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 $caller eq $target;
  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"};
  74                                 push @{$packages_used_array{$caller}},$target;
  75                                 }
  76                         }
  77                 eval { CORE::require "$file"; } or confess $@;
  78                 1;      # Otherwise 'require' would already file above.
  79         }
  80
  81         sub Wuse ($@)
  82         {
  83         my($file,@list)=@_;
  84
  85 #               print STDERR "Wuse $file\n";
  86                 Wrequire $file;
  87                 local $Exporter::ExportLevel=$Exporter::ExportLevel+1;
  88                 $file->import(@list);
  89                 1;
  90         }
  91
  92         sub import
  93         {
  94         my($class,@rest)=@_;
  95
  96                 local $Exporter::ExportLevel=$Exporter::ExportLevel+1;
  97                 Wrequire("$class");
  98                 return $class->SUPER::import(@rest);
  99         }
 100 }
 101
 102 use WebConfig;  # see also below: Wuse 'WebConfig';
 103 require Image::Size;    # for &imgsize
 104 use File::Basename;     # &basename
 105 use Carp qw(cluck confess);
 106 use URI::Escape;
 107 require HTTP::BrowserDetect;
 108 require HTTP::Negotiate;
 109 our $have_Geo_IP; BEGIN { $have_Geo_IP=eval { require Geo::IP; 1; }; }
 110 # Do not: use ModPerl::Util qw(exit);
 111 # to prevent in mod_perl2: "exit" is not exported by the ModPerl::Util module
 112 # I do not know why.
 113 use POSIX qw(strftime);
 114 use Tie::Handle;
 115 use Apache2::Const qw(HTTP_MOVED_TEMPORARILY OK HTTP_OK);
 116 use URI;
 117 use URI::QueryParam;
 118 use Cwd;
 119 require HTTP::Date;
 120 require Storable;
 121 require Digest::MD5;
 122 require Data::Compare;
 123 use Data::Dumper;
 124 require Encode;
 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
 132
 133 #our $W;
 134
 135 sub init ($%)
 136 {
 137 my($class,%args)=@_;
 138
 139         print STDERR "$class->init ".Apache2::RequestUtil->request()->unparsed_uri()."\n";
 140
 141         # We need to track package dependencies, so we need to call it from &init.
 142         # We cannot do it in BEGIN { } block
 143         # as it would not be tracked for each of the toplevel users later.
 144         Wuse 'WebConfig';
 145         Wrequire 'My::Hash';
 146
 147         # $W={} can get somehow created very easily.
 148         # Do not: cluck "W not empty:\n".Dumper($W) if keys(%$W);
 149         # to prevent (of $W->{"headers_in"}): TODO: Enumeration may not be expected.
 150         cluck "W not empty; __PACKAGE__ was: ".$W->{"__PACKAGE__"} if keys(%$W);
 151         $W=My::Hash->new({},"My::Hash::Sub","My::Hash::Push");
 152         bless $W,$class;
 153         %$W=(
 154                 "__PACKAGE__"=>scalar(caller()),
 155                 %WebConfig,
 156                 %args,  # override %WebConfig settings
 157                 );
 158
 159         # {"__PACKAGE__"} is mandatory for mod_perl-2.0;
 160         # $Apache2::Registry::curstash is no longer supported.
 161         do { cluck "No $_" if !$W->{$_}; } for "__PACKAGE__";
 162         exit_hook_start();
 163
 164         do { $W->{$_}=0  if !defined $W->{$_}; } for "detect_ent";
 165         do { $W->{$_}=0  if !defined $W->{$_}; } for "detect_js";
 166         do { $W->{$_}=1  if !defined $W->{$_}; } for "have_css";        # AFAIK it does not hurt anyone.
 167         do { $W->{$_}=0  if !defined $W->{$_}; } for "css_inherit";
 168         do { $W->{$_}=1  if !defined $W->{$_}; } for "footer";
 169         do { $W->{$_}=1  if !defined $W->{$_}; } for "footer_delimit";
 170         do { $W->{$_}=1  if !defined $W->{$_}; } for "footer_ids";
 171         do { $W->{$_}=1  if !defined $W->{$_}; } for "indexme";
 172         do { $W->{$_}="" if !defined $W->{$_}; } for "head";
 173         do { $W->{$_}="" if !defined $W->{$_}; } for "body_attr";
 174         do { $W->{$_}="en-US" if !defined $W->{$_}; } for "language";
 175
 176         my $footer_any=0;
 177         for (qw(footer_ids)) {
 178                 $W->{$_}=0 if !$W->{"footer"};
 179                 $footer_any=1 if $W->{$_};
 180                 }
 181         $W->{"footer"}=0 if !$footer_any;
 182         $W->{"footer_delimit"}=0 if !$W->{"footer"};
 183
 184         $W->{"r"}=Apache2::RequestUtil->request();
 185
 186         $W->{"r"}->push_handlers("PerlCleanupHandler"=>\&cleanup);
 187
 188         $W->{"web_hostname"}||=$W->{"r"}->hostname();
 189
 190         tie *STDOUT,$W->{"r"};
 191         select *STDOUT;
 192         $|=1;
 193
 194         $W->{"headers_in"}=$W->{"r"}->headers_in();
 195         Wrequire 'My::Hash::Merge';
 196         $W->{"headers_in"}=My::Hash::Merge->new(
 197                         $W->{"headers_in"},
 198                         My::Hash::Sub->new({
 199                                 "_remote_ip"=>sub { return $W->{"r"}->connection()->remote_ip(); },
 200                                 }),
 201                         );
 202         Wrequire 'My::Hash::Readonly';
 203         $W->{"headers_in"}=My::Hash::Readonly->new($W->{"headers_in"});
 204
 205         if ($W->{"r"}->method() eq "GET" || $W->{"r"}->method() eq "HEAD") {
 206                 for (\$W->{"http_safe"}) {
 207                         # Do not: # Extend the current ETag system instead if you would need it:
 208                         #         cluck "Explicitely NOT HTTP-Safe for method \"".$W->{"r"}->method()."\"?!?"
 209                         #                       if defined($$_) && !$$_;
 210                         # as sometimes it just does not make sense to cache it.
 211                         $$_=1 if !defined $$_;
 212                         }
 213                 }
 214         else {
 215                 for (\$W->{"http_safe"}) {
 216                         cluck "Undefined HTTP-Safe-ty for method \"".$W->{"r"}->method()."\"!"
 217                                         if !defined($$_);
 218                         $$_=0 if !defined $$_;
 219                         }
 220                 }
 221         if ($W->{"http_safe"}) {
 222                 Wrequire 'My::Hash::RecordKeys';
 223                 $W->{"headers_in_RecordKeys"}=My::Hash::RecordKeys->new($W->{"headers_in"});
 224                 $W->{"headers_in"}=$W->{"headers_in_RecordKeys"};
 225                 }
 226
 227         {
 228                 local $_=$W->{"r"}->args() || "";
 229                 if ($W->{"detect_ent"}) {
 230                                  if (/[&]amp;have_ent/)
 231                                 { $W->{"have_ent"}=0; }
 232                         elsif (    /[&]have_ent/)
 233                                 { $W->{"have_ent"}=1; }
 234                         else
 235                                 { delete $W->{"have_ent"}; }
 236                         if (!defined $W->{"have_ent"} && $W->{"r"}->method() eq "GET") {
 237                                 $W->{"head"}.='<meta http-equiv="Refresh" content="0; URL='
 238                                                 .escapeHTML("http://".$W->{"web_hostname"}."/".($W->{"r"}->uri()=~m#^/*(.*)$#)[0]
 239                                                                 ."?".($_ || "detect_ent_glue=1").'&have_ent=detect')
 240                                                 .'" />'."\n";
 241                                 }
 242                         }
 243                 s/([&])amp;/$1/g;
 244                 $W->{"r"}->args($_);
 245                 }
 246
 247         $W->{"args"}=URI->new("?".$W->{"r"}->args())->query_form_hash();
 248         $W->merge_post_args() if $W->{"r"}->method() eq "POST";
 249         # Prepare '$args' first to (FIXME: Why?) prevent: Not a reference
 250         my $args=$W->{"args"};
 251         $W->{"args_orig"}=Storable::dclone($args);
 252
 253         $W->{"browser"}=sub {
 254                 # Lazy-evaluation, we may not need the "User-Agent" header at all.
 255                 return our $r||=HTTP::BrowserDetect->new($W->{"headers_in"}{"User-Agent"});
 256                 };
 257
 258         if (!defined $W->{"have_style"}) {
 259                 $W->{"have_style"}=sub {
 260                         # Lazy-evaluation, we may not need the "User-Agent" header at all.
 261                         return our $r||=(!$W->{"browser"}->netscape() || ($W->{"browser"}->major() && $W->{"browser"}->major()>4) ? 1 : 0);
 262                         };
 263                 }
 264
 265         $W->{"have_js"}=($W->{"args"}{"have_js"} ? 1 : 0);
 266         if ($W->{"detect_js"} && !$W->{"have_js"}) {
 267                 # Do not: <script />
 268                 # as at least Lynx inhibits any further HTML output.
 269                 # Do not: text/javascript
 270                 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
 271                 # "application/javascript" so far standardized till 2005-12-08 by:
 272                 #       http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
 273                 $W->{"head"}.='<script type="application/javascript" src="'.uri_escaped(path_web('/My/HaveJS.pm')).'"></script>'."\n";
 274                 }
 275
 276         do { _args_check(%$_) if $_; } for ($W->{"args_check"});
 277
 278         $W->{"_init_done"}=1;
 279         return $W;
 280 }
 281
 282 sub form_method($)
 283 {
 284 my($method)=@_;
 285
 286         return q{enctype="application/x-www-form-urlencoded" accept-charset="us-ascii utf-8"} if $method eq "post";
 287         return                                             q{accept-charset="us-ascii utf-8"} if $method eq "get";
 288         cluck "Undefined method: $method";
 289         return ""
 290 }
 291
 292 sub merge_post_args($)
 293 {
 294 my($class)=@_;
 295
 296         my @post_args=$class->read_post_args();
 297         while (@post_args) {
 298                 my $name=shift @post_args;
 299                 my $data=shift @post_args;
 300                 my $ref=\$W->{"args"}{$name};
 301                    if (!defined $$ref)       { $$ref=$data; }
 302                 elsif (!ref $$ref)           { $$ref=[$$ref,$data]; }
 303                 elsif ("ARRAY" eq ref $$ref) { push @$$ref,$data; }
 304                 else {
 305                         cluck "Ignoring POST argument \"$name\", orig is weird:\n",Dumper($$ref);
 306                         }
 307                 }
 308         return;
 309 }
 310
 311 # Do not: use CGI;
 312 # as CGI parsing of POST vs. QUERY_STRING data, multiple-valued keys etc.
 313 # is too dense and causes weird problems, together with mod_perl etc.
 314 sub read_post_args($)
 315 {
 316 my($class)=@_;
 317
 318         local $_=$class->http_headers_in_for("Content-type")->content_type();
 319         return $class->read_multipart_form_data() if $_ eq "multipart/form-data";
 320         return $class->read_application_x_www_form_urlencoded() if $_ eq "application/x-www-form-urlencoded";
 321         cluck "Unknown POST data body, ignored: $_";
 322         return;
 323 }
 324
 325 sub read_application_x_www_form_urlencoded($)
 326 {
 327 my($class)=@_;
 328
 329         my $body="";
 330         for (;;) {
 331                 my $got=$W->{"r"}->read(my($buf),0x1000);
 332                 # Do not: cluck "Error reading POST data: $!" if !defined $got;
 333                 # as it should be done using: APR::Error exceptions
 334                 last if !$got;
 335                 $body.=$buf;
 336                 }
 337         return URI->new("?".$body)->query_form();
 338 }
 339
 340 sub read_multipart_form_data($)
 341 {
 342 my($class)=@_;
 343
 344         my $parser=MIME::Parser->new();
 345         # FIXME: No unlink()s done!
 346         $parser->output_under("/tmp");
 347
 348         local *R_FH;
 349         tie *R_FH,$W->{"r"};
 350         local *FH;
 351         tie *FH,"My::Web::ReadMerged",
 352                         join("",map(($_.": ".$W->{"headers_in"}{$_}."\n"),qw(
 353                                         Content-type
 354                                         )))."\n",
 355                         \*R_FH;
 356         my $body=$parser->parse(\*FH);
 357         cluck "No multipart POST request body?" if !$body->is_multipart();
 358
 359         return map((
 360                         $_->head()->mime_attr("content-disposition.name")
 361                         =>
 362                         join("",@{$_->body()})
 363                         ),$body->parts());
 364
 365         # TODO: Globalize, make it IO::* compatible, split to the merging part + IO::Scalar.
 366         package My::Web::ReadMerged;
 367
 368         require Tie::Handle;
 369         require Exporter;
 370         our @ISA=qw(Tie::Handle Exporter);
 371         use Carp qw(cluck confess);
 372
 373         sub READLINE($)
 374         {
 375         my($self)=@_;
 376
 377                 confess "Slurp not yet implemented" if !defined $/;
 378                 # Apache2::RequestIO does not support 'READLINE'!
 379                 for (;;) {
 380                         if (defined $self->{"data"} && $self->{"data"}=~s{^.*\Q$/\E}{}) {
 381                                 $self->{"offset"}+=length $&;
 382                                 return $&;
 383                                 }
 384                         my $fh_orig=$self->{"fh_orig"};
 385                         if (!$fh_orig) {
 386                                 my $r=$self->{"data"};
 387                                 delete $self->{"data"};
 388                                 $self->{"offset"}+=length $r if defined $r;
 389                                 return $r;
 390                                 }
 391                         my $got=read $fh_orig,my($buf),0x1000;
 392                         cluck "Error reading POST data: $!" if !defined $got;
 393                         delete $self->{"fh_orig"} if !$got;
 394                         cluck "INTERNAL: fh_orig should not exist here" if !defined $self->{"data"};
 395                         $self->{"data"}.=$buf;
 396                         }
 397         }
 398
 399         sub TELL($)
 400         {
 401         my($self)=@_;
 402
 403                 return $self->{"offset"};
 404         }
 405
 406         sub TIEHANDLE($$$)
 407         {
 408         my($class,$data,$fh_orig)=@_;
 409
 410                 my $self=bless {},$class;
 411                 $self->{"data"}=$data;
 412                 $self->{"offset"}=0;
 413                 $self->{"fh_orig"}=$fh_orig;
 414                 return $self;
 415         }
 416 }
 417
 418 sub cleanup($)
 419 {
 420 my($apache_request)=@_;
 421
 422         cluck "CORE::GLOBAL::exit hook not ran" if !$W->{"_exit_done"};
 423         cluck "packages not finalized" if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
 424         cache_finish();
 425         # Sanity protection.
 426         $W=undef();
 427         exit_hook_stop();
 428         return OK;
 429 }
 430
 431 # PerlResponseHandler is RUN_FIRST and &ModPerl::Util::exit returns OK, so no (sane) go.
 432 # PerlLogHandler is already too late to be able to produce any output.
 433 my $exit_orig;
 434 sub exit_hook
 435 {
 436         cluck "Missing ->init while in exit_hook()" if !$W->{"_init_done"};
 437         # &footer will call us recursively!
 438         footer() if !$W->{"_exit_done"}++;
 439         return &{$exit_orig}(@_);
 440 }
 441 sub exit_hook_start
 442 {
 443         do { cluck "exit_hook_start() twice?"; return; } if defined $exit_orig;
 444         $exit_orig=\&CORE::GLOBAL::exit;
 445         # Prevent: Subroutine CORE::GLOBAL::exit redefined
 446         no warnings 'redefine';
 447         *CORE::GLOBAL::exit=\&exit_hook;
 448 }
 449 sub exit_hook_stop
 450 {
 451         do { cluck "exit_hook_stop() without exit_hook_start()?"; return; }
 452                         if \&exit_hook ne \&CORE::GLOBAL::exit;
 453         do { cluck "INTERNAL: exit_orig uninitialized"; return; }
 454                         if !$exit_orig;
 455         # Prevent: Subroutine CORE::GLOBAL::exit redefined
 456         no warnings 'redefine';
 457         *CORE::GLOBAL::exit=$exit_orig;
 458         $exit_orig=undef();
 459 }
 460
 461 # Be aware other parts of code (non-My::Web) will NOT use this function!
 462 # Do not: Wprint $W->{"heading"},"undef"=>1;
 463 # as we would need to undef() it to turn it off and it would get defaulted in such case.
 464 # Do not: exists $W->{"heading"}
 465 # as we use a lot of 'for $W->{"heading"}' which instantiates it with the value: undef()
 466 sub Wprint($%)
 467 {
 468 my($text,%args)=@_;
 469
 470         cluck "undef Wprint" if !defined $text && !$args{"undef"};
 471         delete $args{"undef"};
 472         cluck join(" ","Invalid arguments:",keys(%args)) if keys(%args);
 473         return if !defined $text;
 474         # Do not: cluck "utf-8 untested" if Encode::is_utf8($text);
 475         # as it is valid here.
 476         $W->{"r"}->puts($text);
 477 }
 478
 479 sub request_check(;$)
 480 {
 481 my($self)=@_;
 482
 483         # Use &eval to prevent: Global $r object is not available. Set:\n\tPerlOptions +GlobalRequest\nin ...
 484         confess "Calling sensitive dynamic code from a static code" if !eval { Apache2::RequestUtil->request(); };
 485         # Do not: confess "Calling sensitive dynamic code without My::Web::init" if !$W->{"__PACKAGE__"};
 486         # as it is valid at least while preparing arguments to call: &project::Lib::init
 487 }
 488
 489 # Do not: use CGI;
 490 # as it is too much backward compatible regarding the charset encodings etc.
 491 # and the resulting code is too dense with no additional functionality for the recent content.
 492 sub escapeHTML($)
 493 {
 494 my($text)=@_;
 495
 496         local $_=$text;
 497         s{&}{&amp;}gso;
 498         s{<}{&lt;}gso;
 499         s{>}{&gt;}gso;
 500         s{"}{&quot;}gso;
 501         return $_;
 502 }
 503
 504 # /home/user/www/webdir
 505 sub dir_top_abs_disk()
 506 {
 507         our $dir_top_abs_disk;
 508         if (!$dir_top_abs_disk) {
 509                 my $selfpkg_relpath=__PACKAGE__;
 510                 $selfpkg_relpath=~s{::}{/}g;
 511                 $selfpkg_relpath.=".pm";
 512                 my $selfpkg_abspath=$INC{$selfpkg_relpath} or do {
 513                         cluck "Unable to find self package $selfpkg_relpath";
 514                         return;
 515                         };
 516                 $selfpkg_abspath=~s{/*\Q$selfpkg_relpath\E$}{} or do {
 517                         cluck "Unable to strip myself \"$selfpkg_relpath\" from the abspath: $selfpkg_abspath";
 518                         return;
 519                         };
 520                 cluck "INC{myself} is relative?: $selfpkg_abspath" if $selfpkg_abspath!~m{^/};
 521                 $dir_top_abs_disk=$selfpkg_abspath;
 522                 }
 523         return $dir_top_abs_disk;
 524 }
 525
 526 sub unparsed_uri()
 527 {
 528         request_check();
 529         if (!$W->{"unparsed_uri"}) {
 530                 # Do not: $W->{"r"}
 531                 # as we may be called before &init from: &My::Project::init
 532                 my $r=Apache2::RequestUtil->request();
 533                 cluck "Calling ".'&unparsed_uri'." from a static code, going to fail" if !$r;
 534                 my $uri_string=$r->unparsed_uri() or cluck "Valid 'r' missing unparsed_uri()?";
 535                 my $uri=URI->new_abs($uri_string,"http://".$W->{"web_hostname"}."/");
 536                 $W->{"unparsed_uri"}=$uri;
 537                 }
 538         return $W->{"unparsed_uri"};
 539 }
 540
 541 sub in_to_uri_abs($)
 542 {
 543 my($in)=@_;
 544
 545         # Otherwise we may have been already processed and thus legally relativized.
 546         # FIXME data: Currently disabled, all the data are too violating such rule.
 547         if (0 && !ref $in) {
 548                 my $uri_check=URI->new($in);
 549                 $uri_check->scheme() || $in=~m{^\Q./\E} || $in=~m{^/}
 550                                 or cluck "Use './' or '/' prefix for all the local references: $in";
 551                 }
 552         my $uri=URI->new_abs($in,unparsed_uri());
 553         $uri=$uri->canonical();
 554         return $uri;
 555 }
 556
 557 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
 558 # $args{"abs"}=1;
 559 sub path_web($%)
 560 {
 561 my($in,%args)=@_;
 562
 563         cluck if !$args{"uri_as_in"} && ref $in;
 564         my $uri=in_to_uri_abs($in);
 565         if (uri_is_local($uri)) {
 566                 # Prefer the $uri values over "args_persistent" values.
 567                 # &query_form_hash comes from: URI::QueryParam
 568                 $uri->query_form_hash({
 569                                 map({
 570                                         my $key=$_;
 571                                         my $val=$W->{"args"}{$key};
 572                                         (!defined $val ? () : ($key=>$val));
 573                                         } keys(%{$W->{"args_persistent"}})),
 574                                 %{$uri->query_form_hash()},
 575                                 });
 576                 }
 577         return $uri->abs(unparsed_uri()) if $W->{"args"}{"Wabs"} || $args{"abs"};
 578         return $uri->rel(unparsed_uri());
 579 }
 580
 581 sub path_abs_disk_register($)
 582 {
 583 my($path_abs_disk)=@_;
 584
 585         $W->{"path_abs_disk_register"}{$path_abs_disk}=1;
 586 }
 587
 588 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
 589 sub path_abs_disk($%)
 590 {
 591 my($in,%args)=@_;
 592
 593         cluck if !$args{"uri_as_in"} && ref $in;
 594         my $uri=in_to_uri_abs($in);
 595         cluck if !uri_is_local($uri);
 596         my $path=$uri->path();
 597         cluck "URI compatibility: ->path() not w/leading slash of URI \"$uri\"; path: $path" if $path!~m{^/};
 598         my $r=dir_top_abs_disk().$path;
 599         path_abs_disk_register $r if !defined $args{"register"} || $args{"register"};
 600         return $r;
 601 }
 602
 603 sub fatal (;$);
 604
 605 sub _args_check (%)
 606 {
 607 my(%tmpl)=@_;
 608
 609         while (my($name,$regex)=each(%tmpl)) {
 610                 my $name_html="Parameter <span class=\"quote\">".escapeHTML($name)."</span>";
 611                 $W->{"args"}{$name}="" if !defined $W->{"args"}{$name};
 612                 $W->{"args"}{$name}=[ $W->{"args"}{$name} ] if !ref $W->{"args"}{$name} && ref $regex;
 613                 fatal "$name_html passed as multivar although singlevar expected"
 614                                 if ref $W->{"args"}{$name} && !ref $regex;
 615                 $regex=$regex->[0] if ref $regex;
 616                 for my $val (!ref $W->{"args"}{$name} ? $W->{"args"}{$name} : @{$W->{"args"}{$name}}) {
 617                         $val="" if !defined $val;
 618                         fatal "$name_html <span class=\"quote\">".escapeHTML($val)."</span>"
 619                                                         ." does not match the required regex <span class=\"quote\">".escapeHTML($regex)."</span> "
 620                                         if $regex ne "" && $val!~/$regex/;
 621                         }
 622                 }
 623 }
 624
 625 sub vskip (;$)
 626 {
 627 my($height)=@_;
 628
 629         return '<p'.(!defined $height ? "" : ' style="height: '.$height.';"').'>&nbsp;</p>'."\n";
 630 }
 631
 632 sub fatal (;$)
 633 {
 634 my($msg)=@_;
 635
 636         $msg="UNKNOWN" if !$msg;
 637         cluck "FATAL: $msg";
 638
 639         # Do not send it unconditionally.
 640         # The intial duplicated '<?xml...' crashes Gecko parser.
 641         $W->{"heading_done"}=0 if $W->{"header_only"};
 642         # Do not send it unconditionally.
 643         # Prevents warn: Headers already sent
 644         if (!$W->{"heading_done"}) {
 645                 $W->{"indexme"}=0;      # For the case no heading was sent yet.
 646                 $W->{"header_only"}=0;  # assurance for &heading
 647                 My::Web->heading();
 648                 }
 649         Wprint "\n".vskip("3ex")."<hr /><h1 class=\"error\">FATAL ERROR: $msg!</h1>\n"
 650                         ."<p>You can report this problem's details to"
 651                         ." ".a_href("mailto:".$W->{"admin_mail"},"admin of this website").".</p>\n";
 652         exit;
 653 }
 654
 655 sub footer_packages_used_comments()
 656 {
 657         my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
 658         for my $package (@$packages_used) {
 659                 my $cvs_id=(eval('$'.$package."::CVS_ID")
 660 #                               || $package     # debug
 661                                 );
 662                 Wprint '<!-- '.$package.' - $'.$cvs_id.'$ -->'."\n" if $cvs_id;
 663                 }
 664 }
 665
 666 sub footer()
 667 {
 668         cluck 'Explicit &footer call is deprecated, !_exit_dne' if !$W->{"_exit_done"};
 669         exit if $W->{"footer_done"}++;  # deadlock prevention:
 670         &{$_}() for reverse @{$W->{"footer_sub_push"}};
 671         if ($W->{"header_only"}) {
 672                 $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
 673                 exit;
 674                 }
 675
 676         Wprint vskip if $W->{"footer_delimit"};
 677         &{$_}() for reverse @{$W->{"footing_delimit_sub_push"}};
 678         Wprint "<hr />\n" if $W->{"footer"};
 679
 680         # Never update the package list while we examine it!
 681         $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
 682
 683         my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
 684         if ($W->{"footer_ids"}) {
 685                 Wprint '<p class="cvs-id">';
 686                 Wprint join("<br />\n",map({ my $package=$_;
 687                         my $cvs_id=(eval('$'.$package."::CVS_ID")
 688 #                                       || $package     # debug
 689                                         );
 690                         if (!$cvs_id) {
 691                                 ();
 692                                 }
 693                         else {
 694                                 $cvs_id='$'.$cvs_id.'$';        # Eaten by 'q' operator.
 695                                 my @cvs_id_split=split / +/,$cvs_id;
 696                                 if (@cvs_id_split==8) {
 697                                         my $file=$package;
 698                                         $file=~s#::#/#g;
 699                                         my $ext;
 700                                         my @tried;
 701                                         for (qw(.pm)) {
 702                                                 $ext=$_;
 703                                                 my $path_abs_disk=path_abs_disk("/$file$ext");
 704                                                 push @tried,$path_abs_disk;
 705                                                 last if -r $path_abs_disk;
 706                                                 cluck "Class file $file not found; tried: ".join(" ",@tried) if !$ext;
 707                                                 }
 708                                         $file.=$ext;
 709                                         my $viewcvs;
 710                                         if ((my $file_cvs=$file)=~s{^My/}{}) {
 711                                                 $viewcvs=$W->{"viewcvs_My"}.$file_cvs;
 712                                                 }
 713                                         else {
 714                                                 $viewcvs=$W->{"viewcvs"}.$file;
 715                                                 }
 716                                         $cvs_id_split[2]=""
 717                                                         .a_href((map({ my $s=$_; $s=~s#/viewcvs/#$&~checkout~/#; $s; } $viewcvs))[0]."?rev=".$cvs_id_split[2],
 718                                                                         $cvs_id_split[2]);
 719                                         $cvs_id_split[1]=a_href($viewcvs,($package!~/^Apache2::/ ? $package : $cvs_id_split[1]));
 720                                         $cvs_id_split[5]=&{$W->{"cvs_id_author_sub"}}($cvs_id_split[5]);
 721                                         }
 722                                 join " ",@cvs_id_split;
 723                                 }
 724                         } @$packages_used));
 725                 Wprint "</p>\n";
 726                 }
 727
 728         footer_packages_used_comments();
 729
 730         do { Wprint $_ if $_; } for $W->{"footing"};
 731
 732         Wprint "</body></html>\n";
 733         exit;
 734 }
 735
 736 # Existing entries are overwritten.
 737 sub header(%)
 738 {
 739 my(%pairs)=@_;
 740
 741         while (my($key,$val)=each(%pairs)) {
 742                 do { cluck "Headers already sent"; next; } if $W->{"heading_done"};
 743                 $W->{"r"}->headers_out()->set($key,$val);
 744                 }
 745 }
 746
 747 sub size_display ($)
 748 {
 749 my($size)=@_;
 750
 751            if ($size<4096)
 752                 {}
 753         elsif ($size<1024*1024)
 754                 { $size=sprintf "%.1fK",$size/1024; }
 755         else
 756                 { $size=sprintf "%.1fM",$size/1024/1024; }
 757         $size.="B";
 758         return $size;
 759 }
 760
 761 sub uri_is_local($)
 762 {
 763 my($in)=@_;
 764
 765         my $uri_rel=in_to_uri_abs($in)->rel(unparsed_uri());
 766         # Do not: defined $uri_rel->("userinfo"|"host"|"port")();
 767         # as they fail to be called for schemes not supporting them.
 768         return 0 if $uri_rel->scheme();
 769         return 0 if $uri_rel->authority();
 770         return 1;
 771 }
 772
 773 # &path_web still may be required for &uri_escaped !
 774 sub uri_escaped($)
 775 {
 776 my($uri)=@_;
 777
 778         cluck if !ref $uri;
 779         my $urient=escapeHTML($uri);
 780         return $uri    if $uri eq $urient;
 781         request_check();
 782         return $urient if uri_is_local $uri;
 783         return $uri    if defined $W->{"have_ent"} && !$W->{"have_ent"};        # non-ent client
 784         return $urient if $W->{"have_ent"};     # ent client
 785         # Unknown client, &escapeHTML should not be needed here:
 786         return escapeHTML(path_web('/My/Redirect.pm?location='.uri_escape($uri->abs(unparsed_uri()))));
 787 }
 788
 789 our $a_href_inhibited;
 790 sub a_href($;$%)
 791 {
 792 my($in,$contents,%args)=@_;
 793
 794         request_check();
 795         do { $$_=1 if !defined $$_; } for (\$args{"size"});
 796         if (!defined $contents) {
 797                 $contents=$in;
 798                 $contents=File::Basename::basename($contents) if $args{"basename"};
 799                 $contents=escapeHTML($contents);
 800                 }
 801         $contents=~s#<a\b[^>]*>##gi;
 802         $contents=~s#</a>##gi;
 803         return $contents if $a_href_inhibited;
 804
 805         my $path_web=path_web $in,%args;
 806         my $r="";
 807         $r.='<a href="';
 808         $r.=uri_escaped $path_web;
 809         $r.='"';
 810         do { $r.=" $_" if $_; } for ($args{"attr"});
 811         $r.='>'.$contents.'</a>';
 812         if ($args{"size"} && uri_is_local($in) && ($args{"size"}>=2 || $in=~/[.](?:gz|Z|rpm|zip|deb|lha)/)) {   # Downloadable?
 813                 my $path_abs_disk=path_abs_disk $in,%args;
 814                 cluck "File not readable: $path_abs_disk" if !-r $path_abs_disk;
 815                 $r.='&nbsp;('.size_display((stat($path_abs_disk))[7]).')';
 816                 }
 817         return $r;
 818 }
 819
 820 sub a_href_inhibit($$;@)
 821 {
 822 my($self,$sub,@sub_args)=@_;
 823
 824         local $a_href_inhibited=1;
 825         return &{$sub}(@sub_args);
 826 }
 827
 828 sub input_hidden_persistents()
 829 {
 830         request_check();
 831         return join("",map({
 832                 my $key=$_;
 833                 my $val=$W->{"args"}{$key};
 834                 (!defined $val ? () : '<input type="hidden"'
 835                                 .' name="'.escapeHTML($key).'"'
 836                                 .' value="'.escapeHTML($val).'"'
 837                                 .' />'."\n");
 838                 } (keys(%{$W->{"args_persistent"}}))));
 839 }
 840
 841 sub http_moved($$;$)
 842 {
 843 my($self,$url,$status)=@_;
 844
 845         $url=path_web($url,"abs"=>1);
 846         $status||=HTTP_MOVED_TEMPORARILY;
 847         $W->{"r"}->status($status);
 848         $W->{"r"}->headers_out()->{"Location"}=$url;
 849         $W->{"header_only"}=1;
 850         $W->{"content_type"}=0;
 851         $W->{"charset"}=0;
 852         My::Web->heading();
 853         exit;
 854         die "NOTREACHED";
 855 }
 856
 857 sub remote_ip ()
 858 {
 859         # Do not: PerlModule                 Apache2::ForwardedFor
 860         #         PerlPostReadRequestHandler Apache2::ForwardedFor
 861         # As 'Apache2::ForwardedFor' takes the first of $ENV{"HTTP_X_FORWARDED_FOR"}
 862         # while the contents is '127.0.0.1, 213.220.195.171' if client has its own proxy.
 863         # We must take the last item ourselves.
 864         # Be VERY sure you always retrieve all the headers unconditionally to hit: My::Hash::RecordKeys
 865         my $x_forwarded_for=$W->{"headers_in"}{"X-Forwarded-For"};
 866         $x_forwarded_for=~s/^.*,\s*// if $x_forwarded_for;
 867         my $remote_ip=$W->{"headers_in"}{"_remote_ip"};
 868         my $r;
 869         $r||=$x_forwarded_for;
 870         $r||=$remote_ip;
 871         return $r;
 872 }
 873
 874 # $url={"JP"=>"http://specific",...};
 875 # $url={""=>"http://default",...};
 876 sub a_href_cc($$;%)
 877 {
 878 my($url,$contents,%args)=@_;
 879
 880         # A bit ineffective but we must process all the possibilities to get stable 'headers_in' hits!
 881         my %map=map(($_=>a_href($url->{$_},$contents,%args)),keys(%$url));
 882         my $cc;
 883         $cc||=Geo::IP->new()->country_code_by_addr(remote_ip()) if $have_Geo_IP;
 884         $cc||="";
 885         my $r=$map{$cc};
 886         return $r if $r;
 887         return $contents;
 888 }
 889
 890 sub make ($)
 891 {
 892 my($cmd)=@_;
 893
 894         # FIXME: &alarm, --timeout is now infinite.
 895         # FIXME: Try to remove bash(1).
 896         # FIXME: Use: @PATH_FLOCK@
 897         my @argv=('flock',dir_top_abs_disk(),'bash','-c',$cmd.' >&2');
 898         print STDERR join(" ","SPAWN:",@argv)."\n";
 899         system @argv;
 900 }
 901
 902 sub make_file($$)
 903 {
 904 my($self,$file)=@_;
 905
 906         cluck "Pathname not absolute: $file" if $file!~m{^/};
 907         return if -f $file;
 908         # TODO: Somehow quickly check dependencies?
 909         return make('make -s --no-print-directory'
 910                                         .' -C '."'".File::Basename::dirname($file)."' '".File::Basename::basename($file)."'");
 911 }
 912
 913 sub img_size ($$)
 914 {
 915 my($width,$height)=@_;
 916
 917         cluck if !defined $width || !defined $height;
 918         return ($W->{"have_style"} ? "style=\"border:0;width:${width}px;height:${height}px\"" : "border=\"0\"")
 919                         ." width=\"$width\" height=\"$height\"";
 920 }
 921
 922 sub negotiate_variant (%)
 923 {
 924 my(%args)=@_;
 925
 926         my @fields=("id","qs","content-type","encoding","charset","lang","size");
 927         return [ map(($args{$_}),@fields) ];
 928 }
 929
 930 # Returns: 'HTTP::Headers' instance.
 931 sub http_headers_in_for($@)
 932 {
 933 my($self,@headers)=@_;
 934
 935         # Limit these entries to generate proper 'Vary' header.
 936         return HTTP::Headers->new(map(($_=>$W->{"headers_in"}{$_}),@headers));
 937 }
 938
 939 # Input: $self is required!
 940 # Input: Put the fallback variant as the first one.
 941 # Returns: always only scalar!
 942 sub Negotiate_choose($$)
 943 {
 944 my($self,$variants)=@_;
 945
 946         my $best=HTTP::Negotiate::choose($variants,
 947                         # Do not: $W->{"r"}
 948                         # to prevent: Can't locate object method "scan" via package "Apache2::RequestRec" at HTTP/Negotiate.pm line 84.
 949                         # Do not: $W->{"r"}->headers_in()
 950                         # to prevent: Can't locate object method "scan" via package "APR::Table" at HTTP/Negotiate.pm line 84.
 951                         # Do not: HTTP::Headers->new($W->{"r"}->headers_in());
 952                         # to prevent empty result or even: Odd number of elements in anonymous hash
 953                         $self->http_headers_in_for(qw(
 954                                         Accept
 955                                         Accept-Charset
 956                                         Accept-Encoding
 957                                         Accept-Language
 958                                         )));
 959         $best||=$variants->[0][0];      # $variants->[0]{"id"}; &HTTP::Negotiate::choose failed?
 960         return $best;
 961 }
 962
 963 my @img_variants=(
 964                 { "id"=>"png","qs"=>0.9,"content-type"=>"image/png" },
 965                 { "id"=>"gif","qs"=>0.7,"content-type"=>"image/gif" },
 966                 );
 967 my $img_variants_re='[.](?:'.join('|',"jpeg",map(($_->{"id"}),@img_variants)).')$';
 968
 969 # Returns: ($path_web,$path_abs_disk)
 970 # URI path segments support ignored here. Where it is used? (';' path segment options)
 971 sub _img_src($%)
 972 {
 973 my($in,%args)=@_;
 974
 975         cluck if !uri_is_local $in;
 976         my $uri=in_to_uri_abs $in;
 977         my $path_abs_disk=path_abs_disk $uri,%args,"uri_as_in"=>1,"register"=>0;
 978
 979         # Known image extension?
 980         return path_web($uri,%args,"uri_as_in"=>1),$path_abs_disk if $uri->path()=~m#$img_variants_re#o;
 981
 982         my @nego_variants;
 983         for my $var (@img_variants) {
 984                 my $path_abs_disk_variant=$path_abs_disk.".".$var->{"id"};
 985                 path_abs_disk_register($path_abs_disk_variant);
 986                 __PACKAGE__->make_file($path_abs_disk_variant);
 987                 push @nego_variants,negotiate_variant(
 988                                 %$var,
 989                                 "size"=>(stat $path_abs_disk_variant)[7],
 990                                 );
 991                 }
 992         my $ext=__PACKAGE__->Negotiate_choose(\@nego_variants);
 993
 994         $uri->path($uri->path().".$ext");
 995         return path_web($uri,%args,"uri_as_in"=>1),path_abs_disk($uri,%args,"uri_as_in"=>1);
 996 }
 997
 998 # $args{"attr"}
 999 sub img ($$%)
1000 {
1001 my($in,$alt,%args)=@_;
1002
1003         request_check();
1004         my($path_web,$path_abs_disk)=_img_src($in,%args);
1005         my($width,$height)=Image::Size::imgsize($path_abs_disk);
1006         $alt=~s/<[^>]*>//g;
1007         $alt=escapeHTML($alt);
1008         my $content="<img src=\"".uri_escaped($path_web)."\" alt=\"$alt\" title=\"$alt\" ".img_size($width,$height)
1009                         .(!$args{"attr"} ? "" : " ".$args{"attr"})." />";
1010         do { return a_href((_img_src($_))[0],$content,"uri_as_in"=>1) if $_; } for $args{"a_href_img"};
1011         do { return a_href $_,$content if $_; } for $args{"a_href"};
1012         return $content;
1013 }
1014
1015 sub centerimg
1016 {
1017         my $r="";
1018         $r.='<table border="0" width="100%"><tr>'."\n";
1019         @_=( [@_] ) if !ref $_[0];
1020         for (@_) {
1021                 $r.="\t".'<td align="center">'.&{\&img}(@$_).'</td>'."\n";
1022                 }
1023         $r.='</tr></table>'."\n";
1024         return $r;
1025 }
1026
1027 sub rightimg
1028 {
1029 my($text,@args_img)=@_;
1030
1031         # FIXME: Workaround bug of 'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)':
1032         #        <col width="@{[ (!$W->{"browser"}->ie() ? "1*" : "90%" ) ]}" />
1033         #        <col width="@{[ (!$W->{"browser"}->ie() ? "0*" : "10%" ) ]}" />
1034         # causes whole invisible projects in: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.10) Gecko/20050719 Galeon/1.3.21
1035         return <<"HERE";
1036 <table border="0" width="100%">
1037         <tr>
1038                 <td align="left">
1039                         @{[ $text ]}
1040                 </td>
1041                 <td align="right">
1042                         @{[ &{\&img}(@args_img) ]}
1043                 </td>
1044         </tr>
1045 </table>
1046 HERE
1047 }
1048
1049 sub readfile($$)
1050 {
1051 my($class,$filename)=@_;
1052
1053         local *F;
1054         open F,$filename or cluck "Cannot open \"$filename\": $!";
1055         my $F=do { local $/=undef(); <F>; };
1056         close F or cluck "Cannot close \"$filename\": $!";
1057         return $F;
1058 }
1059
1060 sub _no_cache($)
1061 {
1062 my($self)=@_;
1063
1064         header("Expires"=>HTTP::Date::time2str(1000000000));    # date in the past
1065         header("Last-Modified"=>HTTP::Date::time2str());        # always modified
1066         header("Cache-Control"=>join(", ",
1067                         "no-cache",
1068                         "no-store",
1069                         "must-revalidate",
1070                         "max-age=0",
1071                         "pre-check=0",  # MSIE
1072                         "post-check=0", # MSIE
1073                         ));     # HTTP/1.1
1074         header("Pragma"=>"no-cache");   # HTTP/1.0
1075         header("Vary"=>"*");    # content may ba based on unpredictable sources
1076 }
1077
1078 sub headers_in_filtered(@)
1079 {
1080 my(@keys)=@_;
1081
1082         return map(($_=>$W->{"headers_in"}{$_}),@keys);
1083 }
1084
1085 our %uri_args_frozen_to_headers_in_keys;
1086 our %uri_args_headers_in_frozen_to_headers_out;
1087
1088 sub uri_args_headers_in_frozen_get($)
1089 {
1090 my($headers_in_keys_arrayref)=@_;
1091
1092         my %uri_args_headers_in_hash=(
1093                 "uri_args_frozen"=>$W->{"uri_args_frozen"},
1094                 "headers_in"=>{ headers_in_filtered(@$headers_in_keys_arrayref) },
1095                 );
1096         return do { local $Storable::canonical=1; Storable::freeze(\%uri_args_headers_in_hash); };
1097 }
1098
1099 sub cache_output_filter($)
1100 {
1101 my($f)=@_;
1102
1103         while ($f->read(my $text,0x400)) {
1104                 cluck "utf-8 untested" if Encode::is_utf8($text);       # Possible here at all?
1105                 $f->print($text);
1106                 $W->{"digest-md5"}->add($text);
1107                 }
1108         return OK;
1109 }
1110
1111 sub cache_start()
1112 {
1113         if (!$W->{"http_safe"}) {
1114                 __PACKAGE__->_no_cache();
1115                 return;
1116                 }
1117
1118         {
1119                 # &Wrequire it here even if it will not be later used; to be stable!
1120                 Wrequire 'My::Hash::RestrictTo';
1121                 my %uri_args_hash=(
1122                         "method"=>$W->{"r"}->method(),
1123                         "uri"=>"http://".$W->{"web_hostname"}."/".$W->{"r"}->uri(),
1124                         "args"=>$W->{"args_orig"},
1125                         );
1126                 $W->{"uri_args_frozen"}=do { local $Storable::canonical=1; Storable::freeze(\%uri_args_hash); };
1127                 last if !(my $headers_in_keys_arrayref=$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}});
1128
1129                 # Protection to be sure we are stable:
1130                 $W->{"headers_in"}=My::Hash::RestrictTo->new($W->{"headers_in"},@$headers_in_keys_arrayref);
1131
1132                 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get($headers_in_keys_arrayref);
1133                 last if !(my $headers_out_hashref=$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}});
1134                 header(%$headers_out_hashref);
1135                 my $status;
1136                 {
1137                         # &meets_conditions will always deny the attempt if !2xx status().
1138                         # At least ap_read_request() sets: r->status=HTTP_REQUEST_TIME_OUT;     /* Until we get a request */
1139                         my $status_old=$W->{"r"}->status();
1140                         $W->{"r"}->status(HTTP_OK);
1141                         # Update httpd's 'r->mtime' as the header "Last-Modified" is just not enough for ap_meets_conditions():
1142                         # &update_mtime() argument is really in _secs_, not in _msecs_ as the docs claim.
1143                         # Be aware '*1000000' would overflow Perl integer anyway.
1144                         # &set_last_modified would also override the "Last-Modified" headers_out!
1145                         # &mtime may exist but somehow does not work.
1146                         $W->{"r"}->update_mtime(HTTP::Date::str2time($headers_out_hashref->{"Last-Modified"}));
1147                         $status=$W->{"r"}->meets_conditions();
1148                         $W->{"r"}->status($status_old);
1149                         }
1150                 last if OK==$status;
1151                 $W->{"r"}->status($status);
1152                 $W->{"header_only"}=1;  # Inhibit &footer output.
1153                 exit;
1154                 die "NOTREACHED";
1155                 }
1156
1157         $W->{"digest-md5"}=Digest::MD5->new();
1158         $W->{"cache_active"}=1;
1159         $W->{"r"}->add_output_filter(\&cache_output_filter);
1160 }
1161
1162 sub cache_finish_last_modified()
1163 {
1164         cluck "Not yet done now? W __PACKAGE__: ".$W->{"__PACKAGE__"}
1165                         if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
1166         for my $package_orig (@{$packages_used_array{$W->{"__PACKAGE__"}}}) {
1167                 local $_=$package_orig.".pm";
1168                 s{::}{/}g;
1169                 path_abs_disk "/$_","register"=>1;
1170                 }
1171         my $mtime_newest;
1172         for my $path_abs_disk (keys(%{$W->{"path_abs_disk_register"}})) {
1173                 my $mtime=(stat $path_abs_disk)[9];
1174                 do { cluck "No mtime for: $path_abs_disk"; next; } if !$mtime;
1175                 $mtime_newest=$mtime if !$mtime_newest || $mtime_newest<$mtime;
1176                 }
1177         cluck "No mtime_newest found for the current W __PACKAGE__: ".$W->{"__PACKAGE__"}
1178                         if !$mtime_newest;
1179         return HTTP::Date::time2str($mtime_newest);
1180 }
1181
1182
1183 sub cache_finish()
1184 {
1185         # Do not: return if !$W->{"uri_args_frozen"};
1186         # as we may have just gave 304 and 'exit;' without starting the caching.
1187         return if !$W->{"cache_active"};
1188
1189         # Headers may not be complete in this case; not sure, just trying.
1190         return if $W->{"r"}->connection()->aborted();
1191
1192         # Fill-in/check: %uri_args_frozen_to_headers_in_keys
1193         my $headers_in_keys_stored_arrayref_ref=\$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}};
1194         my @headers_in_keys=tied(%{$W->{"headers_in_RecordKeys"}})->accessed();
1195         if (!$$headers_in_keys_stored_arrayref_ref
1196                         || !Data::Compare::Compare(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)) {
1197                 cluck "Non-matching generated 'headers_in_keys' per 'uri_args_frozen' key:\n"
1198                                                 .Dumper(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)
1199                                 if $$headers_in_keys_stored_arrayref_ref;
1200                 # Build or possibly prevent such further warn dupes:
1201                 $$headers_in_keys_stored_arrayref_ref=\@headers_in_keys;
1202                 # Build or regenerate as obsoleted now:
1203                 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get(\@headers_in_keys);
1204                 }
1205
1206         # Prepare 'headers_out' for the future reusal:
1207         my %headers_out;
1208         # Do not: $W->{"digest-md5"}->b64digest();
1209         # as it will not provide the trailing filling '='s.
1210         # RFC 1864 is not clear if they should be there but its sample provides them.
1211         # Do not try to provide canonical "\r\n" form of newlines as is said by RFC 1864.
1212         # RFC 2068 (HTTP/1.1) section 14.16 says the newlines should NOT be converted for HTTP.
1213         # ',""' to avoid breaking the headers by its default "\n".
1214         $headers_out{"Content-MD5"}=MIME::Base64::encode_base64($W->{"digest-md5"}->digest(),"");
1215         # In fact we could also use MD5 for ETag as if we know ETag we also know MD5.
1216         # But this way we do not need to calculate MD5 and we still can provide such ETag. So.
1217         # $W->{"r"}->set_etag() ?
1218         $headers_out{"ETag"}='"'.Digest::MD5::md5_base64($W->{"uri_args_headers_in_frozen"}).'"';
1219         # $W->{"r"}->set_content_length() ?
1220         $headers_out{"Content-Length"}=$W->{"r"}->bytes_sent();
1221         my %Vary=map(($_=>1),(@headers_in_keys));
1222         for (keys(%Vary)) {
1223                 next if !/^_/;
1224                 $Vary{"*"}=1;
1225                 delete $Vary{$_};
1226                 }
1227         %Vary=("*"=>1) if $Vary{"*"};
1228         $headers_out{"Vary"}=join(", ",sort keys(%Vary)) if keys(%Vary);
1229         # $W->{"r"}->set_last_modified() ?
1230         $headers_out{"Last-Modified"}=cache_finish_last_modified();
1231
1232         # Fill-in/check: %uri_args_headers_in_frozen_to_headers_out
1233         my $headers_out_stored_hashref_ref=\$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}};
1234         if (!$$headers_out_stored_hashref_ref
1235                         || !Data::Compare::Compare(\%headers_out,$$headers_out_stored_hashref_ref)) {
1236                 cluck "Non-matching generated 'headers_out' per 'uri_args_headers_in_frozen' key:\n"
1237                                                 .Dumper(\%headers_out,$$headers_out_stored_hashref_ref)
1238                                 if $$headers_out_stored_hashref_ref;
1239                 # Build or possibly prevent such further warn dupes:
1240                 $$headers_out_stored_hashref_ref=\%headers_out;
1241                 }
1242
1243 ###print STDERR Dumper(\%uri_args_frozen_to_headers_in_keys,\%uri_args_headers_in_frozen_to_headers_out);
1244 }
1245
1246 sub heading()
1247 {
1248 my($class)=@_;
1249
1250         if (!$W->{"header_only"}) {
1251                 header("Content-Style-Type"=>"text/css");
1252                 # Do not: text/javascript
1253                 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
1254                 # "application/javascript" so far standardized till 2005-12-08 by:
1255                 #       http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
1256                 header("Content-Script-Type"=>"application/javascript");
1257                 # $W->{"r"}->content_languages() ?
1258                 do { header("Content-Language"=>$_) if $_; } for $W->{"language"};
1259                 }
1260         # TODO: Support also: private
1261         header("Cache-Control"=>"public");      # HTTP/1.1
1262
1263         # Use $W->{"charset"}=0 to disable charset.
1264         $W->{"charset"}="us-ascii"
1265                         if !defined $W->{"charset"} && (!defined($W->{"content_type"}) || $W->{"content_type"});
1266
1267         # Workaround bug
1268         #   https://bugzilla.mozilla.org/show_bug.cgi?id=120556
1269         # of at least
1270         #   Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1271         # http://validator.w3.org/ does not send ANY "Accept" headers!
1272         if (!defined $W->{"content_type"}) {
1273                 # Be _stable_ for "headers_in".
1274                 my $accept=$W->{"headers_in"}{"Accept"};
1275                 my $user_agent=$W->{"headers_in"}{"User-Agent"}||"";
1276                 $W->{"content_type"}="application/xhtml+xml"
1277                                 if !$accept && $user_agent=~m{^W3C_Validator/}i;
1278                 # Be _stable_:
1279                 my $negotiated=$class->Negotiate_choose([
1280                                 # Put the fallback variant as the first one.
1281                                 # Rate both variants the same to prefer "text/html" for undecided clients.
1282                                 # At least
1283                                 #   Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1284                                 # prefers "application/xhtml+xml" over "text/html" itself:
1285                                 #   text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5
1286                                 negotiate_variant(
1287                                                 "id"=>"text/html",
1288                                                 "content-type"=>"text/html",
1289                                                 "qs"=>0.6,
1290                                                 (!$W->{"charset"} ? () : "charset"=>$W->{"charset"}),
1291                                                 "lang"=>$W->{"language"},
1292                                                 ),
1293                                 negotiate_variant(
1294                                                 "id"=>"application/xhtml+xml",
1295                                                 "content-type"=>"application/xhtml+xml",
1296                                                 "qs"=>0.6,
1297                                                 (!$W->{"charset"} ? () : "charset"=>$W->{"charset"}),
1298                                                 "lang"=>$W->{"language"},
1299                                                 ),
1300                                 # application/xml ?
1301                                 # text/xml ?
1302                                 ]);
1303                 $W->{"content_type"}=$negotiated if !defined $W->{"content_type"};
1304                 }
1305         # mod_perl doc: If you set this header via the headers_out table directly, it
1306         #               will be ignored by Apache. So do not do that.
1307         my $type;
1308         if ($W->{"content_type"}) {
1309                 $type=MIME::Types->new()->type($W->{"content_type"});
1310                 cluck "MIME::Types type '".$W->{"content_type"}."' not known" if !$type;
1311                 }
1312         cluck "charset='".$W->{"charset"}."' does not match content-type='".$W->{"content_type"}."'"
1313                         if ($W->{"charset"} ? 1 : 0) != (!$type ? 0 : $type->isAscii());
1314         $W->{"r"}->content_type($W->{"content_type"}.(!$W->{"charset"} ? "" : "; charset=".$W->{"charset"}))
1315                         if $W->{"content_type"};
1316
1317         cache_start();
1318         # We still can append headers before we put out some text.
1319         # FIXME: It is not clean to still append them without overwriting.
1320         return if $W->{"heading_done"};
1321         Wprint '<?xml version="1.0" encoding="'.$W->{"charset"}.'"?>'."\n"
1322                         if (!$W->{"header_only"} || $W->{"header_only"} eq "xml") && (0
1323                                         || $W->{"content_type"}=~m{^application/\w+[+]xml$}
1324                                         || $W->{"content_type"} eq "text/vnd.wap.wml");
1325         return if $W->{"header_only"};
1326         # Split 'heading_done' for the proper handling of: /project/Rel.pm
1327         $W->{"heading_done"}++;
1328
1329         Wprint '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">'."\n";
1330         Wprint '<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="'.$W->{"language"}.'">'."\n";
1331         my $title=$W->{"title_prefix"}.join("",map({ ': '.$_; } ($W->{"title"} || ())));
1332         # Do not: cluck if $title=~/[<>]/;
1333         # as it is not solved just by: &a_href_inhibit
1334         # as sometimes titles use also: <i>...</i>
1335         $title=~s#<[^>]*>##g;
1336         Wprint "<head>";
1337         Wprint "<title>$title</title>\n";
1338         if ($W->{"have_css"}) {
1339                 # Everything can get overriden later.
1340                 for my $css ("/My/Web.css",@{$W->{"css_push"}}) {
1341                         Wprint <<"HERE";
1342 <link rel="stylesheet" type="text/css" href="@{[ uri_escaped(path_web $css) ]}" />
1343 HERE
1344                         }
1345                 if ($W->{"css_inherit"}) {
1346                         # Do not: <script />
1347                         # as at least Lynx inhibits any further HTML output.
1348                         # Do not: text/javascript
1349                         # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
1350                         # "application/javascript" so far standardized till 2005-12-08 by:
1351                         #       http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
1352                         Wprint <<"HERE";
1353 <script type="application/javascript" src="@{[ uri_escaped(path_web('/My/css_inherit.js')) ]}"></script>
1354 HERE
1355                         }
1356                 }
1357         Wprint '<meta name="robots" content="'.($W->{"indexme"} ? "" : "no" ).'index,follow" />'."\n";
1358         Wprint $W->{"head"};
1359         for my $type (qw(prev next index contents start up)) {
1360                 do { Wprint '<link rel="'.$type.'" href="'.uri_escaped(path_web $_).'" />'."\n" if $_; }
1361                                 for ($W->{"rel_$type"});
1362                 }
1363         Wprint "</head><body";
1364 #       Wprint ' bgcolor="black" text="white" link="aqua" vlink="teal"'
1365 #                       if $W->{"browser"}->netscape() && (!$W->{"browser"}->major() || $W->{"browser"}->major()<=4);
1366         Wprint $W->{"body_attr"};
1367         Wprint ">\n";
1368
1369         do { Wprint $_ if $_; } for $W->{"heading"};
1370 }
1371
1372 BEGIN {
1373         delete $W->{"__My::Web_init"};
1374         }
1375
1376 1;