respond

[git.ikiwiki.info.git] / IkiWiki / Plugin / aggregate.pm

diff --git a/IkiWiki/Plugin/aggregate.pm b/IkiWiki/Plugin/aggregate.pm
index adaa619ab1a255d5904c71bc100c52c2fb10b414..5a9eb433de8311e0a9d77909aa45343c609bdea5 100644 (file)
--- a/IkiWiki/Plugin/aggregate.pm
+++ b/IkiWiki/Plugin/aggregate.pm
@@ -4,7 +4,7 @@ package IkiWiki::Plugin::aggregate;
 
 use warnings;
 use strict;
 
 use warnings;
 use strict;

-use IkiWiki 2.00;
+use IkiWiki 3.00;

 use HTML::Parser;
 use HTML::Tagset;
 use HTML::Entities;
 use HTML::Parser;
 use HTML::Tagset;
 use HTML::Entities;


@@ -14,7 +14,7 @@ use open qw{:utf8 :std};
 my %feeds;
 my %guids;
 
 my %feeds;
 my %guids;
 

-sub import { #{{{
+sub import {

        hook(type => "getopt", id => "aggregate", call => \&getopt);
        hook(type => "getsetup", id => "aggregate", call => \&getsetup);
        hook(type => "checkconfig", id => "aggregate", call => \&checkconfig);
        hook(type => "getopt", id => "aggregate", call => \&getopt);
        hook(type => "getsetup", id => "aggregate", call => \&getsetup);
        hook(type => "checkconfig", id => "aggregate", call => \&checkconfig);


@@ -26,9 +26,9 @@ sub import { #{{{
        if (exists $config{aggregate_webtrigger} && $config{aggregate_webtrigger}) {
                hook(type => "cgi", id => "aggregate", call => \&cgi);
        }
        if (exists $config{aggregate_webtrigger} && $config{aggregate_webtrigger}) {
                hook(type => "cgi", id => "aggregate", call => \&cgi);
        }

-} # }}}
+}

 
 

-sub getopt () { #{{{
+sub getopt () {

         eval q{use Getopt::Long};
        error($@) if $@;
         Getopt::Long::Configure('pass_through');
         eval q{use Getopt::Long};
        error($@) if $@;
         Getopt::Long::Configure('pass_through');


@@ -36,9 +36,9 @@ sub getopt () { #{{{
                "aggregate" => \$config{aggregate},
                "aggregateinternal!" => \$config{aggregateinternal},
        );
                "aggregate" => \$config{aggregate},
                "aggregateinternal!" => \$config{aggregateinternal},
        );

-} #}}}
+}

 
 

-sub getsetup () { #{{{
+sub getsetup () {

        return
                plugin => {
                        safe => 1,
        return
                plugin => {
                        safe => 1,


@@ -46,7 +46,7 @@ sub getsetup () { #{{{
                },
                aggregateinternal => {
                        type => "boolean",
                },
                aggregateinternal => {
                        type => "boolean",

-                       example => 0,
+                       example => 1,

                        description => "enable aggregation to internal pages?",
                        safe => 0, # enabling needs manual transition
                        rebuild => 0,
                        description => "enable aggregation to internal pages?",
                        safe => 0, # enabling needs manual transition
                        rebuild => 0,


@@ -58,16 +58,20 @@ sub getsetup () { #{{{
                        safe => 1,
                        rebuild => 0,
                },
                        safe => 1,
                        rebuild => 0,
                },

-} #}}}
+}
+
+sub checkconfig () {
+       if (! defined $config{aggregateinternal}) {
+               $config{aggregateinternal}=1;
+       }

 
 

-sub checkconfig () { #{{{

        if ($config{aggregate} && ! ($config{post_commit} && 
                                     IkiWiki::commit_hook_enabled())) {
                launchaggregation();
        }
        if ($config{aggregate} && ! ($config{post_commit} && 
                                     IkiWiki::commit_hook_enabled())) {
                launchaggregation();
        }

-} #}}}
+}

 
 

-sub cgi ($) { #{{{
+sub cgi ($) {

        my $cgi=shift;
 
        if (defined $cgi->param('do') &&
        my $cgi=shift;
 
        if (defined $cgi->param('do') &&


@@ -90,9 +94,9 @@ sub cgi ($) { #{{{
                }
                exit 0;
        }
                }
                exit 0;
        }

-} #}}}
+}

 
 

-sub launchaggregation () { #{{{
+sub launchaggregation () {

        # See if any feeds need aggregation.
        loadstate();
        my @feeds=needsaggregate();
        # See if any feeds need aggregation.
        loadstate();
        my @feeds=needsaggregate();


@@ -135,16 +139,16 @@ sub launchaggregation () { #{{{
        unlockaggregate();
 
        return 1;
        unlockaggregate();
 
        return 1;

-} #}}}
+}

 
 #  Pages with extension _aggregated have plain html markup, pass through.
 
 #  Pages with extension _aggregated have plain html markup, pass through.

-sub htmlize (@) { #{{{
+sub htmlize (@) {

        my %params=@_;
        return $params{content};
        my %params=@_;
        return $params{content};

-} #}}}
+}

 
 # Used by ikiwiki-transition aggregateinternal.
 
 # Used by ikiwiki-transition aggregateinternal.

-sub migrate_to_internal { #{{{
+sub migrate_to_internal {

        if (! lockaggregate()) {
                error("an aggregation process is currently running");
        }
        if (! lockaggregate()) {
                error("an aggregation process is currently running");
        }


@@ -190,9 +194,9 @@ sub migrate_to_internal { #{{{
        IkiWiki::unlockwiki;
        
        unlockaggregate();
        IkiWiki::unlockwiki;
        
        unlockaggregate();

-} #}}}
+}

 
 

-sub needsbuild (@) { #{{{
+sub needsbuild (@) {

        my $needsbuild=shift;
        
        loadstate();
        my $needsbuild=shift;
        
        loadstate();


@@ -206,9 +210,9 @@ sub needsbuild (@) { #{{{
                        markunseen($feed->{sourcepage});
                }
        }
                        markunseen($feed->{sourcepage});
                }
        }

-} # }}}
+}

 
 

-sub preprocess (@) { #{{{
+sub preprocess (@) {

        my %params=@_;
 
        foreach my $required (qw{name url}) {
        my %params=@_;
 
        foreach my $required (qw{name url}) {


@@ -245,6 +249,7 @@ sub preprocess (@) { #{{{
        $feed->{template}=$params{template} . ".tmpl";
        delete $feed->{unseen};
        $feed->{lastupdate}=0 unless defined $feed->{lastupdate};
        $feed->{template}=$params{template} . ".tmpl";
        delete $feed->{unseen};
        $feed->{lastupdate}=0 unless defined $feed->{lastupdate};

+       $feed->{lasttry}=$feed->{lastupdate} unless defined $feed->{lasttry};

        $feed->{numposts}=0 unless defined $feed->{numposts};
        $feed->{newposts}=0 unless defined $feed->{newposts};
        $feed->{message}=gettext("new feed") unless defined $feed->{message};
        $feed->{numposts}=0 unless defined $feed->{numposts};
        $feed->{newposts}=0 unless defined $feed->{newposts};
        $feed->{message}=gettext("new feed") unless defined $feed->{message};


@@ -265,9 +270,9 @@ sub preprocess (@) { #{{{
               ($feed->{newposts} ? "; ".$feed->{newposts}.
                                    " ".gettext("new") : "").
               ")";
               ($feed->{newposts} ? "; ".$feed->{newposts}.
                                    " ".gettext("new") : "").
               ")";

-} # }}}
+}

 
 

-sub delete (@) { #{{{
+sub delete (@) {

        my @files=@_;
 
        # Remove feed data for removed pages.
        my @files=@_;
 
        # Remove feed data for removed pages.


@@ -275,9 +280,9 @@ sub delete (@) { #{{{
                my $page=pagename($file);
                markunseen($page);
        }
                my $page=pagename($file);
                markunseen($page);
        }

-} #}}}
+}

 
 

-sub markunseen ($) { #{{{
+sub markunseen ($) {

        my $page=shift;
 
        foreach my $id (keys %feeds) {
        my $page=shift;
 
        foreach my $id (keys %feeds) {


@@ -285,11 +290,11 @@ sub markunseen ($) { #{{{
                        $feeds{$id}->{unseen}=1;
                }
        }
                        $feeds{$id}->{unseen}=1;
                }
        }

-} #}}}
+}

 
 my $state_loaded=0;
 
 
 my $state_loaded=0;
 

-sub loadstate () { #{{{
+sub loadstate () {

        return if $state_loaded;
        $state_loaded=1;
        if (-e "$config{wikistatedir}/aggregate") {
        return if $state_loaded;
        $state_loaded=1;
        if (-e "$config{wikistatedir}/aggregate") {


@@ -323,9 +328,9 @@ sub loadstate () { #{{{
 
                close IN;
        }
 
                close IN;
        }

-} #}}}
+}

 
 

-sub savestate () { #{{{
+sub savestate () {

        return unless $state_loaded;
        garbage_collect();
        my $newfile="$config{wikistatedir}/aggregate.new";
        return unless $state_loaded;
        garbage_collect();
        my $newfile="$config{wikistatedir}/aggregate.new";


@@ -342,7 +347,8 @@ sub savestate () { #{{{
                                push @line, "tag=$_" foreach @{$data->{tags}};
                        }
                        else {
                                push @line, "tag=$_" foreach @{$data->{tags}};
                        }
                        else {

-                               push @line, "$field=".$data->{$field};
+                               push @line, "$field=".$data->{$field}
+                                       if defined $data->{$field};

                        }
                }
                print OUT join(" ", @line)."\n" || error("write $newfile: $!", $cleanup);
                        }
                }
                print OUT join(" ", @line)."\n" || error("write $newfile: $!", $cleanup);


@@ -350,9 +356,9 @@ sub savestate () { #{{{
        close OUT || error("save $newfile: $!", $cleanup);
        rename($newfile, "$config{wikistatedir}/aggregate") ||
                error("rename $newfile: $!", $cleanup);
        close OUT || error("save $newfile: $!", $cleanup);
        rename($newfile, "$config{wikistatedir}/aggregate") ||
                error("rename $newfile: $!", $cleanup);

-} #}}}
+}

 
 

-sub garbage_collect () { #{{{
+sub garbage_collect () {

        foreach my $name (keys %feeds) {
                # remove any feeds that were not seen while building the pages
                # that used to contain them
        foreach my $name (keys %feeds) {
                # remove any feeds that were not seen while building the pages
                # that used to contain them


@@ -375,9 +381,9 @@ sub garbage_collect () { #{{{
                        delete $guid->{md5};
                }
        }
                        delete $guid->{md5};
                }
        }

-} #}}}
+}

 
 

-sub mergestate () { #{{{
+sub mergestate () {

        # Load the current state in from disk, and merge into it
        # values from the state in memory that might have changed
        # during aggregation.
        # Load the current state in from disk, and merge into it
        # values from the state in memory that might have changed
        # during aggregation.


@@ -390,14 +396,15 @@ sub mergestate () { #{{{
        # fields.
        foreach my $name (keys %myfeeds) {
                if (exists $feeds{$name}) {
        # fields.
        foreach my $name (keys %myfeeds) {
                if (exists $feeds{$name}) {

-                       foreach my $field (qw{message lastupdate numposts
-                                             newposts error}) {
+                       foreach my $field (qw{message lastupdate lasttry
+                                             numposts newposts error}) {

                                $feeds{$name}->{$field}=$myfeeds{$name}->{$field};
                        }
                }
        }
 
        # New guids can be created during aggregation.
                                $feeds{$name}->{$field}=$myfeeds{$name}->{$field};
                        }
                }
        }
 
        # New guids can be created during aggregation.

+       # Guids have a few fields that may be updated during aggregation.

        # It's also possible that guids were removed from the on-disk state
        # while the aggregation was in process. That would only happen if
        # their feed was also removed, so any removed guids added back here
        # It's also possible that guids were removed from the on-disk state
        # while the aggregation was in process. That would only happen if
        # their feed was also removed, so any removed guids added back here


@@ -406,16 +413,21 @@ sub mergestate () { #{{{
                if (! exists $guids{$guid}) {
                        $guids{$guid}=$myguids{$guid};
                }
                if (! exists $guids{$guid}) {
                        $guids{$guid}=$myguids{$guid};
                }

+               else {
+                       foreach my $field (qw{md5}) {
+                               $guids{$guid}->{$field}=$myguids{$guid}->{$field};
+                       }
+               }

        }
        }

-} #}}}
+}

 
 

-sub clearstate () { #{{{
+sub clearstate () {

        %feeds=();
        %guids=();
        $state_loaded=0;
        %feeds=();
        %guids=();
        $state_loaded=0;

-} #}}}
+}

 
 

-sub expire () { #{{{
+sub expire () {

        foreach my $feed (values %feeds) {
                next unless $feed->{expireage} || $feed->{expirecount};
                my $count=0;
        foreach my $feed (values %feeds) {
                next unless $feed->{expireage} || $feed->{expirecount};
                my $count=0;


@@ -444,24 +456,24 @@ sub expire () { #{{{
                        }
                }
        }
                        }
                }
        }

-} #}}}
+}

 
 

-sub needsaggregate () { #{{{
+sub needsaggregate () {

        return values %feeds if $config{rebuild};
        return grep { time - $_->{lastupdate} >= $_->{updateinterval} } values %feeds;
        return values %feeds if $config{rebuild};
        return grep { time - $_->{lastupdate} >= $_->{updateinterval} } values %feeds;

-} #}}}
+}

 
 

-sub aggregate (@) { #{{{
+sub aggregate (@) {

        eval q{use XML::Feed};
        error($@) if $@;
        eval q{use URI::Fetch};
        error($@) if $@;
 
        foreach my $feed (@_) {
        eval q{use XML::Feed};
        error($@) if $@;
        eval q{use URI::Fetch};
        error($@) if $@;
 
        foreach my $feed (@_) {

-               $feed->{lastupdate}=time;
+               $feed->{lasttry}=time;

                $feed->{newposts}=0;
                $feed->{message}=sprintf(gettext("last checked %s"),
                $feed->{newposts}=0;
                $feed->{message}=sprintf(gettext("last checked %s"),

-                       displaytime($feed->{lastupdate}));
+                       displaytime($feed->{lasttry}));

                $feed->{error}=0;
 
                debug(sprintf(gettext("checking feed %s ..."), $feed->{name}));
                $feed->{error}=0;
 
                debug(sprintf(gettext("checking feed %s ..."), $feed->{name}));


@@ -483,6 +495,10 @@ sub aggregate (@) { #{{{
                        debug($feed->{message});
                        next;
                }
                        debug($feed->{message});
                        next;
                }

+
+               # lastupdate is only set if we were able to contact the server
+               $feed->{lastupdate}=$feed->{lasttry};
+

                if ($res->status == URI::Fetch::URI_GONE()) {
                        $feed->{message}=gettext("feed not found");
                        $feed->{error}=1;
                if ($res->status == URI::Fetch::URI_GONE()) {
                        $feed->{message}=gettext("feed not found");
                        $feed->{error}=1;


@@ -524,6 +540,11 @@ sub aggregate (@) { #{{{
                }
 
                foreach my $entry ($f->entries) {
                }
 
                foreach my $entry ($f->entries) {

+                       # XML::Feed doesn't work around XML::Atom's bizarre
+                       # API, so we will. Real unicode strings? Yes please.
+                       # See [[bugs/Aggregated_Atom_feeds_are_double-encoded]]
+                       local $XML::Atom::ForceUnicode = 1;
+

                        my $c=$entry->content;
                        # atom feeds may have no content, only a summary
                        if (! defined $c && ref $entry->summary) {
                        my $c=$entry->content;
                        # atom feeds may have no content, only a summary
                        if (! defined $c && ref $entry->summary) {


@@ -542,9 +563,9 @@ sub aggregate (@) { #{{{
                        );
                }
        }
                        );
                }
        }

-} #}}}
+}

 
 

-sub add_page (@) { #{{{
+sub add_page (@) {

        my %params=@_;
        
        my $feed=$params{feed};
        my %params=@_;
        
        my $feed=$params{feed};


@@ -629,27 +650,29 @@ sub add_page (@) { #{{{
                # creation time on record for the new page.
                utime $mtime, $mtime, "$config{srcdir}/".htmlfn($guid->{page});
                # Store it in pagectime for expiry code to use also.
                # creation time on record for the new page.
                utime $mtime, $mtime, "$config{srcdir}/".htmlfn($guid->{page});
                # Store it in pagectime for expiry code to use also.

-               $IkiWiki::pagectime{$guid->{page}}=$mtime;
+               $IkiWiki::pagectime{$guid->{page}}=$mtime
+                       unless exists $IkiWiki::pagectime{$guid->{page}};

        }
        else {
                # Dummy value for expiry code.
        }
        else {
                # Dummy value for expiry code.

-               $IkiWiki::pagectime{$guid->{page}}=time;
+               $IkiWiki::pagectime{$guid->{page}}=time
+                       unless exists $IkiWiki::pagectime{$guid->{page}};

        }
        }

-} #}}}
+}

 
 

-sub wikiescape ($) { #{{{
+sub wikiescape ($) {

        # escape accidental wikilinks and preprocessor stuff
        return encode_entities(shift, '\[\]');
        # escape accidental wikilinks and preprocessor stuff
        return encode_entities(shift, '\[\]');

-} #}}}
+}

 
 

-sub urlabs ($$) { #{{{
+sub urlabs ($$) {

        my $url=shift;
        my $urlbase=shift;
 
        URI->new_abs($url, $urlbase)->as_string;
        my $url=shift;
        my $urlbase=shift;
 
        URI->new_abs($url, $urlbase)->as_string;

-} #}}}
+}

 
 

-sub htmlabs ($$) { #{{{
+sub htmlabs ($$) {

        # Convert links in html from relative to absolute.
        # Note that this is a heuristic, which is not specified by the rss
        # spec and may not be right for all feeds. Also, see Debian
        # Convert links in html from relative to absolute.
        # Note that this is a heuristic, which is not specified by the rss
        # spec and may not be right for all feeds. Also, see Debian


@@ -685,15 +708,15 @@ sub htmlabs ($$) { #{{{
        $p->eof;
 
        return $ret;
        $p->eof;
 
        return $ret;

-} #}}}
+}

 
 

-sub htmlfn ($) { #{{{
+sub htmlfn ($) {

        return shift().".".($config{aggregateinternal} ? "_aggregated" : $config{htmlext});
        return shift().".".($config{aggregateinternal} ? "_aggregated" : $config{htmlext});

-} #}}}
+}

 
 my $aggregatelock;
 
 
 my $aggregatelock;
 

-sub lockaggregate () { #{{{
+sub lockaggregate () {

        # Take an exclusive lock to prevent multiple concurrent aggregators.
        # Returns true if the lock was aquired.
        if (! -d $config{wikistatedir}) {
        # Take an exclusive lock to prevent multiple concurrent aggregators.
        # Returns true if the lock was aquired.
        if (! -d $config{wikistatedir}) {


@@ -706,11 +729,11 @@ sub lockaggregate () { #{{{
                return 0;
        }
        return 1;
                return 0;
        }
        return 1;

-} #}}}
+}

 
 

-sub unlockaggregate () { #{{{
+sub unlockaggregate () {

        return close($aggregatelock) if $aggregatelock;
        return;
        return close($aggregatelock) if $aggregatelock;
        return;

-} #}}}
+}

 
 1
 
 1