X-Git-Url: http://www.git.stargrave.org/?a=blobdiff_plain;f=lib%2FPublicInbox%2FViewDiff.pm;h=124a723a5fb465247963a5d01fd5c40fda593c11;hb=HEAD;hp=a8045687ef48b59c0755f4e7b8f750aa722a07d3;hpb=24d67d4f20decf06e73764ac53b68d16f06c8f12;p=public-inbox.git diff --git a/lib/PublicInbox/ViewDiff.pm b/lib/PublicInbox/ViewDiff.pm index a8045687..124a723a 100644 --- a/lib/PublicInbox/ViewDiff.pm +++ b/lib/PublicInbox/ViewDiff.pm @@ -1,4 +1,4 @@ -# Copyright (C) 2019 all contributors +# Copyright (C) all contributors # License: AGPL-3.0+ # # used by PublicInbox::View @@ -7,59 +7,57 @@ # (or reconstruct) blobs. package PublicInbox::ViewDiff; -use strict; -use warnings; -use base qw(Exporter); -our @EXPORT_OK = qw(flush_diff); +use v5.12; +use parent qw(Exporter); +our @EXPORT_OK = qw(flush_diff uri_escape_path); use URI::Escape qw(uri_escape_utf8); -use PublicInbox::Hval qw(ascii_html); +use PublicInbox::Hval qw(ascii_html to_attr); use PublicInbox::Git qw(git_unquote); -sub DSTATE_INIT () { 0 } -sub DSTATE_STAT () { 1 } # TODO -sub DSTATE_HEAD () { 2 } # /^diff --git /, /^index /, /^--- /, /^\+\+\+ / -sub DSTATE_CTX () { 3 } # /^ / -sub DSTATE_ADD () { 4 } # /^\+/ -sub DSTATE_DEL () { 5 } # /^\-/ -my @state2class = ( - '', # init - '', # stat - 'head', - '', # ctx - 'add', - 'del' -); - -sub UNSAFE () { "^A-Za-z0-9\-\._~/" } - -my $OID_NULL = '0{7,40}'; -my $OID_BLOB = '[a-f0-9]{7,40}'; -my $PATH_A = '"?a/.+|/dev/null'; -my $PATH_B = '"?b/.+|/dev/null'; - -sub to_html ($$) { - $_[0]->linkify_1($_[1]); - $_[0]->linkify_2(ascii_html($_[1])); +my $OID_NULL = '0{7,}'; +my $OID_BLOB = '[a-f0-9]{7,}'; +my $LF = qr!\n!; +my $ANY = qr![^\n]!; +my $FN = qr!(?:"?[^/\n]+/[^\n]+|/dev/null)!; + +# cf. git diff.c :: get_compact_summary +my $DIFFSTAT_COMMENT = + qr/(?: *\((?:new|gone|(?:(?:new|mode) [\+\-][lx]))\))? *\z/s; +my $NULL_TO_BLOB = qr/^(index $OID_NULL\.\.)($OID_BLOB)\b/ms; +my $BLOB_TO_NULL = qr/^index ($OID_BLOB)(\.\.$OID_NULL)\b/ms; +my $BLOB_TO_BLOB = qr/^index ($OID_BLOB)\.\.($OID_BLOB)/ms; +our $EXTRACT_DIFFS = qr/( + (?: # begin header stuff, don't capture filenames, here, + # but instead wait for the --- and +++ lines. + (?:^diff\x20--git\x20$FN\x20$FN$LF) + + # old mode || new mode || copy|rename|deleted|... + (?:^[a-z]$ANY+$LF)* + )? # end of optional stuff, everything below is required + ^index\x20($OID_BLOB)\.\.($OID_BLOB)$ANY*$LF + ^---\x20($FN)$LF + ^\+{3}\x20($FN)$LF)/msx; +our $IS_OID = qr/\A$OID_BLOB\z/s; + +sub uri_escape_path { + # '/' + $URI::Escape::Unsafe{RFC3986} + uri_escape_utf8($_[0], "^A-Za-z0-9\-\._~/"); } # link to line numbers in blobs -sub diff_hunk ($$$$) { - my ($dctx, $spfx, $ca, $cb) = @_; - my $oid_a = $dctx->{oid_a}; - my $oid_b = $dctx->{oid_b}; - - (defined($spfx) && defined($oid_a) && defined($oid_b)) or - return "@@ $ca $cb @@"; - - my ($n) = ($ca =~ /^-(\d+)/); - $n = defined($n) ? do { ++$n; "#n$n" } : ''; - - my $rv = qq(@@ {Q}$n">$ca); - - ($n) = ($cb =~ /^\+(\d+)/); - $n = defined($n) ? do { ++$n; "#n$n" } : ''; - - $rv .= qq( {Q}$n">$cb @@); +sub diff_hunk ($$$) { + my ($dctx, $ca, $cb) = @_; + my ($oid_a, $oid_b, $spfx) = @$dctx{qw(oid_a oid_b spfx)}; + + if (defined($spfx) && defined($oid_a) && defined($oid_b)) { + my $n = ($ca =~ /^-([0-9]+)/) ? "#n$1" : ''; + my $x = qq(@@ {Q}$n">$ca); + + $n = ($cb =~ /^\+([0-9]+)/) ? "#n$1" : ''; + $x .= qq( {Q}$n">$cb @@); + } else { + "@@ $ca $cb @@"; + } } sub oid ($$$) { @@ -67,95 +65,184 @@ sub oid ($$$) { defined($spfx) ? qq({Q}">$oid) : $oid; } -sub to_state ($$$) { - my ($dst, $state, $new_state) = @_; - $$dst .= '' if $state2class[$state]; - $_[1] = $new_state; - my $class = $state2class[$new_state] or return; - $$dst .= qq(); +# returns true if diffstat anchor written, false otherwise +sub anchor0 ($$$) { + my ($ctx, $fn, $rest) = @_; + + my $orig = $fn; + + # normal git diffstat output is impossible to parse reliably + # without --numstat, and that isn't the default for format-patch. + # So only do best-effort handling of renames for common cases; + # which works well in practice. If projects put "=>", or trailing + # spaces in filenames, oh well :P + $fn =~ s/$DIFFSTAT_COMMENT//; + $fn =~ s/\{(?:.+) => (.+)\}/$1/ or $fn =~ s/.* => (.+)/$1/; + $fn = git_unquote($fn); + + # long filenames will require us to check in anchor1() + push(@{$ctx->{-long_path}}, $fn) if $fn =~ s!\A\.\.\./?!!; + + my $attr = to_attr($ctx->{-apfx}.$fn) // return; + $ctx->{-anchors}->{$attr} = 1; + my $spaces = ($orig =~ s/( +)\z//) ? $1 : ''; + print { $ctx->{zfh} } " ", + ascii_html($orig), '', $spaces, + $ctx->{-linkify}->to_html($rest); } -sub flush_diff ($$$$) { - my ($dst, $spfx, $linkify, $diff) = @_; - my $state = DSTATE_INIT; - my $dctx = { Q => '' }; # {}, keys: oid_a, oid_b, path_a, path_b +# returns "diff --git" anchor destination, undef otherwise +sub anchor1 ($$) { + my ($ctx, $pb) = @_; + my $attr = to_attr($ctx->{-apfx}.$pb) // return; + + my $ok = delete $ctx->{-anchors}->{$attr}; + + # unlikely, check the end of long path names we captured, + # assume diffstat and diff output follow the same order, + # and ignore different ordering (could be malicious input) + unless ($ok) { + my $fn = shift(@{$ctx->{-long_path}}) // return; + $pb =~ /\Q$fn\E\z/s or return; + $attr = to_attr($ctx->{-apfx}.$fn) // return; + $ok = delete $ctx->{-anchors}->{$attr} // return; + } + $ok ? "diff --git" : undef +} - foreach my $s (@$diff) { - if ($s =~ /^ /) { - if ($state2class[$state]) { - to_state($dst, $state, DSTATE_CTX); - } - $$dst .= to_html($linkify, $s); - } elsif ($s =~ /^-- $/) { # email signature begins - $state == DSTATE_INIT or - to_state($dst, $state, DSTATE_INIT); - $$dst .= $s; - } elsif ($s =~ m!^diff --git ($PATH_A) ($PATH_B)$!) { - if ($state != DSTATE_HEAD) { - my ($pa, $pb) = ($1, $2); - to_state($dst, $state, DSTATE_HEAD); - $pa = (split('/', git_unquote($pa), 2))[1]; - $pb = (split('/', git_unquote($pb), 2))[1]; - $dctx = { - Q => "?b=".uri_escape_utf8($pb, UNSAFE), - }; - if ($pa ne $pb) { - $dctx->{Q} .= - "&a=".uri_escape_utf8($pa, UNSAFE); +sub diff_header ($$$) { + my ($x, $ctx, $top) = @_; + my (undef, undef, $pa, $pb) = splice(@$top, 0, 4); # ignore oid_{a,b} + my $spfx = $ctx->{-spfx}; + my $dctx = { spfx => $spfx }; + + # get rid of leading "a/" or "b/" (or whatever --{src,dst}-prefix are) + $pa = (split(m'/', git_unquote($pa), 2))[1] if $pa ne '/dev/null'; + $pb = (split(m'/', git_unquote($pb), 2))[1] if $pb ne '/dev/null'; + if ($pa eq $pb && $pb ne '/dev/null') { + $dctx->{Q} = '?b='.uri_escape_path($pb); + } else { + my @q; + push @q, 'b='.uri_escape_path($pb) if $pb ne '/dev/null'; + push @q, 'a='.uri_escape_path($pa) if $pa ne '/dev/null'; + $dctx->{Q} = '?'.join('&', @q); + } + + # linkify early and all at once, since we know the following + # subst ops on $$x won't need further escaping: + $$x = $ctx->{-linkify}->to_html($$x); + + # no need to capture oid_a and oid_b on add/delete, + # we just linkify OIDs directly via s///e in conditional + if ($$x =~ s/$NULL_TO_BLOB/$1 . oid($dctx, $spfx, $2)/e) { + push @{$ctx->{-qry}->{dfpost}}, $2; + } elsif ($$x =~ s/$BLOB_TO_NULL/'index '.oid($dctx, $spfx, $1).$2/e) { + push @{$ctx->{-qry}->{dfpre}}, $1; + } elsif ($$x =~ $BLOB_TO_BLOB) { + # modification-only, not add/delete: + # linkify hunk headers later using oid_a and oid_b + @$dctx{qw(oid_a oid_b)} = ($1, $2); + push @{$ctx->{-qry}->{dfpre}}, $1; + push @{$ctx->{-qry}->{dfpost}}, $2; + } else { + warn "BUG? <$$x> had no ^index line"; + } + $$x =~ s!^diff --git!anchor1($ctx, $pb) // 'diff --git'!ems; + print { $ctx->{zfh} } qq(), $$x, ''; + $dctx; +} + +sub diff_before_or_after ($$) { + my ($ctx, $x) = @_; + if (exists $ctx->{-anchors} && $$x =~ # diffstat lines: + /((?:^\x20(?:[^\n]+?)(?:\x20+\|\x20[^\n]*\n))+) + (\x20[0-9]+\x20files?\x20)changed,/msx) { + my $pre = substr($$x, 0, $-[0]); # (likely) short prefix + substr($$x, 0, $+[0], ''); # sv_chop on $$x ($$x may be long) + my @x = ($2, $1); + my $lnk = $ctx->{-linkify}; + my $zfh = $ctx->{zfh}; + # uninteresting prefix + print $zfh $lnk->to_html($pre); + for my $l (split(/^/m, pop(@x))) { # $2 per-file stat lines + $l =~ /^ (.+)( +\| .*\z)/s and + anchor0($ctx, $1, $2) and next; + print $zfh $lnk->to_html($l); + } + my $ch = $ctx->{changed_href} // '#related'; + print $zfh pop(@x), # $3 /^ \d+ files? / + qq(changed,), + # insertions/deletions, notes, commit message, etc: + $lnk->to_html($$x); + } else { + print { $ctx->{zfh} } $ctx->{-linkify}->to_html($$x); + } +} + +# callers must do CRLF => LF conversion before calling this +sub flush_diff ($$) { + my ($ctx, $cur) = @_; + + my @top = split($EXTRACT_DIFFS, $$cur); + undef $$cur; # free memory + + my $lnk = $ctx->{-linkify}; + my $dctx; # {}, keys: Q, oid_a, oid_b + my $zfh = $ctx->zfh; + + while (defined(my $x = shift @top)) { + if (scalar(@top) >= 4 && + $top[1] =~ $IS_OID && + $top[0] =~ $IS_OID) { + $dctx = diff_header(\$x, $ctx, \@top); + } elsif ($dctx) { + open(my $afh, '>>:utf8', \(my $after='')) or + die "open: $!"; + + # Quiet "Complex regular subexpression recursion limit" + # warning. Perl will truncate matches upon hitting + # that limit, giving us more (and shorter) scalars than + # would be ideal, but otherwise it's harmless. + # + # We could replace the `+' metacharacter with `{1,100}' + # to limit the matches ourselves to 100, but we can + # let Perl do it for us, quietly. + no warnings 'regexp'; + + for my $s (split(/((?:(?:^\+[^\n]*\n)+)| + (?:(?:^-[^\n]*\n)+)| + (?:^@@ [^\n]+\n))/xsm, $x)) { + undef $x; + if (!defined($dctx)) { + print $afh $s; + } elsif ($s =~ s/\A@@ (\S+) (\S+) @@//) { + print $zfh qq(), + diff_hunk($dctx, $1, $2), + $lnk->to_html($s), + ''; + } elsif ($s =~ /\A\+/) { # $s may be huge + print $zfh qq(), + $lnk->to_html($s), + ''; + } elsif ($s =~ /\A-- $/sm) { # email sig starts + $dctx = undef; + print $afh $s; + } elsif ($s =~ /\A-/) { # $s may be huge + print $zfh qq(), + $lnk->to_html($s), + ''; + } else { # $s may be huge + print $zfh $lnk->to_html($s); } } - $$dst .= to_html($linkify, $s); - } elsif ($s =~ s/^(index $OID_NULL\.\.)($OID_BLOB)\b//o) { - $$dst .= $1 . oid($dctx, $spfx, $2); - $dctx = { Q => '' }; - $$dst .= to_html($linkify, $s) ; - } elsif ($s =~ s/^index ($OID_BLOB)(\.\.$OID_NULL)\b//o) { - $$dst .= 'index ' . oid($dctx, $spfx, $1) . $2; - $dctx = { Q => '' }; - $$dst .= to_html($linkify, $s); - } elsif ($s =~ /^index ($OID_BLOB)\.\.($OID_BLOB)/o) { - $dctx->{oid_a} = $1; - $dctx->{oid_b} = $2; - $$dst .= to_html($linkify, $s); - } elsif ($s =~ s/^@@ (\S+) (\S+) @@//) { - $$dst .= '' if $state2class[$state]; - $$dst .= qq(); - $$dst .= diff_hunk($dctx, $spfx, $1, $2); - $$dst .= ''; - $state = DSTATE_CTX; - $$dst .= to_html($linkify, $s); - } elsif ($s =~ m!^--- $PATH_A! || - $s =~ m!^\+{3} $PATH_B!) { - # color only (no oid link) - $state == DSTATE_INIT and - to_state($dst, $state, DSTATE_HEAD); - $$dst .= to_html($linkify, $s); - } elsif ($s =~ /^\+/) { - if ($state != DSTATE_ADD && $state != DSTATE_INIT) { - to_state($dst, $state, DSTATE_ADD); - } - $$dst .= to_html($linkify, $s); - } elsif ($s =~ /^-/) { - if ($state != DSTATE_DEL && $state != DSTATE_INIT) { - to_state($dst, $state, DSTATE_DEL); + if (!$dctx) { + utf8::decode($after); + diff_before_or_after($ctx, \$after); } - $$dst .= to_html($linkify, $s); - # ignore the following lines in headers: - } elsif ($s =~ /^(?:dis)similarity index/ || - $s =~ /^(?:old|new) mode/ || - $s =~ /^(?:deleted|new) file mode/ || - $s =~ /^(?:copy|rename) (?:from|to) / || - $s =~ /^(?:dis)?similarity index /) { - $$dst .= to_html($linkify, $s); } else { - $state == DSTATE_INIT or - to_state($dst, $state, DSTATE_INIT); - $$dst .= to_html($linkify, $s); + diff_before_or_after($ctx, \$x); } } - @$diff = (); - $$dst .= '' if $state2class[$state]; - undef; } 1;