]> git.vanrenterghem.biz Git - git.ikiwiki.info.git/blobdiff - IkiWiki/Render.pm
* Many changes to make ikiwiki very resistant to write failures
[git.ikiwiki.info.git] / IkiWiki / Render.pm
index d7f07c08a1028f7a7353953610daeab7ac236e21..cbba282517ebaa0ce220b05853c6d1907859875d 100644 (file)
@@ -7,57 +7,43 @@ use strict;
 use IkiWiki;
 use Encode;
 
-sub linkify ($$$) { #{{{
-       my $lpage=shift; # the page containing the links
-       my $page=shift; # the page the link will end up on (different for inline)
-       my $content=shift;
-
-       $content =~ s{(\\?)$config{wiki_link_regexp}}{
-               $2 ? ( $1 ? "[[$2|$3]]" : htmllink($lpage, $page, titlepage($3), 0, 0, pagetitle($2)))
-                  : ( $1 ? "[[$3]]" :    htmllink($lpage, $page, titlepage($3)))
-       }eg;
-       
-       return $content;
-} #}}}
-
-sub htmlize ($$) { #{{{
-       my $type=shift;
-       my $content=shift;
-       
-       if (exists $hooks{htmlize}{$type}) {
-               $content=$hooks{htmlize}{$type}{call}->($content);
-       }
-       else {
-               error("htmlization of $type not supported");
+my %backlinks;
+my $backlinks_calculated=0;
+
+sub calculate_backlinks () { #{{{
+       return if $backlinks_calculated;
+       %backlinks=();
+       foreach my $page (keys %links) {
+               foreach my $link (@{$links{$page}}) {
+                       my $bestlink=bestlink($page, $link);
+                       if (length $bestlink && $bestlink ne $page) {
+                               $backlinks{$bestlink}{$page}=1;
+                       }
+               }
        }
-
-       run_hooks(sanitize => sub {
-               $content=shift->($content);
-       });
-       
-       return $content;
+       $backlinks_calculated=1;
 } #}}}
 
 sub backlinks ($) { #{{{
        my $page=shift;
 
+       calculate_backlinks();
+
        my @links;
-       foreach my $p (keys %links) {
-               next if bestlink($page, $p) eq $page;
-               if (grep { length $_ && bestlink($p, $_) eq $page } @{$links{$p}}) {
-                       my $href=abs2rel(htmlpage($p), dirname($page));
+       return unless $backlinks{$page};
+       foreach my $p (keys %{$backlinks{$page}}) {
+               my $href=abs2rel(htmlpage($p), dirname($page));
                        
-                       # Trim common dir prefixes from both pages.
-                       my $p_trimmed=$p;
-                       my $page_trimmed=$page;
-                       my $dir;
-                       1 while (($dir)=$page_trimmed=~m!^([^/]+/)!) &&
-                               defined $dir &&
-                               $p_trimmed=~s/^\Q$dir\E// &&
-                               $page_trimmed=~s/^\Q$dir\E//;
-                                      
-                       push @links, { url => $href, page => pagetitle($p_trimmed) };
-               }
+               # Trim common dir prefixes from both pages.
+               my $p_trimmed=$p;
+               my $page_trimmed=$page;
+               my $dir;
+               1 while (($dir)=$page_trimmed=~m!^([^/]+/)!) &&
+                       defined $dir &&
+                       $p_trimmed=~s/^\Q$dir\E// &&
+                       $page_trimmed=~s/^\Q$dir\E//;
+                              
+               push @links, { url => $href, page => pagetitle($p_trimmed) };
        }
 
        return sort { $a->{page} cmp $b->{page} } @links;
@@ -84,80 +70,6 @@ sub parentlinks ($) { #{{{
        return @ret;
 } #}}}
 
-my %preprocessing;
-sub preprocess ($$$) { #{{{
-       my $page=shift; # the page the data comes from
-       my $destpage=shift; # the page the data will appear in (different for inline)
-       my $content=shift;
-
-       my $handle=sub {
-               my $escape=shift;
-               my $command=shift;
-               my $params=shift;
-               if (length $escape) {
-                       return "[[$command $params]]";
-               }
-               elsif ($preprocessing{$page}++ > 10) {
-                       # Avoid loops of preprocessed pages preprocessing
-                       # other pages that preprocess them, etc.
-                       return "[[$command preprocessing loop detected]]";
-               }
-               elsif (exists $hooks{preprocess}{$command}) {
-                       # Note: preserve order of params, some plugins may
-                       # consider it significant.
-                       my @params;
-                       while ($params =~ /(?:(\w+)=)?(?:"""(.*?)"""|"([^"]+)"|(\S+))(?:\s+|$)/sg) {
-                               my $key=$1;
-                               my $val;
-                               if (defined $2) {
-                                       $val=$2;
-                                       $val=~s/\r\n/\n/mg;
-                                       $val=~s/^\n+//g;
-                                       $val=~s/\n+$//g;
-                               }
-                               elsif (defined $3) {
-                                       $val=$3;
-                               }
-                               elsif (defined $4) {
-                                       $val=$4;
-                               }
-
-                               if (defined $key) {
-                                       push @params, $key, $val;
-                               }
-                               else {
-                                       push @params, $val, '';
-                               }
-                       }
-                       my $ret=$hooks{preprocess}{$command}{call}->(
-                               @params,
-                               page => $page,
-                               destpage => $destpage,
-                       );
-                       $preprocessing{$page}--;
-                       return $ret;
-               }
-               else {
-                       return "[[$command not processed]]";
-               }
-       };
-       
-       $content =~ s{(\\?)\[\[(\w+)\s+((?:(?:\w+=)?(?:""".*?"""|"[^"]+"|[^\s\]]+)\s*)*)\]\]}{$handle->($1, $2, $3)}seg;
-       return $content;
-} #}}}
-
-sub add_depends ($$) { #{{{
-       my $page=shift;
-       my $pagespec=shift;
-       
-       if (! exists $depends{$page}) {
-               $depends{$page}=$pagespec;
-       }
-       else {
-               $depends{$page}=pagespec_merge($depends{$page}, $pagespec);
-       }
-} # }}}
-
 sub genpage ($$$) { #{{{
        my $page=shift;
        my $content=shift;
@@ -182,8 +94,13 @@ sub genpage ($$$) { #{{{
                $actions++;
        }
        if ($config{discussion}) {
-               $template->param(discussionlink => htmllink($page, $page, "Discussion", 1, 1));
-               $actions++;
+               my $discussionlink=gettext("discussion");
+               if ($page !~ /.*\/\Q$discussionlink\E$/ &&
+                  (length $config{cgiurl} ||
+                   exists $links{$page."/".$discussionlink})) {
+                       $template->param(discussionlink => htmllink($page, $page, gettext("Discussion"), 1, 1));
+                       $actions++;
+               }
        }
 
        if ($actions) {
@@ -209,68 +126,54 @@ sub genpage ($$$) { #{{{
        $content=$template->output;
 
        run_hooks(format => sub {
-               $content=shift->($content);
+               $content=shift->(
+                       page => $page,
+                       content => $content,
+               );
        });
 
        return $content;
 } #}}}
 
-sub check_overwrite ($$) { #{{{
-       # Important security check. Make sure to call this before saving
-       # any files to the source directory.
-       my $dest=shift;
-       my $src=shift;
-       
-       if (! exists $renderedfiles{$src} && -e $dest && ! $config{rebuild}) {
-               error("$dest already exists and was not rendered from $src before");
-       }
-} #}}}
-
-sub displaytime ($) { #{{{
-       my $time=shift;
-
-       eval q{use POSIX};
-       # strftime doesn't know about encodings, so make sure
-       # its output is properly treated as utf8
-       return decode_utf8(POSIX::strftime(
-                       $config{timeformat}, localtime($time)));
-} #}}}
-
 sub mtime ($) { #{{{
        my $file=shift;
        
        return (stat($file))[9];
 } #}}}
 
-sub findlinks ($$) { #{{{
-       my $page=shift;
-       my $content=shift;
+sub scan ($) { #{{{
+       my $file=shift;
 
-       my @links;
-       while ($content =~ /(?<!\\)$config{wiki_link_regexp}/g) {
-               push @links, titlepage($2);
-       }
-       if ($config{discussion}) {
-               # Discussion links are a special case since they're not in the
-               # text of the page, but on its template.
-               return @links, "$page/discussion";
+       my $type=pagetype($file);
+       if (defined $type) {
+               my $srcfile=srcfile($file);
+               my $content=readfile($srcfile);
+               my $page=pagename($file);
+               will_render($page, htmlpage($page), 1);
+
+               # Always needs to be done, since filters might add links
+               # to the content.
+               $content=filter($page, $content);
+
+               my @links;
+               while ($content =~ /(?<!\\)$config{wiki_link_regexp}/g) {
+                       push @links, titlepage($2);
+               }
+               if ($config{discussion}) {
+                       # Discussion links are a special case since they're
+                       # not in the text of the page, but on its template.
+                       push @links, $page."/".gettext("discussion");
+               }
+               $links{$page}=\@links;
+               
+               # Preprocess in scan-only mode.
+               preprocess($page, $page, $content, 1);
        }
        else {
-               return @links;
+               will_render($file, $file, 1);
        }
 } #}}}
 
-sub filter ($$) {
-       my $page=shift;
-       my $content=shift;
-
-       run_hooks(filter => sub {
-               $content=shift->(page => $page, content => $content);
-       });
-
-       return $content;
-}
-
 sub render ($) { #{{{
        my $file=shift;
        
@@ -280,29 +183,42 @@ sub render ($) { #{{{
                my $content=readfile($srcfile);
                my $page=pagename($file);
                delete $depends{$page};
+               will_render($page, htmlpage($page), 1);
                
                $content=filter($page, $content);
-               
-               $links{$page}=[findlinks($page, $content)];
-               
                $content=preprocess($page, $page, $content);
                $content=linkify($page, $page, $content);
-               $content=htmlize($type, $content);
+               $content=htmlize($page, $type, $content);
                
-               check_overwrite("$config{destdir}/".htmlpage($page), $page);
                writefile(htmlpage($page), $config{destdir},
                        genpage($page, $content, mtime($srcfile)));
                $oldpagemtime{$page}=time;
-               $renderedfiles{$page}=htmlpage($page);
        }
        else {
-               my $content=readfile($srcfile, 1);
-               $links{$file}=[];
+               my $srcfd=readfile($srcfile, 1, 1);
                delete $depends{$file};
-               check_overwrite("$config{destdir}/$file", $file);
-               writefile($file, $config{destdir}, $content, 1);
+               will_render($file, $file, 1);
+               writefile($file, $config{destdir}, undef, 1, sub {
+                       my $destfd=shift;
+                       my $cleanup=shift;
+
+                       my $blksize = 16384;
+                       my ($len, $buf, $written);
+                       while ($len = sysread $srcfd, $buf, $blksize) {
+                               if (! defined $len) {
+                                       next if $! =~ /^Interrupted/;
+                                       error("failed to read $srcfile: $!", $cleanup);
+                               }
+                               my $offset = 0;
+                               while ($len) {
+                                       defined($written = syswrite $destfd, $buf, $len, $offset)
+                                               or error("failed to write $file: $!", $cleanup);
+                                       $len -= $written;
+                                       $offset += $written;
+                               }
+                       }
+               });
                $oldpagemtime{$file}=time;
-               $renderedfiles{$file}=$file;
        }
 } #}}}
 
@@ -321,17 +237,18 @@ sub refresh () { #{{{
        my %exists;
        my @files;
        eval q{use File::Find};
+       error($@) if $@;
        find({
                no_chdir => 1,
                wanted => sub {
                        $_=decode_utf8($_);
-                       if (/$config{wiki_file_prune_regexp}/) {
+                       if (file_pruned($_, $config{srcdir})) {
                                $File::Find::prune=1;
                        }
                        elsif (! -d $_ && ! -l $_) {
                                my ($f)=/$config{wiki_file_regexp}/; # untaint
                                if (! defined $f) {
-                                       warn("skipping bad filename $_\n");
+                                       warn(sprintf(gettext("skipping bad filename %s"), $_)."\n");
                                }
                                else {
                                        $f=~s/^\Q$config{srcdir}\E\/?//;
@@ -345,22 +262,25 @@ sub refresh () { #{{{
                no_chdir => 1,
                wanted => sub {
                        $_=decode_utf8($_);
-                       if (/$config{wiki_file_prune_regexp}/) {
+                       if (file_pruned($_, $config{underlaydir})) {
                                $File::Find::prune=1;
                        }
                        elsif (! -d $_ && ! -l $_) {
                                my ($f)=/$config{wiki_file_regexp}/; # untaint
                                if (! defined $f) {
-                                       warn("skipping bad filename $_\n");
+                                       warn(sprintf(gettext("skipping bad filename %s"), $_)."\n");
                                }
                                else {
-                                       # Don't add files that are in the
+                                       # Don't add pages that are in the
                                        # srcdir.
                                        $f=~s/^\Q$config{underlaydir}\E\/?//;
                                        if (! -e "$config{srcdir}/$f" && 
                                            ! -l "$config{srcdir}/$f") {
-                                               push @files, $f;
-                                               $exists{pagename($f)}=1;
+                                               my $page=pagename($f);
+                                               if (! $exists{$page}) {
+                                                       push @files, $f;
+                                                       $exists{$page}=1;
+                                               }
                                        }
                                }
                        }
@@ -373,12 +293,10 @@ sub refresh () { #{{{
        my @add;
        foreach my $file (@files) {
                my $page=pagename($file);
+               $pagesources{$page}=$file;
                if (! $oldpagemtime{$page}) {
-                       debug("new page $page") unless exists $pagectime{$page};
                        push @add, $file;
-                       $links{$page}=[];
                        $pagecase{lc $page}=$page;
-                       $pagesources{$page}=$file;
                        if ($config{getctime} && -e "$config{srcdir}/$file") {
                                $pagectime{$page}=rcs_getctime("$config{srcdir}/$file");
                        }
@@ -390,64 +308,64 @@ sub refresh () { #{{{
        my @del;
        foreach my $page (keys %oldpagemtime) {
                if (! $exists{$page}) {
-                       debug("removing old page $page");
+                       debug(sprintf(gettext("removing old page %s"), $page));
                        push @del, $pagesources{$page};
-                       prune($config{destdir}."/".$renderedfiles{$page});
-                       delete $renderedfiles{$page};
+                       $links{$page}=[];
+                       $renderedfiles{$page}=[];
                        $oldpagemtime{$page}=0;
+                       prune($config{destdir}."/".$_)
+                               foreach @{$oldrenderedfiles{$page}};
                        delete $pagesources{$page};
                }
        }
-       
-       # render any updated files
+
+       # scan changed and new files
+       my @changed;
        foreach my $file (@files) {
                my $page=pagename($file);
                
                if (! exists $oldpagemtime{$page} ||
                    mtime(srcfile($file)) > $oldpagemtime{$page} ||
                    $forcerebuild{$page}) {
-                       debug("rendering $file");
-                       render($file);
-                       $rendered{$file}=1;
+                       debug(sprintf(gettext("scanning %s"), $file));
+                       push @changed, $file;
+                       scan($file);
                }
        }
+       calculate_backlinks();
+
+       # render changed and new pages
+       foreach my $file (@changed) {
+               debug(sprintf(gettext("rendering %s"), $file));
+               render($file);
+               $rendered{$file}=1;
+       }
        
-       # if any files were added or removed, check to see if each page
-       # needs an update due to linking to them or inlining them.
-       # TODO: inefficient; pages may get rendered above and again here;
-       # problem is the bestlink may have changed and we won't know until
-       # now
+       # rebuild pages that link to added or removed pages
        if (@add || @del) {
-FILE:          foreach my $file (@files) {
-                       my $page=pagename($file);
-                       foreach my $f (@add, @del) {
-                               my $p=pagename($f);
-                               foreach my $link (@{$links{$page}}) {
-                                       if (bestlink($page, $link) eq $p) {
-                                               debug("rendering $file, which links to $p");
-                                               render($file);
-                                               $rendered{$file}=1;
-                                               next FILE;
-                                       }
-                               }
+               foreach my $f (@add, @del) {
+                       my $p=pagename($f);
+                       foreach my $page (keys %{$backlinks{$p}}) {
+                               my $file=$pagesources{$page};
+                               next if $rendered{$file};
+                               debug(sprintf(gettext("rendering %s, which links to %s"), $file, $p));
+                               render($file);
+                               $rendered{$file}=1;
                        }
                }
        }
 
-       # Handle backlinks; if a page has added/removed links, update the
-       # pages it links to. Also handles rebuilding dependant pages.
-       # TODO: inefficient; pages may get rendered above and again here;
-       # problem is the backlinks could be wrong in the first pass render
-       # above
        if (%rendered || @del) {
+               # rebuild dependant pages
                foreach my $f (@files) {
+                       next if $rendered{$f};
                        my $p=pagename($f);
                        if (exists $depends{$p}) {
                                foreach my $file (keys %rendered, @del) {
                                        next if $f eq $file;
                                        my $page=pagename($file);
-                                       if (pagespec_match($page, $depends{$p})) {
-                                               debug("rendering $f, which depends on $page");
+                                       if (pagespec_match($page, $depends{$p}, $p)) {
+                                               debug(sprintf(gettext("rendering %s, which depends on %s"), $f, $page));
                                                render($f);
                                                $rendered{$f}=1;
                                                last;
@@ -456,6 +374,8 @@ FILE:               foreach my $file (@files) {
                        }
                }
                
+               # handle backlinks; if a page has added/removed links,
+               # update the pages it links to
                my %linkchanged;
                foreach my $file (keys %rendered, @del) {
                        my $page=pagename($file);
@@ -482,13 +402,25 @@ FILE:             foreach my $file (@files) {
                foreach my $link (keys %linkchanged) {
                        my $linkfile=$pagesources{$link};
                        if (defined $linkfile) {
-                               debug("rendering $linkfile, to update its backlinks");
+                               next if $rendered{$linkfile};
+                               debug(sprintf(gettext("rendering %s, to update its backlinks"), $linkfile));
                                render($linkfile);
                                $rendered{$linkfile}=1;
                        }
                }
        }
 
+       # remove no longer rendered files
+       foreach my $src (keys %rendered) {
+               my $page=pagename($src);
+               foreach my $file (@{$oldrenderedfiles{$page}}) {
+                       if (! grep { $_ eq $file } @{$renderedfiles{$page}}) {
+                               debug(sprintf(gettext("removing %s, no longer rendered by %s"), $file, $page));
+                               prune($config{destdir}."/".$file);
+                       }
+               }
+       }
+
        if (@del) {
                run_hooks(delete => sub { shift->(@del) });
        }
@@ -497,4 +429,29 @@ FILE:              foreach my $file (@files) {
        }
 } #}}}
 
+sub commandline_render () { #{{{
+       loadplugins();
+       checkconfig();
+       lockwiki();
+       loadindex();
+       unlockwiki();
+
+       my $srcfile=possibly_foolish_untaint($config{render});
+       my $file=$srcfile;
+       $file=~s/\Q$config{srcdir}\E\/?//;
+
+       my $type=pagetype($file);
+       die sprintf(gettext("ikiwiki: cannot render %s"), $srcfile)."\n" unless defined $type;
+       my $content=readfile($srcfile);
+       my $page=pagename($file);
+       $pagesources{$page}=$file;
+       $content=filter($page, $content);
+       $content=preprocess($page, $page, $content);
+       $content=linkify($page, $page, $content);
+       $content=htmlize($page, $type, $content);
+
+       print genpage($page, $content, mtime($srcfile));
+       exit 0;
+} #}}}
+
 1