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
42 our @ISA=qw(Tie::Handle Exporter);
44 my %packages_used_hash; # $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
45 my %packages_used_array;
49 use Carp qw(cluck confess);
50 $W->{"__My::Web_init"}=1;
56 # print STDERR "Wrequire $file\n";
63 for (my $depth=0;defined caller($depth);$depth++) {
64 $callers{caller($depth)}=1;
66 my $selfpkg=__PACKAGE__;
68 for my $target ($class,__PACKAGE__) {
69 for my $caller (keys(%callers)) {
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';
102 require Image::Size; # for &imgsize
103 use File::Basename; # &basename
104 use Carp qw(cluck confess);
106 require HTTP::BrowserDetect;
107 require HTTP::Negotiate;
108 our $have_Geo_IP; BEGIN { $have_Geo_IP=eval { require Geo::IP; 1; }; }
109 # Do not: use ModPerl::Util qw(exit);
110 # to prevent in mod_perl2: "exit" is not exported by the ModPerl::Util module
112 use POSIX qw(strftime);
114 use Apache2::Const qw(HTTP_MOVED_TEMPORARILY OK HTTP_OK);
121 require Data::Compare;
124 use Apache2::RequestUtil;
126 use Apache2::Connection;
127 require MIME::Base64;
128 use Apache2::ServerUtil;
130 require MIME::Parser;
131 use Apache2::RequestRec;
132 use Apache2::RequestIO;
133 use Apache2::Response;
142 print STDERR "$class->init ".Apache2::RequestUtil->request()->unparsed_uri()."\n";
144 # We need to track package dependencies, so we need to call it from &init.
145 # We cannot do it in BEGIN { } block
146 # as it would not be tracked for each of the toplevel users later.
150 # $W={} can get somehow created very easily.
151 # Do not: cluck "W not empty:\n".Dumper($W) if keys(%$W);
152 # to prevent (of $W->{"headers_in"}): TODO: Enumeration may not be expected.
153 cluck "W not empty; __PACKAGE__ was: ".$W->{"__PACKAGE__"} if keys(%$W);
154 $W=My::Hash->new({},"My::Hash::Sub","My::Hash::Push");
157 "__PACKAGE__"=>scalar(caller()),
159 %args, # override %WebConfig settings
162 # {"__PACKAGE__"} is mandatory for mod_perl-2.0;
163 # $Apache2::Registry::curstash is no longer supported.
164 do { cluck "No $_" if !$W->{$_}; } for "__PACKAGE__";
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 $W->{"headers_in"}=$W->{"r"}->headers_in();
198 Wrequire 'My::Hash::Merge';
199 $W->{"headers_in"}=My::Hash::Merge->new(
202 "_remote_ip"=>sub { return $W->{"r"}->connection()->remote_ip(); },
205 Wrequire 'My::Hash::Readonly';
206 $W->{"headers_in"}=My::Hash::Readonly->new($W->{"headers_in"});
208 if ($W->{"r"}->method() eq "GET" || $W->{"r"}->method() eq "HEAD") {
209 for (\$W->{"http_safe"}) {
210 # Do not: # Extend the current ETag system instead if you would need it:
211 # cluck "Explicitely NOT HTTP-Safe for method \"".$W->{"r"}->method()."\"?!?"
212 # if defined($$_) && !$$_;
213 # as sometimes it just does not make sense to cache it.
214 $$_=1 if !defined $$_;
218 for (\$W->{"http_safe"}) {
219 cluck "Undefined HTTP-Safe-ty for method \"".$W->{"r"}->method()."\"!"
221 $$_=0 if !defined $$_;
224 # Used only if: $W->{"http_safe"}
225 # but we would cause on different method(): Appending to the '_done' package list
226 Wrequire 'My::Hash::RecordKeys';
227 if ($W->{"http_safe"}) {
228 $W->{"headers_in_RecordKeys"}=My::Hash::RecordKeys->new($W->{"headers_in"});
229 $W->{"headers_in"}=$W->{"headers_in_RecordKeys"};
233 local $_=$W->{"r"}->args() || "";
234 if ($W->{"detect_ent"}) {
235 if (/[&]amp;have_ent/)
236 { $W->{"have_ent"}=0; }
237 elsif ( /[&]have_ent/)
238 { $W->{"have_ent"}=1; }
240 { delete $W->{"have_ent"}; }
241 if (!defined $W->{"have_ent"} && $W->{"r"}->method() eq "GET") {
242 $W->{"head"}.='<meta http-equiv="Refresh" content="0; URL='
243 .escapeHTML("http://".$W->{"web_hostname"}."/".($W->{"r"}->uri()=~m#^/*(.*)$#)[0]
244 ."?".($_ || "detect_ent_glue=1").'&have_ent=detect')
252 $W->{"args"}=URI->new("?".$W->{"r"}->args())->query_form_hash();
253 $W->merge_post_args() if $W->{"r"}->method() eq "POST";
254 # Prepare '$args' first to (FIXME: Why?) prevent: Not a reference
255 my $args=$W->{"args"};
256 $W->{"args_orig"}=Storable::dclone($args);
258 $W->{"browser"}=sub {
259 # Lazy-evaluation, we may not need the "User-Agent" header at all.
260 return our $r||=HTTP::BrowserDetect->new($W->{"headers_in"}{"User-Agent"});
263 if (!defined $W->{"have_style"}) {
264 $W->{"have_style"}=sub {
265 # Lazy-evaluation, we may not need the "User-Agent" header at all.
266 return our $r||=(!$W->{"browser"}->netscape() || ($W->{"browser"}->major() && $W->{"browser"}->major()>4) ? 1 : 0);
270 $W->{"have_js"}=($W->{"args"}{"have_js"} ? 1 : 0);
271 if ($W->{"detect_js"} && !$W->{"have_js"}) {
273 # as at least Lynx inhibits any further HTML output.
274 # Do not: text/javascript
275 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
276 # "application/javascript" so far standardized till 2005-12-08 by:
277 # http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
278 $W->{"head"}.='<script type="application/javascript" src="'.uri_escaped(path_web('/My/HaveJS.pm')).'"></script>'."\n";
281 # Required by &_args_check below.
282 $W->{"_init_done"}=1;
284 do { _args_check(%$_) if $_; } for $W->{"args_check"};
293 return q{enctype="application/x-www-form-urlencoded" accept-charset="us-ascii utf-8"} if $method eq "post";
294 return q{accept-charset="us-ascii utf-8"} if $method eq "get";
295 cluck "Undefined method: $method";
299 sub merge_post_args($)
303 my @post_args=$class->read_post_args();
305 my $name=shift @post_args;
306 my $data=shift @post_args;
307 my $ref=\$W->{"args"}{$name};
308 if (!defined $$ref) { $$ref=$data; }
309 elsif (!ref $$ref) { $$ref=[$$ref,$data]; }
310 elsif ("ARRAY" eq ref $$ref) { push @$$ref,$data; }
312 cluck "Ignoring POST argument \"$name\", orig is weird:\n",Dumper($$ref);
319 # as CGI parsing of POST vs. QUERY_STRING data, multiple-valued keys etc.
320 # is too dense and causes weird problems, together with mod_perl etc.
321 sub read_post_args($)
325 local $_=$class->http_headers_in_for("Content-type")->content_type();
326 return $class->read_multipart_form_data() if $_ eq "multipart/form-data";
327 return $class->read_application_x_www_form_urlencoded() if $_ eq "application/x-www-form-urlencoded";
328 cluck "Unknown POST data body, ignored: $_";
332 sub read_application_x_www_form_urlencoded($)
338 my $got=$W->{"r"}->read(my($buf),0x1000);
339 # Do not: cluck "Error reading POST data: $!" if !defined $got;
340 # as it should be done using: APR::Error exceptions
344 return URI->new("?".$body)->query_form();
347 sub read_multipart_form_data($)
351 my $parser=MIME::Parser->new();
352 # FIXME: No unlink()s done!
353 $parser->output_under("/tmp");
358 tie *FH,"My::Web::ReadMerged",
359 join("",map(($_.": ".$W->{"headers_in"}{$_}."\n"),qw(
363 my $body=$parser->parse(\*FH);
364 cluck "No multipart POST request body?" if !$body->is_multipart();
367 $_->head()->mime_attr("content-disposition.name")
369 join("",@{$_->body()})
372 # TODO: Globalize, make it IO::* compatible, split to the merging part + IO::Scalar.
373 package My::Web::ReadMerged;
377 our @ISA=qw(Tie::Handle Exporter);
378 use Carp qw(cluck confess);
384 confess "Slurp not yet implemented" if !defined $/;
385 # Apache2::RequestIO does not support 'READLINE'!
387 if (defined $self->{"data"} && $self->{"data"}=~s{^.*\Q$/\E}{}) {
388 $self->{"offset"}+=length $&;
391 my $fh_orig=$self->{"fh_orig"};
393 my $r=$self->{"data"};
394 delete $self->{"data"};
395 $self->{"offset"}+=length $r if defined $r;
398 my $got=read $fh_orig,my($buf),0x1000;
399 cluck "Error reading POST data: $!" if !defined $got;
400 delete $self->{"fh_orig"} if !$got;
401 cluck "INTERNAL: fh_orig should not exist here" if !defined $self->{"data"};
402 $self->{"data"}.=$buf;
410 return $self->{"offset"};
415 my($class,$data,$fh_orig)=@_;
417 my $self=bless {},$class;
418 $self->{"data"}=$data;
420 $self->{"fh_orig"}=$fh_orig;
427 my($apache_request)=@_;
429 cluck "CORE::GLOBAL::exit hook not ran" if !$W->{"_exit_done"};
430 cluck "packages not finalized" if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
438 # PerlResponseHandler is RUN_FIRST and &ModPerl::Util::exit returns OK, so no (sane) go.
439 # PerlLogHandler is already too late to be able to produce any output.
443 cluck "Missing ->init while in exit_hook()" if !$W->{"_init_done"};
444 # &footer will call us recursively!
445 footer() if !$W->{"_exit_done"}++;
446 return &{$exit_orig}(@_);
450 do { cluck "exit_hook_start() twice?"; return; } if defined $exit_orig;
451 $exit_orig=\&CORE::GLOBAL::exit;
452 # Prevent: Subroutine CORE::GLOBAL::exit redefined
453 no warnings 'redefine';
454 *CORE::GLOBAL::exit=\&exit_hook;
458 do { cluck "exit_hook_stop() without exit_hook_start()?"; return; }
459 if \&exit_hook ne \&CORE::GLOBAL::exit;
460 do { cluck "INTERNAL: exit_orig uninitialized"; return; }
462 # Prevent: Subroutine CORE::GLOBAL::exit redefined
463 no warnings 'redefine';
464 *CORE::GLOBAL::exit=$exit_orig;
468 # Be aware other parts of code (non-My::Web) will NOT use this function!
469 # Do not: Wprint $W->{"heading"},"undef"=>1;
470 # as we would need to undef() it to turn it off and it would get defaulted in such case.
471 # Do not: exists $W->{"heading"}
472 # as we use a lot of 'for $W->{"heading"}' which instantiates it with the value: undef()
477 cluck "undef Wprint" if !defined $text && !$args{"undef"};
478 delete $args{"undef"};
479 cluck join(" ","Invalid arguments:",keys(%args)) if keys(%args);
480 return if !defined $text;
481 # Do not: cluck "utf-8 untested" if Encode::is_utf8($text);
482 # as it is valid here.
483 $W->{"r"}->puts($text);
486 sub request_check(;$)
490 # Use &eval to prevent: Global $r object is not available. Set:\n\tPerlOptions +GlobalRequest\nin ...
491 confess "Calling sensitive dynamic code from a static code" if !eval { Apache2::RequestUtil->request(); };
492 # Do not: confess "Calling sensitive dynamic code without My::Web::init" if !$W->{"__PACKAGE__"};
493 # as it is valid at least while preparing arguments to call: &project::Lib::init
497 # as it is too much backward compatible regarding the charset encodings etc.
498 # and the resulting code is too dense with no additional functionality for the recent content.
511 # /home/user/www/webdir
512 sub dir_top_abs_disk()
514 our $dir_top_abs_disk;
515 if (!$dir_top_abs_disk) {
516 my $selfpkg_relpath=__PACKAGE__;
517 $selfpkg_relpath=~s{::}{/}g;
518 $selfpkg_relpath.=".pm";
519 my $selfpkg_abspath=$INC{$selfpkg_relpath} or do {
520 cluck "Unable to find self package $selfpkg_relpath";
523 $selfpkg_abspath=~s{/*\Q$selfpkg_relpath\E$}{} or do {
524 cluck "Unable to strip myself \"$selfpkg_relpath\" from the abspath: $selfpkg_abspath";
527 cluck "INC{myself} is relative?: $selfpkg_abspath" if $selfpkg_abspath!~m{^/};
528 $dir_top_abs_disk=$selfpkg_abspath;
530 return $dir_top_abs_disk;
536 if (!$W->{"unparsed_uri"}) {
538 # as we may be called before &init from: &My::Project::init
539 my $r=Apache2::RequestUtil->request();
540 cluck "Calling ".'&unparsed_uri'." from a static code, going to fail" if !$r;
541 my $uri_string=$r->unparsed_uri() or cluck "Valid 'r' missing unparsed_uri()?";
542 my $uri=URI->new_abs($uri_string,"http://".$W->{"web_hostname"}."/");
543 $W->{"unparsed_uri"}=$uri;
545 return $W->{"unparsed_uri"};
552 # Otherwise we may have been already processed and thus legally relativized.
553 # FIXME data: Currently disabled, all the data are too violating such rule.
555 my $uri_check=URI->new($in);
556 $uri_check->scheme() || $in=~m{^\Q./\E} || $in=~m{^/}
557 or cluck "Use './' or '/' prefix for all the local references: $in";
559 my $uri=URI->new_abs($in,unparsed_uri());
560 $uri=$uri->canonical();
564 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
570 cluck if !$args{"uri_as_in"} && ref $in;
571 my $uri=in_to_uri_abs($in);
572 if (uri_is_local($uri)) {
573 # Prefer the $uri values over "args_persistent" values.
574 # &query_form_hash comes from: URI::QueryParam
575 $uri->query_form_hash({
578 my $val=$W->{"args"}{$key};
579 (!defined $val ? () : ($key=>$val));
580 } keys(%{$W->{"args_persistent"}})),
581 %{$uri->query_form_hash()},
584 return $uri->abs(unparsed_uri()) if $W->{"args"}{"Wabs"} || $args{"abs"};
585 return $uri->rel(unparsed_uri());
588 sub path_abs_disk_register($)
590 my($path_abs_disk)=@_;
592 $W->{"path_abs_disk_register"}{$path_abs_disk}=1;
595 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
596 sub path_abs_disk($%)
600 cluck if !$args{"uri_as_in"} && ref $in;
601 my $uri=in_to_uri_abs($in);
602 cluck if !uri_is_local($uri);
603 my $path=$uri->path();
604 cluck "URI compatibility: ->path() not w/leading slash of URI \"$uri\"; path: $path" if $path!~m{^/};
605 my $r=dir_top_abs_disk().$path;
606 path_abs_disk_register $r if !defined $args{"register"} || $args{"register"};
616 while (my($name,$regex)=each(%tmpl)) {
617 my $name_html="Parameter <span class=\"quote\">".escapeHTML($name)."</span>";
618 $W->{"args"}{$name}="" if !defined $W->{"args"}{$name};
619 $W->{"args"}{$name}=[ $W->{"args"}{$name} ] if !ref $W->{"args"}{$name} && ref $regex;
620 fatal "$name_html passed as multivar although singlevar expected"
621 if ref $W->{"args"}{$name} && !ref $regex;
622 $regex=$regex->[0] if ref $regex;
623 for my $val (!ref $W->{"args"}{$name} ? $W->{"args"}{$name} : @{$W->{"args"}{$name}}) {
624 $val="" if !defined $val;
625 fatal "$name_html <span class=\"quote\">".escapeHTML($val)."</span>"
626 ." does not match the required regex <span class=\"quote\">".escapeHTML($regex)."</span> "
627 if $regex ne "" && $val!~/$regex/;
636 return '<p'.(!defined $height ? "" : ' style="height: '.$height.';"').'> </p>'."\n";
643 $msg="UNKNOWN" if !$msg;
646 # Do not send it unconditionally.
647 # The intial duplicated '<?xml...' crashes Gecko parser.
648 $W->{"heading_done"}=0 if $W->{"header_only"};
649 # Do not send it unconditionally.
650 # Prevents warn: Headers already sent
651 if (!$W->{"heading_done"}) {
652 $W->{"indexme"}=0; # For the case no heading was sent yet.
653 $W->{"header_only"}=0; # assurance for &heading
654 $W->{"content_type"}="text/html"; # Force HTML and avoid strictly checked XHTML.
657 Wprint "\n".vskip("3ex")."<hr /><h1 class=\"error\">FATAL ERROR: $msg!</h1>\n"
658 ."<p>You can report this problem's details to"
659 ." ".a_href("mailto:".$W->{"admin_mail"},"admin of this website").".</p>\n";
663 sub footer_packages_used_comments()
665 my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
666 for my $package (@$packages_used) {
667 my $cvs_id=(eval('$'.$package."::CVS_ID")
668 # || $package # debug
670 Wprint '<!-- '.$package.' - $'.$cvs_id.'$ -->'."\n" if $cvs_id;
676 cluck 'Explicit &footer call is deprecated, !_exit_dne' if !$W->{"_exit_done"};
677 exit if $W->{"footer_done"}++; # deadlock prevention:
678 &{$_}() for reverse @{$W->{"footer_sub_push"}};
679 if ($W->{"header_only"}) {
680 $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
684 Wprint vskip if $W->{"footer_delimit"};
685 &{$_}() for reverse @{$W->{"footing_delimit_sub_push"}};
686 Wprint "<hr />\n" if $W->{"footer"};
688 # Never update the package list while we examine it!
689 $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
691 my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
692 if ($W->{"footer_ids"}) {
693 Wprint '<p class="cvs-id">';
694 Wprint join("<br />\n",map({ my $package=$_;
695 my $cvs_id=(eval('$'.$package."::CVS_ID")
696 # || $package # debug
702 $cvs_id='$'.$cvs_id.'$'; # Eaten by 'q' operator.
703 my @cvs_id_split=split / +/,$cvs_id;
704 if (@cvs_id_split==8) {
711 my $path_abs_disk=path_abs_disk("/$file$ext");
712 push @tried,$path_abs_disk;
713 last if -r $path_abs_disk;
714 cluck "Class file $file not found; tried: ".join(" ",@tried) if !$ext;
718 if ((my $file_cvs=$file)=~s{^My/}{}) {
719 $viewcvs=$W->{"viewcvs_My"}.$file_cvs;
722 $viewcvs=$W->{"viewcvs"}.$file;
725 .a_href((map({ my $s=$_; $s=~s#/viewcvs/#$&~checkout~/#; $s; } $viewcvs))[0]."?rev=".$cvs_id_split[2],
727 $cvs_id_split[1]=a_href($viewcvs,($package!~/^Apache2::/ ? $package : $cvs_id_split[1]));
728 $cvs_id_split[5]=&{$W->{"cvs_id_author_sub"}}($cvs_id_split[5]);
730 join " ",@cvs_id_split;
736 footer_packages_used_comments();
738 do { Wprint $_ if $_; } for $W->{"footing"};
740 Wprint "</body></html>\n";
744 # Existing entries are overwritten.
749 while (my($key,$val)=each(%pairs)) {
750 do { cluck "Headers already sent"; next; } if $W->{"heading_done"};
751 $W->{"r"}->headers_out()->set($key,$val);
761 elsif ($size<1024*1024)
762 { $size=sprintf "%.1fK",$size/1024; }
764 { $size=sprintf "%.1fM",$size/1024/1024; }
773 my $uri_rel=in_to_uri_abs($in)->rel(unparsed_uri());
774 # Do not: defined $uri_rel->("userinfo"|"host"|"port")();
775 # as they fail to be called for schemes not supporting them.
776 return 0 if $uri_rel->scheme();
777 return 0 if $uri_rel->authority();
781 # &path_web still may be required for &uri_escaped !
787 my $urient=escapeHTML($uri);
788 return $uri if $uri eq $urient;
790 return $urient if uri_is_local $uri;
791 return $uri if defined $W->{"have_ent"} && !$W->{"have_ent"}; # non-ent client
792 return $urient if $W->{"have_ent"}; # ent client
793 # Unknown client, &escapeHTML should not be needed here:
794 return escapeHTML(path_web('/My/Redirect.pm?location='.uri_escape($uri->abs(unparsed_uri()))));
797 our $a_href_inhibited;
800 my($in,$contents,%args)=@_;
803 do { $$_=1 if !defined $$_; } for (\$args{"size"});
804 if (!defined $contents) {
806 $contents=File::Basename::basename($contents) if $args{"basename"};
807 $contents=escapeHTML($contents);
809 $contents=~s#<a\b[^>]*>##gi;
810 $contents=~s#</a>##gi;
811 return $contents if $a_href_inhibited;
813 my $path_web=path_web $in,%args;
816 $r.=uri_escaped $path_web;
818 do { $r.=" $_" if $_; } for ($args{"attr"});
819 $r.='>'.$contents.'</a>';
820 if ($args{"size"} && uri_is_local($in) && ($args{"size"}>=2 || $in=~/[.](?:gz|Z|rpm|zip|deb|lha)/)) { # Downloadable?
821 my $path_abs_disk=path_abs_disk $in,%args;
822 cluck "File not readable: $path_abs_disk" if !-r $path_abs_disk;
823 $r.=' ('.size_display((stat($path_abs_disk))[7]).')';
828 sub a_href_inhibit($$;@)
830 my($self,$sub,@sub_args)=@_;
832 local $a_href_inhibited=1;
833 return &{$sub}(@sub_args);
836 sub input_hidden_persistents()
841 my $val=$W->{"args"}{$key};
842 (!defined $val ? () : '<input type="hidden"'
843 .' name="'.escapeHTML($key).'"'
844 .' value="'.escapeHTML($val).'"'
846 } (keys(%{$W->{"args_persistent"}}))));
851 my($self,$url,$status)=@_;
853 $url=path_web($url,"abs"=>1);
854 $status||=HTTP_MOVED_TEMPORARILY;
855 $W->{"r"}->status($status);
856 $W->{"r"}->headers_out()->{"Location"}=$url;
857 $W->{"header_only"}=1;
858 $W->{"content_type"}=0;
867 # Do not: PerlModule Apache2::ForwardedFor
868 # PerlPostReadRequestHandler Apache2::ForwardedFor
869 # As 'Apache2::ForwardedFor' takes the first of $ENV{"HTTP_X_FORWARDED_FOR"}
870 # while the contents is '127.0.0.1, 213.220.195.171' if client has its own proxy.
871 # We must take the last item ourselves.
872 # Be VERY sure you always retrieve all the headers unconditionally to hit: My::Hash::RecordKeys
873 my $x_forwarded_for=$W->{"headers_in"}{"X-Forwarded-For"};
874 $x_forwarded_for=~s/^.*,\s*// if $x_forwarded_for;
875 my $remote_ip=$W->{"headers_in"}{"_remote_ip"};
877 $r||=$x_forwarded_for;
882 # $url={"JP"=>"http://specific",...};
883 # $url={""=>"http://default",...};
886 my($url,$contents,%args)=@_;
888 # A bit ineffective but we must process all the possibilities to get stable 'headers_in' hits!
889 my %map=map(($_=>a_href($url->{$_},$contents,%args)),keys(%$url));
891 $cc||=Geo::IP->new()->country_code_by_addr(remote_ip()) if $have_Geo_IP;
902 # FIXME: &alarm, --timeout is now infinite.
903 # FIXME: Try to remove bash(1).
904 # FIXME: Use: @PATH_FLOCK@
905 my @argv=('flock',dir_top_abs_disk(),'bash','-c',$cmd.' >&2');
906 print STDERR join(" ","SPAWN:",@argv)."\n";
914 cluck "Pathname not absolute: $file" if $file!~m{^/};
916 # TODO: Somehow quickly check dependencies?
917 return make('make -s --no-print-directory'
918 .' -C '."'".File::Basename::dirname($file)."' '".File::Basename::basename($file)."'");
923 my($width,$height)=@_;
925 cluck if !defined $width || !defined $height;
926 return ($W->{"have_style"} ? "style=\"border:0;width:${width}px;height:${height}px\"" : "border=\"0\"")
927 ." width=\"$width\" height=\"$height\"";
930 sub negotiate_variant (%)
934 my @fields=("id","qs","content-type","encoding","charset","lang","size");
935 return [ map(($args{$_}),@fields) ];
938 # Returns: 'HTTP::Headers' instance.
939 sub http_headers_in_for($@)
941 my($self,@headers)=@_;
943 # Limit these entries to generate proper 'Vary' header.
944 return HTTP::Headers->new(map(($_=>$W->{"headers_in"}{$_}),@headers));
947 # Input: $self is required!
948 # Input: Put the fallback variant as the first one.
949 # Returns: always only scalar!
950 sub Negotiate_choose($$)
952 my($self,$variants)=@_;
954 my $best=HTTP::Negotiate::choose($variants,
956 # to prevent: Can't locate object method "scan" via package "Apache2::RequestRec" at HTTP/Negotiate.pm line 84.
957 # Do not: $W->{"r"}->headers_in()
958 # to prevent: Can't locate object method "scan" via package "APR::Table" at HTTP/Negotiate.pm line 84.
959 # Do not: HTTP::Headers->new($W->{"r"}->headers_in());
960 # to prevent empty result or even: Odd number of elements in anonymous hash
961 $self->http_headers_in_for(qw(
967 $best||=$variants->[0][0]; # $variants->[0]{"id"}; &HTTP::Negotiate::choose failed?
972 { "id"=>"png","qs"=>0.9,"content-type"=>"image/png" },
973 { "id"=>"gif","qs"=>0.7,"content-type"=>"image/gif" },
975 my $img_variants_re='[.](?:'.join('|',"jpeg",map(($_->{"id"}),@img_variants)).')$';
977 # Returns: ($path_web,$path_abs_disk)
978 # URI path segments support ignored here. Where it is used? (';' path segment options)
983 cluck if !uri_is_local $in;
984 my $uri=in_to_uri_abs $in;
985 my $path_abs_disk=path_abs_disk $uri,%args,"uri_as_in"=>1,"register"=>0;
987 # Known image extension?
988 return path_web($uri,%args,"uri_as_in"=>1),$path_abs_disk if $uri->path()=~m#$img_variants_re#o;
991 for my $var (@img_variants) {
992 my $path_abs_disk_variant=$path_abs_disk.".".$var->{"id"};
993 path_abs_disk_register($path_abs_disk_variant);
994 __PACKAGE__->make_file($path_abs_disk_variant);
995 push @nego_variants,negotiate_variant(
997 "size"=>(stat $path_abs_disk_variant)[7],
1000 my $ext=__PACKAGE__->Negotiate_choose(\@nego_variants);
1002 $uri->path($uri->path().".$ext");
1003 return path_web($uri,%args,"uri_as_in"=>1),path_abs_disk($uri,%args,"uri_as_in"=>1);
1009 my($in,$alt,%args)=@_;
1012 my($path_web,$path_abs_disk)=_img_src($in,%args);
1013 my($width,$height)=Image::Size::imgsize($path_abs_disk);
1015 $alt=escapeHTML($alt);
1016 my $content="<img src=\"".uri_escaped($path_web)."\" alt=\"$alt\" title=\"$alt\" ".img_size($width,$height)
1017 .(!$args{"attr"} ? "" : " ".$args{"attr"})." />";
1018 do { return a_href((_img_src($_))[0],$content,"uri_as_in"=>1) if $_; } for $args{"a_href_img"};
1019 do { return a_href $_,$content if $_; } for $args{"a_href"};
1026 $r.='<table border="0" width="100%"><tr>'."\n";
1027 @_=( [@_] ) if !ref $_[0];
1029 $r.="\t".'<td align="center">'.&{\&img}(@$_).'</td>'."\n";
1031 $r.='</tr></table>'."\n";
1037 my($text,@args_img)=@_;
1039 # FIXME: Workaround bug of 'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)':
1040 # <col width="@{[ (!$W->{"browser"}->ie() ? "1*" : "90%" ) ]}" />
1041 # <col width="@{[ (!$W->{"browser"}->ie() ? "0*" : "10%" ) ]}" />
1042 # causes whole invisible projects in: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.10) Gecko/20050719 Galeon/1.3.21
1044 <table border="0" width="100%">
1050 @{[ &{\&img}(@args_img) ]}
1059 my($class,$filename)=@_;
1062 open F,$filename or cluck "Cannot open \"$filename\": $!";
1063 my $F=do { local $/=undef(); <F>; };
1064 close F or cluck "Cannot close \"$filename\": $!";
1072 header("Expires"=>HTTP::Date::time2str(1000000000)); # date in the past
1073 header("Last-Modified"=>HTTP::Date::time2str()); # always modified
1074 header("Cache-Control"=>join(", ",
1079 "pre-check=0", # MSIE
1080 "post-check=0", # MSIE
1082 header("Pragma"=>"no-cache"); # HTTP/1.0
1083 header("Vary"=>"*"); # content may ba based on unpredictable sources
1086 sub headers_in_filtered(@)
1090 return map(($_=>$W->{"headers_in"}{$_}),@keys);
1093 our %uri_args_frozen_to_headers_in_keys;
1094 our %uri_args_headers_in_frozen_to_headers_out;
1096 sub uri_args_headers_in_frozen_get($)
1098 my($headers_in_keys_arrayref)=@_;
1100 my %uri_args_headers_in_hash=(
1101 "uri_args_frozen"=>$W->{"uri_args_frozen"},
1102 "headers_in"=>{ headers_in_filtered(@$headers_in_keys_arrayref) },
1104 return do { local $Storable::canonical=1; Storable::freeze(\%uri_args_headers_in_hash); };
1107 sub cache_output_filter($)
1111 while ($f->read(my $text,0x400)) {
1112 cluck "utf-8 untested" if Encode::is_utf8($text); # Possible here at all?
1114 $W->{"digest-md5"}->add($text);
1121 # Used only if: !$W->{"http_safe"}
1122 # but we would cause on different method(): Appending to the '_done' package list
1123 # &Wrequire it here even if it will not be later used; to be stable!
1124 Wrequire 'My::Hash::RestrictTo';
1125 if (!$W->{"http_safe"}) {
1126 __PACKAGE__->_no_cache();
1132 "method"=>$W->{"r"}->method(),
1133 "uri"=>"http://".$W->{"web_hostname"}."/".$W->{"r"}->uri(),
1134 "args"=>$W->{"args_orig"},
1136 $W->{"uri_args_frozen"}=do { local $Storable::canonical=1; Storable::freeze(\%uri_args_hash); };
1137 last if !(my $headers_in_keys_arrayref=$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}});
1139 # Protection to be sure we are stable:
1140 $W->{"headers_in"}=My::Hash::RestrictTo->new($W->{"headers_in"},@$headers_in_keys_arrayref);
1142 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get($headers_in_keys_arrayref);
1143 last if !(my $headers_out_hashref=$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}});
1144 header(%$headers_out_hashref);
1147 # &meets_conditions will always deny the attempt if !2xx status().
1148 # At least ap_read_request() sets: r->status=HTTP_REQUEST_TIME_OUT; /* Until we get a request */
1149 my $status_old=$W->{"r"}->status();
1150 $W->{"r"}->status(HTTP_OK);
1151 # Update httpd's 'r->mtime' as the header "Last-Modified" is just not enough for ap_meets_conditions():
1152 # &update_mtime() argument is really in _secs_, not in _msecs_ as the docs claim.
1153 # Be aware '*1000000' would overflow Perl integer anyway.
1154 # &set_last_modified would also override the "Last-Modified" headers_out!
1155 # &mtime may exist but somehow does not work.
1156 $W->{"r"}->update_mtime(HTTP::Date::str2time($headers_out_hashref->{"Last-Modified"}));
1157 $status=$W->{"r"}->meets_conditions();
1158 $W->{"r"}->status($status_old);
1160 last if OK==$status;
1161 $W->{"r"}->status($status);
1162 $W->{"header_only"}=1; # Inhibit &footer output.
1167 $W->{"digest-md5"}=Digest::MD5->new();
1168 $W->{"cache_active"}=1;
1169 $W->{"r"}->add_output_filter(\&cache_output_filter);
1172 sub cache_finish_last_modified()
1174 cluck "Not yet done now? W __PACKAGE__: ".$W->{"__PACKAGE__"}
1175 if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
1176 for my $package_orig (@{$packages_used_array{$W->{"__PACKAGE__"}}}) {
1177 local $_=$package_orig.".pm";
1179 path_abs_disk "/$_","register"=>1;
1182 for my $path_abs_disk (keys(%{$W->{"path_abs_disk_register"}})) {
1183 my $mtime=(stat $path_abs_disk)[9];
1184 do { cluck "No mtime for: $path_abs_disk"; next; } if !$mtime;
1185 $mtime_newest=$mtime if !$mtime_newest || $mtime_newest<$mtime;
1187 cluck "No mtime_newest found for the current W __PACKAGE__: ".$W->{"__PACKAGE__"}
1189 return HTTP::Date::time2str($mtime_newest);
1195 # Do not: return if !$W->{"uri_args_frozen"};
1196 # as we may have just gave 304 and 'exit;' without starting the caching.
1197 return if !$W->{"cache_active"};
1199 # Headers may not be complete in this case; not sure, just trying.
1200 return if $W->{"r"}->connection()->aborted();
1202 # Fill-in/check: %uri_args_frozen_to_headers_in_keys
1203 my $headers_in_keys_stored_arrayref_ref=\$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}};
1204 my @headers_in_keys=tied(%{$W->{"headers_in_RecordKeys"}})->accessed();
1205 if (!$$headers_in_keys_stored_arrayref_ref
1206 || !Data::Compare::Compare(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)) {
1207 cluck "Non-matching generated 'headers_in_keys' per 'uri_args_frozen' key:\n"
1208 .Dumper(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)
1209 if $$headers_in_keys_stored_arrayref_ref;
1210 # Build or possibly prevent such further warn dupes:
1211 $$headers_in_keys_stored_arrayref_ref=\@headers_in_keys;
1212 # Build or regenerate as obsoleted now:
1213 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get(\@headers_in_keys);
1216 # Prepare 'headers_out' for the future reusal:
1218 # Do not: $W->{"digest-md5"}->b64digest();
1219 # as it will not provide the trailing filling '='s.
1220 # RFC 1864 is not clear if they should be there but its sample provides them.
1221 # Do not try to provide canonical "\r\n" form of newlines as is said by RFC 1864.
1222 # RFC 2068 (HTTP/1.1) section 14.16 says the newlines should NOT be converted for HTTP.
1223 # ',""' to avoid breaking the headers by its default "\n".
1224 $headers_out{"Content-MD5"}=MIME::Base64::encode_base64($W->{"digest-md5"}->digest(),"");
1225 # In fact we could also use MD5 for ETag as if we know ETag we also know MD5.
1226 # But this way we do not need to calculate MD5 and we still can provide such ETag. So.
1227 # $W->{"r"}->set_etag() ?
1228 $headers_out{"ETag"}='"'.Digest::MD5::md5_base64($W->{"uri_args_headers_in_frozen"}).'"';
1229 # $W->{"r"}->set_content_length() ?
1230 $headers_out{"Content-Length"}=$W->{"r"}->bytes_sent();
1231 my %Vary=map(($_=>1),(@headers_in_keys));
1237 %Vary=("*"=>1) if $Vary{"*"};
1238 $headers_out{"Vary"}=join(", ",sort keys(%Vary)) if keys(%Vary);
1239 # $W->{"r"}->set_last_modified() ?
1240 $headers_out{"Last-Modified"}=cache_finish_last_modified();
1242 # Fill-in/check: %uri_args_headers_in_frozen_to_headers_out
1243 my $headers_out_stored_hashref_ref=\$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}};
1244 if (!$$headers_out_stored_hashref_ref
1245 || !Data::Compare::Compare(\%headers_out,$$headers_out_stored_hashref_ref)) {
1246 cluck "Non-matching generated 'headers_out' per 'uri_args_headers_in_frozen' key:\n"
1247 .Dumper(\%headers_out,$$headers_out_stored_hashref_ref)
1248 if $$headers_out_stored_hashref_ref;
1249 # Build or possibly prevent such further warn dupes:
1250 $$headers_out_stored_hashref_ref=\%headers_out;
1253 ###print STDERR Dumper(\%uri_args_frozen_to_headers_in_keys,\%uri_args_headers_in_frozen_to_headers_out);
1260 if (!$W->{"header_only"}) {
1261 header("Content-Style-Type"=>"text/css");
1262 # Do not: text/javascript
1263 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
1264 # "application/javascript" so far standardized till 2005-12-08 by:
1265 # http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
1266 header("Content-Script-Type"=>"application/javascript");
1267 # $W->{"r"}->content_languages() ?
1268 do { header("Content-Language"=>$_) if $_; } for $W->{"language"};
1270 # TODO: Support also: private
1271 header("Cache-Control"=>"public"); # HTTP/1.1
1273 # Use $W->{"charset"}=0 to disable charset.
1274 $W->{"charset"}="us-ascii"
1275 if !defined $W->{"charset"} && (!defined($W->{"content_type"}) || $W->{"content_type"});
1278 # https://bugzilla.mozilla.org/show_bug.cgi?id=120556
1280 # Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1281 # http://validator.w3.org/ does not send ANY "Accept" headers!
1282 if (!defined $W->{"content_type"}) {
1283 # Be _stable_ for "headers_in".
1284 my $accept=$W->{"headers_in"}{"Accept"};
1285 my $user_agent=$W->{"headers_in"}{"User-Agent"}||"";
1286 $W->{"content_type"}="application/xhtml+xml"
1287 if !$accept && $user_agent=~m{^W3C_Validator/}i;
1289 my $negotiated=$class->Negotiate_choose([
1290 # Put the fallback variant as the first one.
1291 # Rate both variants the same to prefer "text/html" for undecided clients.
1293 # Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1294 # prefers "application/xhtml+xml" over "text/html" itself:
1295 # text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5
1298 "content-type"=>"text/html",
1300 (!$W->{"charset"} ? () : "charset"=>$W->{"charset"}),
1301 "lang"=>$W->{"language"},
1304 "id"=>"application/xhtml+xml",
1305 "content-type"=>"application/xhtml+xml",
1307 (!$W->{"charset"} ? () : "charset"=>$W->{"charset"}),
1308 "lang"=>$W->{"language"},
1313 $W->{"content_type"}=$negotiated if !defined $W->{"content_type"};
1315 # mod_perl doc: If you set this header via the headers_out table directly, it
1316 # will be ignored by Apache. So do not do that.
1318 if ($W->{"content_type"}) {
1319 $type=MIME::Types->new()->type($W->{"content_type"});
1320 cluck "MIME::Types type '".$W->{"content_type"}."' not known" if !$type;
1322 cluck "charset='".$W->{"charset"}."' does not match content-type='".$W->{"content_type"}."'"
1323 if ($W->{"charset"} ? 1 : 0) != (!$type ? 0 : $type->isAscii());
1324 $W->{"r"}->content_type($W->{"content_type"}.(!$W->{"charset"} ? "" : "; charset=".$W->{"charset"}))
1325 if $W->{"content_type"};
1328 # We still can append headers before we put out some text.
1329 # FIXME: It is not clean to still append them without overwriting.
1330 return if $W->{"heading_done"};
1331 Wprint '<?xml version="1.0" encoding="'.$W->{"charset"}.'"?>'."\n"
1332 if (!$W->{"header_only"} || $W->{"header_only"} eq "xml") && (0
1333 || $W->{"content_type"}=~m{^application/\w+[+]xml$}
1334 || $W->{"content_type"} eq "text/vnd.wap.wml");
1335 return if $W->{"header_only"};
1336 # Split 'heading_done' for the proper handling of: /project/Rel.pm
1337 $W->{"heading_done"}++;
1339 Wprint '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">'."\n";
1340 Wprint '<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="'.$W->{"language"}.'">'."\n";
1341 my $title=$W->{"title_prefix"}.join("",map({ ': '.$_; } ($W->{"title"} || ())));
1342 # Do not: cluck if $title=~/[<>]/;
1343 # as it is not solved just by: &a_href_inhibit
1344 # as sometimes titles use also: <i>...</i>
1345 $title=~s#<[^>]*>##g;
1347 Wprint "<title>$title</title>\n";
1348 if ($W->{"have_css"}) {
1349 # Everything can get overriden later.
1350 for my $css ("/My/Web.css",@{$W->{"css_push"}}) {
1352 <link rel="stylesheet" type="text/css" href="@{[ uri_escaped(path_web $css) ]}" />
1355 if ($W->{"css_inherit"}) {
1356 # Do not: <script />
1357 # as at least Lynx inhibits any further HTML output.
1358 # Do not: text/javascript
1359 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
1360 # "application/javascript" so far standardized till 2005-12-08 by:
1361 # http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
1363 <script type="application/javascript" src="@{[ uri_escaped(path_web('/My/css_inherit.js')) ]}"></script>
1367 Wprint '<meta name="robots" content="'.($W->{"indexme"} ? "" : "no" ).'index,follow" />'."\n";
1368 Wprint $W->{"head"};
1369 for my $type (qw(prev next index contents start up)) {
1370 do { Wprint '<link rel="'.$type.'" href="'.uri_escaped(path_web $_).'" />'."\n" if $_; }
1371 for ($W->{"rel_$type"});
1373 Wprint "</head><body";
1374 # Wprint ' bgcolor="black" text="white" link="aqua" vlink="teal"'
1375 # if $W->{"browser"}->netscape() && (!$W->{"browser"}->major() || $W->{"browser"}->major()<=4);
1376 Wprint $W->{"body_attr"};
1379 do { Wprint $_ if $_; } for $W->{"heading"};
1383 delete $W->{"__My::Web_init"};