use warnings;
use strict;
use IkiWiki;
-use Encode;
-my (%backlinks, %rendered);
+my (%backlinks, %rendered, %scanned);
our %brokenlinks;
my $links_calculated=0;
}
if (defined $config{historyurl} && length $config{historyurl}) {
my $u=$config{historyurl};
- $u=~s/\[\[file\]\]/$pagesources{$page}/g;
+ my $p=uri_escape_utf8($pagesources{$page}, '^A-Za-z0-9\-\._~/');
+ $u=~s/\[\[file\]\]/$p/g;
$template->param(historyurl => $u);
$actions++;
}
sub scan ($) {
my $file=shift;
+ return if $phase > PHASE_SCAN || $scanned{$file};
+ $scanned{$file}=1;
debug(sprintf(gettext("scanning %s"), $file));
}
delete $typedlinks{$page};
+ # Preprocess in scan-only mode.
+ preprocess($page, $page, $content, 1);
+
run_hooks(scan => sub {
shift->(
page => $page,
content => $content,
);
});
-
- # Preprocess in scan-only mode.
- preprocess($page, $page, $content, 1);
}
else {
will_render($file, $file, 1);
}
}
-sub prune ($) {
+sub prune ($;$) {
my $file=shift;
+ my $up_to=shift;
unlink($file);
my $dir=dirname($file);
- while (rmdir($dir)) {
+ while ((! defined $up_to || $dir =~ m{^\Q$up_to\E\/}) && rmdir($dir)) {
$dir=dirname($dir);
}
}
}
-sub find_src_files () {
+# Finds all files in the srcdir, and the underlaydirs.
+# Returns the files, and their corresponding pages.
+#
+# When run in only_underlay mode, adds only the underlay files to
+# the files and pages passed in.
+sub find_src_files (;$$$) {
+ my $only_underlay=shift;
my @files;
+ if (defined $_[0]) {
+ @files=@{shift()};
+ }
my %pages;
+ if (defined $_[0]) {
+ %pages=%{shift()};
+ }
+
eval q{use File::Find};
error($@) if $@;
- my ($page, $dir, $underlay);
+ eval q{use Cwd};
+ die $@ if $@;
+ my $origdir=getcwd();
+ my $abssrcdir=Cwd::abs_path($config{srcdir});
+
+ @IkiWiki::underlayfiles=();
+
+ my ($page, $underlay);
my $helper=sub {
my $file=decode_utf8($_);
return if -l $file || -d _;
- $file=~s/^\Q$dir\E\/?//;
+ $file=~s/^\.\///;
return if ! length $file;
$page = pagename($file);
if (! exists $pagesources{$page} &&
if ($underlay) {
# avoid underlaydir override attacks; see security.mdwn
- if (! -l "$config{srcdir}/$f" && ! -e _) {
+ if (! -l "$abssrcdir/$f" && ! -e _) {
if (! $pages{$page}) {
push @files, $f;
+ push @IkiWiki::underlayfiles, $f;
$pages{$page}=1;
}
}
}
};
- find({
- no_chdir => 1,
- wanted => $helper,
- }, $dir=$config{srcdir});
- $underlay=1;
- foreach (@{$config{underlaydirs}}, $config{underlaydir}) {
+ unless ($only_underlay) {
+ chdir($config{srcdir}) || die "chdir $config{srcdir}: $!";
find({
no_chdir => 1,
wanted => $helper,
- }, $dir=$_);
+ }, '.');
+ chdir($origdir) || die "chdir $origdir: $!";
+ }
+
+ $underlay=1;
+ foreach (@{$config{underlaydirs}}, $config{underlaydir}) {
+ if (chdir($_)) {
+ find({
+ no_chdir => 1,
+ wanted => $helper,
+ }, '.');
+ chdir($origdir) || die "chdir: $!";
+ }
};
+
return \@files, \%pages;
}
+# Given a hash of files that have changed, and a hash of files that were
+# deleted, should return the same results as find_src_files, with the same
+# sanity checks. But a lot faster!
+sub process_changed_files ($$) {
+ my $changed_raw=shift;
+ my $deleted_raw=shift;
+
+ my @files;
+ my %pages;
+
+ foreach my $file (keys %$changed_raw) {
+ my $page = pagename($file);
+ next if ! exists $pagesources{$page} && file_pruned($file);
+ my ($f) = $file =~ /$config{wiki_file_regexp}/; # untaint
+ if (! defined $f) {
+ warn(sprintf(gettext("skipping bad filename %s"), $file)."\n");
+ next;
+ }
+ push @files, $f;
+ if ($pages{$page}) {
+ debug(sprintf(gettext("%s has multiple possible source pages"), $page));
+ }
+ $pages{$page}=1;
+ }
+
+ # So far, we only have the changed files. Now add in all the old
+ # files that were not changed or deleted, excluding ones that came
+ # from the underlay.
+ my %old_underlay;
+ foreach my $f (@IkiWiki::underlayfiles) {
+ $old_underlay{$f}=1;
+ }
+ foreach my $page (keys %pagesources) {
+ my $f=$pagesources{$page};
+ unless ($old_underlay{$f} || exists $pages{$page} || exists $deleted_raw->{$f}) {
+ $pages{$page}=1;
+ push @files, $f;
+ }
+ }
+
+ # add in the underlay
+ find_src_files(1, \@files, \%pages);
+}
+
sub find_new_files ($) {
my $files=shift;
my @new;
foreach my $file (@$files) {
my $page=pagename($file);
+
+ if ($config{rcs} && $config{gettime} &&
+ -e "$config{srcdir}/$file") {
+ if (! $times_noted) {
+ debug(sprintf(gettext("querying %s for file creation and modification times.."), $config{rcs}));
+ $times_noted=1;
+ }
+
+ eval {
+ my $ctime=rcs_getctime($file);
+ if ($ctime > 0) {
+ $pagectime{$page}=$ctime;
+ }
+ };
+ if ($@) {
+ print STDERR $@;
+ }
+ my $mtime;
+ eval {
+ $mtime=rcs_getmtime($file);
+ };
+ if ($@) {
+ print STDERR $@;
+ }
+ elsif ($mtime > 0) {
+ utime($mtime, $mtime, "$config{srcdir}/$file");
+ }
+ }
+
if (exists $pagesources{$page} && $pagesources{$page} ne $file) {
# the page has changed its type
$forcerebuild{$page}=1;
if (isinternal($page)) {
push @internal_new, $file;
}
- elsif ($config{rcs}) {
+ else {
push @new, $file;
- if ($config{gettime} && -e "$config{srcdir}/$file") {
- if (! $times_noted) {
- debug(sprintf(gettext("querying %s for file creation and modification times.."), $config{rcs}));
- $times_noted=1;
- }
-
- eval {
- my $ctime=rcs_getctime("$config{srcdir}/$file");
- if ($ctime > 0) {
- $pagectime{$page}=$ctime;
- }
- };
- if ($@) {
- print STDERR $@;
- }
- my $mtime;
- eval {
- $mtime=rcs_getmtime("$config{srcdir}/$file");
- };
- if ($@) {
- print STDERR $@;
- }
- elsif ($mtime > 0) {
- utime($mtime, $mtime, "$config{srcdir}/$file");
- }
- }
}
$pagecase{lc $page}=$page;
if (! exists $pagectime{$page}) {
}
foreach my $old (@{$oldrenderedfiles{$page}}) {
- prune($config{destdir}."/".$old);
+ prune($config{destdir}."/".$old, $config{destdir});
}
foreach my $source (keys %destsources) {
}
delete $pagecase{lc $page};
+ $delpagesources{$page}=$pagesources{$page};
delete $pagesources{$page};
}
}
$pagemtime{$page}=$stat[9];
if (isinternal($page)) {
+ my $content = readfile($srcfile);
+
# Preprocess internal page in scan-only mode.
- preprocess($page, $page, readfile($srcfile), 1);
+ preprocess($page, $page, $content, 1);
+
+ run_hooks(scan => sub {
+ shift->(
+ page => $page,
+ content => $content,
+ );
+ });
+
push @internal_changed, $file;
}
else {
foreach my $file (@{$oldrenderedfiles{$page}}) {
if (! grep { $_ eq $file } @{$renderedfiles{$page}}) {
debug(sprintf(gettext("removing %s, no longer built by %s"), $file, $page));
- prune($config{destdir}."/".$file);
+ prune($config{destdir}."/".$file, $config{destdir});
}
}
}
return 1;
}
-
sub refresh () {
+ $phase = PHASE_SCAN;
+
srcdir_check();
run_hooks(refresh => sub { shift->() });
- my ($files, $pages)=find_src_files();
- my ($new, $internal_new)=find_new_files($files);
- my ($del, $internal_del)=find_del_files($pages);
- my ($changed, $internal_changed)=find_changed($files);
- run_hooks(needsbuild => sub { shift->($changed) });
+ my ($files, $pages, $new, $internal_new, $del, $internal_del, $changed, $internal_changed);
+ my $want_find_changes=$config{only_committed_changes} &&
+ exists $IkiWiki::hooks{rcs}{rcs_find_changes} &&
+ exists $IkiWiki::hooks{rcs}{rcs_get_current_rev};
+ if (! $config{rebuild} && $want_find_changes && defined $IkiWiki::lastrev && length $IkiWiki::lastrev) {
+ my ($changed_raw, $del_raw);
+ ($changed_raw, $del_raw, $IkiWiki::lastrev) = $IkiWiki::hooks{rcs}{rcs_find_changes}{call}->($IkiWiki::lastrev);
+ ($files, $pages)=process_changed_files($changed_raw, $del_raw);
+ }
+ else {
+ ($files, $pages)=find_src_files();
+ }
+ if ($want_find_changes) {
+ if (! defined($IkiWiki::lastrev) || ! length $IkiWiki::lastrev) {
+ $IkiWiki::lastrev=$IkiWiki::hooks{rcs}{rcs_get_current_rev}{call}->();
+ }
+ }
+ ($new, $internal_new)=find_new_files($files);
+ ($del, $internal_del)=find_del_files($pages);
+ ($changed, $internal_changed)=find_changed($files);
+ my %existingfiles;
+ run_hooks(needsbuild => sub {
+ my $ret=shift->($changed, [@$del, @$internal_del]);
+ if (ref $ret eq 'ARRAY' && $ret != $changed) {
+ if (! %existingfiles) {
+ foreach my $f (@$files) {
+ $existingfiles{$f}=1;
+ }
+ }
+ @$changed=grep $existingfiles{$_}, @$ret;
+ }
+ });
my $oldlink_targets=calculate_old_links($changed, $del);
foreach my $file (@$changed) {
}
calculate_links();
-
+
+ # At this point it becomes OK to start matching pagespecs.
+ $phase = PHASE_RENDER;
+ # Save some memory: we no longer need to keep track of which pages
+ # we've scanned
+ %scanned = ();
+
remove_del(@$del, @$internal_del);
foreach my $file (@$changed) {
derender_internal($file);
}
+ run_hooks(build_affected => sub {
+ my %affected = shift->();
+ while (my ($page, $message) = each %affected) {
+ next unless exists $pagesources{$page};
+ render($pagesources{$page}, $message);
+ }
+ });
+
my ($backlinkchanged, $linkchangers)=calculate_changed_links($changed,
$del, $oldlink_targets);
run_hooks(delete => sub { shift->(@$del, @$internal_del) });
}
if (%rendered) {
- run_hooks(change => sub { shift->(keys %rendered) });
+ run_hooks(rendered => sub { shift->(keys %rendered) });
+ run_hooks(change => sub { shift->(keys %rendered) }); # back-compat
}
+ my %all_changed = map { $_ => 1 }
+ @$new, @$changed, @$del,
+ @$internal_new, @$internal_changed, @$internal_del;
+ run_hooks(changes => sub { shift->(keys %all_changed) });
}
sub clean_rendered {
remove_unrendered();
foreach my $page (keys %oldrenderedfiles) {
foreach my $file (@{$oldrenderedfiles{$page}}) {
- prune($config{destdir}."/".$file);
+ prune($config{destdir}."/".$file, $config{destdir});
}
}
}
loadindex();
unlockwiki();
+ # This function behaves as though it's in the render phase;
+ # all other files are assumed to have been scanned last time.
+ $phase = PHASE_RENDER;
+
my $srcfile=possibly_foolish_untaint($config{render});
my $file=$srcfile;
$file=~s/\Q$config{srcdir}\E\/?//;