#!/usr/bin/perl -w
#
# i18n-diff-auditor
# Copyright (C)2018 Ian Jackson
# GPLv3+, NO WARRANTY, see below.
#
#
# Usage:
#  something like this
#  git-log -n1 -p | ./i18n-diff-auditor -D 2>&1 |less -j10 +/'^!.*'
#
# -D is for debug.  Currently only one level.
#
# Output is the relevant diff hunks, with each line prepended with
# space for ok lines and ! for questionable ones, and with relevant
# diff lines prepended with lines starting !! (and lines starting #
# for debug output), so ovrall:
#
#   !! <message>   reasoning for subsequent questionable diff line(s)
#   !+             diff line found to be questionable
#   !-             diff line found to be questionable
#    @@@ etc.      diff furniture
#    +             diff line checked and ok
#    -             diff line checked and ok
#   # <stuff>      debug output (normally precedes relevant output)
#
# Changes are generally marked as ok if they correspond to a known
# intended code change pattern.  (That includes changing error calls
# to different error calls.)  If they don't correspond to any known
# pattern, they are "questionable" and the first thing that doesn't
# match the most common pattern is reported.
#
# Might be useful for projects other than dgit, provided it uses
# the same gettext aliases (__ f_ i_) and similar error calls
# (die, confess, fail).
#
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program.  If not, see <https://www.gnu.org/licenses/>.

use strict;
use Carp;
use Data::Dumper;
use Getopt::Long;

our $debug = 0;
GetOptions("debug|D+" => \$debug
	   );

our @debug;
sub debug ($$) {
    my ($i,$s) = @_;
    push @{ $debug[$i] }, $s if $debug;
}

my @d = <>;
unshift @d, "# dummy line to make line 1 index 1 in \@d\n";

our $i_last_l_ok = -1;
our $count_i_last_l_ok;

sub l_ok ($) {
    my ($i) = @_;

    if ($i == $i_last_l_ok) {
	confess $i if $count_i_last_l_ok++ > 50;
    } else {
	$count_i_last_l_ok = 0;
	$i_last_l_ok = $i;
    }

    return unless $i < @d;
    $_ = $d[$i];
    #print STDERR "L $i\n";
    1;
}

sub l ($) {
    my ($i) = @_;
    confess $i unless l_ok $i;
};

our $perlop_text = <<'END'; # c&p from man perlop
           left        terms and list operators (leftward)
           left        ->
           nonassoc    ++ --
           right       **
           right       ! ~ \ and unary + and -
           left        =~ !~
           left        * / % x
           left        + - .
           left        << >>
           nonassoc    named unary operators
           nonassoc    < > <= >= lt gt le ge
           nonassoc    == != <=> eq ne cmp ~~
           left        &
           left        | ^
           left        &&
           left        || //
           nonassoc    ..  ...
           right       ?:
           right       = += -= *= etc. goto last next redo dump
           left        , =>
           nonassoc    list operators (rightward)
           right       not
           left        and
           left        or xor

           **=    +=    *=    &=    &.=    <<=    &&=
                  -=    /=    |=    |.=    >>=    ||=
                  .=    %=    ^=    ^.=           //=
                        x=
END

our $perlop_re;

sub prep_perlop () {
    my @ops;
    foreach (split /\n/, $perlop_text) {
	next unless m{\S};
	s{\s+$}{};
	s{^\s+}{};
	s{^(?: left | right | nonassoc ) \s+}{}x;
	next if m{^terms and list operators};
	next if m{^named unary};
	next if m{^list operators};
	s{ and unary.*}{};
	s{ etc\. }{ };
	s{\?\:}{ ? : };
	foreach my $op (split /\s+/) {
	    next unless length $op;
	    next if $op =~ m{^\w+$};
	    $op =~ s/\W/\\$&/g;
	    push @ops, $op;
	}
    }
    $perlop_re = '(?: '.(join ' | ', @ops).' )';
    $perlop_re = qr{$perlop_re}x;
    #print STDERR "$perlop_re\n";
}

prep_perlop();

our ($ifilehead, $ifirsthunkhead);
our ($ihunkhead, $ihunkend);
our ($ichunkstart, $ichunkend);
our ($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)
    #   P     something to print in messages
    #   V     value, only for: heredoc string
    #   Q     quote characcter, only for: heredoc string
    for (;;) {
	s{^\s+}{};
	if (s{^[\$\@\%]?[_0-9a-zA-Z]+}{}) {
	    push @o, { T => 'ident', E => $&, P => $& };
	} elsif (s{^\<\<(['"]?)([A-Z_]+)\1}{}) {
	    my ($q,$d) = ($1,$2);
	    $q ||= '"';
	    push @o, { T => 'heredoc', Q => $q, Delim => $d,
		       E => $&, P => "<<$q$d$q" };
	    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 ) [^"]
                              )*
                       ) \1 }{}x) {
	    my ($q,$v) = ($1,$2);
	    push @o, { T => 'string', E => $&, P => "$q$q",
		       Q => $q, V => $v};
	} elsif (s{^$perlop_re|^\;}{}) {
	    push @o, { T => 'op', E => $&, P => $& };
	} elsif (s/^[[{(]//) {
	    push @o, { T => 'bra', E => $&, P => $& };
	} elsif (s/^[]})]//) {
	    push @o, { T => 'ket', E => $&, P => $& };
	} elsif (s/^( [\$\@\%] )( \{ )//x) {
	    push @o, { T => 'deref', E => $1, P => $1 },
		     { T => 'bra',   E => $2, P => $2 };
	} elsif (s/^ [\$\@\%] [^[^{] //x) {
	    push @o, { T => 'specvar', E => $&, P => $& };
	} elsif (!length) {
	    last;
	} elsif (s{^\#.*\n}{}) {
	} else {
	    m{^.{0,10}};
	    die "cannot tokenise \`$&'";
	}
    }
    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;
	next if grep { $_->{Invented} } @inputs;
	my $new = { T => 'joinedstrings',
		    E => (join '.', map { $_->{E} } @inputs),
		    P => (join '.', map { $_->{P} } @inputs),
		    V => (join '',  map { $_->{V} } @inputs),
		    Q => $q,
		  };
	@o = (@o[0..$i-1], $new, @o[$i+3..$#o]);
	$i--; # counteracts $i++
    }
    debug $ichunkstart, "semiparsed: ".join ' ', map { $_->{P} } @o;
    # debug $ichunkstart, "semiparsed V: ".join ' ', map { defined $_->{V} ? ">$_->{V}<" : '-' } @o;
    return @o;
}	    

our @analysed_x;
our @analysed_y;

sub analyse_chunk_core () {
    $before //= '';
    die "plain deletion\n" unless defined $after;
    my @xs = semiparse $before;
    my @ys = semiparse $after;
    @analysed_x = @analysed_y = ();
    my $next_something = sub {
	my ($ary,$anal,$var,$what) = @_;
	die "ran out of $what\n" unless @$ary;
	my $r = shift @$ary;
	push @$anal, $r->{P};
	$$var = $r;
    };
    my ($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 = ();
  ANALYSE:
    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';
	next if $x->{E} eq 'die'     and $y->{E} eq 'fail';
	foreach my $with_fh (qw(0 1)) {
	    next unless $x->{E} eq 'printf';
	    next unless $y->{E} eq 'print';
	    next unless @xs >= $with_fh;
	    next unless @ys >  $with_fh;
	    if ($with_fh) {
		next unless $xs[0]{E} eq $ys[0]{E};
		next unless
		    $xs[0]{E} =~ m{^[A-Z]+$} or
		    $xs[0]{T} eq 'ident' && $xs[0]{E} =~ m{^\$};
	    }
	    next unless $ys[$with_fh]{E} eq 'f_';
	    # yay!
	    $next_x->() if $with_fh;
	    $next_y->() if $with_fh;
	    $next_y->(); # f_
	    next ANALYSE;
	}
	if ($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 = $y->{E};
	if (is_trans($y)) {
	    $next_y->();
	    die "__ on non-string $y->{P}\n"     unless is_string($y);
	    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{(?<!\\) [\$\@]};
	    }
	    eval {
		die "__ string changed\n"       unless $y->{V} eq $x->{V};
		die "__ string quote changed\n" unless $y->{Q} eq $x->{Q};
	    };
	    $string_changed = $@;
	}
	if ($ye eq '__') {
	    $_ = $y->{V};
	    die "percent $& in __ ' string\n" if m{\%};
	    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{\%};
	    next unless $string_changed;
	    die "f_ old string '-quoted\n" if $x->{Q} ne '"';
	    my $xs = $x->{V};
	    my $exactly = sub {
		my ($lit, $what) = @_;
		my $xl = substr($xs, 0, length($lit));
		if ($xl ne $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)');
		$fmt = $';
		if ($& eq '%%') { $exactly->('%', '%%'); next; }
		elsif ($& 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);
		    die "\@... => $y->{P}" if $y->{Q} ne '"';
		    $exactly->($y->{V}, $y->{P});
		    next;
		}
		my $bras = 0;
		for (;;) {
		    if (!$bras and !@ys) {
			last;
		    }
		    $next_y->();
		    if (!$bras and
			(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+}{} 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 '?') {
			$bras++;
		    } elsif ($y->{T} eq 'ket' or $y->{E} eq ':') {
			die "too many kets at $y->{E}\n" unless $bras;
			$bras--;
		    }
		}
	    }
	    next;
	}
	die "mismatch $x->{P} => $y->{P}\n";
    }
}

sub analyse_chunk () {
    for (;;) {
	eval { analyse_chunk_core(); };
	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;
	} else {
	    die $@;
	}
    }
}

our @report;
our $last_filehead = -1;

sub report_on_hunk () {
    return unless @report;
    if ($last_filehead != $ifilehead) {
	foreach (my $i=$ifilehead; $i<$ifirsthunkhead; $i++) {
	    print $d[$i];
	}
	$last_filehead = $ifilehead;
    }
    my $dummy_r = { S => (scalar @d)+1, E => (scalar @d)+1 };
    my $r;
    for (my $i=$ihunkhead; ; $i++) {
	for (;;) {
	    $r //= shift @report;
	    $r //= $dummy_r;
	    last if $i < $r->{E};
	    confess unless $r->{Done} == 03;
	    $r = undef;
	}

	last unless $i<$ihunkend;

	foreach my $ds (@{ $debug[$i] }) {
	    print "# $ds\n";
	}

	if ($i == $r->{S}) {
	    print "!! $r->{M}";
	    $r->{Done} |= 01;
	}
	if ($i >= $r->{S}) {
	    print "!";
	    $r->{Done} |= 02;
	} else {
	    print " ";
	}
	print $d[$i];
    }
    confess unless $r = $dummy_r;
}

for ($ifilehead = 0; l_ok $ifilehead; $ifilehead++) {
    m{^diff} or next;
    $ifirsthunkhead = $ifilehead;
    while (l_ok $ifirsthunkhead and
	   m{^diff|^index|^---|^\Q+++\E}) {
	$ifirsthunkhead++
    }
    $ihunkhead = $ifirsthunkhead;
    while (l_ok $ihunkhead) {
	m{^\@\@} or confess "$ihunkhead $_ ?";
	my $i = $ihunkhead + 1;
	for (; ; $i++) {
	    if (!l_ok $i or m{^ } or m{^\@\@}) {
		if (defined $ichunkstart) {
		    $ichunkend = $i;
		    eval { analyse_chunk(); 1; };
		    if (length $@) {
			debug $ichunkstart, "done x: @analysed_x";
			debug $ichunkstart, "done y: @analysed_y";
			push @report, { M => $@,
					S => $ichunkstart,
					E => $ichunkend };
		    }
		    $ichunkstart = $ichunkend = $before = $after = undef;
		}
		l_ok $i or last;
		m{^\@\@} and last;
	    } elsif (m{^[-+]}) {
		my $which = $& eq '-' ? \$before : \$after;
		$ichunkstart //= $i;
		$$which //= '';
		$$which .= $';
	    } else {
		confess "$i $_ ?";
	    }
	}
	$ihunkend = $i;
	report_on_hunk();
	$ichunkend = $i;
	$ihunkhead = $i;
    }
}