X-Git-Url: http://git.vanrenterghem.biz/git.ikiwiki.info.git/blobdiff_plain/d01334b8e9c229895ecb6d0b0a7acda86515df71..15250ee6bfdb66cd823fb856bda84e372647f61f:/IkiWiki.pm diff --git a/IkiWiki.pm b/IkiWiki.pm index 140f7f740..efb48293a 100644 --- a/IkiWiki.pm +++ b/IkiWiki.pm @@ -5,6 +5,7 @@ package IkiWiki; use warnings; use strict; use Encode; +use Fcntl q{:flock}; use URI::Escape q{uri_escape_utf8}; use POSIX (); use Storable; @@ -14,15 +15,15 @@ use vars qw{%config %links %oldlinks %pagemtime %pagectime %pagecase %pagestate %wikistate %renderedfiles %oldrenderedfiles %pagesources %delpagesources %destsources %depends %depends_simple @mass_depends %hooks %forcerebuild %loaded_plugins %typedlinks - %oldtypedlinks %autofiles}; + %oldtypedlinks %autofiles @underlayfiles $lastrev $phase}; use Exporter q{import}; our @EXPORT = qw(hook debug error htmlpage template template_depends deptype add_depends pagespec_match pagespec_match_list bestlink htmllink readfile writefile pagetype srcfile pagename - displaytime will_render gettext ngettext urlto targetpage + displaytime strftime_utf8 will_render gettext ngettext urlto targetpage add_underlay pagetitle titlepage linkpage newpagefile - inject add_link add_autofile + inject add_link add_autofile useragent %config %links %pagestate %wikistate %renderedfiles %pagesources %destsources %typedlinks); our $VERSION = 3.00; # plugin interface version, next is ikiwiki version @@ -34,6 +35,11 @@ our $DEPEND_CONTENT=1; our $DEPEND_PRESENCE=2; our $DEPEND_LINKS=4; +# Phases of processing. +sub PHASE_SCAN () { 0 } +sub PHASE_RENDER () { 1 } +$phase = PHASE_SCAN; + # Optimisation. use Memoize; memoize("abs2rel"); @@ -103,6 +109,14 @@ sub getsetup () { safe => 1, rebuild => 1, }, + reverse_proxy => { + type => "boolean", + default => 0, + description => "do not adjust cgiurl if CGI is accessed via different URL", + advanced => 0, + safe => 1, + rebuild => 0, # only affects CGI requests + }, cgi_wrapper => { type => "string", default => '', @@ -118,6 +132,29 @@ sub getsetup () { safe => 0, rebuild => 0, }, + cgi_overload_delay => { + type => "string", + default => '', + example => "10", + description => "number of seconds to delay CGI requests when overloaded", + safe => 1, + rebuild => 0, + }, + cgi_overload_message => { + type => "string", + default => '', + example => "Please wait", + description => "message to display when overloaded (may contain html)", + safe => 1, + rebuild => 0, + }, + only_committed_changes => { + type => "boolean", + default => 0, + description => "enable optimization of only refreshing committed changes?", + safe => 1, + rebuild => 0, + }, rcs => { type => "string", default => '', @@ -129,7 +166,8 @@ sub getsetup () { type => "internal", default => [qw{mdwn link inline meta htmlscrubber passwordauth openid signinedit lockedit conditional - recentchanges parentlinks editpage}], + recentchanges parentlinks editpage + templatebody}], description => "plugins to enable by default", safe => 0, rebuild => 1, @@ -237,8 +275,8 @@ sub getsetup () { html5 => { type => "boolean", default => 0, - description => "generate HTML5? (experimental)", - advanced => 1, + description => "use elements new in HTML5 like
?", + advanced => 0, safe => 1, rebuild => 1, }, @@ -305,9 +343,9 @@ sub getsetup () { rebuild => 0, }, umask => { - type => "integer", - example => "022", - description => "force ikiwiki to use a particular umask", + type => "string", + example => "public", + description => "force ikiwiki to use a particular umask (keywords public, group or private, or a number)", advanced => 1, safe => 0, # paranoia rebuild => 0, @@ -320,11 +358,20 @@ sub getsetup () { safe => 0, # paranoia rebuild => 0, }, + libdirs => { + type => "string", + default => [], + example => ["$ENV{HOME}/.local/share/ikiwiki"], + description => "extra library and plugin directories", + advanced => 1, + safe => 0, # directory + rebuild => 0, + }, libdir => { type => "string", default => "", example => "$ENV{HOME}/.ikiwiki/", - description => "extra library and plugin directory", + description => "extra library and plugin directory (searched after libdirs)", advanced => 1, safe => 0, # directory rebuild => 0, @@ -336,6 +383,14 @@ sub getsetup () { safe => 0, # paranoia rebuild => 0, }, + timezone => { + type => "string", + default => "", + example => "US/Eastern", + description => "time zone name", + safe => 1, + rebuild => 1, + }, include => { type => "string", default => undef, @@ -477,7 +532,7 @@ sub getsetup () { }, setuptype => { type => "internal", - default => "Standard", + default => "Yaml", description => "perl class to use to dump setup file", safe => 0, rebuild => 0, @@ -489,6 +544,47 @@ sub getsetup () { safe => 0, rebuild => 0, }, + cookiejar => { + type => "string", + default => { file => "$ENV{HOME}/.ikiwiki/cookies" }, + description => "cookie control", + safe => 0, # hooks into perl module internals + rebuild => 0, + }, + useragent => { + type => "string", + default => "ikiwiki/$version", + example => "Wget/1.13.4 (linux-gnu)", + description => "set custom user agent string for outbound HTTP requests e.g. when fetching aggregated RSS feeds", + safe => 0, + rebuild => 0, + }, + responsive_layout => { + type => "boolean", + default => 1, + description => "theme has a responsive layout? (mobile-optimized)", + safe => 1, + rebuild => 1, + }, + deterministic => { + type => "boolean", + default => 0, + description => "try harder to produce deterministic output", + safe => 1, + rebuild => 1, + advanced => 1, + }, +} + +sub getlibdirs () { + my @libdirs; + if ($config{libdirs}) { + @libdirs = @{$config{libdirs}}; + } + if (length $config{libdir}) { + push @libdirs, $config{libdir}; + } + return @libdirs; } sub defaultconfig () { @@ -497,7 +593,6 @@ sub defaultconfig () { foreach my $key (keys %s) { push @ret, $key, $s{$key}->{default}; } - use Data::Dumper; return @ret; } @@ -529,6 +624,23 @@ sub checkconfig () { $ENV{$val}=$config{ENV}{$val}; } } + if (defined $config{timezone} && length $config{timezone}) { + $ENV{TZ}=$config{timezone}; + } + elsif (defined $ENV{TZ} && length $ENV{TZ}) { + $config{timezone}=$ENV{TZ}; + } + else { + eval q{use Config qw()}; + error($@) if $@; + + if ($Config::Config{d_gnulibc} && -e '/etc/localtime') { + $config{timezone}=$ENV{TZ}=':/etc/localtime'; + } + else { + $config{timezone}=$ENV{TZ}='GMT'; + } + } if ($config{w3mmode}) { eval q{use Cwd q{abs_path}}; @@ -544,7 +656,7 @@ sub checkconfig () { error(gettext("Must specify url to wiki with --url when using --cgi")); } - if (length $config{url}) { + if (defined $config{url} && length $config{url}) { eval q{use URI}; my $baseurl = URI->new($config{url}); @@ -556,12 +668,39 @@ sub checkconfig () { $local_cgiurl = $cgiurl->path; - if ($cgiurl->scheme ne $baseurl->scheme or - $cgiurl->authority ne $baseurl->authority) { + if ($cgiurl->scheme eq 'https' && + $baseurl->scheme eq 'http') { + # We assume that the same content is available + # over both http and https, because if it + # wasn't, accessing the static content + # from the CGI would be mixed-content, + # which would be a security flaw. + + if ($cgiurl->authority ne $baseurl->authority) { + # use protocol-relative URL for + # static content + $local_url = "$config{url}/"; + $local_url =~ s{^http://}{//}; + } + # else use host-relative URL for static content + + # either way, CGI needs to be absolute + $local_cgiurl = $config{cgiurl}; + } + elsif ($cgiurl->scheme ne $baseurl->scheme) { # too far apart, fall back to absolute URLs $local_url = "$config{url}/"; $local_cgiurl = $config{cgiurl}; } + elsif ($cgiurl->authority ne $baseurl->authority) { + # slightly too far apart, fall back to + # protocol-relative URLs + $local_url = "$config{url}/"; + $local_url =~ s{^https?://}{//}; + $local_cgiurl = $config{cgiurl}; + $local_cgiurl =~ s{^https?://}{//}; + } + # else keep host-relative URLs } $local_url =~ s{//$}{/}; @@ -574,7 +713,23 @@ sub checkconfig () { unless exists $config{wikistatedir} && defined $config{wikistatedir}; if (defined $config{umask}) { - umask(possibly_foolish_untaint($config{umask})); + my $u = possibly_foolish_untaint($config{umask}); + + if ($u =~ m/^\d+$/) { + umask($u); + } + elsif ($u eq 'private') { + umask(077); + } + elsif ($u eq 'group') { + umask(027); + } + elsif ($u eq 'public') { + umask(022); + } + else { + error(sprintf(gettext("unsupported umask setting %s"), $u)); + } } run_hooks(checkconfig => sub { shift->() }); @@ -585,14 +740,14 @@ sub checkconfig () { sub listplugins () { my %ret; - foreach my $dir (@INC, $config{libdir}) { + foreach my $dir (@INC, getlibdirs()) { next unless defined $dir && length $dir; foreach my $file (glob("$dir/IkiWiki/Plugin/*.pm")) { my ($plugin)=$file=~/.*\/(.*)\.pm$/; $ret{$plugin}=1; } } - foreach my $dir ($config{libdir}, "$installdir/lib/ikiwiki") { + foreach my $dir (getlibdirs(), "$installdir/lib/ikiwiki") { next unless defined $dir && length $dir; foreach my $file (glob("$dir/plugins/*")) { $ret{basename($file)}=1 if -x $file; @@ -603,8 +758,8 @@ sub listplugins () { } sub loadplugins () { - if (defined $config{libdir} && length $config{libdir}) { - unshift @INC, possibly_foolish_untaint($config{libdir}); + foreach my $dir (getlibdirs()) { + unshift @INC, possibly_foolish_untaint($dir); } foreach my $plugin (@{$config{default_plugins}}, @{$config{add_plugins}}) { @@ -637,8 +792,8 @@ sub loadplugin ($;$) { return if ! $force && grep { $_ eq $plugin} @{$config{disable_plugins}}; - foreach my $dir (defined $config{libdir} ? possibly_foolish_untaint($config{libdir}) : undef, - "$installdir/lib/ikiwiki") { + foreach my $possiblytainteddir (getlibdirs(), "$installdir/lib/ikiwiki") { + my $dir = possibly_foolish_untaint($possiblytainteddir); if (defined $dir && -x "$dir/plugins/$plugin") { eval { require IkiWiki::Plugin::external }; if ($@) { @@ -676,6 +831,7 @@ sub debug ($) { } my $log_open=0; +my $log_failed=0; sub log_message ($$) { my $type=shift; @@ -686,9 +842,17 @@ sub log_message ($$) { Sys::Syslog::openlog('ikiwiki', '', 'user'); $log_open=1; } - return eval { - Sys::Syslog::syslog($type, "[$config{wikiname}] %s", join(" ", @_)); + eval { + my $message = "[$config{wikiname}] ".join(" ", @_); + utf8::encode($message); + Sys::Syslog::syslog($type, "%s", $message); }; + if ($@) { + print STDERR "failed to syslog: $@" unless $log_failed; + $log_failed=1; + print STDERR "@_\n"; + } + return $@; } elsif (! $config{cgi}) { return print "@_\n"; @@ -810,17 +974,23 @@ sub srcfile ($;$) { return (srcfile_stat(@_))[0]; } -sub add_underlay ($) { +sub add_literal_underlay ($) { my $dir=shift; - if ($dir !~ /^\//) { - $dir="$config{underlaydirbase}/$dir"; - } - if (! grep { $_ eq $dir } @{$config{underlaydirs}}) { unshift @{$config{underlaydirs}}, $dir; } +} + +sub add_underlay ($) { + my $dir = shift; + + if ($dir !~ /^\//) { + $dir="$config{underlaydirbase}/$dir"; + } + add_literal_underlay($dir); + # why does it return 1? we just don't know return 1; } @@ -1007,7 +1177,7 @@ sub bestlink ($$) { sub isinlinableimage ($) { my $file=shift; - return $file =~ /\.(png|gif|jpg|jpeg)$/i; + return $file =~ /\.(png|gif|jpg|jpeg|svg)$/i; } sub pagetitle ($;$) { @@ -1054,7 +1224,25 @@ sub cgiurl (@) { } return $cgiurl."?". - join("&", map $_."=".uri_escape_utf8($params{$_}), keys %params); + join("&", map $_."=".uri_escape_utf8($params{$_}), sort(keys %params)); +} + +sub cgiurl_abs (@) { + eval q{use URI}; + URI->new_abs(cgiurl(@_), $config{cgiurl}); +} + +# Same as cgiurl_abs, but when the user connected using https, +# will be a https url even if the cgiurl is normally a http url. +# +# This should be used for anything involving emailing a login link, +# because a https session cookie will not be sent over http. +sub cgiurl_abs_samescheme (@) { + my $u=cgiurl_abs(@_); + if (($ENV{HTTPS} && lc $ENV{HTTPS} ne "off")) { + $u=~s/^http:/https:/i; + } + return $u } sub baseurl (;$) { @@ -1068,6 +1256,16 @@ sub baseurl (;$) { return $page; } +sub urlabs ($$) { + my $url=shift; + my $urlbase=shift; + + return $url unless defined $urlbase && length $urlbase; + + eval q{use URI}; + URI->new_abs($url, $urlbase)->as_string; +} + sub abs2rel ($$) { # Work around very innefficient behavior in File::Spec if abs2rel # is passed two relative paths. It's much faster if paths are @@ -1103,9 +1301,25 @@ sub formattime ($;$) { $format=$config{timeformat}; } - # strftime doesn't know about encodings, so make sure - # its output is properly treated as utf8 - return decode_utf8(POSIX::strftime($format, localtime($time))); + return strftime_utf8($format, localtime($time)); +} + +my $strftime_encoding; +sub strftime_utf8 { + # strftime didn't know about encodings in older Perl, so make sure + # its output is properly treated as utf8. + # Note that this does not handle utf-8 in the format string. + my $result = POSIX::strftime(@_); + + if (Encode::is_utf8($result)) { + return $result; + } + + ($strftime_encoding) = POSIX::setlocale(&POSIX::LC_TIME) =~ m#\.([^@]+)# + unless defined $strftime_encoding; + $strftime_encoding + ? Encode::decode($strftime_encoding, $result) + : $result; } sub date_3339 ($) { @@ -1134,13 +1348,13 @@ sub beautify_urlpath ($) { return $url; } -sub urlto ($$;$) { +sub urlto ($;$$) { my $to=shift; my $from=shift; my $absolute=shift; if (! length $to) { - return beautify_urlpath(baseurl($from)."index.$config{htmlext}"); + $to = 'index'; } if (! $destsources{$to}) { @@ -1152,7 +1366,7 @@ sub urlto ($$;$) { } if (! defined $from) { - my $u = $local_url; + my $u = $local_url || ''; $u =~ s{/$}{}; return $u.beautify_urlpath("/".$to); } @@ -1208,7 +1422,7 @@ sub htmllink ($$$;@) { $cgilink = " "create", - page => lc($link), + page => $link, from => $lpage )."\" rel=\"nofollow\">?"; } @@ -1242,10 +1456,11 @@ sub userpage ($) { return length $config{userdir} ? "$config{userdir}/$user" : $user; } +# Username to display for openid accounts. sub openiduser ($) { my $user=shift; - if ($user =~ m!^https?://! && + if (defined $user && $user =~ m!^https?://! && eval q{use Net::OpenID::VerifiedIdentity; 1} && !$@) { my $display; @@ -1276,6 +1491,36 @@ sub openiduser ($) { return; } +# Username to display for emailauth accounts. +sub emailuser ($) { + my $user=shift; + if (defined $user && $user =~ m/(.+)@/) { + my $nick=$1; + # remove any characters from not allowed in wiki files + # support use w/o %config set + my $chars = defined $config{wiki_file_chars} ? $config{wiki_file_chars} : "-[:alnum:]+/.:_"; + $nick=~s/[^$chars]/_/g; + return $nick; + } + return; +} + +# Some user information should not be exposed in commit metadata, etc. +# This generates a cloaked form of such information. +sub cloak ($) { + my $user=shift; + # cloak email address using http://xmlns.com/foaf/spec/#term_mbox_sha1sum + if ($user=~m/(.+)@/) { + my $nick=$1; + eval q{use Digest::SHA}; + return $user if $@; + return $nick.'@'.Digest::SHA::sha1_hex("mailto:$user"); + } + else { + return $user; + } +} + sub htmlize ($$$$) { my $page=shift; my $destpage=shift; @@ -1357,16 +1602,21 @@ sub preprocess ($$$;$$) { # consider it significant. my @params; while ($params =~ m{ - (?:([-\w]+)=)? # 1: named parameter key? + (?:([-.\w]+)=)? # 1: named parameter key? (?: """(.*?)""" # 2: triple-quoted value | "([^"]*?)" # 3: single-quoted value | - (\S+) # 4: unquoted value + '''(.*?)''' # 4: triple-single-quote + | + <<([a-zA-Z]+)\n # 5: heredoc start + (.*?)\n\5 # 6: heredoc value + | + (\S+) # 7: unquoted value ) (?:\s+|$) # delimiter to next param - }sgx) { + }msgx) { my $key=$1; my $val; if (defined $2) { @@ -1381,6 +1631,12 @@ sub preprocess ($$$;$$) { elsif (defined $4) { $val=$4; } + elsif (defined $7) { + $val=$7; + } + elsif (defined $6) { + $val=$6; + } if (defined $key) { push @params, $key, $val; @@ -1389,7 +1645,7 @@ sub preprocess ($$$;$$) { push @params, $val, ''; } } - if ($preprocessing{$page}++ > 3) { + if ($preprocessing{$page}++ > 8) { # Avoid loops of preprocessed pages preprocessing # other pages that preprocess them, etc. return "[[!$command ". @@ -1410,6 +1666,11 @@ sub preprocess ($$$;$$) { if ($@) { my $error=$@; chomp $error; + eval q{use HTML::Entities}; + # Also encode most ASCII punctuation + # as entities so that error messages + # are not interpreted as Markdown etc. + $error = encode_entities($error, '^-A-Za-z0-9+_,./:;= '."'"); $ret="[[!$command ". gettext("Error").": $error"."]]"; } @@ -1443,12 +1704,17 @@ sub preprocess ($$$;$$) { ( # 4: the parameters.. \s+ # Must have space if parameters present (?: - (?:[-\w]+=)? # named parameter key? + (?:[-.\w]+=)? # named parameter key? (?: """.*?""" # triple-quoted value | "[^"]*?" # single-quoted value | + '''.*?''' # triple-single-quote + | + <<([a-zA-Z]+)\n # 5: heredoc start + (?:.*?)\n\5 # heredoc value + | [^"\s\]]+ # unquoted value ) \s* # whitespace or end @@ -1466,12 +1732,17 @@ sub preprocess ($$$;$$) { \s+ ( # 4: the parameters.. (?: - (?:[-\w]+=)? # named parameter key? + (?:[-.\w]+=)? # named parameter key? (?: """.*?""" # triple-quoted value | "[^"]*?" # single-quoted value | + '''.*?''' # triple-single-quote + | + <<([a-zA-Z]+)\n # 5: heredoc start + (?:.*?)\n\5 # heredoc value + | [^"\s\]]+ # unquoted value ) \s* # whitespace or end @@ -1577,7 +1848,7 @@ sub check_canchange (@) { $file=possibly_foolish_untaint($file); if (! defined $file || ! length $file || file_pruned($file)) { - error(gettext("bad file name %s"), $file); + error(sprintf(gettext("bad file name %s"), $file)); } my $type=pagetype($file); @@ -1636,8 +1907,11 @@ sub lockwiki () { } open($wikilock, '>', "$config{wikistatedir}/lockfile") || error ("cannot write to $config{wikistatedir}/lockfile: $!"); - if (! flock($wikilock, 2)) { # LOCK_EX - error("failed to get lock"); + if (! flock($wikilock, LOCK_EX | LOCK_NB)) { + debug("failed to get lock; waiting..."); + if (! flock($wikilock, LOCK_EX)) { + error("failed to get lock"); + } } return 1; } @@ -1677,7 +1951,8 @@ sub enable_commit_hook () { sub loadindex () { %oldrenderedfiles=%pagectime=(); - if (! $config{rebuild}) { + my $rebuild=$config{rebuild}; + if (! $rebuild) { %pagesources=%pagemtime=%oldlinks=%links=%depends= %destsources=%renderedfiles=%pagecase=%pagestate= %depends_simple=%typedlinks=%oldtypedlinks=(); @@ -1689,7 +1964,8 @@ sub loadindex () { open ($in, "<", "$config{wikistatedir}/indexdb") || return; } else { - $config{gettime}=1; # first build + # gettime on first build + $config{gettime}=1 unless defined $config{gettime}; return; } } @@ -1717,10 +1993,16 @@ sub loadindex () { foreach my $src (keys %$pages) { my $d=$pages->{$src}; - my $page=pagename($src); + my $page; + if (exists $d->{page} && ! $rebuild) { + $page=$d->{page}; + } + else { + $page=pagename($src); + } $pagectime{$page}=$d->{ctime}; $pagesources{$page}=$src; - if (! $config{rebuild}) { + if (! $rebuild) { $pagemtime{$page}=$d->{mtime}; $renderedfiles{$page}=$d->{dest}; if (exists $d->{links} && ref $d->{links}) { @@ -1770,6 +2052,8 @@ sub loadindex () { foreach my $page (keys %renderedfiles) { $destsources{$_}=$page foreach @{$renderedfiles{$page}}; } + $lastrev=$index->{lastrev}; + @underlayfiles=@{$index->{underlayfiles}} if ref $index->{underlayfiles}; return close($in); } @@ -1791,6 +2075,7 @@ sub saveindex () { my $src=$pagesources{$page}; $index{page}{$src}={ + page => $page, ctime => $pagectime{$page}, mtime => $pagemtime{$page}, dest => $renderedfiles{$page}, @@ -1810,11 +2095,7 @@ sub saveindex () { } if (exists $pagestate{$page}) { - foreach my $id (@plugins) { - foreach my $key (keys %{$pagestate{$page}{$id}}) { - $index{page}{$src}{state}{$id}{$key}=$pagestate{$page}{$id}{$key}; - } - } + $index{page}{$src}{state}=$pagestate{$page}; } } @@ -1826,6 +2107,9 @@ sub saveindex () { } } + $index{lastrev}=$lastrev; + $index{underlayfiles}=\@underlayfiles; + $index{version}="3"; my $ret=Storable::nstore_fd(\%index, $out); return if ! defined $ret || ! $ret; @@ -1883,14 +2167,23 @@ sub template_depends ($$;@) { if (defined $page && defined $tpage) { add_depends($page, $tpage); } - + my @opts=( filter => sub { my $text_ref = shift; ${$text_ref} = decode_utf8(${$text_ref}); + run_hooks(readtemplate => sub { + ${$text_ref} = shift->( + id => $name, + page => $tpage, + content => ${$text_ref}, + untrusted => $untrusted, + ); + }); }, loop_context_vars => 1, die_on_bad_params => 0, + parent_global_vars => 1, filename => $filename, @_, ($untrusted ? (no_includes => 1) : ()), @@ -1905,39 +2198,6 @@ sub template ($;@) { template_depends(shift, undef, @_); } -sub misctemplate ($$;@) { - my $title=shift; - my $content=shift; - my %params=@_; - - my $template=template("page.tmpl"); - - my $page=""; - if (exists $params{page}) { - $page=delete $params{page}; - } - run_hooks(pagetemplate => sub { - shift->( - page => $page, - destpage => $page, - template => $template, - ); - }); - templateactions($template, ""); - - $template->param( - dynamic => 1, - title => $title, - wikiname => $config{wikiname}, - content => $content, - baseurl => baseurl(), - html5 => $config{html5}, - %params, - ); - - return $template->output; -} - sub templateactions ($$) { my $template=shift; my $page=shift; @@ -2032,7 +2292,7 @@ sub rcs_recentchanges ($) { $hooks{rcs}{rcs_recentchanges}{call}->(@_); } -sub rcs_diff ($) { +sub rcs_diff ($;$) { $hooks{rcs}{rcs_diff}{call}->(@_); } @@ -2209,6 +2469,133 @@ sub add_autofile ($$$) { $autofiles{$file}{generator}=$generator; } +sub useragent (@) { + my %params = @_; + my $for_url = delete $params{for_url}; + # Fail safe, in case a plugin calling this function is relying on + # a future parameter to make the UA more strict + foreach my $key (keys %params) { + error "Internal error: useragent(\"$key\" => ...) not understood"; + } + + eval q{use LWP}; + error($@) if $@; + + my %args = ( + agent => $config{useragent}, + cookie_jar => $config{cookiejar}, + env_proxy => 0, + protocols_allowed => [qw(http https)], + ); + my %proxies; + + if (defined $for_url) { + # We know which URL we're going to fetch, so we can choose + # whether it's going to go through a proxy or not. + # + # We reimplement http_proxy, https_proxy and no_proxy here, so + # that we are not relying on LWP implementing them exactly the + # same way we do. + + eval q{use URI}; + error($@) if $@; + + my $proxy; + my $uri = URI->new($for_url); + + if ($uri->scheme eq 'http') { + $proxy = $ENV{http_proxy}; + # HTTP_PROXY is deliberately not implemented + # because the HTTP_* namespace is also used by CGI + } + elsif ($uri->scheme eq 'https') { + $proxy = $ENV{https_proxy}; + $proxy = $ENV{HTTPS_PROXY} unless defined $proxy; + } + else { + $proxy = undef; + } + + foreach my $var (qw(no_proxy NO_PROXY)) { + my $no_proxy = $ENV{$var}; + if (defined $no_proxy) { + foreach my $domain (split /\s*,\s*/, $no_proxy) { + if ($domain =~ s/^\*?\.//) { + # no_proxy="*.example.com" or + # ".example.com": match suffix + # against .example.com + if ($uri->host =~ m/(^|\.)\Q$domain\E$/i) { + $proxy = undef; + } + } + else { + # no_proxy="example.com": + # match exactly example.com + if (lc $uri->host eq lc $domain) { + $proxy = undef; + } + } + } + } + } + + if (defined $proxy) { + $proxies{$uri->scheme} = $proxy; + # Paranoia: make sure we can't bypass the proxy + $args{protocols_allowed} = [$uri->scheme]; + } + } + else { + # The plugin doesn't know yet which URL(s) it's going to + # fetch, so we have to make some conservative assumptions. + my $http_proxy = $ENV{http_proxy}; + my $https_proxy = $ENV{https_proxy}; + $https_proxy = $ENV{HTTPS_PROXY} unless defined $https_proxy; + + # We don't respect no_proxy here: if we are not using the + # paranoid user-agent, then we need to give the proxy the + # opportunity to reject undesirable requests. + + # If we have one, we need the other: otherwise, neither + # LWPx::ParanoidAgent nor the proxy would have the + # opportunity to filter requests for the other protocol. + if (defined $https_proxy && defined $http_proxy) { + %proxies = (http => $http_proxy, https => $https_proxy); + } + elsif (defined $https_proxy) { + %proxies = (http => $https_proxy, https => $https_proxy); + } + elsif (defined $http_proxy) { + %proxies = (http => $http_proxy, https => $http_proxy); + } + + } + + if (scalar keys %proxies) { + # The configured proxy is responsible for deciding which + # URLs are acceptable to fetch and which URLs are not. + my $ua = LWP::UserAgent->new(%args); + foreach my $scheme (@{$ua->protocols_allowed}) { + unless ($proxies{$scheme}) { + error "internal error: $scheme is allowed but has no proxy"; + } + } + # We can't pass the proxies in %args because that only + # works since LWP 6.24. + foreach my $scheme (keys %proxies) { + $ua->proxy($scheme, $proxies{$scheme}); + } + return $ua; + } + + eval q{use LWPx::ParanoidAgent}; + if ($@) { + print STDERR "warning: installing LWPx::ParanoidAgent is recommended\n"; + return LWP::UserAgent->new(%args); + } + return LWPx::ParanoidAgent->new(%args); +} + sub sortspec_translate ($$) { my $spec = shift; my $reverse = shift; @@ -2345,6 +2732,19 @@ sub pagespec_match ($$;@) { return $sub->($page, @params); } +# e.g. @pages = sort_pages("title", \@pages, reverse => "yes") +# +# Not exported yet, but could be in future if it is generally useful. +# Note that this signature is not the same as IkiWiki::SortSpec::sort_pages, +# which is "more internal". +sub sort_pages ($$;@) { + my $sort = shift; + my $list = shift; + my %params = @_; + $sort = sortspec_translate($sort, $params{reverse}); + return IkiWiki::SortSpec::sort_pages($sort, @$list); +} + sub pagespec_match_list ($$;@) { my $page=shift; my $pagespec=shift; @@ -2450,21 +2850,48 @@ our @ISA = 'IkiWiki::SuccessReason'; package IkiWiki::SuccessReason; +# A blessed array-ref: +# +# [0]: human-readable reason for success (or, in FailReason subclass, failure) +# [1]{""}: +# - if absent or false, the influences of this evaluation are "static", +# see the influences_static method +# - if true, they are dynamic (not static) +# [1]{any other key}: +# the dependency types of influences, as returned by the influences method + use overload ( + # in string context, it's the human-readable reason '""' => sub { $_[0][0] }, + # in boolean context, SuccessReason is 1 and FailReason is 0 '0+' => sub { 1 }, + # negating a result gives the opposite result with the same influences '!' => sub { bless $_[0], 'IkiWiki::FailReason'}, + # A & B = (A ? B : A) with the influences of both '&' => sub { $_[1]->merge_influences($_[0], 1); $_[1] }, + # A | B = (A ? A : B) with the influences of both '|' => sub { $_[0]->merge_influences($_[1]); $_[0] }, fallback => 1, ); +# SuccessReason->new("human-readable reason", page => deptype, ...) + sub new { my $class = shift; my $value = shift; return bless [$value, {@_}], $class; } +# influences(): return a reference to a copy of the hash +# { page => dependency type } describing the pages that indirectly influenced +# this result, but would not cause a dependency through ikiwiki's core +# dependency logic. +# +# See [[todo/dependency_types]] for extensive discussion of what this means. +# +# influences(page => deptype, ...): remove all influences, replace them +# with the arguments, and return a reference to a copy of the new influences. + sub influences { my $this=shift; $this->[1]={@_} if @_; @@ -2473,15 +2900,46 @@ sub influences { return \%i; } +# True if this result has the same influences whichever page it matches, +# For instance, whether bar matches backlink(foo) is influenced only by +# the set of links in foo, so its only influence is { foo => DEPEND_LINKS }, +# which does not mention bar anywhere. +# +# False if this result would have different influences when matching +# different pages. For instance, when testing whether link(foo) matches bar, +# { bar => DEPEND_LINKS } is an influence on that result, because changing +# bar's links could change the outcome; so its influences are not the same +# as when testing whether link(foo) matches baz. +# +# Static influences are one of the things that make pagespec_match_list +# more efficient than repeated calls to pagespec_match. + sub influences_static { return ! $_[0][1]->{""}; } +# Change the influences of $this to be the influences of "$this & $other" +# or "$this | $other". +# +# If both $this and $other are either successful or have influences, +# or this is an "or" operation, the result has all the influences from +# either of the arguments. It has dynamic influences if either argument +# has dynamic influences. +# +# If this is an "and" operation, and at least one argument is a +# FailReason with no influences, the result has no influences, and they +# are not dynamic. For instance, link(foo) matching bar is influenced +# by bar, but enabled(ddate) has no influences. Suppose ddate is disabled; +# then (link(foo) and enabled(ddate)) not matching bar is not influenced by +# bar, because it would be false however often you edit bar. + sub merge_influences { my $this=shift; my $other=shift; my $anded=shift; + # This "if" is odd because it needs to avoid negating $this + # or $other, which would alter the objects in-place. Be careful. if (! $anded || (($this || %{$this->[1]}) && ($other || %{$other->[1]}))) { foreach my $influence (keys %{$other->[1]}) { @@ -2494,6 +2952,8 @@ sub merge_influences { } } +# Change $this so it is not considered to be influenced by $torm. + sub remove_influence { my $this=shift; my $torm=shift; @@ -2613,8 +3073,14 @@ sub match_link ($$;@) { } sub match_backlink ($$;@) { - my $ret=match_link($_[1], $_[0], @_); - $ret->influences($_[1] => $IkiWiki::DEPEND_LINKS); + my $page=shift; + my $testpage=shift; + my %params=@_; + if ($testpage eq '.') { + $testpage = $params{'location'} + } + my $ret=match_link($testpage, $page, @_); + $ret->influences($testpage => $IkiWiki::DEPEND_LINKS); return $ret; } @@ -2705,12 +3171,12 @@ sub match_user ($$;@) { my $user=shift; my %params=@_; - my $regexp=IkiWiki::glob2re($user); - if (! exists $params{user}) { return IkiWiki::ErrorReason->new("no user specified"); } + my $regexp=IkiWiki::glob2re($user); + if (defined $params{user} && $params{user}=~$regexp) { return IkiWiki::SuccessReason->new("user is $user"); } @@ -2750,8 +3216,10 @@ sub match_ip ($$;@) { if (! exists $params{ip}) { return IkiWiki::ErrorReason->new("no IP specified"); } + + my $regexp=IkiWiki::glob2re(lc $ip); - if (defined $params{ip} && lc $params{ip} eq lc $ip) { + if (defined $params{ip} && lc $params{ip}=~$regexp) { return IkiWiki::SuccessReason->new("IP is $ip"); } else { @@ -2774,6 +3242,7 @@ sub cmp_title { IkiWiki::pagetitle(IkiWiki::basename($b)) } +sub cmp_path { IkiWiki::pagetitle($a) cmp IkiWiki::pagetitle($b) } sub cmp_mtime { $IkiWiki::pagemtime{$b} <=> $IkiWiki::pagemtime{$a} } sub cmp_age { $IkiWiki::pagectime{$b} <=> $IkiWiki::pagectime{$a} }