2 # Common functions for HTML/XHTML output generation
3 # Copyright (C) 2003-2005 Jan Kratochvil <project-www.jankratochvil.net@jankratochvil.net>
5 # This program is free software; you can redistribute it and/or modify
6 # it under the terms of the GNU General Public License as published by
7 # the Free Software Foundation; exactly version 2 of June 1991 is required
9 # This program is distributed in the hope that it will be useful,
10 # but WITHOUT ANY WARRANTY; without even the implied warranty of
11 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 # GNU General Public License for more details.
14 # You should have received a copy of the GNU General Public License
15 # along with this program; if not, write to the Free Software
16 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20 require 5.6.0; # at least 'use warnings;' but we need some 5.6.0+ modules anyway
21 our $VERSION=do { my @r=(q$Revision$=~/\d+/g); sprintf "%d.".("%03d"x$#r),@r; };
32 &path_web &path_abs_disk
36 &img ¢erimg &rightimg
38 &input_hidden_persistents
41 our @ISA=qw(Tie::Handle Exporter);
43 my %packages_used_hash; # $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
44 my %packages_used_array;
48 use Carp qw(cluck confess);
49 $W->{"__My::Web_init"}=1;
55 # print STDERR "Wrequire $file\n";
62 for (my $depth=0;defined caller($depth);$depth++) {
63 $callers{caller($depth)}=1;
65 my $selfpkg=__PACKAGE__;
67 for my $target ($class,__PACKAGE__) {
68 for my $caller (keys(%callers)) {
69 next if $caller eq $target;
70 next if $packages_used_hash{$caller}{$target}++;
71 cluck "Appending to the '_done' package list: caller=$caller,target=$target"
72 if $packages_used_hash{$caller}{"_done"};
73 push @{$packages_used_array{$caller}},$target;
76 eval { CORE::require "$file"; } or confess $@;
77 1; # Otherwise 'require' would already file above.
84 # print STDERR "Wuse $file\n";
86 local $Exporter::ExportLevel=$Exporter::ExportLevel+1;
95 local $Exporter::ExportLevel=$Exporter::ExportLevel+1;
97 return $class->SUPER::import(@rest);
101 use WebConfig; # see also below: Wuse 'WebConfig';
103 require Image::Size; # for &imgsize
104 use File::Basename; # &basename
105 use Carp qw(cluck confess);
107 require HTTP::BrowserDetect;
108 require HTTP::Negotiate;
109 our $have_Geo_IP; BEGIN { $have_Geo_IP=eval { require Geo::IP; 1; }; }
110 # Do not: use ModPerl::Util qw(exit);
111 # to prevent in mod_perl2: "exit" is not exported by the ModPerl::Util module
113 use POSIX qw(strftime);
115 use Apache2::Const qw(HTTP_MOVED_TEMPORARILY OK HTTP_OK);
122 require Data::Compare;
126 use Apache2::Connection;
127 require MIME::Base64;
128 use Apache2::ServerUtil;
138 print STDERR "$class->init ".Apache2::RequestUtil->request()->unparsed_uri()."\n";
140 # We need to track package dependencies, so we need to call it from &init.
141 # We cannot do it in BEGIN { } block
142 # as it would not be tracked for each of the toplevel users later.
146 # $W={} can get somehow created very easily.
147 # Do not: cluck "W not empty:\n".Dumper($W) if keys(%$W);
148 # to prevent (of $W->{"headers_in"}): TODO: Enumeration may not be expected.
149 cluck "W not empty; __PACKAGE__ was: ".$W->{"__PACKAGE__"} if keys(%$W);
150 $W=My::Hash->new({},"My::Hash::Sub","My::Hash::Push");
153 "__PACKAGE__"=>scalar(caller()),
155 %args, # override %WebConfig settings
158 # {"__PACKAGE__"} is mandatory for mod_perl-2.0;
159 # $Apache2::Registry::curstash is no longer supported.
160 do { cluck "No $_" if !$W->{$_}; } for "__PACKAGE__";
164 do { cluck "charset==$_, expecting ISO-8859-1" if $_ ne "ISO-8859-1"; } for CGI::charset();
165 CGI::charset("utf-8");
167 do { $W->{$_}=0 if !defined $W->{$_}; } for "detect_ent";
168 do { $W->{$_}=0 if !defined $W->{$_}; } for "detect_js";
169 do { $W->{$_}=1 if !defined $W->{$_}; } for "have_css"; # AFAIK it does not hurt anyone.
170 do { $W->{$_}=0 if !defined $W->{$_}; } for "css_inherit";
171 do { $W->{$_}=1 if !defined $W->{$_}; } for "footer";
172 do { $W->{$_}=1 if !defined $W->{$_}; } for "footer_delimit";
173 do { $W->{$_}=1 if !defined $W->{$_}; } for "footer_ids";
174 do { $W->{$_}=1 if !defined $W->{$_}; } for "indexme";
175 do { $W->{$_}="" if !defined $W->{$_}; } for "head";
176 do { $W->{$_}="" if !defined $W->{$_}; } for "body_attr";
177 do { $W->{$_}="en-US" if !defined $W->{$_}; } for "language";
180 for (qw(footer_ids)) {
181 $W->{$_}=0 if !$W->{"footer"};
182 $footer_any=1 if $W->{$_};
184 $W->{"footer"}=0 if !$footer_any;
185 $W->{"footer_delimit"}=0 if !$W->{"footer"};
187 $W->{"r"}=Apache2::RequestUtil->request();
189 $W->{"r"}->push_handlers("PerlCleanupHandler"=>\&cleanup);
191 $W->{"web_hostname"}||=$W->{"r"}->hostname();
193 tie *STDOUT,$W->{"r"};
197 $W->{"QUERY_STRING"}=$W->{"r"}->args() || "";
198 if ($W->{"detect_ent"}) {
199 if ($W->{"QUERY_STRING"}=~/[&]amp;have_ent/)
200 { $W->{"have_ent"}=0; }
201 elsif ($W->{"QUERY_STRING"}=~ /[&]have_ent/)
202 { $W->{"have_ent"}=1; }
204 { delete $W->{"have_ent"}; }
205 if (!defined $W->{"have_ent"} && $W->{"r"}->method() eq "GET") {
206 $W->{"head"}.='<meta http-equiv="Refresh" content="0; URL='
207 .escapeHTML("http://".$W->{"web_hostname"}."/".($W->{"r"}->uri()=~m#^/*(.*)$#)[0]
208 ."?".($W->{"QUERY_STRING"} || "detect_ent_glue=1").'&have_ent=detect')
212 $W->{"QUERY_STRING"}=~s/([&])amp;/$1/g;
213 $W->{"r"}->args($W->{"QUERY_STRING"});
214 # Workaround: &CGI::Vars behaves weird if strings passed both as POST data and in: $QUERY_STRING
215 do { $W->{"r"}->args(""); delete $ENV{"QUERY_STRING"}; } if $W->{"r"}->method() eq "POST";
216 # Do not: $W->{"r"}->args()
217 # as it parses only QUERY_STRING (not POST data).
218 $W->{"args_orig_array"}=[ CGI->new($W->{"r"})->Vars() ];
219 $W->{"args"}={ @{$W->{"args_orig_array"}} };
220 for my $name (keys(%{$W->{"args"}})) {
221 my @vals=split /\x00/,$W->{"args"}{$name};
223 $W->{"args"}{$name}=[@vals];
226 $W->{"headers_in"}=$W->{"r"}->headers_in();
227 Wrequire 'My::Hash::Merge';
228 $W->{"headers_in"}=My::Hash::Merge->new(
231 "_remote_ip"=>sub { return $W->{"r"}->connection()->remote_ip(); },
234 Wrequire 'My::Hash::Readonly';
235 $W->{"headers_in"}=My::Hash::Readonly->new($W->{"headers_in"});
237 if ($W->{"r"}->method() eq "GET" || $W->{"r"}->method() eq "HEAD") {
238 for (\$W->{"http_safe"}) {
239 # Do not: # Extend the current ETag system instead if you would need it:
240 # cluck "Explicitely NOT HTTP-Safe for method \"".$W->{"r"}->method()."\"?!?"
241 # if defined($$_) && !$$_;
242 # as sometimes it just does not make sense to cache it.
243 $$_=1 if !defined $$_;
247 for (\$W->{"http_safe"}) {
248 cluck "Undefined HTTP-Safe-ty for method \"".$W->{"r"}->method()."\"!"
250 $$_=0 if !defined $$_;
253 if ($W->{"http_safe"}) {
254 Wrequire 'My::Hash::RecordKeys';
255 $W->{"headers_in_RecordKeys"}=My::Hash::RecordKeys->new($W->{"headers_in"});
256 $W->{"headers_in"}=$W->{"headers_in_RecordKeys"};
259 $W->{"browser"}=sub {
260 # Lazy-evaluation, we may not need the "User-Agent" header at all.
261 return our $r||=HTTP::BrowserDetect->new($W->{"headers_in"}{"User-Agent"});
264 if (!defined $W->{"have_style"}) {
265 $W->{"have_style"}=sub {
266 # Lazy-evaluation, we may not need the "User-Agent" header at all.
267 return our $r||=(!$W->{"browser"}->netscape() || ($W->{"browser"}->major() && $W->{"browser"}->major()>4) ? 1 : 0);
271 $W->{"have_js"}=($W->{"args"}{"have_js"} ? 1 : 0);
272 if ($W->{"detect_js"} && !$W->{"have_js"}) {
274 # as at least Lynx inhibits any further HTML output.
275 # Do not: text/javascript
276 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
277 # "application/javascript" so far standardized till 2005-12-08 by:
278 # http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
279 $W->{"head"}.='<script type="application/javascript" src="'.uri_escaped(path_web('/My/HaveJS.pm')).'"></script>'."\n";
282 do { _args_check(%$_) if $_; } for ($W->{"args_check"});
284 $W->{"_init_done"}=1;
290 my($apache_request)=@_;
292 cluck "CORE::GLOBAL::exit hook not ran" if !$W->{"_exit_done"};
293 cluck "packages not finalized" if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
301 # PerlResponseHandler is RUN_FIRST and &ModPerl::Util::exit returns OK, so no (sane) go.
302 # PerlLogHandler is already too late to be able to produce any output.
306 cluck "Missing ->init while in exit_hook()" if !$W->{"_init_done"};
307 # &footer will call us recursively!
308 footer() if !$W->{"_exit_done"}++;
309 return &{$exit_orig}(@_);
313 do { cluck "exit_hook_start() twice?"; return; } if defined $exit_orig;
314 $exit_orig=\&CORE::GLOBAL::exit;
315 # Prevent: Subroutine CORE::GLOBAL::exit redefined
316 no warnings 'redefine';
317 *CORE::GLOBAL::exit=\&exit_hook;
321 do { cluck "exit_hook_stop() without exit_hook_start()?"; return; }
322 if \&exit_hook ne \&CORE::GLOBAL::exit;
323 do { cluck "INTERNAL: exit_orig uninitialized"; return; }
325 # Prevent: Subroutine CORE::GLOBAL::exit redefined
326 no warnings 'redefine';
327 *CORE::GLOBAL::exit=$exit_orig;
331 # Be aware other parts of code (non-My::Web) will NOT use this function!
332 # Do not: Wprint $W->{"heading"},"undef"=>1;
333 # as we would need to undef() it to turn it off and it would get defaulted in such case.
334 # Do not: exists $W->{"heading"}
335 # as we use a lot of 'for $W->{"heading"}' which instantiates it with the value: undef()
340 cluck "undef Wprint" if !defined $text && !$args{"undef"};
341 delete $args{"undef"};
342 cluck join(" ","Invalid arguments:",keys(%args)) if keys(%args);
343 return if !defined $text;
344 # Do not: cluck "utf-8 untested" if Encode::is_utf8($text);
345 # as it is valid here.
346 $W->{"r"}->puts($text);
349 sub request_check(;$)
353 # Use &eval to prevent: Global $r object is not available. Set:\n\tPerlOptions +GlobalRequest\nin ...
354 # CGI requires valid "r": check it beforehand here.
355 confess "Calling sensitive dynamic code from a static code" if !eval { Apache2::RequestUtil->request(); };
356 # Do not: confess "Calling sensitive dynamic code without My::Web::init" if !$W->{"__PACKAGE__"};
357 # as it is valid at least while preparing arguments to call: &project::Lib::init
364 # Prevent &CGI::escapeHTML breaking utf-8 strings like: \xC4\x9B eq \x{11B}
365 # Prevent case if we run under mod_perl but still just initializing:
366 request_check() if $ENV{"MOD_PERL"};
367 # Generally we are initialized from &init but we may be used without it without mod_perl
368 # and in such case check the change on all non-first invocations.
370 if (!$ENV{"MOD_PERL"} && $init++) {
371 do { cluck "charset==$_" if $_ ne "utf-8"; } for CGI::charset();
373 CGI::charset("utf-8");
375 return CGI::escapeHTML($text);
378 # /home/user/www/webdir
379 sub dir_top_abs_disk()
381 our $dir_top_abs_disk;
382 if (!$dir_top_abs_disk) {
383 my $selfpkg_relpath=__PACKAGE__;
384 $selfpkg_relpath=~s{::}{/}g;
385 $selfpkg_relpath.=".pm";
386 my $selfpkg_abspath=$INC{$selfpkg_relpath} or do {
387 cluck "Unable to find self package $selfpkg_relpath";
390 $selfpkg_abspath=~s{/*\Q$selfpkg_relpath\E$}{} or do {
391 cluck "Unable to strip myself \"$selfpkg_relpath\" from the abspath: $selfpkg_abspath";
394 cluck "INC{myself} is relative?: $selfpkg_abspath" if $selfpkg_abspath!~m{^/};
395 $dir_top_abs_disk=$selfpkg_abspath;
397 return $dir_top_abs_disk;
403 if (!$W->{"unparsed_uri"}) {
405 # as we may be called before &init from: &My::Project::init
406 my $r=Apache2::RequestUtil->request();
407 cluck "Calling ".'&unparsed_uri'." from a static code, going to fail" if !$r;
408 my $uri_string=$r->unparsed_uri() or cluck "Valid 'r' missing unparsed_uri()?";
409 my $uri=URI->new_abs($uri_string,"http://".$W->{"web_hostname"}."/");
410 $W->{"unparsed_uri"}=$uri;
412 return $W->{"unparsed_uri"};
419 # Otherwise we may have been already processed and thus legally relativized.
420 # FIXME data: Currently disabled, all the data are too violating such rule.
422 my $uri_check=URI->new($in);
423 $uri_check->scheme() || $in=~m{^\Q./\E} || $in=~m{^/}
424 or cluck "Use './' or '/' prefix for all the local references: $in";
426 my $uri=URI->new_abs($in,unparsed_uri());
427 $uri=$uri->canonical();
431 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
437 cluck if !$args{"uri_as_in"} && ref $in;
438 my $uri=in_to_uri_abs($in);
439 if (uri_is_local($uri)) {
440 # Prefer the $uri values over "args_persistent" values.
441 $uri->query_form_hash({
444 my $val=$W->{"args"}{$key};
445 (!defined $val ? () : ($key=>$val));
446 } keys(%{$W->{"args_persistent"}})),
447 %{$uri->query_form_hash()},
450 return $uri->abs(unparsed_uri()) if $W->{"args"}{"Wabs"} || $args{"abs"};
451 return $uri->rel(unparsed_uri());
454 sub path_abs_disk_register($)
456 my($path_abs_disk)=@_;
458 $W->{"path_abs_disk_register"}{$path_abs_disk}=1;
461 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
462 sub path_abs_disk($%)
466 cluck if !$args{"uri_as_in"} && ref $in;
467 my $uri=in_to_uri_abs($in);
468 cluck if !uri_is_local($uri);
469 my $path=$uri->path();
470 cluck "URI compatibility: ->path() not w/leading slash of URI \"$uri\"; path: $path" if $path!~m{^/};
471 my $r=dir_top_abs_disk().$path;
472 path_abs_disk_register $r if !defined $args{"register"} || $args{"register"};
482 while (my($name,$regex)=each(%tmpl)) {
483 my $name_html="Parameter <span class=\"quote\">".escapeHTML($name)."</span>";
484 $W->{"args"}{$name}="" if !defined $W->{"args"}{$name};
485 $W->{"args"}{$name}=[ $W->{"args"}{$name} ] if !ref $W->{"args"}{$name} && ref $regex;
486 fatal "$name_html passed as multivar although singlevar expected"
487 if ref $W->{"args"}{$name} && !ref $regex;
488 $regex=$regex->[0] if ref $regex;
489 for my $val (!ref $W->{"args"}{$name} ? $W->{"args"}{$name} : @{$W->{"args"}{$name}}) {
490 $val="" if !defined $val;
491 fatal "$name_html <span class=\"quote\">".escapeHTML($val)."</span>"
492 ." does not match the required regex <span class=\"quote\">".escapeHTML($regex)."</span> "
493 if $regex ne "" && $val!~/$regex/;
502 return '<p'.(!defined $height ? "" : ' style="height: '.$height.';"').'> </p>'."\n";
509 $msg="UNKNOWN" if !$msg;
512 # Do not send it unconditionally.
513 # The intial duplicated '<?xml...' crashes Gecko parser.
514 $W->{"heading_done"}=0 if $W->{"header_only"};
515 # Do not send it unconditionally.
516 # Prevents warn: Headers already sent
517 if (!$W->{"heading_done"}) {
518 $W->{"indexme"}=0; # For the case no heading was sent yet.
519 $W->{"header_only"}=0; # assurance for &heading
522 Wprint "\n".vskip("3ex")."<hr /><h1 class=\"error\">FATAL ERROR: $msg!</h1>\n"
523 ."<p>You can report this problem's details to"
524 ." ".a_href("mailto:".$W->{"admin_mail"},"admin of this website").".</p>\n";
528 sub footer_packages_used_comments()
530 my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
531 for my $package (@$packages_used) {
532 my $cvs_id=(eval('$'.$package."::CVS_ID")
533 # || $package # debug
535 Wprint '<!-- '.$package.' - $'.$cvs_id.'$ -->'."\n" if $cvs_id;
541 cluck 'Explicit &footer call is deprecated, !_exit_dne' if !$W->{"_exit_done"};
542 exit if $W->{"footer_done"}++; # deadlock prevention:
543 &{$_}() for reverse @{$W->{"footer_sub_push"}};
544 if ($W->{"header_only"}) {
545 $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
549 Wprint vskip if $W->{"footer_delimit"};
550 &{$_}() for reverse @{$W->{"footing_delimit_sub_push"}};
551 Wprint "<hr />\n" if $W->{"footer"};
553 # Never update the package list while we examine it!
554 $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
556 my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
557 if ($W->{"footer_ids"}) {
558 Wprint '<p class="cvs-id">';
559 Wprint join("<br />\n",map({ my $package=$_;
560 my $cvs_id=(eval('$'.$package."::CVS_ID")
561 # || $package # debug
567 $cvs_id='$'.$cvs_id.'$'; # Eaten by 'q' operator.
568 my @cvs_id_split=split / +/,$cvs_id;
569 if (@cvs_id_split==8) {
576 my $path_abs_disk=path_abs_disk("/$file$ext");
577 push @tried,$path_abs_disk;
578 last if -r $path_abs_disk;
579 cluck "Class file $file not found; tried: ".join(" ",@tried) if !$ext;
583 if ((my $file_cvs=$file)=~s{^My/}{}) {
584 $viewcvs=$W->{"viewcvs_My"}.$file_cvs;
587 $viewcvs=$W->{"viewcvs"}.$file;
590 .a_href((map({ my $s=$_; $s=~s#/viewcvs/#$&~checkout~/#; $s; } $viewcvs))[0]."?rev=".$cvs_id_split[2],
592 $cvs_id_split[1]=a_href($viewcvs,($package!~/^Apache2::/ ? $package : $cvs_id_split[1]));
593 $cvs_id_split[5]=&{$W->{"cvs_id_author_sub"}}($cvs_id_split[5]);
595 join " ",@cvs_id_split;
601 footer_packages_used_comments();
603 do { Wprint $_ if $_; } for $W->{"footing"};
605 Wprint "</body></html>\n";
609 # Existing entries are overwritten.
614 while (my($key,$val)=each(%pairs)) {
615 do { cluck "Headers already sent"; next; } if $W->{"heading_done"};
616 $W->{"r"}->headers_out()->set($key,$val);
626 elsif ($size<1024*1024)
627 { $size=sprintf "%.1fK",$size/1024; }
629 { $size=sprintf "%.1fM",$size/1024/1024; }
638 my $uri_rel=in_to_uri_abs($in)->rel(unparsed_uri());
639 # Do not: defined $uri_rel->("userinfo"|"host"|"port")();
640 # as they fail to be called for schemes not supporting them.
641 return 0 if $uri_rel->scheme();
642 return 0 if $uri_rel->authority();
646 # &path_web still may be required for &uri_escaped !
652 my $urient=escapeHTML($uri);
653 return $uri if $uri eq $urient;
655 return $urient if uri_is_local $uri;
656 return $uri if defined $W->{"have_ent"} && !$W->{"have_ent"}; # non-ent client
657 return $urient if $W->{"have_ent"}; # ent client
658 # Unknown client, &escapeHTML should not be needed here:
659 return escapeHTML(path_web('/My/Redirect.pm?location='.uri_escape($uri->abs(unparsed_uri()))));
662 our $a_href_inhibited;
665 my($in,$contents,%args)=@_;
668 do { $$_=1 if !defined $$_; } for (\$args{"size"});
669 if (!defined $contents) {
671 $contents=File::Basename::basename($contents) if $args{"basename"};
672 $contents=escapeHTML($contents);
674 $contents=~s#<a\b[^>]*>##gi;
675 $contents=~s#</a>##gi;
676 return $contents if $a_href_inhibited;
678 my $path_web=path_web $in,%args;
681 $r.=uri_escaped $path_web;
683 do { $r.=" $_" if $_; } for ($args{"attr"});
684 $r.='>'.$contents.'</a>';
685 if ($args{"size"} && uri_is_local($in) && ($args{"size"}>=2 || $in=~/[.](?:gz|Z|rpm|zip|deb|lha)/)) { # Downloadable?
686 my $path_abs_disk=path_abs_disk $in,%args;
687 cluck "File not readable: $path_abs_disk" if !-r $path_abs_disk;
688 $r.=' ('.size_display((stat($path_abs_disk))[7]).')';
693 sub a_href_inhibit($$;@)
695 my($self,$sub,@sub_args)=@_;
697 local $a_href_inhibited=1;
698 return &{$sub}(@sub_args);
701 sub input_hidden_persistents()
706 my $val=$W->{"args"}{$key};
707 (!defined $val ? () : '<input type="hidden"'
708 .' name="'.escapeHTML($key).'"'
709 .' value="'.escapeHTML($val).'"'
711 } (keys(%{$W->{"args_persistent"}}))));
716 my($self,$url,$status)=@_;
718 $url=path_web($url,"abs"=>1);
719 $status||=HTTP_MOVED_TEMPORARILY;
720 $W->{"r"}->status($status);
721 $W->{"r"}->headers_out()->{"Location"}=$url;
722 $W->{"header_only"}=1;
723 $W->{"content_type"}=0;
732 # Do not: PerlModule Apache2::ForwardedFor
733 # PerlPostReadRequestHandler Apache2::ForwardedFor
734 # As 'Apache2::ForwardedFor' takes the first of $ENV{"HTTP_X_FORWARDED_FOR"}
735 # while the contents is '127.0.0.1, 213.220.195.171' if client has its own proxy.
736 # We must take the last item ourselves.
737 # Be VERY sure you always retrieve all the headers unconditionally to hit: My::Hash::RecordKeys
738 my $x_forwarded_for=$W->{"headers_in"}{"X-Forwarded-For"};
739 $x_forwarded_for=~s/^.*,\s*// if $x_forwarded_for;
740 my $remote_ip=$W->{"headers_in"}{"_remote_ip"};
742 $r||=$x_forwarded_for;
747 # $url={"JP"=>"http://specific",...};
748 # $url={""=>"http://default",...};
751 my($url,$contents,%args)=@_;
753 # A bit ineffective but we must process all the possibilities to get stable 'headers_in' hits!
754 my %map=map(($_=>a_href($url->{$_},$contents,%args)),keys(%$url));
756 $cc||=Geo::IP->new()->country_code_by_addr(remote_ip()) if $have_Geo_IP;
767 # FIXME: &alarm, --timeout is now infinite.
768 # FIXME: Try to remove bash(1).
769 # FIXME: Use: @PATH_FLOCK@
770 my @argv=('flock',dir_top_abs_disk(),'bash','-c',$cmd.' >&2');
771 print STDERR join(" ","SPAWN:",@argv)."\n";
779 cluck "Pathname not absolute: $file" if $file!~m{^/};
781 # TODO: Somehow quickly check dependencies?
782 return make('make -s --no-print-directory'
783 .' -C '."'".File::Basename::dirname($file)."' '".File::Basename::basename($file)."'");
788 my($width,$height)=@_;
790 cluck if !defined $width || !defined $height;
791 return ($W->{"have_style"} ? "style=\"border:0;width:${width}px;height:${height}px\"" : "border=\"0\"")
792 ." width=\"$width\" height=\"$height\"";
795 sub negotiate_variant (%)
799 my @fields=("id","qs","content-type","encoding","charset","lang","size");
800 return [ map(($args{$_}),@fields) ];
803 # Input: $self is required!
804 # Input: Put the fallback variant as the first one.
805 # Returns: always only scalar!
806 sub Negotiate_choose($$)
808 my($self,$variants)=@_;
810 # Limit these entries to generate proper 'Vary' header.
811 my %hash=(map(($_=>$W->{"headers_in"}{$_}),qw(
817 my $best=HTTP::Negotiate::choose($variants,
819 # to prevent: Can't locate object method "scan" via package "Apache2::RequestRec" at HTTP/Negotiate.pm line 84.
820 # Do not: $W->{"r"}->headers_in()
821 # to prevent: Can't locate object method "scan" via package "APR::Table" at HTTP/Negotiate.pm line 84.
822 # Do not: HTTP::Headers->new($W->{"r"}->headers_in());
823 # to prevent empty result or even: Odd number of elements in anonymous hash
824 HTTP::Headers->new(%hash));
825 $best||=$variants->[0][0]; # $variants->[0]{"id"}; &HTTP::Negotiate::choose failed?
830 { "id"=>"png","qs"=>0.9,"content-type"=>"image/png" },
831 { "id"=>"gif","qs"=>0.7,"content-type"=>"image/gif" },
833 my $img_variants_re='[.](?:'.join('|',"jpeg",map(($_->{"id"}),@img_variants)).')$';
835 # Returns: ($path_web,$path_abs_disk)
836 # URI path segments support ignored here. Where it is used? (';' path segment options)
841 cluck if !uri_is_local $in;
842 my $uri=in_to_uri_abs $in;
843 my $path_abs_disk=path_abs_disk $uri,%args,"uri_as_in"=>1,"register"=>0;
845 # Known image extension?
846 return path_web($uri,%args,"uri_as_in"=>1),$path_abs_disk if $uri->path()=~m#$img_variants_re#o;
849 for my $var (@img_variants) {
850 my $path_abs_disk_variant=$path_abs_disk.".".$var->{"id"};
851 path_abs_disk_register($path_abs_disk_variant);
852 __PACKAGE__->make_file($path_abs_disk_variant);
853 push @nego_variants,negotiate_variant(
855 "size"=>(stat $path_abs_disk_variant)[7],
858 my $ext=__PACKAGE__->Negotiate_choose(\@nego_variants);
860 $uri->path($uri->path().".$ext");
861 return path_web($uri,%args,"uri_as_in"=>1),path_abs_disk($uri,%args,"uri_as_in"=>1);
867 my($in,$alt,%args)=@_;
870 my($path_web,$path_abs_disk)=_img_src($in,%args);
871 my($width,$height)=Image::Size::imgsize($path_abs_disk);
873 $alt=escapeHTML($alt);
874 my $content="<img src=\"".uri_escaped($path_web)."\" alt=\"$alt\" title=\"$alt\" ".img_size($width,$height)
875 .(!$args{"attr"} ? "" : " ".$args{"attr"})." />";
876 do { return a_href((_img_src($_))[0],$content,"uri_as_in"=>1) if $_; } for $args{"a_href_img"};
877 do { return a_href $_,$content if $_; } for $args{"a_href"};
884 $r.='<table border="0" width="100%"><tr>'."\n";
885 @_=( [@_] ) if !ref $_[0];
887 $r.="\t".'<td align="center">'.&{\&img}(@$_).'</td>'."\n";
889 $r.='</tr></table>'."\n";
895 my($text,@args_img)=@_;
897 # FIXME: Workaround bug of 'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)':
898 # <col width="@{[ (!$W->{"browser"}->ie() ? "1*" : "90%" ) ]}" />
899 # <col width="@{[ (!$W->{"browser"}->ie() ? "0*" : "10%" ) ]}" />
900 # causes whole invisible projects in: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.10) Gecko/20050719 Galeon/1.3.21
902 <table border="0" width="100%">
908 @{[ &{\&img}(@args_img) ]}
917 my($class,$filename)=@_;
920 open F,$filename or cluck "Cannot open \"$filename\": $!";
921 my $F=do { local $/=undef(); <F>; };
922 close F or cluck "Cannot close \"$filename\": $!";
930 header("Expires"=>HTTP::Date::time2str(1000000000)); # date in the past
931 header("Last-Modified"=>HTTP::Date::time2str()); # always modified
932 header("Cache-Control"=>join(", ",
937 "pre-check=0", # MSIE
938 "post-check=0", # MSIE
940 header("Pragma"=>"no-cache"); # HTTP/1.0
941 header("Vary"=>"*"); # content may ba based on unpredictable sources
944 sub headers_in_filtered(@)
948 return map(($_=>$W->{"headers_in"}{$_}),@keys);
951 our %uri_args_frozen_to_headers_in_keys;
952 our %uri_args_headers_in_frozen_to_headers_out;
954 sub uri_args_headers_in_frozen_get($)
956 my($headers_in_keys_arrayref)=@_;
958 my %uri_args_headers_in_hash=(
959 "uri_args_frozen"=>$W->{"uri_args_frozen"},
960 "headers_in"=>{ headers_in_filtered(@$headers_in_keys_arrayref) },
962 return do { local $Storable::canonical=1; Storable::freeze(\%uri_args_headers_in_hash); };
965 sub cache_output_filter($)
969 while ($f->read(my $text,0x400)) {
970 cluck "utf-8 untested" if Encode::is_utf8($text); # Possible here at all?
972 $W->{"digest-md5"}->add($text);
979 if (!$W->{"http_safe"}) {
980 __PACKAGE__->_no_cache();
985 # &Wrequire it here even if it will not be later used; to be stable!
986 Wrequire 'My::Hash::RestrictTo';
988 "uri"=>"http://".$W->{"web_hostname"}."/".$W->{"r"}->uri(),
989 "args"=>$W->{"args_orig_array"},
991 $W->{"uri_args_frozen"}=do { local $Storable::canonical=1; Storable::freeze(\%uri_args_hash); };
992 last if !(my $headers_in_keys_arrayref=$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}});
994 # Protection to be sure we are stable:
995 $W->{"headers_in"}=My::Hash::RestrictTo->new($W->{"headers_in"},@$headers_in_keys_arrayref);
997 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get($headers_in_keys_arrayref);
998 last if !(my $headers_out_hashref=$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}});
999 header(%$headers_out_hashref);
1002 # &meets_conditions will always deny the attempt if !2xx status().
1003 # At least ap_read_request() sets: r->status=HTTP_REQUEST_TIME_OUT; /* Until we get a request */
1004 my $status_old=$W->{"r"}->status();
1005 $W->{"r"}->status(HTTP_OK);
1006 # Update httpd's 'r->mtime' as the header "Last-Modified" is just not enough for ap_meets_conditions():
1007 # &update_mtime() argument is really in _secs_, not in _msecs_ as the docs claim.
1008 # Be aware '*1000000' would overflow Perl integer anyway.
1009 # &set_last_modified would also override the "Last-Modified" headers_out!
1010 # &mtime may exist but somehow does not work.
1011 $W->{"r"}->update_mtime(HTTP::Date::str2time($headers_out_hashref->{"Last-Modified"}));
1012 $status=$W->{"r"}->meets_conditions();
1013 $W->{"r"}->status($status_old);
1015 last if OK==$status;
1016 $W->{"r"}->status($status);
1017 $W->{"header_only"}=1; # Inhibit &footer output.
1022 $W->{"digest-md5"}=Digest::MD5->new();
1023 $W->{"cache_active"}=1;
1024 $W->{"r"}->add_output_filter(\&cache_output_filter);
1027 sub cache_finish_last_modified()
1029 cluck "Not yet done now? W __PACKAGE__: ".$W->{"__PACKAGE__"}
1030 if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
1031 for my $package_orig (@{$packages_used_array{$W->{"__PACKAGE__"}}}) {
1032 local $_=$package_orig.".pm";
1034 path_abs_disk "/$_","register"=>1;
1037 for my $path_abs_disk (keys(%{$W->{"path_abs_disk_register"}})) {
1038 my $mtime=(stat $path_abs_disk)[9];
1039 do { cluck "No mtime for: $path_abs_disk"; next; } if !$mtime;
1040 $mtime_newest=$mtime if !$mtime_newest || $mtime_newest<$mtime;
1042 cluck "No mtime_newest found for the current W __PACKAGE__: ".$W->{"__PACKAGE__"}
1044 return HTTP::Date::time2str($mtime_newest);
1050 # Do not: return if !$W->{"uri_args_frozen"};
1051 # as we may have just gave 304 and 'exit;' without starting the caching.
1052 return if !$W->{"cache_active"};
1054 # Headers may not be complete in this case; not sure, just trying.
1055 return if $W->{"r"}->connection()->aborted();
1057 # Fill-in/check: %uri_args_frozen_to_headers_in_keys
1058 my $headers_in_keys_stored_arrayref_ref=\$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}};
1059 my @headers_in_keys=tied(%{$W->{"headers_in_RecordKeys"}})->accessed();
1060 if (!$$headers_in_keys_stored_arrayref_ref
1061 || !Data::Compare::Compare(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)) {
1062 cluck "Non-matching generated 'headers_in_keys' per 'uri_args_frozen' key:\n"
1063 .Dumper(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)
1064 if $$headers_in_keys_stored_arrayref_ref;
1065 # Build or possibly prevent such further warn dupes:
1066 $$headers_in_keys_stored_arrayref_ref=\@headers_in_keys;
1067 # Build or regenerate as obsoleted now:
1068 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get(\@headers_in_keys);
1071 # Prepare 'headers_out' for the future reusal:
1073 # Do not: $W->{"digest-md5"}->b64digest();
1074 # as it will not provide the trailing filling '='s.
1075 # RFC 1864 is not clear if they should be there but its sample provides them.
1076 # Do not try to provide canonical "\r\n" form of newlines as is said by RFC 1864.
1077 # RFC 2068 (HTTP/1.1) section 14.16 says the newlines should NOT be converted for HTTP.
1078 # ',""' to avoid breaking the headers by its default "\n".
1079 $headers_out{"Content-MD5"}=MIME::Base64::encode_base64($W->{"digest-md5"}->digest(),"");
1080 # In fact we could also use MD5 for ETag as if we know ETag we also know MD5.
1081 # But this way we do not need to calculate MD5 and we still can provide such ETag. So.
1082 # $W->{"r"}->set_etag() ?
1083 $headers_out{"ETag"}='"'.Digest::MD5::md5_base64($W->{"uri_args_headers_in_frozen"}).'"';
1084 # $W->{"r"}->set_content_length() ?
1085 $headers_out{"Content-Length"}=$W->{"r"}->bytes_sent();
1086 my %Vary=map(($_=>1),(@headers_in_keys));
1092 %Vary=("*"=>1) if $Vary{"*"};
1093 $headers_out{"Vary"}=join(", ",sort keys(%Vary)) if keys(%Vary);
1094 # $W->{"r"}->set_last_modified() ?
1095 $headers_out{"Last-Modified"}=cache_finish_last_modified();
1097 # Fill-in/check: %uri_args_headers_in_frozen_to_headers_out
1098 my $headers_out_stored_hashref_ref=\$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}};
1099 if (!$$headers_out_stored_hashref_ref
1100 || !Data::Compare::Compare(\%headers_out,$$headers_out_stored_hashref_ref)) {
1101 cluck "Non-matching generated 'headers_out' per 'uri_args_headers_in_frozen' key:\n"
1102 .Dumper(\%headers_out,$$headers_out_stored_hashref_ref)
1103 if $$headers_out_stored_hashref_ref;
1104 # Build or possibly prevent such further warn dupes:
1105 $$headers_out_stored_hashref_ref=\%headers_out;
1108 ###print STDERR Dumper(\%uri_args_frozen_to_headers_in_keys,\%uri_args_headers_in_frozen_to_headers_out);
1115 if (!$W->{"header_only"}) {
1116 header("Content-Style-Type"=>"text/css");
1117 # Do not: text/javascript
1118 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
1119 # "application/javascript" so far standardized till 2005-12-08 by:
1120 # http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
1121 header("Content-Script-Type"=>"application/javascript");
1122 # $W->{"r"}->content_languages() ?
1123 do { header("Content-Language"=>$_) if $_; } for $W->{"language"};
1125 # TODO: Support also: private
1126 header("Cache-Control"=>"public"); # HTTP/1.1
1128 # Use $W->{"charset"}=0 to disable charset.
1129 $W->{"charset"}="us-ascii"
1130 if !defined $W->{"charset"} && (!defined($W->{"content_type"}) || $W->{"content_type"});
1133 # https://bugzilla.mozilla.org/show_bug.cgi?id=120556
1135 # Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1136 # http://validator.w3.org/ does not send ANY "Accept" headers!
1137 if (!defined $W->{"content_type"}) {
1138 # Be _stable_ for "headers_in".
1139 my $accept=$W->{"headers_in"}{"Accept"};
1140 my $user_agent=$W->{"headers_in"}{"User-Agent"}||"";
1141 $W->{"content_type"}="application/xhtml+xml"
1142 if !$accept && $user_agent=~m{^W3C_Validator/}i;
1144 my $negotiated=$class->Negotiate_choose([
1145 # Put the fallback variant as the first one.
1146 # Rate both variants the same to prefer "text/html" for undecided clients.
1148 # Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1149 # prefers "application/xhtml+xml" over "text/html" itself:
1150 # text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5
1153 "content-type"=>"text/html",
1155 (!$W->{"charset"} ? () : "charset"=>$W->{"charset"}),
1156 "lang"=>$W->{"language"},
1159 "id"=>"application/xhtml+xml",
1160 "content-type"=>"application/xhtml+xml",
1162 (!$W->{"charset"} ? () : "charset"=>$W->{"charset"}),
1163 "lang"=>$W->{"language"},
1168 $W->{"content_type"}=$negotiated if !defined $W->{"content_type"};
1170 # mod_perl doc: If you set this header via the headers_out table directly, it
1171 # will be ignored by Apache. So do not do that.
1173 if ($W->{"content_type"}) {
1174 $type=MIME::Types->new()->type($W->{"content_type"});
1175 cluck "MIME::Types type '".$W->{"content_type"}."' not known" if !$type;
1177 cluck "charset='".$W->{"charset"}."' does not match content-type='".$W->{"content_type"}."'"
1178 if ($W->{"charset"} ? 1 : 0) != (!$type ? 0 : $type->isAscii());
1179 $W->{"r"}->content_type($W->{"content_type"}.(!$W->{"charset"} ? "" : "; charset=".$W->{"charset"}))
1180 if $W->{"content_type"};
1183 # We still can append headers before we put out some text.
1184 # FIXME: It is not clean to still append them without overwriting.
1185 return if $W->{"heading_done"};
1186 Wprint '<?xml version="1.0" encoding="'.$W->{"charset"}.'"?>'."\n"
1187 if (!$W->{"header_only"} || $W->{"header_only"} eq "xml") && (0
1188 || $W->{"content_type"}=~m{^application/\w+[+]xml$}
1189 || $W->{"content_type"} eq "text/vnd.wap.wml");
1190 return if $W->{"header_only"};
1191 # Split 'heading_done' for the proper handling of: /project/Rel.pm
1192 $W->{"heading_done"}++;
1194 Wprint '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">'."\n";
1195 Wprint '<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="'.$W->{"language"}.'">'."\n";
1196 my $title=$W->{"title_prefix"}.join("",map({ ': '.$_; } ($W->{"title"} || ())));
1197 # Do not: cluck if $title=~/[<>]/;
1198 # as it is not solved just by: &a_href_inhibit
1199 # as sometimes titles use also: <i>...</i>
1200 $title=~s#<[^>]*>##g;
1202 Wprint "<title>$title</title>\n";
1203 if ($W->{"have_css"}) {
1204 # Everything can get overriden later.
1205 for my $css ("/My/Web.css",@{$W->{"css_push"}}) {
1207 <link rel="stylesheet" type="text/css" href="@{[ uri_escaped(path_web $css) ]}" />
1210 if ($W->{"css_inherit"}) {
1211 # Do not: <script />
1212 # as at least Lynx inhibits any further HTML output.
1213 # Do not: text/javascript
1214 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
1215 # "application/javascript" so far standardized till 2005-12-08 by:
1216 # http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
1218 <script type="application/javascript" src="@{[ uri_escaped(path_web('/My/css_inherit.js')) ]}"></script>
1222 Wprint '<meta name="robots" content="'.($W->{"indexme"} ? "" : "no" ).'index,follow" />'."\n";
1223 Wprint $W->{"head"};
1224 for my $type (qw(prev next index contents start up)) {
1225 do { Wprint '<link rel="'.$type.'" href="'.uri_escaped(path_web $_).'" />'."\n" if $_; }
1226 for ($W->{"rel_$type"});
1228 Wprint "</head><body";
1229 # Wprint ' bgcolor="black" text="white" link="aqua" vlink="teal"'
1230 # if $W->{"browser"}->netscape() && (!$W->{"browser"}->major() || $W->{"browser"}->major()<=4);
1231 Wprint $W->{"body_attr"};
1234 do { Wprint $_ if $_; } for $W->{"heading"};
1238 delete $W->{"__My::Web_init"};