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;
57 # print STDERR "Wrequire $file\n";
64 for (my $depth=0;defined caller($depth);$depth++) {
65 $callers{caller($depth)}=1;
67 my $selfpkg=__PACKAGE__;
69 for my $target ($class,__PACKAGE__) {
70 for my $caller (keys(%callers)) {
71 next if $packages_used_hash{$caller}{$target}++;
72 cluck "Appending to the '_done' package list: caller=$caller,target=$target"
73 if $packages_used_hash{$caller}{"_done"};
75 unshift @{$packages_used_array{$caller}},$target;
78 push @{$packages_used_array{$caller}},$target;
82 eval { CORE::require "$file"; } or confess $@;
83 1; # Otherwise 'require' would already file above.
90 # print STDERR "Wuse $file\n";
92 local $Exporter::ExportLevel=$Exporter::ExportLevel+1;
101 local $Exporter::ExportLevel=$Exporter::ExportLevel+1;
103 return $class->SUPER::import(@rest);
107 use WebConfig; # see also below: Wuse 'WebConfig';
108 require Image::Size; # for &imgsize
109 use File::Basename; # &basename
110 use Carp qw(cluck confess);
112 require HTTP::BrowserDetect;
113 require HTTP::Negotiate;
114 our $have_Geo_IP; BEGIN { $have_Geo_IP=eval { require Geo::IP; 1; }; }
115 # Do not: use ModPerl::Util qw(exit);
116 # to prevent in mod_perl2: "exit" is not exported by the ModPerl::Util module
118 use POSIX qw(strftime);
120 use Apache2::Const qw(HTTP_MOVED_TEMPORARILY OK HTTP_OK);
127 require Data::Compare;
130 use Apache2::RequestUtil;
132 use Apache2::Connection;
133 require MIME::Base64;
134 use Apache2::ServerUtil;
136 require MIME::Parser;
137 use Apache2::RequestRec;
138 use Apache2::RequestIO;
139 use Apache2::Response;
148 print STDERR "$class->init ".Apache2::RequestUtil->request()->unparsed_uri()."\n";
150 # We need to track package dependencies, so we need to call it from &init.
151 # We cannot do it in BEGIN { } block
152 # as it would not be tracked for each of the toplevel users later.
156 # $W={} can get somehow created very easily.
157 # Do not: cluck "W not empty:\n".Dumper($W) if keys(%$W);
158 # to prevent (of $W->{"headers_in"}): TODO: Enumeration may not be expected.
159 cluck "W not empty; __PACKAGE__ was: ".$W->{"__PACKAGE__"} if keys(%$W);
160 $W=My::Hash->new({},"My::Hash::Sub","My::Hash::Push");
163 "__PACKAGE__"=>scalar(caller()),
165 %args, # override %WebConfig settings
168 # {"__PACKAGE__"} is mandatory for mod_perl-2.0;
169 # $Apache2::Registry::curstash is no longer supported.
170 do { cluck "No $_" if !$W->{$_}; } for "__PACKAGE__";
172 # Package dependencies tracking only:
173 Wrequire $W->{"__PACKAGE__"},"first"=>1;
175 do { $W->{$_}=0 if !defined $W->{$_}; } for "detect_ent";
176 do { $W->{$_}=0 if !defined $W->{$_}; } for "detect_js";
177 do { $W->{$_}=1 if !defined $W->{$_}; } for "have_css"; # AFAIK it does not hurt anyone.
178 do { $W->{$_}=0 if !defined $W->{$_}; } for "css_inherit";
179 do { $W->{$_}=1 if !defined $W->{$_}; } for "footer";
180 do { $W->{$_}=1 if !defined $W->{$_}; } for "footer_delimit";
181 do { $W->{$_}=1 if !defined $W->{$_}; } for "footer_ids";
182 do { $W->{$_}=1 if !defined $W->{$_}; } for "indexme";
183 do { $W->{$_}="" if !defined $W->{$_}; } for "head";
184 do { $W->{$_}="" if !defined $W->{$_}; } for "body_attr";
185 do { $W->{$_}="en-US" if !defined $W->{$_}; } for "language";
188 for (qw(footer_ids)) {
189 $W->{$_}=0 if !$W->{"footer"};
190 $footer_any=1 if $W->{$_};
192 $W->{"footer"}=0 if !$footer_any;
193 $W->{"footer_delimit"}=0 if !$W->{"footer"};
195 $W->{"r"}=Apache2::RequestUtil->request();
197 $W->{"r"}->push_handlers("PerlCleanupHandler"=>\&cleanup);
199 $W->{"web_hostname"}||=$W->{"r"}->hostname();
201 tie *STDOUT,$W->{"r"};
205 $W->{"headers_in"}=$W->{"r"}->headers_in();
206 Wrequire 'My::Hash::Merge';
207 $W->{"headers_in"}=My::Hash::Merge->new(
210 "_remote_ip"=>sub { return $W->{"r"}->connection()->remote_ip(); },
213 Wrequire 'My::Hash::Readonly';
214 $W->{"headers_in"}=My::Hash::Readonly->new($W->{"headers_in"});
216 if ($W->{"r"}->method() eq "GET" || $W->{"r"}->method() eq "HEAD") {
217 for (\$W->{"http_safe"}) {
218 # Do not: # Extend the current ETag system instead if you would need it:
219 # cluck "Explicitely NOT HTTP-Safe for method \"".$W->{"r"}->method()."\"?!?"
220 # if defined($$_) && !$$_;
221 # as sometimes it just does not make sense to cache it.
222 $$_=1 if !defined $$_;
226 for (\$W->{"http_safe"}) {
227 cluck "Undefined HTTP-Safe-ty for method \"".$W->{"r"}->method()."\"!"
229 $$_=0 if !defined $$_;
232 # Used only if: $W->{"http_safe"}
233 # but we would cause on different method(): Appending to the '_done' package list
234 Wrequire 'My::Hash::RecordKeys';
235 if ($W->{"http_safe"}) {
236 $W->{"headers_in_RecordKeys"}=My::Hash::RecordKeys->new($W->{"headers_in"});
237 $W->{"headers_in"}=$W->{"headers_in_RecordKeys"};
241 local $_=$W->{"r"}->args() || "";
242 if ($W->{"detect_ent"}) {
243 if (/[&]amp;have_ent/)
244 { $W->{"have_ent"}=0; }
245 elsif ( /[&]have_ent/)
246 { $W->{"have_ent"}=1; }
248 { delete $W->{"have_ent"}; }
249 if (!defined $W->{"have_ent"} && $W->{"r"}->method() eq "GET") {
250 $W->{"head"}.='<meta http-equiv="Refresh" content="0; URL='
251 .escapeHTML("http://".$W->{"web_hostname"}."/".($W->{"r"}->uri()=~m#^/*(.*)$#)[0]
252 ."?".($_ || "detect_ent_glue=1").'&have_ent=detect')
260 $W->{"args"}=URI->new("?".$W->{"r"}->args())->query_form_hash();
261 $W->merge_post_args() if $W->{"r"}->method() eq "POST";
262 # Prepare '$args' first to (FIXME: Why?) prevent: Not a reference
263 my $args=$W->{"args"};
264 $W->{"args_orig"}=Storable::dclone($args);
266 $W->{"browser"}=sub {
267 # Lazy-evaluation, we may not need the "User-Agent" header at all.
268 return our $r||=HTTP::BrowserDetect->new($W->{"headers_in"}{"User-Agent"});
271 if (!defined $W->{"have_style"}) {
272 $W->{"have_style"}=sub {
273 # Lazy-evaluation, we may not need the "User-Agent" header at all.
274 return our $r||=(!$W->{"browser"}->netscape() || ($W->{"browser"}->major() && $W->{"browser"}->major()>4) ? 1 : 0);
278 $W->{"have_js"}=($W->{"args"}{"have_js"} ? 1 : 0);
279 if ($W->{"detect_js"} && !$W->{"have_js"}) {
281 # as at least Lynx inhibits any further HTML output.
282 # Do not: text/javascript
283 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
284 # "application/javascript" so far standardized till 2005-12-08 by:
285 # http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
286 $W->{"head"}.='<script type="application/javascript" src="'.uri_escaped(path_web('/My/HaveJS.pm')).'"></script>'."\n";
289 # Required by &_args_check below.
290 $W->{"_init_done"}=1;
292 do { _args_check(%$_) if $_; } for $W->{"args_check"};
301 return q{enctype="application/x-www-form-urlencoded" accept-charset="us-ascii utf-8"} if $method eq "post";
302 return q{accept-charset="us-ascii utf-8"} if $method eq "get";
303 cluck "Undefined method: $method";
307 sub merge_post_args($)
311 my @post_args=$class->read_post_args();
313 my $name=shift @post_args;
314 my $data=shift @post_args;
315 my $ref=\$W->{"args"}{$name};
316 if (!defined $$ref) { $$ref=$data; }
317 elsif (!ref $$ref) { $$ref=[$$ref,$data]; }
318 elsif ("ARRAY" eq ref $$ref) { push @$$ref,$data; }
320 cluck "Ignoring POST argument \"$name\", orig is weird:\n",Dumper($$ref);
327 # as CGI parsing of POST vs. QUERY_STRING data, multiple-valued keys etc.
328 # is too dense and causes weird problems, together with mod_perl etc.
329 sub read_post_args($)
333 local $_=$class->http_headers_in_for("Content-type")->content_type();
334 return $class->read_multipart_form_data() if $_ eq "multipart/form-data";
335 return $class->read_application_x_www_form_urlencoded() if $_ eq "application/x-www-form-urlencoded";
336 cluck "Unknown POST data body, ignored: $_";
340 sub read_application_x_www_form_urlencoded($)
346 my $got=$W->{"r"}->read(my($buf),0x1000);
347 # Do not: cluck "Error reading POST data: $!" if !defined $got;
348 # as it should be done using: APR::Error exceptions
352 return URI->new("?".$body)->query_form();
355 sub read_multipart_form_data($)
359 my $parser=MIME::Parser->new();
360 # FIXME: No unlink()s done!
361 $parser->output_under("/tmp");
366 tie *FH,"My::Web::ReadMerged",
367 join("",map(($_.": ".$W->{"headers_in"}{$_}."\n"),qw(
371 my $body=$parser->parse(\*FH);
372 cluck "No multipart POST request body?" if !$body->is_multipart();
375 $_->head()->mime_attr("content-disposition.name")
377 join("",@{$_->body()})
380 # TODO: Globalize, make it IO::* compatible, split to the merging part + IO::Scalar.
381 package My::Web::ReadMerged;
385 our @ISA=qw(Tie::Handle Exporter);
386 use Carp qw(cluck confess);
392 confess "Slurp not yet implemented" if !defined $/;
393 # Apache2::RequestIO does not support 'READLINE'!
395 if (defined $self->{"data"} && $self->{"data"}=~s{^.*\Q$/\E}{}) {
396 $self->{"offset"}+=length $&;
399 my $fh_orig=$self->{"fh_orig"};
401 my $r=$self->{"data"};
402 delete $self->{"data"};
403 $self->{"offset"}+=length $r if defined $r;
406 my $got=read $fh_orig,my($buf),0x1000;
407 cluck "Error reading POST data: $!" if !defined $got;
408 delete $self->{"fh_orig"} if !$got;
409 cluck "INTERNAL: fh_orig should not exist here" if !defined $self->{"data"};
410 $self->{"data"}.=$buf;
418 return $self->{"offset"};
423 my($class,$data,$fh_orig)=@_;
425 my $self=bless {},$class;
426 $self->{"data"}=$data;
428 $self->{"fh_orig"}=$fh_orig;
435 my($apache_request)=@_;
437 cluck "CORE::GLOBAL::exit hook not ran" if !$W->{"_exit_done"};
438 cluck "packages not finalized" if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
446 # PerlResponseHandler is RUN_FIRST and &ModPerl::Util::exit returns OK, so no (sane) go.
447 # PerlLogHandler is already too late to be able to produce any output.
451 cluck "Missing ->init while in exit_hook()" if !$W->{"_init_done"};
452 # &footer will call us recursively!
453 footer() if !$W->{"_exit_done"}++;
454 return &{$exit_orig}(@_);
458 do { cluck "exit_hook_start() twice?"; return; } if defined $exit_orig;
459 $exit_orig=\&CORE::GLOBAL::exit;
460 # Prevent: Subroutine CORE::GLOBAL::exit redefined
461 no warnings 'redefine';
462 *CORE::GLOBAL::exit=\&exit_hook;
466 do { cluck "exit_hook_stop() without exit_hook_start()?"; return; }
467 if \&exit_hook ne \&CORE::GLOBAL::exit;
468 do { cluck "INTERNAL: exit_orig uninitialized"; return; }
470 # Prevent: Subroutine CORE::GLOBAL::exit redefined
471 no warnings 'redefine';
472 *CORE::GLOBAL::exit=$exit_orig;
476 # Be aware other parts of code (non-My::Web) will NOT use this function!
477 # Do not: Wprint $W->{"heading"},"undef"=>1;
478 # as we would need to undef() it to turn it off and it would get defaulted in such case.
479 # Do not: exists $W->{"heading"}
480 # as we use a lot of 'for $W->{"heading"}' which instantiates it with the value: undef()
485 cluck "undef Wprint" if !defined $text && !$args{"undef"};
486 delete $args{"undef"};
487 cluck join(" ","Invalid arguments:",keys(%args)) if keys(%args);
488 return if !defined $text;
489 # Do not: cluck "utf-8 untested" if Encode::is_utf8($text);
490 # as it is valid here.
491 $W->{"r"}->puts($text);
494 sub request_check(;$)
498 # Use &eval to prevent: Global $r object is not available. Set:\n\tPerlOptions +GlobalRequest\nin ...
499 confess "Calling sensitive dynamic code from a static code" if !eval { Apache2::RequestUtil->request(); };
500 # Do not: confess "Calling sensitive dynamic code without My::Web::init" if !$W->{"__PACKAGE__"};
501 # as it is valid at least while preparing arguments to call: &project::Lib::init
505 # as it is too much backward compatible regarding the charset encodings etc.
506 # and the resulting code is too dense with no additional functionality for the recent content.
519 # /home/user/www/webdir
520 sub dir_top_abs_disk()
522 our $dir_top_abs_disk;
523 if (!$dir_top_abs_disk) {
524 my $selfpkg_relpath=__PACKAGE__;
525 $selfpkg_relpath=~s{::}{/}g;
526 $selfpkg_relpath.=".pm";
527 my $selfpkg_abspath=$INC{$selfpkg_relpath} or do {
528 cluck "Unable to find self package $selfpkg_relpath";
531 $selfpkg_abspath=~s{/*\Q$selfpkg_relpath\E$}{} or do {
532 cluck "Unable to strip myself \"$selfpkg_relpath\" from the abspath: $selfpkg_abspath";
535 cluck "INC{myself} is relative?: $selfpkg_abspath" if $selfpkg_abspath!~m{^/};
536 $dir_top_abs_disk=$selfpkg_abspath;
538 return $dir_top_abs_disk;
544 if (!$W->{"unparsed_uri"}) {
546 # as we may be called before &init from: &My::Project::init
547 my $r=Apache2::RequestUtil->request();
548 cluck "Calling ".'&unparsed_uri'." from a static code, going to fail" if !$r;
549 my $uri_string=$r->unparsed_uri() or cluck "Valid 'r' missing unparsed_uri()?";
550 my $uri=URI->new_abs($uri_string,"http://".$W->{"web_hostname"}."/");
551 $W->{"unparsed_uri"}=$uri;
553 return $W->{"unparsed_uri"};
560 # Otherwise we may have been already processed and thus legally relativized.
561 # FIXME data: Currently disabled, all the data are too violating such rule.
563 my $uri_check=URI->new($in);
564 $uri_check->scheme() || $in=~m{^\Q./\E} || $in=~m{^/}
565 or cluck "Use './' or '/' prefix for all the local references: $in";
567 my $uri=URI->new_abs($in,unparsed_uri());
568 $uri=$uri->canonical();
572 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
578 cluck if !$args{"uri_as_in"} && ref $in;
579 my $uri=in_to_uri_abs($in);
580 if (uri_is_local($uri)) {
581 # Prefer the $uri values over "args_persistent" values.
582 # &query_form_hash comes from: URI::QueryParam
583 $uri->query_form_hash({
586 my $val=$W->{"args"}{$key};
587 (!defined $val ? () : ($key=>$val));
588 } keys(%{$W->{"args_persistent"}})),
589 %{$uri->query_form_hash()},
592 return $uri->abs(unparsed_uri()) if $W->{"args"}{"Wabs"} || $args{"abs"};
593 return $uri->rel(unparsed_uri());
596 sub path_abs_disk_register($)
598 my($path_abs_disk)=@_;
600 $W->{"path_abs_disk_register"}{$path_abs_disk}=1;
603 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
604 sub path_abs_disk($%)
608 cluck if !$args{"uri_as_in"} && ref $in;
609 my $uri=in_to_uri_abs($in);
610 cluck if !uri_is_local($uri);
611 my $path=$uri->path();
612 cluck "URI compatibility: ->path() not w/leading slash of URI \"$uri\"; path: $path" if $path!~m{^/};
613 my $r=dir_top_abs_disk().$path;
614 path_abs_disk_register $r if !defined $args{"register"} || $args{"register"};
624 while (my($name,$regex)=each(%tmpl)) {
625 my $name_html="Parameter <span class=\"quote\">".escapeHTML($name)."</span>";
626 $W->{"args"}{$name}="" if !defined $W->{"args"}{$name};
627 $W->{"args"}{$name}=[ $W->{"args"}{$name} ] if !ref $W->{"args"}{$name} && ref $regex;
628 fatal "$name_html passed as multivar although singlevar expected"
629 if ref $W->{"args"}{$name} && !ref $regex;
630 $regex=$regex->[0] if ref $regex;
631 for my $val (!ref $W->{"args"}{$name} ? $W->{"args"}{$name} : @{$W->{"args"}{$name}}) {
632 $val="" if !defined $val;
633 fatal "$name_html <span class=\"quote\">".escapeHTML($val)."</span>"
634 ." does not match the required regex <span class=\"quote\">".escapeHTML($regex)."</span> "
635 if $regex ne "" && $val!~/$regex/;
644 return '<p'.(!defined $height ? "" : ' style="height: '.$height.';"').'> </p>'."\n";
651 $msg="UNKNOWN" if !$msg;
654 # Do not send it unconditionally.
655 # The intial duplicated '<?xml...' crashes Gecko parser.
656 $W->{"heading_done"}=0 if $W->{"header_only"};
657 # Do not send it unconditionally.
658 # Prevents warn: Headers already sent
659 if (!$W->{"heading_done"}) {
660 $W->{"indexme"}=0; # For the case no heading was sent yet.
661 $W->{"header_only"}=0; # assurance for &heading
662 $W->{"content_type"}="text/html"; # Force HTML and avoid strictly checked XHTML.
665 Wprint "\n".vskip("3ex")."<hr /><h1 class=\"error\">FATAL ERROR: $msg!</h1>\n"
666 ."<p>You can report this problem's details to"
667 ." ".a_href("mailto:".$W->{"admin_mail"},"admin of this website").".</p>\n";
671 sub footer_packages_used_comments()
673 my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
674 for my $package (@$packages_used) {
675 my $cvs_id=(eval('$'.$package."::CVS_ID")
676 # || $package # debug
678 Wprint '<!-- '.$package.' - $'.$cvs_id.'$ -->'."\n" if $cvs_id;
684 cluck 'Explicit &footer call is deprecated, !_exit_dne' if !$W->{"_exit_done"};
685 exit if $W->{"footer_done"}++; # deadlock prevention:
686 &{$_}() for reverse @{$W->{"footer_sub_push"}};
687 if ($W->{"header_only"}) {
688 $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
692 Wprint vskip if $W->{"footer_delimit"};
693 &{$_}() for reverse @{$W->{"footing_delimit_sub_push"}};
694 Wprint "<hr />\n" if $W->{"footer"};
696 # Never update the package list while we examine it!
697 $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
699 my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
700 if ($W->{"footer_ids"}) {
701 Wprint '<p class="cvs-id">';
702 Wprint join("<br />\n",map({ my $package=$_;
703 my $cvs_id=(eval('$'.$package."::CVS_ID")
704 # || $package # debug
710 $cvs_id='$'.$cvs_id.'$'; # Eaten by 'q' operator.
711 my @cvs_id_split=split / +/,$cvs_id;
712 if (@cvs_id_split==8) {
719 my $path_abs_disk=path_abs_disk("/$file$ext");
720 push @tried,$path_abs_disk;
721 last if -r $path_abs_disk;
722 cluck "Class file $file not found; tried: ".join(" ",@tried) if !$ext;
726 if ((my $file_cvs=$file)=~s{^My/}{}) {
727 $viewcvs=$W->{"viewcvs_My"}.$file_cvs;
730 $viewcvs=$W->{"viewcvs"}.$file;
733 .a_href((map({ my $s=$_; $s=~s#/viewcvs/#$&~checkout~/#; $s; } $viewcvs))[0]."?rev=".$cvs_id_split[2],
735 $cvs_id_split[1]=a_href($viewcvs,($package!~/^Apache2::/ ? $package : $cvs_id_split[1]));
736 $cvs_id_split[5]=&{$W->{"cvs_id_author_sub"}}($cvs_id_split[5]);
738 join " ",@cvs_id_split;
744 footer_packages_used_comments();
746 do { Wprint $_ if $_; } for $W->{"footing"};
748 Wprint "</body></html>\n";
752 # Existing entries are overwritten.
757 while (my($key,$val)=each(%pairs)) {
758 do { cluck "Headers already sent"; next; } if $W->{"heading_done"};
759 $W->{"r"}->headers_out()->set($key,$val);
769 elsif ($size<1024*1024)
770 { $size=sprintf "%.1fK",$size/1024; }
772 { $size=sprintf "%.1fM",$size/1024/1024; }
781 my $uri_rel=in_to_uri_abs($in)->rel(unparsed_uri());
782 # Do not: defined $uri_rel->("userinfo"|"host"|"port")();
783 # as they fail to be called for schemes not supporting them.
784 return 0 if $uri_rel->scheme();
785 return 0 if $uri_rel->authority();
789 # &path_web still may be required for &uri_escaped !
795 my $urient=escapeHTML($uri);
796 return $uri if $uri eq $urient;
798 return $urient if uri_is_local $uri;
799 return $uri if defined $W->{"have_ent"} && !$W->{"have_ent"}; # non-ent client
800 return $urient if $W->{"have_ent"}; # ent client
801 # Unknown client, &escapeHTML should not be needed here:
802 return escapeHTML(path_web('/My/Redirect.pm?location='.uri_escape($uri->abs(unparsed_uri()))));
805 our $a_href_inhibited;
808 my($in,$contents,%args)=@_;
811 do { $$_=1 if !defined $$_; } for (\$args{"size"});
812 if (!defined $contents) {
814 $contents=File::Basename::basename($contents) if $args{"basename"};
815 $contents=escapeHTML($contents);
817 $contents=~s#<a\b[^>]*>##gi;
818 $contents=~s#</a>##gi;
819 return $contents if $a_href_inhibited;
821 my $path_web=path_web $in,%args;
824 $r.=uri_escaped $path_web;
826 do { $r.=" $_" if $_; } for ($args{"attr"});
827 $r.='>'.$contents.'</a>';
828 if ($args{"size"} && uri_is_local($in) && ($args{"size"}>=2 || $in=~/[.](?:gz|Z|rpm|zip|deb|lha)/)) { # Downloadable?
829 my $path_abs_disk=path_abs_disk $in,%args;
830 cluck "File not readable: $path_abs_disk" if !-r $path_abs_disk;
831 $r.=' ('.size_display((stat($path_abs_disk))[7]).')';
836 sub a_href_inhibit($$;@)
838 my($self,$sub,@sub_args)=@_;
840 local $a_href_inhibited=1;
841 return &{$sub}(@sub_args);
844 sub input_hidden_persistents()
849 my $val=$W->{"args"}{$key};
850 (!defined $val ? () : '<input type="hidden"'
851 .' name="'.escapeHTML($key).'"'
852 .' value="'.escapeHTML($val).'"'
854 } (keys(%{$W->{"args_persistent"}}))));
859 my($self,$url,$status)=@_;
861 $url=path_web($url,"abs"=>1);
862 $status||=HTTP_MOVED_TEMPORARILY;
863 $W->{"r"}->status($status);
864 $W->{"r"}->headers_out()->{"Location"}=$url;
865 $W->{"header_only"}=1;
866 $W->{"content_type"}=0;
875 # Do not: PerlModule Apache2::ForwardedFor
876 # PerlPostReadRequestHandler Apache2::ForwardedFor
877 # As 'Apache2::ForwardedFor' takes the first of $ENV{"HTTP_X_FORWARDED_FOR"}
878 # while the contents is '127.0.0.1, 213.220.195.171' if client has its own proxy.
879 # We must take the last item ourselves.
880 # Be VERY sure you always retrieve all the headers unconditionally to hit: My::Hash::RecordKeys
881 my $x_forwarded_for=$W->{"headers_in"}{"X-Forwarded-For"};
882 $x_forwarded_for=~s/^.*,\s*// if $x_forwarded_for;
883 my $remote_ip=$W->{"headers_in"}{"_remote_ip"};
885 $r||=$x_forwarded_for;
890 # $url={"JP"=>"http://specific",...};
891 # $url={""=>"http://default",...};
894 my($url,$contents,%args)=@_;
896 # A bit ineffective but we must process all the possibilities to get stable 'headers_in' hits!
897 my %map=map(($_=>a_href($url->{$_},$contents,%args)),keys(%$url));
899 $cc||=Geo::IP->new()->country_code_by_addr(remote_ip()) if $have_Geo_IP;
910 # FIXME: &alarm, --timeout is now infinite.
911 # FIXME: Try to remove bash(1).
912 # FIXME: Use: @PATH_FLOCK@
913 my @argv=('flock',dir_top_abs_disk(),'bash','-c',$cmd.' >&2');
914 print STDERR join(" ","SPAWN:",@argv)."\n";
922 cluck "Pathname not absolute: $file" if $file!~m{^/};
924 # TODO: Somehow quickly check dependencies?
925 return make('make -s --no-print-directory'
926 .' -C '."'".File::Basename::dirname($file)."' '".File::Basename::basename($file)."'");
931 my($width,$height)=@_;
933 cluck if !defined $width || !defined $height;
934 return ($W->{"have_style"} ? "style=\"border:0;width:${width}px;height:${height}px\"" : "border=\"0\"")
935 ." width=\"$width\" height=\"$height\"";
938 sub negotiate_variant (%)
942 my @fields=("id","qs","content-type","encoding","charset","lang","size");
943 return [ map(($args{$_}),@fields) ];
946 # Returns: 'HTTP::Headers' instance.
947 sub http_headers_in_for($@)
949 my($self,@headers)=@_;
951 # Limit these entries to generate proper 'Vary' header.
952 return HTTP::Headers->new(map(($_=>$W->{"headers_in"}{$_}),@headers));
955 # Input: $self is required!
956 # Input: Put the fallback variant as the first one.
957 # Returns: always only scalar!
958 sub Negotiate_choose($$)
960 my($self,$variants)=@_;
962 my $best=HTTP::Negotiate::choose($variants,
964 # to prevent: Can't locate object method "scan" via package "Apache2::RequestRec" at HTTP/Negotiate.pm line 84.
965 # Do not: $W->{"r"}->headers_in()
966 # to prevent: Can't locate object method "scan" via package "APR::Table" at HTTP/Negotiate.pm line 84.
967 # Do not: HTTP::Headers->new($W->{"r"}->headers_in());
968 # to prevent empty result or even: Odd number of elements in anonymous hash
969 $self->http_headers_in_for(qw(
975 $best||=$variants->[0][0]; # $variants->[0]{"id"}; &HTTP::Negotiate::choose failed?
980 { "id"=>"png","qs"=>0.9,"content-type"=>"image/png" },
981 { "id"=>"gif","qs"=>0.7,"content-type"=>"image/gif" },
983 my $img_variants_re='[.](?:'.join('|',"jpeg",map(($_->{"id"}),@img_variants)).')$';
985 # Returns: ($path_web,$path_abs_disk)
986 # URI path segments support ignored here. Where it is used? (';' path segment options)
991 cluck if !uri_is_local $in;
992 my $uri=in_to_uri_abs $in;
993 my $path_abs_disk=path_abs_disk $uri,%args,"uri_as_in"=>1,"register"=>0;
995 # Known image extension?
996 return path_web($uri,%args,"uri_as_in"=>1),$path_abs_disk if $uri->path()=~m#$img_variants_re#o;
999 for my $var (@img_variants) {
1000 my $path_abs_disk_variant=$path_abs_disk.".".$var->{"id"};
1001 path_abs_disk_register($path_abs_disk_variant);
1002 __PACKAGE__->make_file($path_abs_disk_variant);
1003 push @nego_variants,negotiate_variant(
1005 "size"=>(stat $path_abs_disk_variant)[7],
1008 my $ext=__PACKAGE__->Negotiate_choose(\@nego_variants);
1010 $uri->path($uri->path().".$ext");
1011 return path_web($uri,%args,"uri_as_in"=>1),path_abs_disk($uri,%args,"uri_as_in"=>1);
1017 my($in,$alt,%args)=@_;
1020 my($path_web,$path_abs_disk)=_img_src($in,%args);
1021 my($width,$height)=Image::Size::imgsize($path_abs_disk);
1023 $alt=escapeHTML($alt);
1024 my $content="<img src=\"".uri_escaped($path_web)."\" alt=\"$alt\" title=\"$alt\" ".img_size($width,$height)
1025 .(!$args{"attr"} ? "" : " ".$args{"attr"})." />";
1026 do { return a_href((_img_src($_))[0],$content,"uri_as_in"=>1) if $_; } for $args{"a_href_img"};
1027 do { return a_href $_,$content if $_; } for $args{"a_href"};
1034 $r.='<table border="0" width="100%"><tr>'."\n";
1035 @_=( [@_] ) if !ref $_[0];
1037 $r.="\t".'<td align="center">'.&{\&img}(@$_).'</td>'."\n";
1039 $r.='</tr></table>'."\n";
1045 my($text,@args_img)=@_;
1047 # FIXME: Workaround bug of 'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)':
1048 # <col width="@{[ (!$W->{"browser"}->ie() ? "1*" : "90%" ) ]}" />
1049 # <col width="@{[ (!$W->{"browser"}->ie() ? "0*" : "10%" ) ]}" />
1050 # causes whole invisible projects in: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.10) Gecko/20050719 Galeon/1.3.21
1052 <table border="0" width="100%">
1058 @{[ &{\&img}(@args_img) ]}
1067 my($class,$filename)=@_;
1070 open F,$filename or cluck "Cannot open \"$filename\": $!";
1071 my $F=do { local $/=undef(); <F>; };
1072 close F or cluck "Cannot close \"$filename\": $!";
1080 header("Expires"=>HTTP::Date::time2str(1000000000)); # date in the past
1081 header("Last-Modified"=>HTTP::Date::time2str()); # always modified
1082 header("Cache-Control"=>join(", ",
1087 "pre-check=0", # MSIE
1088 "post-check=0", # MSIE
1090 header("Pragma"=>"no-cache"); # HTTP/1.0
1091 header("Vary"=>"*"); # content may ba based on unpredictable sources
1094 sub headers_in_filtered(@)
1098 return map(($_=>$W->{"headers_in"}{$_}),@keys);
1101 our %uri_args_frozen_to_headers_in_keys;
1102 our %uri_args_headers_in_frozen_to_headers_out;
1104 sub uri_args_headers_in_frozen_get($)
1106 my($headers_in_keys_arrayref)=@_;
1108 my %uri_args_headers_in_hash=(
1109 "uri_args_frozen"=>$W->{"uri_args_frozen"},
1110 "headers_in"=>{ headers_in_filtered(@$headers_in_keys_arrayref) },
1112 return do { local $Storable::canonical=1; Storable::freeze(\%uri_args_headers_in_hash); };
1115 sub cache_output_filter($)
1119 while ($f->read(my $text,0x400)) {
1120 cluck "utf-8 untested" if Encode::is_utf8($text); # Possible here at all?
1122 $W->{"digest-md5"}->add($text);
1129 # Used only if: !$W->{"http_safe"}
1130 # but we would cause on different method(): Appending to the '_done' package list
1131 # &Wrequire it here even if it will not be later used; to be stable!
1132 Wrequire 'My::Hash::RestrictTo';
1133 if (!$W->{"http_safe"}) {
1134 __PACKAGE__->_no_cache();
1140 "method"=>$W->{"r"}->method(),
1141 "uri"=>"http://".$W->{"web_hostname"}."/".$W->{"r"}->uri(),
1142 "args"=>$W->{"args_orig"},
1144 $W->{"uri_args_frozen"}=do { local $Storable::canonical=1; Storable::freeze(\%uri_args_hash); };
1145 last if !(my $headers_in_keys_arrayref=$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}});
1147 # Protection to be sure we are stable:
1148 $W->{"headers_in"}=My::Hash::RestrictTo->new($W->{"headers_in"},@$headers_in_keys_arrayref);
1150 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get($headers_in_keys_arrayref);
1151 last if !(my $headers_out_hashref=$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}});
1152 header(%$headers_out_hashref);
1155 # &meets_conditions will always deny the attempt if !2xx status().
1156 # At least ap_read_request() sets: r->status=HTTP_REQUEST_TIME_OUT; /* Until we get a request */
1157 my $status_old=$W->{"r"}->status();
1158 $W->{"r"}->status(HTTP_OK);
1159 # Update httpd's 'r->mtime' as the header "Last-Modified" is just not enough for ap_meets_conditions():
1160 # &update_mtime() argument is really in _secs_, not in _msecs_ as the docs claim.
1161 # Be aware '*1000000' would overflow Perl integer anyway.
1162 # &set_last_modified would also override the "Last-Modified" headers_out!
1163 # &mtime may exist but somehow does not work.
1164 $W->{"r"}->update_mtime(HTTP::Date::str2time($headers_out_hashref->{"Last-Modified"}));
1165 $status=$W->{"r"}->meets_conditions();
1166 $W->{"r"}->status($status_old);
1168 last if OK==$status;
1169 $W->{"r"}->status($status);
1170 $W->{"header_only"}=1; # Inhibit &footer output.
1175 $W->{"digest-md5"}=Digest::MD5->new();
1176 $W->{"cache_active"}=1;
1177 $W->{"r"}->add_output_filter(\&cache_output_filter);
1180 sub cache_finish_last_modified()
1182 cluck "Not yet done now? W __PACKAGE__: ".$W->{"__PACKAGE__"}
1183 if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
1184 for my $package_orig (@{$packages_used_array{$W->{"__PACKAGE__"}}}) {
1185 local $_=$package_orig.".pm";
1187 path_abs_disk "/$_","register"=>1;
1190 for my $path_abs_disk (keys(%{$W->{"path_abs_disk_register"}})) {
1191 my $mtime=(stat $path_abs_disk)[9];
1192 do { cluck "No mtime for: $path_abs_disk"; next; } if !$mtime;
1193 $mtime_newest=$mtime if !$mtime_newest || $mtime_newest<$mtime;
1195 cluck "No mtime_newest found for the current W __PACKAGE__: ".$W->{"__PACKAGE__"}
1197 return HTTP::Date::time2str($mtime_newest);
1203 # Do not: return if !$W->{"uri_args_frozen"};
1204 # as we may have just gave 304 and 'exit;' without starting the caching.
1205 return if !$W->{"cache_active"};
1207 # Headers may not be complete in this case; not sure, just trying.
1208 return if $W->{"r"}->connection()->aborted();
1210 # Fill-in/check: %uri_args_frozen_to_headers_in_keys
1211 my $headers_in_keys_stored_arrayref_ref=\$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}};
1212 my @headers_in_keys=tied(%{$W->{"headers_in_RecordKeys"}})->accessed();
1213 if (!$$headers_in_keys_stored_arrayref_ref
1214 || !Data::Compare::Compare(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)) {
1215 cluck "Non-matching generated 'headers_in_keys' per 'uri_args_frozen' key:\n"
1216 .Dumper(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)
1217 if $$headers_in_keys_stored_arrayref_ref;
1218 # Build or possibly prevent such further warn dupes:
1219 $$headers_in_keys_stored_arrayref_ref=\@headers_in_keys;
1220 # Build or regenerate as obsoleted now:
1221 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get(\@headers_in_keys);
1224 # Prepare 'headers_out' for the future reusal:
1226 # Do not: $W->{"digest-md5"}->b64digest();
1227 # as it will not provide the trailing filling '='s.
1228 # RFC 1864 is not clear if they should be there but its sample provides them.
1229 # Do not try to provide canonical "\r\n" form of newlines as is said by RFC 1864.
1230 # RFC 2068 (HTTP/1.1) section 14.16 says the newlines should NOT be converted for HTTP.
1231 # ',""' to avoid breaking the headers by its default "\n".
1232 $headers_out{"Content-MD5"}=MIME::Base64::encode_base64($W->{"digest-md5"}->digest(),"");
1233 # In fact we could also use MD5 for ETag as if we know ETag we also know MD5.
1234 # But this way we do not need to calculate MD5 and we still can provide such ETag. So.
1235 # $W->{"r"}->set_etag() ?
1236 $headers_out{"ETag"}='"'.Digest::MD5::md5_base64($W->{"uri_args_headers_in_frozen"}).'"';
1237 # $W->{"r"}->set_content_length() ?
1238 $headers_out{"Content-Length"}=$W->{"r"}->bytes_sent();
1239 my %Vary=map(($_=>1),(@headers_in_keys));
1245 %Vary=("*"=>1) if $Vary{"*"};
1246 $headers_out{"Vary"}=join(", ",sort keys(%Vary)) if keys(%Vary);
1247 # $W->{"r"}->set_last_modified() ?
1248 $headers_out{"Last-Modified"}=cache_finish_last_modified();
1250 # Fill-in/check: %uri_args_headers_in_frozen_to_headers_out
1251 my $headers_out_stored_hashref_ref=\$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}};
1252 if (!$$headers_out_stored_hashref_ref
1253 || !Data::Compare::Compare(\%headers_out,$$headers_out_stored_hashref_ref)) {
1254 cluck "Non-matching generated 'headers_out' per 'uri_args_headers_in_frozen' key:\n"
1255 .Dumper(\%headers_out,$$headers_out_stored_hashref_ref)
1256 if $$headers_out_stored_hashref_ref;
1257 # Build or possibly prevent such further warn dupes:
1258 $$headers_out_stored_hashref_ref=\%headers_out;
1261 ###print STDERR Dumper(\%uri_args_frozen_to_headers_in_keys,\%uri_args_headers_in_frozen_to_headers_out);
1268 if (!$W->{"header_only"}) {
1269 header("Content-Style-Type"=>"text/css");
1270 # Do not: text/javascript
1271 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
1272 # "application/javascript" so far standardized till 2005-12-08 by:
1273 # http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
1274 header("Content-Script-Type"=>"application/javascript");
1275 # $W->{"r"}->content_languages() ?
1276 do { header("Content-Language"=>$_) if $_; } for $W->{"language"};
1278 # TODO: Support also: private
1279 header("Cache-Control"=>"public"); # HTTP/1.1
1281 # Use $W->{"charset"}=0 to disable charset.
1282 $W->{"charset"}="us-ascii"
1283 if !defined $W->{"charset"} && (!defined($W->{"content_type"}) || $W->{"content_type"});
1286 # https://bugzilla.mozilla.org/show_bug.cgi?id=120556
1288 # Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1289 # http://validator.w3.org/ does not send ANY "Accept" headers!
1290 if (!defined $W->{"content_type"}) {
1291 # Be _stable_ for "headers_in".
1292 my $accept=$W->{"headers_in"}{"Accept"};
1293 my $user_agent=$W->{"headers_in"}{"User-Agent"}||"";
1294 $W->{"content_type"}="application/xhtml+xml"
1295 if !$accept && $user_agent=~m{^W3C_Validator/}i;
1297 my $negotiated=$class->Negotiate_choose([
1298 # Put the fallback variant as the first one.
1299 # Rate both variants the same to prefer "text/html" for undecided clients.
1301 # Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1302 # prefers "application/xhtml+xml" over "text/html" itself:
1303 # text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5
1306 "content-type"=>"text/html",
1308 (!$W->{"charset"} ? () : "charset"=>$W->{"charset"}),
1309 "lang"=>$W->{"language"},
1312 "id"=>"application/xhtml+xml",
1313 "content-type"=>"application/xhtml+xml",
1315 (!$W->{"charset"} ? () : "charset"=>$W->{"charset"}),
1316 "lang"=>$W->{"language"},
1321 $W->{"content_type"}=$negotiated if !defined $W->{"content_type"};
1323 # mod_perl doc: If you set this header via the headers_out table directly, it
1324 # will be ignored by Apache. So do not do that.
1326 if ($W->{"content_type"}) {
1327 $type=MIME::Types->new()->type($W->{"content_type"});
1328 cluck "MIME::Types type '".$W->{"content_type"}."' not known" if !$type;
1330 cluck "charset='".$W->{"charset"}."' does not match content-type='".$W->{"content_type"}."'"
1331 if ($W->{"charset"} ? 1 : 0) != (!$type ? 0 : $type->isAscii());
1332 $W->{"r"}->content_type($W->{"content_type"}.(!$W->{"charset"} ? "" : "; charset=".$W->{"charset"}))
1333 if $W->{"content_type"};
1336 # We still can append headers before we put out some text.
1337 # FIXME: It is not clean to still append them without overwriting.
1338 return if $W->{"heading_done"};
1339 Wprint '<?xml version="1.0" encoding="'.$W->{"charset"}.'"?>'."\n"
1340 if (!$W->{"header_only"} || $W->{"header_only"} eq "xml") && (0
1341 || $W->{"content_type"}=~m{^application/\w+[+]xml$}
1342 || $W->{"content_type"} eq "text/vnd.wap.wml");
1343 return if $W->{"header_only"};
1344 # Split 'heading_done' for the proper handling of: /project/Rel.pm
1345 $W->{"heading_done"}++;
1347 Wprint '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">'."\n";
1348 Wprint '<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="'.$W->{"language"}.'">'."\n";
1349 my $title=$W->{"title_prefix"}.join("",map({ ': '.$_; } ($W->{"title"} || ())));
1350 # Do not: cluck if $title=~/[<>]/;
1351 # as it is not solved just by: &a_href_inhibit
1352 # as sometimes titles use also: <i>...</i>
1353 $title=~s#<[^>]*>##g;
1355 Wprint "<title>$title</title>\n";
1356 if ($W->{"have_css"}) {
1357 # Everything can get overriden later.
1358 for my $css ("/My/Web.css",@{$W->{"css_push"}}) {
1360 <link rel="stylesheet" type="text/css" href="@{[ uri_escaped(path_web $css) ]}" />
1363 if ($W->{"css_inherit"}) {
1364 # Do not: <script />
1365 # as at least Lynx inhibits any further HTML output.
1366 # Do not: text/javascript
1367 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
1368 # "application/javascript" so far standardized till 2005-12-08 by:
1369 # http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
1371 <script type="application/javascript" src="@{[ uri_escaped(path_web('/My/css_inherit.js')) ]}"></script>
1375 Wprint '<meta name="robots" content="'.($W->{"indexme"} ? "" : "no" ).'index,follow" />'."\n";
1376 Wprint $W->{"head"};
1377 for my $type (qw(prev next index contents start up)) {
1378 do { Wprint '<link rel="'.$type.'" href="'.uri_escaped(path_web $_).'" />'."\n" if $_; }
1379 for ($W->{"rel_$type"});
1381 Wprint "</head><body";
1382 # Wprint ' bgcolor="black" text="white" link="aqua" vlink="teal"'
1383 # if $W->{"browser"}->netscape() && (!$W->{"browser"}->major() || $W->{"browser"}->major()<=4);
1384 Wprint $W->{"body_attr"};
1387 do { Wprint $_ if $_; } for $W->{"heading"};
1391 delete $W->{"__My::Web_init"};