i18n: i18n-diff-auditor: fix bra and ket regexps

[dgit.git] / i18n-diff-auditor

diff --git a/i18n-diff-auditor b/i18n-diff-auditor
index b4fc5968ad66bde2bae169b9457c9dc831b4b3bb..42ebe00d9dbc85f6c1937ad1df92ee89d29921cc 100755 (executable)
--- 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 ($) {
     ($_) = @_;
@@ -143,9 +144,9 @@ sub semiparse ($) {
                       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 $q } @inputs;
@@ -172,7 +174,7 @@ sub semiparse ($) {
                    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;
     return @o;
@@ -208,17 +210,19 @@ sub analyse_chunk_core () {
        $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 $ys[0]{E} eq '('
-           and ($ys[1]{E} eq '__' or $ys[2]{E} eq 'f_')) {
+           and is_trans($ys[1])) {
            $next_y->(); # (
            $next_y->(); # __ f_
            @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);
@@ -238,6 +242,10 @@ 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 $fmt =~ m{\%};
@@ -248,8 +256,8 @@ sub analyse_chunk_core () {
                my ($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";
+                   debug $ichunkstart, "not exactly x: \"..$xs\"";
+                   debug $ichunkstart, "not exactly y:   $lit";
                    my $next = @ys ? $ys[0]{P} : '(end)';
                    die "string contents mismatch near $what before $next\n";
                }
@@ -267,6 +275,15 @@ sub analyse_chunk_core () {
                elsif ($& 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);
@@ -289,9 +306,10 @@ sub analyse_chunk_core () {
                        # lookahead shows close of containing scope
                        # or lower precedence operator
                        unshift @ys, $y;
+                       pop @analysed_y;
                        last;
                    }
-                   $xs =~ s{^\s+}{};
+                   $xs =~ s{^\s+}{} if $bras;
                    $exactly->($y->{E}, $y->{P});
                    if ($y->{T} eq 'bra' or $y->{E} eq '?') {
                        $bras++;