparky sparky at pld-linux.org
Tue Mar 13 04:22:47 CET 2007


Author: sparky                       Date: Tue Mar 13 03:22:47 2007 GMT
Module: admin                         Tag: DEVEL
---- Log message:
- annother trial - robust brute-force parsing: tries all %if-%else-%endif
  possibilities
  for opera.spec gives only 31 results, but it takes 18 seconds (@1.3GHz)
  to parse (there are 6192 posibilities)

---- Files affected:
admin/distfiles:
   specparser.pl (1.13.2.2 -> 1.13.2.3) 

---- Diffs:

================================================================
Index: admin/distfiles/specparser.pl
diff -u admin/distfiles/specparser.pl:1.13.2.2 admin/distfiles/specparser.pl:1.13.2.3
--- admin/distfiles/specparser.pl:1.13.2.2	Mon Mar 12 23:24:17 2007
+++ admin/distfiles/specparser.pl	Tue Mar 13 04:22:42 2007
@@ -16,23 +16,23 @@
 use strict;
 use warnings;
 
-my %macro;
+my %no_source;
 my $spec;
 my $base_spec;
+my @spec;
+my @sources;
 
 sub next_spec($)
 {
 	$spec = shift;
+	@spec = ();
 	$base_spec = $spec;
 	$base_spec =~ s|.*/||;
-	%macro = ( "nil" => [""] );
-	#$err_cnt = 0;
 }
 
 sub error($)
 {
-	#$err_cnt++;
-	print "ERROR: $base_spec: $_[0]\n";
+	print_once( "ERROR: $base_spec: $_[0]" );
 }
 
 sub warning($)
@@ -50,11 +50,26 @@
 	return $v;
 }
 
-sub expand_tr($)
+# expand macros in string
+sub expand($$) # {{{
 {
-	my $v = shift;
-
+	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";
+		}
+		$v =~ s/\%\{([^\}]+)\}/$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, "");
@@ -76,83 +91,105 @@
 		return $v if (length $v > 1000 or $cnt-- <= 0)
 	}
 
+	error("unexpanded macros in $v")
+		if ($v =~ /\%[^0-9]/);
+	
 	return $v;
-}
+} # }}}
 
-# recursive expansion
-sub expand_r($$);
-sub expand_r($$)
-{
-	my ($to_expand, $level) = @_;
-	return $to_expand if $level < 0;
-	$level--;
-
-	if ($to_expand =~ /\%\{([^\}]+)\}/) {
-		my $key = quotemeta $1;
-		my $values;
-		if (defined $macro{$1}) {
-			$values = $macro{$1};
-		} else {
-			error("undefined macro $1");
-			$values = ["UNDEFINED"];
-		}
-		my @vs;
-		foreach my $value (@{$values}) {
-			(my $vs = $to_expand) =~ s/\%\{($key)\}/$value/;
-			push @vs, expand_r($vs, $level);
-		}
+sub preparse_spec($) # {{{
+{
+	@spec = ("");
 
-		return @vs;
-	} else {
-		return $to_expand;
+	open(F, "< $_[0]") or die;
+	while (<F>) {
+		chomp;
+		if (/^\s*(\%(description|package|prep|install|pre|post|files)|BuildRoot|URL)/) {
+			last;
+		} elsif (/^\s*(\%if|\%else|\%endif|\%define|Version|Name)/) {
+			if ($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{$1} = 1;
+		}
 	}
-}
+	close(F);
 
-# expand macros in string
-sub expand($)
-{
-	my $v = trim_spaces(shift);
-	my $done = [];
+	shift @spec;
+} # }}}
 
-	foreach my $e ( expand_r($v, 20) ) {
-		$e = expand_tr($e);
-		push @{$done}, $e;
-		error("unexpanded macros in $e")
-			if ($e =~ /\%[^0-9]/);
-	}
 
-	return $done;
-}
 
-# define given macro
-sub define($$)
+sub find_closing($$)
 {
-	my ($n, $v) = @_;
-	$macro{$n} = [] unless exists $macro{$n};
-	push @{$macro{$n}}, trim_spaces($v);
+	my ($spec, $end) = @_;
+	my $level = 0;
+	local $_;
+	while ($_ = shift @{$spec}) {
+		if (/^\s*\%($end)\s*$/ and $level <= 0) {
+			return;
+		} elsif (/^\s*\%if/) {
+			$level++;
+		} elsif (/^\s*\%endif\s*$/) {
+			$level--;
+		}
+	}
 }
 
-# sets hash of macros defined with %define or %global
-# also define %{name}, %{version} and %{source_N}
-sub parse_defines($)
+my $total = 0;
+
+sub cont($$);
+sub cont($$) # {{{
 {
-	open(F, "< $_[0]") or die;
-	while (<F>) {
-		chomp;
-		if (/^\s*\%(define|global)\s+([^\s]+)\s+([^\s].*)$/) {
-			define($2, $3);
+	my ($spec, $macros) = @_;
+	local $_;
+	while ($_ = shift @{$spec}) {
+		if (/^\s*\%if/) { # if, ifarch, ifos
+
+			# split spec parsing
+			my @speccopy = @{$spec};
+			my %macroscopy = %{$macros};
+			cont(\@speccopy, \%macroscopy);
+
+			find_closing($spec, "else|endif");
+			# continue parsing
+			
+		} elsif (/^\s*\%else\s*$/) {
+
+			# %else happens only when %if was interpreted
+			# so skip until %endif
+			
+			find_closing($spec, "endif");
+
+		} elsif (/^\s*\%(define|global)\s+([^\s]+)\s+([^\s].*)$/) {
+			$macros->{$2} = $3;
 		} elsif (/^Version\s*:\s*(.*)/i) {
-			define("version", $1);
+			$macros->{"version"} = $1;
 		} elsif (/^Name\s*:\s*(.*)/i) {
-			define("name", $1);
-		} elsif (/^Patch(\d+)\s*:\s*(.*)/i) {
-			define("patch_$1", expand($2));
-		} elsif (/^NoSource\s*:\s*(\d+)\s*$/i) {
-			define("no_source_$1", "1");
+			$macros->{"name"} = $1;
+		} elsif (!/\%(endif|undef|bcond)/) {
+			warn "unrecognised line: $_\n";
 		}
 	}
-	close(F);
-}
+
+	# the end, yuppie !
+	foreach my $s (@sources) {
+		print_source( $s->[0], $s->[1], expand( $s->[2], $macros ) );
+	}
+	
+	if (++$total > 10000) {
+		error("maximum number of bcond posibilities exceeded");
+		exit 0;
+	}
+} # }}}
 
 my %printed;
 sub print_once($)
@@ -164,8 +201,10 @@
 	}
 }
 
-sub print_source($$$) {
+sub print_source($$$) # {{{
+{
 	my ($no, $md5, $s) = @_;
+
 	if ($s =~ /^([a-z0-9A-Z:\=\?\@\+\~\.\-\/_]|\%[0-9])+$/) {
 		if ($s =~ /^(ftp|http|https):\/\//) {
 			if ($s =~ /\/$/) {
@@ -185,10 +224,9 @@
 	} else {
 		error("source $no url $s is ill-formatted");
 	}
+} # }}}
 
-}
-
-sub print_md5($)
+sub add_md5_to_print($) # {{{
 {
 	open(F, "< $_[0]") or die;
 	my $sourceno = undef;
@@ -197,32 +235,34 @@
 		chomp;
 		if (/^Source(\d+)\s*:\s*(.*)/i) {
 			$sourceno = $1;
-			$source = expand($2);
+			$source = $2;
 		} elsif (/^\s*#\s*source(\d+)-md5\s*:\s*([a-f0-9]{32})/i) {
 			my $no = $1;
 			my $md5 = $2;
-			if (defined $macro{"no_source_$no"}) {
+			if (defined $no_source{$no}) {
 				error("both NoSource: $no and md5 given");
-			} elsif (defined $sourceno and ($sourceno == $no)) {
-				foreach my $s (@{$source}) {
-					print_source($no, $md5, $s);
+			} if (defined $sourceno) {
+				if ($sourceno == $no) {
+					push @sources, [$no, $md5, $source];
+				} else {
+					error("found md5 for source $no, but last defined source is $sourceno");
 				}
-			} elsif (defined $sourceno) {
-				error("found md5 for source $no, but last defined source is $sourceno");
 			} else {
 				error("source $no not defined");
 			}
+
 			$sourceno = undef;
 			$source = undef;
 		}
 	}
 	close(F);
-}
+} # }}}
 
 next_spec(shift);
-parse_defines($spec);
-print_md5($spec);
+preparse_spec($spec);
+add_md5_to_print($spec);
+cont( \@spec, { "nil" => "" } );
 
 exit(0);
 
-# vim: ts=4:sw=4
+# vim: ts=4:sw=4:fdm=marker
================================================================

---- CVS-web:
    http://cvs.pld-linux.org/admin/distfiles/specparser.pl?r1=1.13.2.2&r2=1.13.2.3&f=u



More information about the pld-cvs-commit mailing list