i18n: i18n-diff-auditor: support die -> fail

[dgit.git] / i18n-diff-auditor

diff --git a/i18n-diff-auditor b/i18n-diff-auditor
index e8914c004e0d20e0de81675d913430d2ac0d924f..e2c010142d334bc42b439bf8b9c590e7a9c95802 100755 (executable)
--- a/i18n-diff-auditor
+++ b/i18n-diff-auditor
@@ -110,9 +110,19 @@ our ($before, $after);
 sub is_string ($) { $_[0]{T} =~ m/heredoc|string/; }
 sub is_trans ($) { grep { $_[0]{E} eq $_ } qw(__ f_ i_); }
 
+sub qp ($) {
+    my ($p) = @_;
+    $p =~ s{\\}{\\\\}g;
+    $p =~ s{\'}{\\'}g;
+    $p =~ s{\n}{\\n}g;
+    $p =~ s{\t}{\\t}g;
+    return "'$p'";
+};
+
 sub semiparse ($) {
     ($_) = @_;
     my @o;
+    #my $in = $_;
     # entries contain
     #   T     type
     #   E     exact input text (does not contain here doc contents)
@@ -128,12 +138,18 @@ sub semiparse ($) {
            $q ||= '"';
            push @o, { T => 'heredoc', Q => $q, Delim => $d,
                       E => $&, P => "<<$q$d$q" };
-           s{^
-                (             .* \n    )
-                ( (?: (?! $d) .* \n )* )
-                          $d     \n
-              }{ $1 }x or die "missing end of here doc $d\n";
-           $o[$#o]{V} = $2;
+           if (s{^
+                   (                 .* \n     )
+                   ( (?: (?! $d \n ) .* \n )*? )
+                             $d         \n
+                }{ $1 }xe) {
+               $o[$#o]{V} = $2;
+            } else {
+               m{^.*\n} or confess;
+               $_ = $&;
+               $o[$#o]{V} = $';
+               $o[$#o]{Invented} = 1;
+           }
        } elsif (s{^ (["'])( (?: [^\\'"]
                                | \\ [^"']
                                | (?! \1 ) [^"]
@@ -144,9 +160,9 @@ 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 => $1 },
@@ -167,6 +183,7 @@ sub semiparse ($) {
        next if grep { !is_string($_) } @inputs;
        my $q = $inputs[0]{Q};
        next if grep { $_->{Q} ne $q } @inputs;
+       next if grep { $_->{Invented} } @inputs;
        my $new = { T => 'joinedstrings',
                    E => (join '.', map { $_->{E} } @inputs),
                    P => (join '.', map { $_->{P} } @inputs),
@@ -177,6 +194,7 @@ sub semiparse ($) {
        $i--; # counteracts $i++
     }
     debug $ichunkstart, "semiparsed: ".join ' ', map { $_->{P} } @o;
+    # debug $ichunkstart, "semiparsed V: ".join ' ', map { defined $_->{V} ? ">$_->{V}<" : '-' } @o;
     return @o;
 }          
 
@@ -210,12 +228,23 @@ 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';
+       next if $x->{E} eq 'die'     and $y->{E} eq 'fail';
        if ($y->{E} eq '+'
            and @ys >= 3
            and $ys[0]{E} eq '('
            and is_trans($ys[1])) {
            $next_y->(); # (
-           $next_y->(); # __ f_
+           $next_y->(); # __ f_ i_
+           @y_expect_suffix = ')';
+       } elsif ($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 $string_changed;
@@ -223,7 +252,7 @@ sub analyse_chunk_core () {
        if (is_trans($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 $y->{V} =~ m{(?<!\\) [\$\@]};
@@ -254,17 +283,20 @@ sub analyse_chunk_core () {
                my ($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 $ichunkstart, "not exactly x: ..".qp($xs);
+                   debug $ichunkstart, "not exactly y:   ".qp($lit);
                    my $next = @ys ? $ys[0]{P} : '(end)';
                    die "string contents mismatch near $what before $next\n";
                }
                $xs = substr($xs, length($lit));
            };
            for (;;) {
+               #print STDERR Dumper($fmt, $xs, \@xs, @ys);
                if ($fmt !~ m{\%[^\%]}) {
                    $exactly->($fmt, '(tail)');
                    $fmt = '';
+                   die "text deleted from end of string: ".qp($xs)."\n"
+                       if length $xs;
                    last;
                }
                $exactly->($`, '(literal)');
@@ -273,6 +305,15 @@ 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);
@@ -298,7 +339,11 @@ sub analyse_chunk_core () {
                        pop @analysed_y;
                        last;
                    }
-                   $xs =~ s{^\s+}{};
+                   $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++;
@@ -320,6 +365,8 @@ sub analyse_chunk () {
        return unless length $@;
        if ($@ =~ m{^missing end of here doc (\S+)\n}) {
            # fudge this
+           # (this never happens now, but in the future we might
+           # want this code again eg to try adding to the chunk)
            $before .= "\n$1\n";
            $after .= "\n$1\n";
            next;