X-Git-Url: http://www.chiark.greenend.org.uk/ucgi/~ianmdlvl/git?p=dgit.git;a=blobdiff_plain;f=i18n-diff-auditor;h=a92db5e1172ed3822210351bfdf3d9863fa86b7c;hp=61f877ed9719e9db09152dbb3b885e03ddfe0c35;hb=21600a2e332fd1074540b6529ea9e1a53f34a6df;hpb=2a2b0931f9b29bf372136451d317eb81649424e2

diff --git a/i18n-diff-auditor b/i18n-diff-auditor
index 61f877ed..a92db5e1 100755
--- a/i18n-diff-auditor
+++ b/i18n-diff-auditor
@@ -107,7 +107,8 @@ our ($ihunkhead, $ihunkend);
 our ($ichunkstart, $ichunkend);
 our ($before, $after);
 
-sub is_string ($) { $_[0]{T} =~ m/heredoc|string/; };
+sub is_string ($) { $_[0]{T} =~ m/heredoc|string/; }
+sub is_trans ($) { grep { $_[0]{E} eq $_ } qw(__ f_ i_); }
 
 sub semiparse ($) {
     ($_) = @_;
@@ -131,7 +132,7 @@ sub semiparse ($) {
                 (             .* \n    )
                 ( (?: (?! $d) .* \n )* )
                           $d     \n
-              }{ $1 }x or die "missing end of here doc $d\n";
+              }{ $1 }xe or die "missing end of here doc $d\n";
 	    $o[$#o]{V} = $2;
 	} elsif (s{^ (["'])( (?: [^\\'"]
                                | \\ [^"']
@@ -139,13 +140,13 @@ sub semiparse ($) {
                               )*
                        ) \1 }{}x) {
 	    my ($q,$v) = ($1,$2);
-	    push @o, { T => 'string', E => $&, P => "$q-string",
+	    push @o, { T => 'string', E => $&, P => "$q$q",
 		       Q => $q, V => $v};
 	} elsif (s{^$perlop_re|^\;}{}) {
 	    push @o, { T => 'op', E => $&, P => $& };
-	} elsif (s/[[{(]//) {
+	} elsif (s/^[[{(]//) {
 	    push @o, { T => 'bra', E => $&, P => $& };
-	} elsif (s/[]})]//) {
+	} elsif (s/^[]})]//) {
 	    push @o, { T => 'ket', E => $&, P => $& };
 	} elsif (s/^( [\$\@\%] )( \{ )//x) {
 	    push @o, { T => 'deref', E => $1, P => $1 },
@@ -159,9 +160,10 @@ sub semiparse ($) {
 	    die "cannot tokenise \`$&'";
 	}
     }
-    for (my $i=@o-2; $i>0; --$i) {
+    for (my $i=0; $i+2 < @o; $i++) {
 	next unless $o[$i+1]{E} eq '.';
 	my @inputs = @o[$i, $i+2];
+	#print STDERR Dumper(\@inputs);
 	next if grep { !is_string($_) } @inputs;
 	my $q = $inputs[0]{Q};
 	next if grep { $_->{Q} ne $q } @inputs;
@@ -172,9 +174,10 @@ sub semiparse ($) {
 		    Q => $q,
 		  };
 	@o = (@o[0..$i-1], $new, @o[$i+3..$#o]);
-	print STDERR Dumper(\@o);
+	$i--; # counteracts $i++
     }
     debug $ichunkstart, "semiparsed: ".join ' ', map { $_->{P} } @o;
+    # debug $ichunkstart, "semiparsed V: ".join ' ', map { defined $_->{V} ? ">$_->{V}<" : '-' } @o;
     return @o;
 }	    
 
@@ -197,17 +200,41 @@ sub analyse_chunk_core () {
     my ($x,$y);
     my $next_x = sub { $next_something->(\@xs, \@analysed_x, \$x, 'before'); };
     my $next_y = sub { $next_something->(\@ys, \@analysed_y, \$y, 'after' ); };
+    our @y_expect_suffix = ();
     for (;;) {
+	while (my $e = shift @y_expect_suffix) {
+	    $next_y->();
+	    $y->{E} eq $e
+		or die "suffix mismatch, expected $e got $y->{E}\n";
+	}
 	last unless @xs or @ys;
 	$next_x->();
 	$next_y->();
 	next if $x->{E} eq $y->{E};
+	next if $x->{E} eq 'sprintf' and $y->{E} eq 'f_';
+	next if $x->{E} eq 'die'     and $y->{E} eq 'confess';
+	if ($y->{E} eq '+'
+	    and @ys >= 3
+	    and $ys[0]{E} eq '('
+	    and is_trans($ys[1])) {
+	    $next_y->(); # (
+	    $next_y->(); # __ f_ i_
+	    @y_expect_suffix = ')';
+	} elsif ($y->{E} eq '('
+	    and @ys > 2
+	    and is_trans($ys[0])
+	    and @analysed_y
+	    and (grep { $_ eq $analysed_y[-1] } (qw( => [ { ? : . ),
+						 '(', ',') )) {
+	    $next_y->(); # __ f_ i_
+	    @y_expect_suffix = ')';
+	}
 	my $string_changed;
 	my $ye = $y->{E};
-	if ($ye eq '__' or $ye eq 'f_') {
+	if (is_trans($y)) {
 	    $next_y->();
 	    die "__ on non-string $y->{P}\n"     unless is_string($y);
-	    die "__ on was non-string $y->{P}\n" unless is_string($x);
+	    die "__ on was non-string $x->{P}\n" unless is_string($x);
 	    if ($y->{Q} ne "'") {
 		die "var subst in new string\n"
 		    if $y->{V} =~ m{(?<!\\) [\$\@]};
@@ -224,6 +251,10 @@ sub analyse_chunk_core () {
 	    die $string_changed if length $string_changed;
 	    next;
 	}
+	if ($ye eq 'i_') {
+	    die $string_changed if length $string_changed;
+	    next;
+	}
 	if ($ye eq 'f_') {
 	    my $fmt = $y->{V};
 	    die "no percent in _f string\n" unless $fmt =~ m{\%};
@@ -234,8 +265,10 @@ sub analyse_chunk_core () {
 		my ($lit, $what) = @_;
 		my $xl = substr($xs, 0, length($lit));
 		if ($xl ne $lit) {
-		    debug $ichunkstart, "not exactly x: $xs";
-		    debug $ichunkstart, "not exactly y: $lit";
+		    my $xsp  = $xs;  $xsp  =~ s/\n/\\n/g;
+		    my $litp = $lit; $litp =~ s/\n/\\n/g;
+		    debug $ichunkstart, "not exactly x: ..\"$xsp\"";
+		    debug $ichunkstart, "not exactly y:  '$litp'";
 		    my $next = @ys ? $ys[0]{P} : '(end)';
 		    die "string contents mismatch near $what before $next\n";
 		}
@@ -245,6 +278,7 @@ sub analyse_chunk_core () {
 		if ($fmt !~ m{\%[^\%]}) {
 		    $exactly->($fmt, '(tail)');
 		    $fmt = '';
+		    die "text deleted from end of string\n" if length $xs;
 		    last;
 		}
 		$exactly->($`, '(literal)');
@@ -253,6 +287,15 @@ sub analyse_chunk_core () {
 		elsif ($& ne '%s') { die "unhandled %-subst $&\n"; }
 		$next_y->();
 		die "expected comma, got $y->{P}\n" unless $y->{E} eq ',';
+		if (!length $fmt and
+		    !length $xs and
+		    @xs and
+		    $xs[0]{E} eq '.') {
+		    # X has   "<earlier>" .                <something>
+		    # Y has   "<earlier>%s" [other args] , <something>
+		    $next_x->(); # eat the '.'
+		    next;
+		}
 		if ($xs =~ m{^\@}) {
 		    $next_y->();
 		    die "\@... => not string" unless is_string($y);
@@ -267,16 +310,22 @@ sub analyse_chunk_core () {
 		    }
 		    $next_y->();
 		    if (!$bras and
-			(grep { $y->{E} eq $_ } qw( or xor and not ; : , )
+			(grep { $y->{E} eq $_ } qw( or xor and not ; :
+						    if unless while when )
+			 or $y->{E} eq ','
 			 or $y->{T} eq 'ket'
 			)) {
 			# lookahead shows close of containing scope
 			# or lower precedence operator
 			unshift @ys, $y;
+			pop @analysed_y;
 			last;
 		    }
-		    $xs =~ s{^\s+}{};
-		    #debug $ichunkstart, "TOKEN $y->{P}\n";
+		    $xs =~ s{^\s+}{} if $bras;
+		    if (is_string($y) and $y->{Q} eq '"') {
+			$exactly->($y->{V}, $y->{P});
+			next;
+		    }
 		    $exactly->($y->{E}, $y->{P});
 		    if ($y->{T} eq 'bra' or $y->{E} eq '?') {
 			$bras++;