Web.pm

   1 # $Id$
   2 # Common functions for HTML/XHTML output generation
   3 # Copyright (C) 2003-2005 Jan Kratochvil <project-www.jankratochvil.net@jankratochvil.net>
   4 #
   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
   8 #
   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.
  13 #
  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
  17
  18
  19 package My::Web;
  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; };
  22 our $CVS_ID=q$Id$;
  23 use strict;
  24 use warnings;
  25
  26 use Exporter;
  27 sub Wrequire($);
  28 sub Wuse($@);
  29 our $W;
  30 our @EXPORT=qw(
  31                 &Wrequire &Wuse
  32                 &path_web &path_abs_disk
  33                 &uri_escaped
  34                 &a_href &a_href_cc
  35                 &vskip
  36                 &img &centerimg &rightimg
  37                 $W
  38                 &input_hidden_persistents
  39                 &escapeHTML
  40                 &form_method
  41                 );
  42 our @ISA=qw(Tie::Handle Exporter);
  43
  44 my %packages_used_hash; # $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
  45 my %packages_used_array;
  46
  47 BEGIN
  48 {
  49         use Carp qw(cluck confess);
  50         $W->{"__My::Web_init"}=1;
  51
  52         sub Wrequire ($)
  53         {
  54         my($file)=@_;
  55
  56 #               print STDERR "Wrequire $file\n";
  57                 $file=~s#/#::#g;
  58                 $file=~s/[.]pm$//;
  59                 my $class=$file;
  60                 $file=~s#::#/#g;
  61                 $file.=".pm";
  62                 my %callers;
  63                 for (my $depth=0;defined caller($depth);$depth++) {
  64                         $callers{caller($depth)}=1;
  65                         }
  66                 my $selfpkg=__PACKAGE__;
  67                 $callers{$selfpkg}=1;
  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}},$target;
  74                                 }
  75                         }
  76                 eval { CORE::require "$file"; } or confess $@;
  77                 1;      # Otherwise 'require' would already file above.
  78         }
  79
  80         sub Wuse ($@)
  81         {
  82         my($file,@list)=@_;
  83
  84 #               print STDERR "Wuse $file\n";
  85                 Wrequire $file;
  86                 local $Exporter::ExportLevel=$Exporter::ExportLevel+1;
  87                 $file->import(@list);
  88                 1;
  89         }
  90
  91         sub import
  92         {
  93         my($class,@rest)=@_;
  94
  95                 local $Exporter::ExportLevel=$Exporter::ExportLevel+1;
  96                 Wrequire("$class");
  97                 return $class->SUPER::import(@rest);
  98         }
  99 }
 100
 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);
 105 use URI::Escape;
 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
 111 # I do not know why.
 112 use POSIX qw(strftime);
 113 use Tie::Handle;
 114 use Apache2::Const qw(HTTP_MOVED_TEMPORARILY OK HTTP_OK);
 115 use URI;
 116 use URI::QueryParam;
 117 use Cwd;
 118 require HTTP::Date;
 119 require Storable;
 120 require Digest::MD5;
 121 require Data::Compare;
 122 use Data::Dumper;
 123 require Encode;
 124 use Apache2::RequestUtil;
 125 use Apache2::Filter;
 126 use Apache2::Connection;
 127 require MIME::Base64;
 128 use Apache2::ServerUtil;
 129 require MIME::Types;
 130 require MIME::Parser;
 131 use Apache2::RequestRec;
 132 use Apache2::RequestIO;
 133 use Apache2::Response;
 134
 135
 136 #our $W;
 137
 138 sub init ($%)
 139 {
 140 my($class,%args)=@_;
 141
 142         print STDERR "$class->init ".Apache2::RequestUtil->request()->unparsed_uri()."\n";
 143
 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.
 147         Wuse 'WebConfig';
 148         Wrequire 'My::Hash';
 149
 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");
 155         bless $W,$class;
 156         %$W=(
 157                 "__PACKAGE__"=>scalar(caller()),
 158                 %WebConfig,
 159                 %args,  # override %WebConfig settings
 160                 );
 161
 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__";
 165         exit_hook_start();
 166
 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";
 178
 179         my $footer_any=0;
 180         for (qw(footer_ids)) {
 181                 $W->{$_}=0 if !$W->{"footer"};
 182                 $footer_any=1 if $W->{$_};
 183                 }
 184         $W->{"footer"}=0 if !$footer_any;
 185         $W->{"footer_delimit"}=0 if !$W->{"footer"};
 186
 187         $W->{"r"}=Apache2::RequestUtil->request();
 188
 189         $W->{"r"}->push_handlers("PerlCleanupHandler"=>\&cleanup);
 190
 191         $W->{"web_hostname"}||=$W->{"r"}->hostname();
 192
 193         tie *STDOUT,$W->{"r"};
 194         select *STDOUT;
 195         $|=1;
 196
 197         $W->{"headers_in"}=$W->{"r"}->headers_in();
 198         Wrequire 'My::Hash::Merge';
 199         $W->{"headers_in"}=My::Hash::Merge->new(
 200                         $W->{"headers_in"},
 201                         My::Hash::Sub->new({
 202                                 "_remote_ip"=>sub { return $W->{"r"}->connection()->remote_ip(); },
 203                                 }),
 204                         );
 205         Wrequire 'My::Hash::Readonly';
 206         $W->{"headers_in"}=My::Hash::Readonly->new($W->{"headers_in"});
 207
 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                         $$_=1 if !defined $$_;
 215                         }
 216                 }
 217         else {
 218                 for (\$W->{"http_safe"}) {
 219                         cluck "Undefined HTTP-Safe-ty for method \"".$W->{"r"}->method()."\"!"
 220                                         if !defined($$_);
 221                         $$_=0 if !defined $$_;
 222                         }
 223                 }
 224         if ($W->{"http_safe"}) {
 225                 Wrequire 'My::Hash::RecordKeys';
 226                 $W->{"headers_in_RecordKeys"}=My::Hash::RecordKeys->new($W->{"headers_in"});
 227                 $W->{"headers_in"}=$W->{"headers_in_RecordKeys"};
 228                 }
 229
 230         {
 231                 local $_=$W->{"r"}->args() || "";
 232                 if ($W->{"detect_ent"}) {
 233                                  if (/[&]amp;have_ent/)
 234                                 { $W->{"have_ent"}=0; }
 235                         elsif (    /[&]have_ent/)
 236                                 { $W->{"have_ent"}=1; }
 237                         else
 238                                 { delete $W->{"have_ent"}; }
 239                         if (!defined $W->{"have_ent"} && $W->{"r"}->method() eq "GET") {
 240                                 $W->{"head"}.='<meta http-equiv="Refresh" content="0; URL='
 241                                                 .escapeHTML("http://".$W->{"web_hostname"}."/".($W->{"r"}->uri()=~m#^/*(.*)$#)[0]
 242                                                                 ."?".($_ || "detect_ent_glue=1").'&have_ent=detect')
 243                                                 .'" />'."\n";
 244                                 }
 245                         }
 246                 s/([&])amp;/$1/g;
 247                 $W->{"r"}->args($_);
 248                 }
 249
 250         $W->{"args"}=URI->new("?".$W->{"r"}->args())->query_form_hash();
 251         $W->merge_post_args() if $W->{"r"}->method() eq "POST";
 252         # Prepare '$args' first to (FIXME: Why?) prevent: Not a reference
 253         my $args=$W->{"args"};
 254         $W->{"args_orig"}=Storable::dclone($args);
 255
 256         $W->{"browser"}=sub {
 257                 # Lazy-evaluation, we may not need the "User-Agent" header at all.
 258                 return our $r||=HTTP::BrowserDetect->new($W->{"headers_in"}{"User-Agent"});
 259                 };
 260
 261         if (!defined $W->{"have_style"}) {
 262                 $W->{"have_style"}=sub {
 263                         # Lazy-evaluation, we may not need the "User-Agent" header at all.
 264                         return our $r||=(!$W->{"browser"}->netscape() || ($W->{"browser"}->major() && $W->{"browser"}->major()>4) ? 1 : 0);
 265                         };
 266                 }
 267
 268         $W->{"have_js"}=($W->{"args"}{"have_js"} ? 1 : 0);
 269         if ($W->{"detect_js"} && !$W->{"have_js"}) {
 270                 # Do not: <script />
 271                 # as at least Lynx inhibits any further HTML output.
 272                 # Do not: text/javascript
 273                 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
 274                 # "application/javascript" so far standardized till 2005-12-08 by:
 275                 #       http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
 276                 $W->{"head"}.='<script type="application/javascript" src="'.uri_escaped(path_web('/My/HaveJS.pm')).'"></script>'."\n";
 277                 }
 278
 279         do { _args_check(%$_) if $_; } for ($W->{"args_check"});
 280
 281         $W->{"_init_done"}=1;
 282         return $W;
 283 }
 284
 285 sub form_method($)
 286 {
 287 my($method)=@_;
 288
 289         return q{enctype="application/x-www-form-urlencoded" accept-charset="us-ascii utf-8"} if $method eq "post";
 290         return                                             q{accept-charset="us-ascii utf-8"} if $method eq "get";
 291         cluck "Undefined method: $method";
 292         return ""
 293 }
 294
 295 sub merge_post_args($)
 296 {
 297 my($class)=@_;
 298
 299         my @post_args=$class->read_post_args();
 300         while (@post_args) {
 301                 my $name=shift @post_args;
 302                 my $data=shift @post_args;
 303                 my $ref=\$W->{"args"}{$name};
 304                    if (!defined $$ref)       { $$ref=$data; }
 305                 elsif (!ref $$ref)           { $$ref=[$$ref,$data]; }
 306                 elsif ("ARRAY" eq ref $$ref) { push @$$ref,$data; }
 307                 else {
 308                         cluck "Ignoring POST argument \"$name\", orig is weird:\n",Dumper($$ref);
 309                         }
 310                 }
 311         return;
 312 }
 313
 314 # Do not: use CGI;
 315 # as CGI parsing of POST vs. QUERY_STRING data, multiple-valued keys etc.
 316 # is too dense and causes weird problems, together with mod_perl etc.
 317 sub read_post_args($)
 318 {
 319 my($class)=@_;
 320
 321         local $_=$class->http_headers_in_for("Content-type")->content_type();
 322         return $class->read_multipart_form_data() if $_ eq "multipart/form-data";
 323         return $class->read_application_x_www_form_urlencoded() if $_ eq "application/x-www-form-urlencoded";
 324         cluck "Unknown POST data body, ignored: $_";
 325         return;
 326 }
 327
 328 sub read_application_x_www_form_urlencoded($)
 329 {
 330 my($class)=@_;
 331
 332         my $body="";
 333         for (;;) {
 334                 my $got=$W->{"r"}->read(my($buf),0x1000);
 335                 # Do not: cluck "Error reading POST data: $!" if !defined $got;
 336                 # as it should be done using: APR::Error exceptions
 337                 last if !$got;
 338                 $body.=$buf;
 339                 }
 340         return URI->new("?".$body)->query_form();
 341 }
 342
 343 sub read_multipart_form_data($)
 344 {
 345 my($class)=@_;
 346
 347         my $parser=MIME::Parser->new();
 348         # FIXME: No unlink()s done!
 349         $parser->output_under("/tmp");
 350
 351         local *R_FH;
 352         tie *R_FH,$W->{"r"};
 353         local *FH;
 354         tie *FH,"My::Web::ReadMerged",
 355                         join("",map(($_.": ".$W->{"headers_in"}{$_}."\n"),qw(
 356                                         Content-type
 357                                         )))."\n",
 358                         \*R_FH;
 359         my $body=$parser->parse(\*FH);
 360         cluck "No multipart POST request body?" if !$body->is_multipart();
 361
 362         return map((
 363                         $_->head()->mime_attr("content-disposition.name")
 364                         =>
 365                         join("",@{$_->body()})
 366                         ),$body->parts());
 367
 368         # TODO: Globalize, make it IO::* compatible, split to the merging part + IO::Scalar.
 369         package My::Web::ReadMerged;
 370
 371         require Tie::Handle;
 372         require Exporter;
 373         our @ISA=qw(Tie::Handle Exporter);
 374         use Carp qw(cluck confess);
 375
 376         sub READLINE($)
 377         {
 378         my($self)=@_;
 379
 380                 confess "Slurp not yet implemented" if !defined $/;
 381                 # Apache2::RequestIO does not support 'READLINE'!
 382                 for (;;) {
 383                         if (defined $self->{"data"} && $self->{"data"}=~s{^.*\Q$/\E}{}) {
 384                                 $self->{"offset"}+=length $&;
 385                                 return $&;
 386                                 }
 387                         my $fh_orig=$self->{"fh_orig"};
 388                         if (!$fh_orig) {
 389                                 my $r=$self->{"data"};
 390                                 delete $self->{"data"};
 391                                 $self->{"offset"}+=length $r if defined $r;
 392                                 return $r;
 393                                 }
 394                         my $got=read $fh_orig,my($buf),0x1000;
 395                         cluck "Error reading POST data: $!" if !defined $got;
 396                         delete $self->{"fh_orig"} if !$got;
 397                         cluck "INTERNAL: fh_orig should not exist here" if !defined $self->{"data"};
 398                         $self->{"data"}.=$buf;
 399                         }
 400         }
 401
 402         sub TELL($)
 403         {
 404         my($self)=@_;
 405
 406                 return $self->{"offset"};
 407         }
 408
 409         sub TIEHANDLE($$$)
 410         {
 411         my($class,$data,$fh_orig)=@_;
 412
 413                 my $self=bless {},$class;
 414                 $self->{"data"}=$data;
 415                 $self->{"offset"}=0;
 416                 $self->{"fh_orig"}=$fh_orig;
 417                 return $self;
 418         }
 419 }
 420
 421 sub cleanup($)
 422 {
 423 my($apache_request)=@_;
 424
 425         cluck "CORE::GLOBAL::exit hook not ran" if !$W->{"_exit_done"};
 426         cluck "packages not finalized" if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
 427         cache_finish();
 428         # Sanity protection.
 429         $W=undef();
 430         exit_hook_stop();
 431         return OK;
 432 }
 433
 434 # PerlResponseHandler is RUN_FIRST and &ModPerl::Util::exit returns OK, so no (sane) go.
 435 # PerlLogHandler is already too late to be able to produce any output.
 436 my $exit_orig;
 437 sub exit_hook
 438 {
 439         cluck "Missing ->init while in exit_hook()" if !$W->{"_init_done"};
 440         # &footer will call us recursively!
 441         footer() if !$W->{"_exit_done"}++;
 442         return &{$exit_orig}(@_);
 443 }
 444 sub exit_hook_start
 445 {
 446         do { cluck "exit_hook_start() twice?"; return; } if defined $exit_orig;
 447         $exit_orig=\&CORE::GLOBAL::exit;
 448         # Prevent: Subroutine CORE::GLOBAL::exit redefined
 449         no warnings 'redefine';
 450         *CORE::GLOBAL::exit=\&exit_hook;
 451 }
 452 sub exit_hook_stop
 453 {
 454         do { cluck "exit_hook_stop() without exit_hook_start()?"; return; }
 455                         if \&exit_hook ne \&CORE::GLOBAL::exit;
 456         do { cluck "INTERNAL: exit_orig uninitialized"; return; }
 457                         if !$exit_orig;
 458         # Prevent: Subroutine CORE::GLOBAL::exit redefined
 459         no warnings 'redefine';
 460         *CORE::GLOBAL::exit=$exit_orig;
 461         $exit_orig=undef();
 462 }
 463
 464 # Be aware other parts of code (non-My::Web) will NOT use this function!
 465 # Do not: Wprint $W->{"heading"},"undef"=>1;
 466 # as we would need to undef() it to turn it off and it would get defaulted in such case.
 467 # Do not: exists $W->{"heading"}
 468 # as we use a lot of 'for $W->{"heading"}' which instantiates it with the value: undef()
 469 sub Wprint($%)
 470 {
 471 my($text,%args)=@_;
 472
 473         cluck "undef Wprint" if !defined $text && !$args{"undef"};
 474         delete $args{"undef"};
 475         cluck join(" ","Invalid arguments:",keys(%args)) if keys(%args);
 476         return if !defined $text;
 477         # Do not: cluck "utf-8 untested" if Encode::is_utf8($text);
 478         # as it is valid here.
 479         $W->{"r"}->puts($text);
 480 }
 481
 482 sub request_check(;$)
 483 {
 484 my($self)=@_;
 485
 486         # Use &eval to prevent: Global $r object is not available. Set:\n\tPerlOptions +GlobalRequest\nin ...
 487         confess "Calling sensitive dynamic code from a static code" if !eval { Apache2::RequestUtil->request(); };
 488         # Do not: confess "Calling sensitive dynamic code without My::Web::init" if !$W->{"__PACKAGE__"};
 489         # as it is valid at least while preparing arguments to call: &project::Lib::init
 490 }
 491
 492 # Do not: use CGI;
 493 # as it is too much backward compatible regarding the charset encodings etc.
 494 # and the resulting code is too dense with no additional functionality for the recent content.
 495 sub escapeHTML($)
 496 {
 497 my($text)=@_;
 498
 499         local $_=$text;
 500         s{&}{&amp;}gso;
 501         s{<}{&lt;}gso;
 502         s{>}{&gt;}gso;
 503         s{"}{&quot;}gso;
 504         return $_;
 505 }
 506
 507 # /home/user/www/webdir
 508 sub dir_top_abs_disk()
 509 {
 510         our $dir_top_abs_disk;
 511         if (!$dir_top_abs_disk) {
 512                 my $selfpkg_relpath=__PACKAGE__;
 513                 $selfpkg_relpath=~s{::}{/}g;
 514                 $selfpkg_relpath.=".pm";
 515                 my $selfpkg_abspath=$INC{$selfpkg_relpath} or do {
 516                         cluck "Unable to find self package $selfpkg_relpath";
 517                         return;
 518                         };
 519                 $selfpkg_abspath=~s{/*\Q$selfpkg_relpath\E$}{} or do {
 520                         cluck "Unable to strip myself \"$selfpkg_relpath\" from the abspath: $selfpkg_abspath";
 521                         return;
 522                         };
 523                 cluck "INC{myself} is relative?: $selfpkg_abspath" if $selfpkg_abspath!~m{^/};
 524                 $dir_top_abs_disk=$selfpkg_abspath;
 525                 }
 526         return $dir_top_abs_disk;
 527 }
 528
 529 sub unparsed_uri()
 530 {
 531         request_check();
 532         if (!$W->{"unparsed_uri"}) {
 533                 # Do not: $W->{"r"}
 534                 # as we may be called before &init from: &My::Project::init
 535                 my $r=Apache2::RequestUtil->request();
 536                 cluck "Calling ".'&unparsed_uri'." from a static code, going to fail" if !$r;
 537                 my $uri_string=$r->unparsed_uri() or cluck "Valid 'r' missing unparsed_uri()?";
 538                 my $uri=URI->new_abs($uri_string,"http://".$W->{"web_hostname"}."/");
 539                 $W->{"unparsed_uri"}=$uri;
 540                 }
 541         return $W->{"unparsed_uri"};
 542 }
 543
 544 sub in_to_uri_abs($)
 545 {
 546 my($in)=@_;
 547
 548         # Otherwise we may have been already processed and thus legally relativized.
 549         # FIXME data: Currently disabled, all the data are too violating such rule.
 550         if (0 && !ref $in) {
 551                 my $uri_check=URI->new($in);
 552                 $uri_check->scheme() || $in=~m{^\Q./\E} || $in=~m{^/}
 553                                 or cluck "Use './' or '/' prefix for all the local references: $in";
 554                 }
 555         my $uri=URI->new_abs($in,unparsed_uri());
 556         $uri=$uri->canonical();
 557         return $uri;
 558 }
 559
 560 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
 561 # $args{"abs"}=1;
 562 sub path_web($%)
 563 {
 564 my($in,%args)=@_;
 565
 566         cluck if !$args{"uri_as_in"} && ref $in;
 567         my $uri=in_to_uri_abs($in);
 568         if (uri_is_local($uri)) {
 569                 # Prefer the $uri values over "args_persistent" values.
 570                 # &query_form_hash comes from: URI::QueryParam
 571                 $uri->query_form_hash({
 572                                 map({
 573                                         my $key=$_;
 574                                         my $val=$W->{"args"}{$key};
 575                                         (!defined $val ? () : ($key=>$val));
 576                                         } keys(%{$W->{"args_persistent"}})),
 577                                 %{$uri->query_form_hash()},
 578                                 });
 579                 }
 580         return $uri->abs(unparsed_uri()) if $W->{"args"}{"Wabs"} || $args{"abs"};
 581         return $uri->rel(unparsed_uri());
 582 }
 583
 584 sub path_abs_disk_register($)
 585 {
 586 my($path_abs_disk)=@_;
 587
 588         $W->{"path_abs_disk_register"}{$path_abs_disk}=1;
 589 }
 590
 591 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
 592 sub path_abs_disk($%)
 593 {
 594 my($in,%args)=@_;
 595
 596         cluck if !$args{"uri_as_in"} && ref $in;
 597         my $uri=in_to_uri_abs($in);
 598         cluck if !uri_is_local($uri);
 599         my $path=$uri->path();
 600         cluck "URI compatibility: ->path() not w/leading slash of URI \"$uri\"; path: $path" if $path!~m{^/};
 601         my $r=dir_top_abs_disk().$path;
 602         path_abs_disk_register $r if !defined $args{"register"} || $args{"register"};
 603         return $r;
 604 }
 605
 606 sub fatal (;$);
 607
 608 sub _args_check (%)
 609 {
 610 my(%tmpl)=@_;
 611
 612         while (my($name,$regex)=each(%tmpl)) {
 613                 my $name_html="Parameter <span class=\"quote\">".escapeHTML($name)."</span>";
 614                 $W->{"args"}{$name}="" if !defined $W->{"args"}{$name};
 615                 $W->{"args"}{$name}=[ $W->{"args"}{$name} ] if !ref $W->{"args"}{$name} && ref $regex;
 616                 fatal "$name_html passed as multivar although singlevar expected"
 617                                 if ref $W->{"args"}{$name} && !ref $regex;
 618                 $regex=$regex->[0] if ref $regex;
 619                 for my $val (!ref $W->{"args"}{$name} ? $W->{"args"}{$name} : @{$W->{"args"}{$name}}) {
 620                         $val="" if !defined $val;
 621                         fatal "$name_html <span class=\"quote\">".escapeHTML($val)."</span>"
 622                                                         ." does not match the required regex <span class=\"quote\">".escapeHTML($regex)."</span> "
 623                                         if $regex ne "" && $val!~/$regex/;
 624                         }
 625                 }
 626 }
 627
 628 sub vskip (;$)
 629 {
 630 my($height)=@_;
 631
 632         return '<p'.(!defined $height ? "" : ' style="height: '.$height.';"').'>&nbsp;</p>'."\n";
 633 }
 634
 635 sub fatal (;$)
 636 {
 637 my($msg)=@_;
 638
 639         $msg="UNKNOWN" if !$msg;
 640         cluck "FATAL: $msg";
 641
 642         # Do not send it unconditionally.
 643         # The intial duplicated '<?xml...' crashes Gecko parser.
 644         $W->{"heading_done"}=0 if $W->{"header_only"};
 645         # Do not send it unconditionally.
 646         # Prevents warn: Headers already sent
 647         if (!$W->{"heading_done"}) {
 648                 $W->{"indexme"}=0;      # For the case no heading was sent yet.
 649                 $W->{"header_only"}=0;  # assurance for &heading
 650                 My::Web->heading();
 651                 }
 652         Wprint "\n".vskip("3ex")."<hr /><h1 class=\"error\">FATAL ERROR: $msg!</h1>\n"
 653                         ."<p>You can report this problem's details to"
 654                         ." ".a_href("mailto:".$W->{"admin_mail"},"admin of this website").".</p>\n";
 655         exit;
 656 }
 657
 658 sub footer_packages_used_comments()
 659 {
 660         my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
 661         for my $package (@$packages_used) {
 662                 my $cvs_id=(eval('$'.$package."::CVS_ID")
 663 #                               || $package     # debug
 664                                 );
 665                 Wprint '<!-- '.$package.' - $'.$cvs_id.'$ -->'."\n" if $cvs_id;
 666                 }
 667 }
 668
 669 sub footer()
 670 {
 671         cluck 'Explicit &footer call is deprecated, !_exit_dne' if !$W->{"_exit_done"};
 672         exit if $W->{"footer_done"}++;  # deadlock prevention:
 673         &{$_}() for reverse @{$W->{"footer_sub_push"}};
 674         if ($W->{"header_only"}) {
 675                 $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
 676                 exit;
 677                 }
 678
 679         Wprint vskip if $W->{"footer_delimit"};
 680         &{$_}() for reverse @{$W->{"footing_delimit_sub_push"}};
 681         Wprint "<hr />\n" if $W->{"footer"};
 682
 683         # Never update the package list while we examine it!
 684         $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
 685
 686         my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
 687         if ($W->{"footer_ids"}) {
 688                 Wprint '<p class="cvs-id">';
 689                 Wprint join("<br />\n",map({ my $package=$_;
 690                         my $cvs_id=(eval('$'.$package."::CVS_ID")
 691 #                                       || $package     # debug
 692                                         );
 693                         if (!$cvs_id) {
 694                                 ();
 695                                 }
 696                         else {
 697                                 $cvs_id='$'.$cvs_id.'$';        # Eaten by 'q' operator.
 698                                 my @cvs_id_split=split / +/,$cvs_id;
 699                                 if (@cvs_id_split==8) {
 700                                         my $file=$package;
 701                                         $file=~s#::#/#g;
 702                                         my $ext;
 703                                         my @tried;
 704                                         for (qw(.pm)) {
 705                                                 $ext=$_;
 706                                                 my $path_abs_disk=path_abs_disk("/$file$ext");
 707                                                 push @tried,$path_abs_disk;
 708                                                 last if -r $path_abs_disk;
 709                                                 cluck "Class file $file not found; tried: ".join(" ",@tried) if !$ext;
 710                                                 }
 711                                         $file.=$ext;
 712                                         my $viewcvs;
 713                                         if ((my $file_cvs=$file)=~s{^My/}{}) {
 714                                                 $viewcvs=$W->{"viewcvs_My"}.$file_cvs;
 715                                                 }
 716                                         else {
 717                                                 $viewcvs=$W->{"viewcvs"}.$file;
 718                                                 }
 719                                         $cvs_id_split[2]=""
 720                                                         .a_href((map({ my $s=$_; $s=~s#/viewcvs/#$&~checkout~/#; $s; } $viewcvs))[0]."?rev=".$cvs_id_split[2],
 721                                                                         $cvs_id_split[2]);
 722                                         $cvs_id_split[1]=a_href($viewcvs,($package!~/^Apache2::/ ? $package : $cvs_id_split[1]));
 723                                         $cvs_id_split[5]=&{$W->{"cvs_id_author_sub"}}($cvs_id_split[5]);
 724                                         }
 725                                 join " ",@cvs_id_split;
 726                                 }
 727                         } @$packages_used));
 728                 Wprint "</p>\n";
 729                 }
 730
 731         footer_packages_used_comments();
 732
 733         do { Wprint $_ if $_; } for $W->{"footing"};
 734
 735         Wprint "</body></html>\n";
 736         exit;
 737 }
 738
 739 # Existing entries are overwritten.
 740 sub header(%)
 741 {
 742 my(%pairs)=@_;
 743
 744         while (my($key,$val)=each(%pairs)) {
 745                 do { cluck "Headers already sent"; next; } if $W->{"heading_done"};
 746                 $W->{"r"}->headers_out()->set($key,$val);
 747                 }
 748 }
 749
 750 sub size_display ($)
 751 {
 752 my($size)=@_;
 753
 754            if ($size<4096)
 755                 {}
 756         elsif ($size<1024*1024)
 757                 { $size=sprintf "%.1fK",$size/1024; }
 758         else
 759                 { $size=sprintf "%.1fM",$size/1024/1024; }
 760         $size.="B";
 761         return $size;
 762 }
 763
 764 sub uri_is_local($)
 765 {
 766 my($in)=@_;
 767
 768         my $uri_rel=in_to_uri_abs($in)->rel(unparsed_uri());
 769         # Do not: defined $uri_rel->("userinfo"|"host"|"port")();
 770         # as they fail to be called for schemes not supporting them.
 771         return 0 if $uri_rel->scheme();
 772         return 0 if $uri_rel->authority();
 773         return 1;
 774 }
 775
 776 # &path_web still may be required for &uri_escaped !
 777 sub uri_escaped($)
 778 {
 779 my($uri)=@_;
 780
 781         cluck if !ref $uri;
 782         my $urient=escapeHTML($uri);
 783         return $uri    if $uri eq $urient;
 784         request_check();
 785         return $urient if uri_is_local $uri;
 786         return $uri    if defined $W->{"have_ent"} && !$W->{"have_ent"};        # non-ent client
 787         return $urient if $W->{"have_ent"};     # ent client
 788         # Unknown client, &escapeHTML should not be needed here:
 789         return escapeHTML(path_web('/My/Redirect.pm?location='.uri_escape($uri->abs(unparsed_uri()))));
 790 }
 791
 792 our $a_href_inhibited;
 793 sub a_href($;$%)
 794 {
 795 my($in,$contents,%args)=@_;
 796
 797         request_check();
 798         do { $$_=1 if !defined $$_; } for (\$args{"size"});
 799         if (!defined $contents) {
 800                 $contents=$in;
 801                 $contents=File::Basename::basename($contents) if $args{"basename"};
 802                 $contents=escapeHTML($contents);
 803                 }
 804         $contents=~s#<a\b[^>]*>##gi;
 805         $contents=~s#</a>##gi;
 806         return $contents if $a_href_inhibited;
 807
 808         my $path_web=path_web $in,%args;
 809         my $r="";
 810         $r.='<a href="';
 811         $r.=uri_escaped $path_web;
 812         $r.='"';
 813         do { $r.=" $_" if $_; } for ($args{"attr"});
 814         $r.='>'.$contents.'</a>';
 815         if ($args{"size"} && uri_is_local($in) && ($args{"size"}>=2 || $in=~/[.](?:gz|Z|rpm|zip|deb|lha)/)) {   # Downloadable?
 816                 my $path_abs_disk=path_abs_disk $in,%args;
 817                 cluck "File not readable: $path_abs_disk" if !-r $path_abs_disk;
 818                 $r.='&nbsp;('.size_display((stat($path_abs_disk))[7]).')';
 819                 }
 820         return $r;
 821 }
 822
 823 sub a_href_inhibit($$;@)
 824 {
 825 my($self,$sub,@sub_args)=@_;
 826
 827         local $a_href_inhibited=1;
 828         return &{$sub}(@sub_args);
 829 }
 830
 831 sub input_hidden_persistents()
 832 {
 833         request_check();
 834         return join("",map({
 835                 my $key=$_;
 836                 my $val=$W->{"args"}{$key};
 837                 (!defined $val ? () : '<input type="hidden"'
 838                                 .' name="'.escapeHTML($key).'"'
 839                                 .' value="'.escapeHTML($val).'"'
 840                                 .' />'."\n");
 841                 } (keys(%{$W->{"args_persistent"}}))));
 842 }
 843
 844 sub http_moved($$;$)
 845 {
 846 my($self,$url,$status)=@_;
 847
 848         $url=path_web($url,"abs"=>1);
 849         $status||=HTTP_MOVED_TEMPORARILY;
 850         $W->{"r"}->status($status);
 851         $W->{"r"}->headers_out()->{"Location"}=$url;
 852         $W->{"header_only"}=1;
 853         $W->{"content_type"}=0;
 854         $W->{"charset"}=0;
 855         My::Web->heading();
 856         exit;
 857         die "NOTREACHED";
 858 }
 859
 860 sub remote_ip ()
 861 {
 862         # Do not: PerlModule                 Apache2::ForwardedFor
 863         #         PerlPostReadRequestHandler Apache2::ForwardedFor
 864         # As 'Apache2::ForwardedFor' takes the first of $ENV{"HTTP_X_FORWARDED_FOR"}
 865         # while the contents is '127.0.0.1, 213.220.195.171' if client has its own proxy.
 866         # We must take the last item ourselves.
 867         # Be VERY sure you always retrieve all the headers unconditionally to hit: My::Hash::RecordKeys
 868         my $x_forwarded_for=$W->{"headers_in"}{"X-Forwarded-For"};
 869         $x_forwarded_for=~s/^.*,\s*// if $x_forwarded_for;
 870         my $remote_ip=$W->{"headers_in"}{"_remote_ip"};
 871         my $r;
 872         $r||=$x_forwarded_for;
 873         $r||=$remote_ip;
 874         return $r;
 875 }
 876
 877 # $url={"JP"=>"http://specific",...};
 878 # $url={""=>"http://default",...};
 879 sub a_href_cc($$;%)
 880 {
 881 my($url,$contents,%args)=@_;
 882
 883         # A bit ineffective but we must process all the possibilities to get stable 'headers_in' hits!
 884         my %map=map(($_=>a_href($url->{$_},$contents,%args)),keys(%$url));
 885         my $cc;
 886         $cc||=Geo::IP->new()->country_code_by_addr(remote_ip()) if $have_Geo_IP;
 887         $cc||="";
 888         my $r=$map{$cc};
 889         return $r if $r;
 890         return $contents;
 891 }
 892
 893 sub make ($)
 894 {
 895 my($cmd)=@_;
 896
 897         # FIXME: &alarm, --timeout is now infinite.
 898         # FIXME: Try to remove bash(1).
 899         # FIXME: Use: @PATH_FLOCK@
 900         my @argv=('flock',dir_top_abs_disk(),'bash','-c',$cmd.' >&2');
 901         print STDERR join(" ","SPAWN:",@argv)."\n";
 902         system @argv;
 903 }
 904
 905 sub make_file($$)
 906 {
 907 my($self,$file)=@_;
 908
 909         cluck "Pathname not absolute: $file" if $file!~m{^/};
 910         return if -f $file;
 911         # TODO: Somehow quickly check dependencies?
 912         return make('make -s --no-print-directory'
 913                                         .' -C '."'".File::Basename::dirname($file)."' '".File::Basename::basename($file)."'");
 914 }
 915
 916 sub img_size ($$)
 917 {
 918 my($width,$height)=@_;
 919
 920         cluck if !defined $width || !defined $height;
 921         return ($W->{"have_style"} ? "style=\"border:0;width:${width}px;height:${height}px\"" : "border=\"0\"")
 922                         ." width=\"$width\" height=\"$height\"";
 923 }
 924
 925 sub negotiate_variant (%)
 926 {
 927 my(%args)=@_;
 928
 929         my @fields=("id","qs","content-type","encoding","charset","lang","size");
 930         return [ map(($args{$_}),@fields) ];
 931 }
 932
 933 # Returns: 'HTTP::Headers' instance.
 934 sub http_headers_in_for($@)
 935 {
 936 my($self,@headers)=@_;
 937
 938         # Limit these entries to generate proper 'Vary' header.
 939         return HTTP::Headers->new(map(($_=>$W->{"headers_in"}{$_}),@headers));
 940 }
 941
 942 # Input: $self is required!
 943 # Input: Put the fallback variant as the first one.
 944 # Returns: always only scalar!
 945 sub Negotiate_choose($$)
 946 {
 947 my($self,$variants)=@_;
 948
 949         my $best=HTTP::Negotiate::choose($variants,
 950                         # Do not: $W->{"r"}
 951                         # to prevent: Can't locate object method "scan" via package "Apache2::RequestRec" at HTTP/Negotiate.pm line 84.
 952                         # Do not: $W->{"r"}->headers_in()
 953                         # to prevent: Can't locate object method "scan" via package "APR::Table" at HTTP/Negotiate.pm line 84.
 954                         # Do not: HTTP::Headers->new($W->{"r"}->headers_in());
 955                         # to prevent empty result or even: Odd number of elements in anonymous hash
 956                         $self->http_headers_in_for(qw(
 957                                         Accept
 958                                         Accept-Charset
 959                                         Accept-Encoding
 960                                         Accept-Language
 961                                         )));
 962         $best||=$variants->[0][0];      # $variants->[0]{"id"}; &HTTP::Negotiate::choose failed?
 963         return $best;
 964 }
 965
 966 my @img_variants=(
 967                 { "id"=>"png","qs"=>0.9,"content-type"=>"image/png" },
 968                 { "id"=>"gif","qs"=>0.7,"content-type"=>"image/gif" },
 969                 );
 970 my $img_variants_re='[.](?:'.join('|',"jpeg",map(($_->{"id"}),@img_variants)).')$';
 971
 972 # Returns: ($path_web,$path_abs_disk)
 973 # URI path segments support ignored here. Where it is used? (';' path segment options)
 974 sub _img_src($%)
 975 {
 976 my($in,%args)=@_;
 977
 978         cluck if !uri_is_local $in;
 979         my $uri=in_to_uri_abs $in;
 980         my $path_abs_disk=path_abs_disk $uri,%args,"uri_as_in"=>1,"register"=>0;
 981
 982         # Known image extension?
 983         return path_web($uri,%args,"uri_as_in"=>1),$path_abs_disk if $uri->path()=~m#$img_variants_re#o;
 984
 985         my @nego_variants;
 986         for my $var (@img_variants) {
 987                 my $path_abs_disk_variant=$path_abs_disk.".".$var->{"id"};
 988                 path_abs_disk_register($path_abs_disk_variant);
 989                 __PACKAGE__->make_file($path_abs_disk_variant);
 990                 push @nego_variants,negotiate_variant(
 991                                 %$var,
 992                                 "size"=>(stat $path_abs_disk_variant)[7],
 993                                 );
 994                 }
 995         my $ext=__PACKAGE__->Negotiate_choose(\@nego_variants);
 996
 997         $uri->path($uri->path().".$ext");
 998         return path_web($uri,%args,"uri_as_in"=>1),path_abs_disk($uri,%args,"uri_as_in"=>1);
 999 }
1000
1001 # $args{"attr"}
1002 sub img ($$%)
1003 {
1004 my($in,$alt,%args)=@_;
1005
1006         request_check();
1007         my($path_web,$path_abs_disk)=_img_src($in,%args);
1008         my($width,$height)=Image::Size::imgsize($path_abs_disk);
1009         $alt=~s/<[^>]*>//g;
1010         $alt=escapeHTML($alt);
1011         my $content="<img src=\"".uri_escaped($path_web)."\" alt=\"$alt\" title=\"$alt\" ".img_size($width,$height)
1012                         .(!$args{"attr"} ? "" : " ".$args{"attr"})." />";
1013         do { return a_href((_img_src($_))[0],$content,"uri_as_in"=>1) if $_; } for $args{"a_href_img"};
1014         do { return a_href $_,$content if $_; } for $args{"a_href"};
1015         return $content;
1016 }
1017
1018 sub centerimg
1019 {
1020         my $r="";
1021         $r.='<table border="0" width="100%"><tr>'."\n";
1022         @_=( [@_] ) if !ref $_[0];
1023         for (@_) {
1024                 $r.="\t".'<td align="center">'.&{\&img}(@$_).'</td>'."\n";
1025                 }
1026         $r.='</tr></table>'."\n";
1027         return $r;
1028 }
1029
1030 sub rightimg
1031 {
1032 my($text,@args_img)=@_;
1033
1034         # FIXME: Workaround bug of 'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)':
1035         #        <col width="@{[ (!$W->{"browser"}->ie() ? "1*" : "90%" ) ]}" />
1036         #        <col width="@{[ (!$W->{"browser"}->ie() ? "0*" : "10%" ) ]}" />
1037         # causes whole invisible projects in: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.10) Gecko/20050719 Galeon/1.3.21
1038         return <<"HERE";
1039 <table border="0" width="100%">
1040         <tr>
1041                 <td align="left">
1042                         @{[ $text ]}
1043                 </td>
1044                 <td align="right">
1045                         @{[ &{\&img}(@args_img) ]}
1046                 </td>
1047         </tr>
1048 </table>
1049 HERE
1050 }
1051
1052 sub readfile($$)
1053 {
1054 my($class,$filename)=@_;
1055
1056         local *F;
1057         open F,$filename or cluck "Cannot open \"$filename\": $!";
1058         my $F=do { local $/=undef(); <F>; };
1059         close F or cluck "Cannot close \"$filename\": $!";
1060         return $F;
1061 }
1062
1063 sub _no_cache($)
1064 {
1065 my($self)=@_;
1066
1067         header("Expires"=>HTTP::Date::time2str(1000000000));    # date in the past
1068         header("Last-Modified"=>HTTP::Date::time2str());        # always modified
1069         header("Cache-Control"=>join(", ",
1070                         "no-cache",
1071                         "no-store",
1072                         "must-revalidate",
1073                         "max-age=0",
1074                         "pre-check=0",  # MSIE
1075                         "post-check=0", # MSIE
1076                         ));     # HTTP/1.1
1077         header("Pragma"=>"no-cache");   # HTTP/1.0
1078         header("Vary"=>"*");    # content may ba based on unpredictable sources
1079 }
1080
1081 sub headers_in_filtered(@)
1082 {
1083 my(@keys)=@_;
1084
1085         return map(($_=>$W->{"headers_in"}{$_}),@keys);
1086 }
1087
1088 our %uri_args_frozen_to_headers_in_keys;
1089 our %uri_args_headers_in_frozen_to_headers_out;
1090
1091 sub uri_args_headers_in_frozen_get($)
1092 {
1093 my($headers_in_keys_arrayref)=@_;
1094
1095         my %uri_args_headers_in_hash=(
1096                 "uri_args_frozen"=>$W->{"uri_args_frozen"},
1097                 "headers_in"=>{ headers_in_filtered(@$headers_in_keys_arrayref) },
1098                 );
1099         return do { local $Storable::canonical=1; Storable::freeze(\%uri_args_headers_in_hash); };
1100 }
1101
1102 sub cache_output_filter($)
1103 {
1104 my($f)=@_;
1105
1106         while ($f->read(my $text,0x400)) {
1107                 cluck "utf-8 untested" if Encode::is_utf8($text);       # Possible here at all?
1108                 $f->print($text);
1109                 $W->{"digest-md5"}->add($text);
1110                 }
1111         return OK;
1112 }
1113
1114 sub cache_start()
1115 {
1116         if (!$W->{"http_safe"}) {
1117                 __PACKAGE__->_no_cache();
1118                 return;
1119                 }
1120
1121         {
1122                 # &Wrequire it here even if it will not be later used; to be stable!
1123                 Wrequire 'My::Hash::RestrictTo';
1124                 my %uri_args_hash=(
1125                         "method"=>$W->{"r"}->method(),
1126                         "uri"=>"http://".$W->{"web_hostname"}."/".$W->{"r"}->uri(),
1127                         "args"=>$W->{"args_orig"},
1128                         );
1129                 $W->{"uri_args_frozen"}=do { local $Storable::canonical=1; Storable::freeze(\%uri_args_hash); };
1130                 last if !(my $headers_in_keys_arrayref=$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}});
1131
1132                 # Protection to be sure we are stable:
1133                 $W->{"headers_in"}=My::Hash::RestrictTo->new($W->{"headers_in"},@$headers_in_keys_arrayref);
1134
1135                 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get($headers_in_keys_arrayref);
1136                 last if !(my $headers_out_hashref=$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}});
1137                 header(%$headers_out_hashref);
1138                 my $status;
1139                 {
1140                         # &meets_conditions will always deny the attempt if !2xx status().
1141                         # At least ap_read_request() sets: r->status=HTTP_REQUEST_TIME_OUT;     /* Until we get a request */
1142                         my $status_old=$W->{"r"}->status();
1143                         $W->{"r"}->status(HTTP_OK);
1144                         # Update httpd's 'r->mtime' as the header "Last-Modified" is just not enough for ap_meets_conditions():
1145                         # &update_mtime() argument is really in _secs_, not in _msecs_ as the docs claim.
1146                         # Be aware '*1000000' would overflow Perl integer anyway.
1147                         # &set_last_modified would also override the "Last-Modified" headers_out!
1148                         # &mtime may exist but somehow does not work.
1149                         $W->{"r"}->update_mtime(HTTP::Date::str2time($headers_out_hashref->{"Last-Modified"}));
1150                         $status=$W->{"r"}->meets_conditions();
1151                         $W->{"r"}->status($status_old);
1152                         }
1153                 last if OK==$status;
1154                 $W->{"r"}->status($status);
1155                 $W->{"header_only"}=1;  # Inhibit &footer output.
1156                 exit;
1157                 die "NOTREACHED";
1158                 }
1159
1160         $W->{"digest-md5"}=Digest::MD5->new();
1161         $W->{"cache_active"}=1;
1162         $W->{"r"}->add_output_filter(\&cache_output_filter);
1163 }
1164
1165 sub cache_finish_last_modified()
1166 {
1167         cluck "Not yet done now? W __PACKAGE__: ".$W->{"__PACKAGE__"}
1168                         if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
1169         for my $package_orig (@{$packages_used_array{$W->{"__PACKAGE__"}}}) {
1170                 local $_=$package_orig.".pm";
1171                 s{::}{/}g;
1172                 path_abs_disk "/$_","register"=>1;
1173                 }
1174         my $mtime_newest;
1175         for my $path_abs_disk (keys(%{$W->{"path_abs_disk_register"}})) {
1176                 my $mtime=(stat $path_abs_disk)[9];
1177                 do { cluck "No mtime for: $path_abs_disk"; next; } if !$mtime;
1178                 $mtime_newest=$mtime if !$mtime_newest || $mtime_newest<$mtime;
1179                 }
1180         cluck "No mtime_newest found for the current W __PACKAGE__: ".$W->{"__PACKAGE__"}
1181                         if !$mtime_newest;
1182         return HTTP::Date::time2str($mtime_newest);
1183 }
1184
1185
1186 sub cache_finish()
1187 {
1188         # Do not: return if !$W->{"uri_args_frozen"};
1189         # as we may have just gave 304 and 'exit;' without starting the caching.
1190         return if !$W->{"cache_active"};
1191
1192         # Headers may not be complete in this case; not sure, just trying.
1193         return if $W->{"r"}->connection()->aborted();
1194
1195         # Fill-in/check: %uri_args_frozen_to_headers_in_keys
1196         my $headers_in_keys_stored_arrayref_ref=\$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}};
1197         my @headers_in_keys=tied(%{$W->{"headers_in_RecordKeys"}})->accessed();
1198         if (!$$headers_in_keys_stored_arrayref_ref
1199                         || !Data::Compare::Compare(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)) {
1200                 cluck "Non-matching generated 'headers_in_keys' per 'uri_args_frozen' key:\n"
1201                                                 .Dumper(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)
1202                                 if $$headers_in_keys_stored_arrayref_ref;
1203                 # Build or possibly prevent such further warn dupes:
1204                 $$headers_in_keys_stored_arrayref_ref=\@headers_in_keys;
1205                 # Build or regenerate as obsoleted now:
1206                 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get(\@headers_in_keys);
1207                 }
1208
1209         # Prepare 'headers_out' for the future reusal:
1210         my %headers_out;
1211         # Do not: $W->{"digest-md5"}->b64digest();
1212         # as it will not provide the trailing filling '='s.
1213         # RFC 1864 is not clear if they should be there but its sample provides them.
1214         # Do not try to provide canonical "\r\n" form of newlines as is said by RFC 1864.
1215         # RFC 2068 (HTTP/1.1) section 14.16 says the newlines should NOT be converted for HTTP.
1216         # ',""' to avoid breaking the headers by its default "\n".
1217         $headers_out{"Content-MD5"}=MIME::Base64::encode_base64($W->{"digest-md5"}->digest(),"");
1218         # In fact we could also use MD5 for ETag as if we know ETag we also know MD5.
1219         # But this way we do not need to calculate MD5 and we still can provide such ETag. So.
1220         # $W->{"r"}->set_etag() ?
1221         $headers_out{"ETag"}='"'.Digest::MD5::md5_base64($W->{"uri_args_headers_in_frozen"}).'"';
1222         # $W->{"r"}->set_content_length() ?
1223         $headers_out{"Content-Length"}=$W->{"r"}->bytes_sent();
1224         my %Vary=map(($_=>1),(@headers_in_keys));
1225         for (keys(%Vary)) {
1226                 next if !/^_/;
1227                 $Vary{"*"}=1;
1228                 delete $Vary{$_};
1229                 }
1230         %Vary=("*"=>1) if $Vary{"*"};
1231         $headers_out{"Vary"}=join(", ",sort keys(%Vary)) if keys(%Vary);
1232         # $W->{"r"}->set_last_modified() ?
1233         $headers_out{"Last-Modified"}=cache_finish_last_modified();
1234
1235         # Fill-in/check: %uri_args_headers_in_frozen_to_headers_out
1236         my $headers_out_stored_hashref_ref=\$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}};
1237         if (!$$headers_out_stored_hashref_ref
1238                         || !Data::Compare::Compare(\%headers_out,$$headers_out_stored_hashref_ref)) {
1239                 cluck "Non-matching generated 'headers_out' per 'uri_args_headers_in_frozen' key:\n"
1240                                                 .Dumper(\%headers_out,$$headers_out_stored_hashref_ref)
1241                                 if $$headers_out_stored_hashref_ref;
1242                 # Build or possibly prevent such further warn dupes:
1243                 $$headers_out_stored_hashref_ref=\%headers_out;
1244                 }
1245
1246 ###print STDERR Dumper(\%uri_args_frozen_to_headers_in_keys,\%uri_args_headers_in_frozen_to_headers_out);
1247 }
1248
1249 sub heading()
1250 {
1251 my($class)=@_;
1252
1253         if (!$W->{"header_only"}) {
1254                 header("Content-Style-Type"=>"text/css");
1255                 # Do not: text/javascript
1256                 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
1257                 # "application/javascript" so far standardized till 2005-12-08 by:
1258                 #       http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
1259                 header("Content-Script-Type"=>"application/javascript");
1260                 # $W->{"r"}->content_languages() ?
1261                 do { header("Content-Language"=>$_) if $_; } for $W->{"language"};
1262                 }
1263         # TODO: Support also: private
1264         header("Cache-Control"=>"public");      # HTTP/1.1
1265
1266         # Use $W->{"charset"}=0 to disable charset.
1267         $W->{"charset"}="us-ascii"
1268                         if !defined $W->{"charset"} && (!defined($W->{"content_type"}) || $W->{"content_type"});
1269
1270         # Workaround bug
1271         #   https://bugzilla.mozilla.org/show_bug.cgi?id=120556
1272         # of at least
1273         #   Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1274         # http://validator.w3.org/ does not send ANY "Accept" headers!
1275         if (!defined $W->{"content_type"}) {
1276                 # Be _stable_ for "headers_in".
1277                 my $accept=$W->{"headers_in"}{"Accept"};
1278                 my $user_agent=$W->{"headers_in"}{"User-Agent"}||"";
1279                 $W->{"content_type"}="application/xhtml+xml"
1280                                 if !$accept && $user_agent=~m{^W3C_Validator/}i;
1281                 # Be _stable_:
1282                 my $negotiated=$class->Negotiate_choose([
1283                                 # Put the fallback variant as the first one.
1284                                 # Rate both variants the same to prefer "text/html" for undecided clients.
1285                                 # At least
1286                                 #   Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1287                                 # prefers "application/xhtml+xml" over "text/html" itself:
1288                                 #   text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5
1289                                 negotiate_variant(
1290                                                 "id"=>"text/html",
1291                                                 "content-type"=>"text/html",
1292                                                 "qs"=>0.6,
1293                                                 (!$W->{"charset"} ? () : "charset"=>$W->{"charset"}),
1294                                                 "lang"=>$W->{"language"},
1295                                                 ),
1296                                 negotiate_variant(
1297                                                 "id"=>"application/xhtml+xml",
1298                                                 "content-type"=>"application/xhtml+xml",
1299                                                 "qs"=>0.6,
1300                                                 (!$W->{"charset"} ? () : "charset"=>$W->{"charset"}),
1301                                                 "lang"=>$W->{"language"},
1302                                                 ),
1303                                 # application/xml ?
1304                                 # text/xml ?
1305                                 ]);
1306                 $W->{"content_type"}=$negotiated if !defined $W->{"content_type"};
1307                 }
1308         # mod_perl doc: If you set this header via the headers_out table directly, it
1309         #               will be ignored by Apache. So do not do that.
1310         my $type;
1311         if ($W->{"content_type"}) {
1312                 $type=MIME::Types->new()->type($W->{"content_type"});
1313                 cluck "MIME::Types type '".$W->{"content_type"}."' not known" if !$type;
1314                 }
1315         cluck "charset='".$W->{"charset"}."' does not match content-type='".$W->{"content_type"}."'"
1316                         if ($W->{"charset"} ? 1 : 0) != (!$type ? 0 : $type->isAscii());
1317         $W->{"r"}->content_type($W->{"content_type"}.(!$W->{"charset"} ? "" : "; charset=".$W->{"charset"}))
1318                         if $W->{"content_type"};
1319
1320         cache_start();
1321         # We still can append headers before we put out some text.
1322         # FIXME: It is not clean to still append them without overwriting.
1323         return if $W->{"heading_done"};
1324         Wprint '<?xml version="1.0" encoding="'.$W->{"charset"}.'"?>'."\n"
1325                         if (!$W->{"header_only"} || $W->{"header_only"} eq "xml") && (0
1326                                         || $W->{"content_type"}=~m{^application/\w+[+]xml$}
1327                                         || $W->{"content_type"} eq "text/vnd.wap.wml");
1328         return if $W->{"header_only"};
1329         # Split 'heading_done' for the proper handling of: /project/Rel.pm
1330         $W->{"heading_done"}++;
1331
1332         Wprint '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">'."\n";
1333         Wprint '<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="'.$W->{"language"}.'">'."\n";
1334         my $title=$W->{"title_prefix"}.join("",map({ ': '.$_; } ($W->{"title"} || ())));
1335         # Do not: cluck if $title=~/[<>]/;
1336         # as it is not solved just by: &a_href_inhibit
1337         # as sometimes titles use also: <i>...</i>
1338         $title=~s#<[^>]*>##g;
1339         Wprint "<head>";
1340         Wprint "<title>$title</title>\n";
1341         if ($W->{"have_css"}) {
1342                 # Everything can get overriden later.
1343                 for my $css ("/My/Web.css",@{$W->{"css_push"}}) {
1344                         Wprint <<"HERE";
1345 <link rel="stylesheet" type="text/css" href="@{[ uri_escaped(path_web $css) ]}" />
1346 HERE
1347                         }
1348                 if ($W->{"css_inherit"}) {
1349                         # Do not: <script />
1350                         # as at least Lynx inhibits any further HTML output.
1351                         # Do not: text/javascript
1352                         # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
1353                         # "application/javascript" so far standardized till 2005-12-08 by:
1354                         #       http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
1355                         Wprint <<"HERE";
1356 <script type="application/javascript" src="@{[ uri_escaped(path_web('/My/css_inherit.js')) ]}"></script>
1357 HERE
1358                         }
1359                 }
1360         Wprint '<meta name="robots" content="'.($W->{"indexme"} ? "" : "no" ).'index,follow" />'."\n";
1361         Wprint $W->{"head"};
1362         for my $type (qw(prev next index contents start up)) {
1363                 do { Wprint '<link rel="'.$type.'" href="'.uri_escaped(path_web $_).'" />'."\n" if $_; }
1364                                 for ($W->{"rel_$type"});
1365                 }
1366         Wprint "</head><body";
1367 #       Wprint ' bgcolor="black" text="white" link="aqua" vlink="teal"'
1368 #                       if $W->{"browser"}->netscape() && (!$W->{"browser"}->major() || $W->{"browser"}->major()<=4);
1369         Wprint $W->{"body_attr"};
1370         Wprint ">\n";
1371
1372         do { Wprint $_ if $_; } for $W->{"heading"};
1373 }
1374
1375 BEGIN {
1376         delete $W->{"__My::Web_init"};
1377         }
1378
1379 1;