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 &text_cc
  35                 &vskip
  36                 &img &centerimg &rightimg &leftimg
  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         # $args{"first"}=1
  53         sub Wrequire ($%)
  54         {
  55         my($file,%args)=@_;
  56
  57 #               print STDERR "Wrequire $file\n";
  58                 $file=~s#/#::#g;
  59                 $file=~s/[.]pm$//;
  60                 my $class=$file;
  61                 $file=~s#::#/#g;
  62                 $file.=".pm";
  63                 my %callers;
  64                 for (my $depth=0;defined caller($depth);$depth++) {
  65                         $callers{caller($depth)}=1;
  66                         }
  67                 my $selfpkg=__PACKAGE__;
  68                 $callers{$selfpkg}=1;
  69                 for my $target ($class,__PACKAGE__) {
  70                         for my $caller (keys(%callers)) {
  71                                 next if $packages_used_hash{$caller}{$target}++;
  72                                 cluck "Appending to the '_done' package list: caller=$caller,target=$target"
  73                                                 if $packages_used_hash{$caller}{"_done"};
  74                                 if ($args{"first"}) {
  75                                         unshift @{$packages_used_array{$caller}},$target;
  76                                         }
  77                                 else {
  78                                         push @{$packages_used_array{$caller}},$target;
  79                                         }
  80                                 }
  81                         }
  82                 eval { CORE::require "$file"; } or confess $@;
  83                 1;      # Otherwise 'require' would already file above.
  84         }
  85
  86         sub Wuse ($@)
  87         {
  88         my($file,@list)=@_;
  89
  90 #               print STDERR "Wuse $file\n";
  91                 Wrequire $file;
  92                 local $Exporter::ExportLevel=$Exporter::ExportLevel+1;
  93                 $file->import(@list);
  94                 1;
  95         }
  96
  97         sub import
  98         {
  99         my($class,@rest)=@_;
 100
 101                 local $Exporter::ExportLevel=$Exporter::ExportLevel+1;
 102                 Wrequire("$class");
 103                 return $class->SUPER::import(@rest);
 104         }
 105 }
 106
 107 use WebConfig;  # see also below: Wuse 'WebConfig';
 108 require Image::Size;    # for &imgsize
 109 use File::Basename;     # &basename
 110 use Carp qw(cluck confess);
 111 use URI::Escape;
 112 require HTTP::BrowserDetect;
 113 require HTTP::Negotiate;
 114 our $have_Geo_IP; BEGIN { $have_Geo_IP=eval { require Geo::IP; 1; }; }
 115 # Do not: use ModPerl::Util qw(exit);
 116 # to prevent in mod_perl2: "exit" is not exported by the ModPerl::Util module
 117 # I do not know why.
 118 use POSIX qw(strftime);
 119 use Tie::Handle;
 120 use Apache2::Const qw(HTTP_MOVED_TEMPORARILY OK HTTP_OK);
 121 use URI;
 122 use URI::QueryParam;
 123 use Cwd;
 124 require HTTP::Date;
 125 require Storable;
 126 require Digest::MD5;
 127 require Data::Compare;
 128 use Data::Dumper;
 129 require Encode;
 130 use Apache2::RequestUtil;
 131 use Apache2::Filter;
 132 use Apache2::Connection;
 133 require MIME::Base64;
 134 use Apache2::ServerUtil;
 135 require MIME::Types;
 136 require MIME::Parser;
 137 use Apache2::RequestRec;
 138 use Apache2::RequestIO;
 139 use Apache2::Response;
 140
 141
 142 #our $W;
 143
 144 sub init ($%)
 145 {
 146 my($class,%args)=@_;
 147
 148         print STDERR "$class->init ".Apache2::RequestUtil->request()->unparsed_uri()."\n";
 149
 150         # We need to track package dependencies, so we need to call it from &init.
 151         # We cannot do it in BEGIN { } block
 152         # as it would not be tracked for each of the toplevel users later.
 153         Wuse 'WebConfig';
 154         Wrequire 'My::Hash';
 155
 156         # $W={} can get somehow created very easily.
 157         # Do not: cluck "W not empty:\n".Dumper($W) if keys(%$W);
 158         # to prevent (of $W->{"headers_in"}): TODO: Enumeration may not be expected.
 159         cluck "W not empty; __PACKAGE__ was: ".$W->{"__PACKAGE__"} if keys(%$W);
 160         $W=My::Hash->new({},"My::Hash::Sub","My::Hash::Push");
 161         bless $W,$class;
 162         %$W=(
 163                 "__PACKAGE__"=>scalar(caller()),
 164                 %WebConfig,
 165                 %args,  # override %WebConfig settings
 166                 );
 167
 168         # {"__PACKAGE__"} is mandatory for mod_perl-2.0;
 169         # $Apache2::Registry::curstash is no longer supported.
 170         do { cluck "No $_" if !$W->{$_}; } for "__PACKAGE__";
 171         exit_hook_start();
 172         # Package dependencies tracking only:
 173         Wrequire $W->{"__PACKAGE__"},"first"=>1;
 174
 175         do { $W->{$_}=0  if !defined $W->{$_}; } for "detect_ent";
 176         do { $W->{$_}=0  if !defined $W->{$_}; } for "detect_js";
 177         do { $W->{$_}=1  if !defined $W->{$_}; } for "have_css";        # AFAIK it does not hurt anyone.
 178         do { $W->{$_}=0  if !defined $W->{$_}; } for "css_inherit";
 179         do { $W->{$_}=1  if !defined $W->{$_}; } for "footer";
 180         do { $W->{$_}=1  if !defined $W->{$_}; } for "footer_delimit";
 181         do { $W->{$_}=1  if !defined $W->{$_}; } for "footer_ids";
 182         do { $W->{$_}=1  if !defined $W->{$_}; } for "indexme";
 183         do { $W->{$_}="" if !defined $W->{$_}; } for "body_attr";
 184         do { $W->{$_}="en-US" if !defined $W->{$_}; } for "language";
 185
 186         my $footer_any=0;
 187         for (qw(footer_ids)) {
 188                 $W->{$_}=0 if !$W->{"footer"};
 189                 $footer_any=1 if $W->{$_};
 190                 }
 191         $W->{"footer"}=0 if !$footer_any;
 192         $W->{"footer_delimit"}=0 if !$W->{"footer"};
 193
 194         $W->{"r"}=Apache2::RequestUtil->request();
 195
 196         $W->{"r"}->push_handlers("PerlCleanupHandler"=>\&cleanup);
 197
 198         $W->{"web_hostname"}||=$W->{"r"}->hostname();
 199
 200         tie *STDOUT,$W->{"r"};
 201         select *STDOUT;
 202         $|=1;
 203
 204         $W->{"headers_in"}=$W->{"r"}->headers_in();
 205         Wrequire 'My::Hash::Merge';
 206         $W->{"headers_in"}=My::Hash::Merge->new(
 207                         $W->{"headers_in"},
 208                         My::Hash::Sub->new({
 209                                 "_remote_ip"=>sub { return $W->{"r"}->connection()->remote_ip(); },
 210                                 }),
 211                         );
 212         Wrequire 'My::Hash::Readonly';
 213         $W->{"headers_in"}=My::Hash::Readonly->new($W->{"headers_in"});
 214
 215         if ($W->{"r"}->method() eq "GET" || $W->{"r"}->method() eq "HEAD") {
 216                 for (\$W->{"http_safe"}) {
 217                         # Do not: # Extend the current ETag system instead if you would need it:
 218                         #         cluck "Explicitely NOT HTTP-Safe for method \"".$W->{"r"}->method()."\"?!?"
 219                         #                       if defined($$_) && !$$_;
 220                         # as sometimes it just does not make sense to cache it.
 221                         $$_=1 if !defined $$_;
 222                         }
 223                 }
 224         else {
 225                 for (\$W->{"http_safe"}) {
 226                         cluck "Undefined HTTP-Safe-ty for method \"".$W->{"r"}->method()."\"!"
 227                                         if !defined($$_);
 228                         $$_=0 if !defined $$_;
 229                         }
 230                 }
 231         # Used only if: $W->{"http_safe"}
 232         # but we would cause on different method(): Appending to the '_done' package list
 233         Wrequire 'My::Hash::RecordKeys';
 234         if ($W->{"http_safe"}) {
 235                 $W->{"headers_in_RecordKeys"}=My::Hash::RecordKeys->new($W->{"headers_in"});
 236                 $W->{"headers_in"}=$W->{"headers_in_RecordKeys"};
 237                 }
 238
 239         {
 240                 local $_=$W->{"r"}->args() || "";
 241                 if ($W->{"detect_ent"}) {
 242                                  if (/[&]amp;have_ent/)
 243                                 { $W->{"have_ent"}=0; }
 244                         elsif (    /[&]have_ent/)
 245                                 { $W->{"have_ent"}=1; }
 246                         else
 247                                 { delete $W->{"have_ent"}; }
 248                         if (!defined $W->{"have_ent"} && $W->{"r"}->method() eq "GET") {
 249                                 $W->{"head_push"}='<meta http-equiv="Refresh" content="0; URL='
 250                                                 .escapeHTML("http://".$W->{"web_hostname"}."/".($W->{"r"}->uri()=~m#^/*(.*)$#)[0]
 251                                                                 ."?".($_ || "detect_ent_glue=1").'&have_ent=detect')
 252                                                 .'" />'."\n";
 253                                 }
 254                         }
 255                 s/([&])amp;/$1/g;
 256                 $W->{"r"}->args($_);
 257                 }
 258
 259         $W->{"args"}=URI->new("?".$W->{"r"}->args())->query_form_hash();
 260         $W->merge_post_args() if $W->{"r"}->method() eq "POST";
 261         # Prepare '$args' first to (FIXME: Why?) prevent: Not a reference
 262         my $args=$W->{"args"};
 263         $W->{"args_orig"}=Storable::dclone($args);
 264
 265         $W->{"browser"}=sub {
 266                 # Lazy-evaluation, we may not need the "User-Agent" header at all.
 267                 return our $r||=HTTP::BrowserDetect->new($W->{"headers_in"}{"User-Agent"});
 268                 };
 269
 270         if (!defined $W->{"have_style"}) {
 271                 $W->{"have_style"}=sub {
 272                         # Lazy-evaluation, we may not need the "User-Agent" header at all.
 273                         return our $r||=(!$W->{"browser"}->netscape() || ($W->{"browser"}->major() && $W->{"browser"}->major()>4) ? 1 : 0);
 274                         };
 275                 }
 276
 277         $W->{"have_js"}=($W->{"args"}{"have_js"} ? 1 : 0);
 278         if ($W->{"detect_js"} && !$W->{"have_js"}) {
 279                 # Do not: <script />
 280                 # as at least Lynx inhibits any further HTML output.
 281                 # Do not: text/javascript
 282                 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
 283                 # "application/javascript" so far standardized till 2005-12-08 by:
 284                 #       http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
 285                 $W->{"head_push"}='<script type="application/javascript" src="'.uri_escaped(path_web('/My/HaveJS.pm')).'"></script>'."\n";
 286                 }
 287
 288         # Required by &_args_check below.
 289         $W->{"_init_done"}=1;
 290
 291         do { _args_check(%$_) if $_; } for $W->{"args_check"};
 292
 293         return $W;
 294 }
 295
 296 sub form_method($)
 297 {
 298 my($method)=@_;
 299
 300         my $r=q{method="}.$method.q{"};
 301         return $r." ".q{enctype="application/x-www-form-urlencoded" accept-charset="us-ascii utf-8"} if $method eq "post";
 302         return $r." ".                                            q{accept-charset="us-ascii utf-8"} if $method eq "get";
 303         cluck "Undefined method: $method";
 304         return $r;
 305 }
 306
 307 sub merge_post_args($)
 308 {
 309 my($class)=@_;
 310
 311         my @post_args=$class->read_post_args();
 312         while (@post_args) {
 313                 my $name=shift @post_args;
 314                 my $data=shift @post_args;
 315                 my $ref=\$W->{"args"}{$name};
 316                    if (!defined $$ref)       { $$ref=$data; }
 317                 elsif (!ref $$ref)           { $$ref=[$$ref,$data]; }
 318                 elsif ("ARRAY" eq ref $$ref) { push @$$ref,$data; }
 319                 else {
 320                         cluck "Ignoring POST argument \"$name\", orig is weird:\n",Dumper($$ref);
 321                         }
 322                 }
 323         return;
 324 }
 325
 326 # Do not: use CGI;
 327 # as CGI parsing of POST vs. QUERY_STRING data, multiple-valued keys etc.
 328 # is too dense and causes weird problems, together with mod_perl etc.
 329 sub read_post_args($)
 330 {
 331 my($class)=@_;
 332
 333         local $_=$class->http_headers_in_for("Content-type")->content_type();
 334         return $class->read_multipart_form_data() if $_ eq "multipart/form-data";
 335         return $class->read_application_x_www_form_urlencoded() if $_ eq "application/x-www-form-urlencoded";
 336         cluck "Unknown POST data body, ignored: $_";
 337         return;
 338 }
 339
 340 sub read_application_x_www_form_urlencoded($)
 341 {
 342 my($class)=@_;
 343
 344         my $body="";
 345         for (;;) {
 346                 my $got=$W->{"r"}->read(my($buf),0x1000);
 347                 # Do not: cluck "Error reading POST data: $!" if !defined $got;
 348                 # as it should be done using: APR::Error exceptions
 349                 last if !$got;
 350                 $body.=$buf;
 351                 }
 352         return URI->new("?".$body)->query_form();
 353 }
 354
 355 sub read_multipart_form_data($)
 356 {
 357 my($class)=@_;
 358
 359         my $parser=MIME::Parser->new();
 360         # FIXME: No unlink()s done!
 361         $parser->output_under("/tmp");
 362
 363         local *R_FH;
 364         tie *R_FH,$W->{"r"};
 365         local *FH;
 366         tie *FH,"My::Web::ReadMerged",
 367                         join("",map(($_.": ".$W->{"headers_in"}{$_}."\n"),qw(
 368                                         Content-type
 369                                         )))."\n",
 370                         \*R_FH;
 371         my $body=$parser->parse(\*FH);
 372         cluck "No multipart POST request body?" if !$body->is_multipart();
 373
 374         return map((
 375                         $_->head()->mime_attr("content-disposition.name")
 376                         =>
 377                         join("",@{$_->body()})
 378                         ),$body->parts());
 379
 380         # TODO: Globalize, make it IO::* compatible, split to the merging part + IO::Scalar.
 381         package My::Web::ReadMerged;
 382
 383         require Tie::Handle;
 384         require Exporter;
 385         our @ISA=qw(Tie::Handle Exporter);
 386         use Carp qw(cluck confess);
 387
 388         sub READLINE($)
 389         {
 390         my($self)=@_;
 391
 392                 confess "Slurp not yet implemented" if !defined $/;
 393                 # Apache2::RequestIO does not support 'READLINE'!
 394                 for (;;) {
 395                         if (defined $self->{"data"} && $self->{"data"}=~s{^.*\Q$/\E}{}) {
 396                                 $self->{"offset"}+=length $&;
 397                                 return $&;
 398                                 }
 399                         my $fh_orig=$self->{"fh_orig"};
 400                         if (!$fh_orig) {
 401                                 my $r=$self->{"data"};
 402                                 delete $self->{"data"};
 403                                 $self->{"offset"}+=length $r if defined $r;
 404                                 return $r;
 405                                 }
 406                         my $got=read $fh_orig,my($buf),0x1000;
 407                         cluck "Error reading POST data: $!" if !defined $got;
 408                         delete $self->{"fh_orig"} if !$got;
 409                         cluck "INTERNAL: fh_orig should not exist here" if !defined $self->{"data"};
 410                         $self->{"data"}.=$buf;
 411                         }
 412         }
 413
 414         sub TELL($)
 415         {
 416         my($self)=@_;
 417
 418                 return $self->{"offset"};
 419         }
 420
 421         sub TIEHANDLE($$$)
 422         {
 423         my($class,$data,$fh_orig)=@_;
 424
 425                 my $self=bless {},$class;
 426                 $self->{"data"}=$data;
 427                 $self->{"offset"}=0;
 428                 $self->{"fh_orig"}=$fh_orig;
 429                 return $self;
 430         }
 431 }
 432
 433 sub cleanup($)
 434 {
 435 my($apache_request)=@_;
 436
 437         cluck "CORE::GLOBAL::exit hook not ran" if !$W->{"_exit_done"};
 438         cluck "packages not finalized" if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
 439         cache_finish();
 440         # Sanity protection.
 441         $W=undef();
 442         exit_hook_stop();
 443         return OK;
 444 }
 445
 446 # PerlResponseHandler is RUN_FIRST and &ModPerl::Util::exit returns OK, so no (sane) go.
 447 # PerlLogHandler is already too late to be able to produce any output.
 448 my $exit_orig;
 449 sub exit_hook
 450 {
 451         cluck "Missing ->init while in exit_hook()" if !$W->{"_init_done"};
 452         # &footer will call us recursively!
 453         footer() if !$W->{"_exit_done"}++;
 454         return &{$exit_orig}(@_);
 455 }
 456 sub exit_hook_start
 457 {
 458         do { cluck "exit_hook_start() twice?"; return; } if defined $exit_orig;
 459         $exit_orig=\&CORE::GLOBAL::exit;
 460         # Prevent: Subroutine CORE::GLOBAL::exit redefined
 461         no warnings 'redefine';
 462         *CORE::GLOBAL::exit=\&exit_hook;
 463 }
 464 sub exit_hook_stop
 465 {
 466         do { cluck "exit_hook_stop() without exit_hook_start()?"; return; }
 467                         if \&exit_hook ne \&CORE::GLOBAL::exit;
 468         do { cluck "INTERNAL: exit_orig uninitialized"; return; }
 469                         if !$exit_orig;
 470         # Prevent: Subroutine CORE::GLOBAL::exit redefined
 471         no warnings 'redefine';
 472         *CORE::GLOBAL::exit=$exit_orig;
 473         $exit_orig=undef();
 474 }
 475
 476 # Be aware other parts of code (non-My::Web) will NOT use this function!
 477 # Do not: Wprint $W->{"heading"},"undef"=>1;
 478 # as we would need to undef() it to turn it off and it would get defaulted in such case.
 479 # Do not: exists $W->{"heading"}
 480 # as we use a lot of 'for $W->{"heading"}' which instantiates it with the value: undef()
 481 sub Wprint($%)
 482 {
 483 my($text,%args)=@_;
 484
 485         cluck "undef Wprint" if !defined $text && !$args{"undef"};
 486         delete $args{"undef"};
 487         cluck join(" ","Invalid arguments:",keys(%args)) if keys(%args);
 488         return if !defined $text;
 489         # Do not: cluck "utf-8 untested" if Encode::is_utf8($text);
 490         # as it is valid here.
 491         $W->{"r"}->puts($text);
 492 }
 493
 494 sub request_check(;$)
 495 {
 496 my($self)=@_;
 497
 498         # Use &eval to prevent: Global $r object is not available. Set:\n\tPerlOptions +GlobalRequest\nin ...
 499         confess "Calling sensitive dynamic code from a static code" if !eval { Apache2::RequestUtil->request(); };
 500         # Do not: confess "Calling sensitive dynamic code without My::Web::init" if !$W->{"__PACKAGE__"};
 501         # as it is valid at least while preparing arguments to call: &project::Lib::init
 502 }
 503
 504 # Do not: use CGI;
 505 # as it is too much backward compatible regarding the charset encodings etc.
 506 # and the resulting code is too dense with no additional functionality for the recent content.
 507 sub escapeHTML($)
 508 {
 509 my($text)=@_;
 510
 511         local $_=$text;
 512         s{&}{&amp;}gso;
 513         s{<}{&lt;}gso;
 514         s{>}{&gt;}gso;
 515         s{"}{&quot;}gso;
 516         return $_;
 517 }
 518
 519 # /home/user/www/webdir
 520 sub dir_top_abs_disk()
 521 {
 522         our $dir_top_abs_disk;
 523         if (!$dir_top_abs_disk) {
 524                 my $selfpkg_relpath=__PACKAGE__;
 525                 $selfpkg_relpath=~s{::}{/}g;
 526                 $selfpkg_relpath.=".pm";
 527                 my $selfpkg_abspath=$INC{$selfpkg_relpath} or do {
 528                         cluck "Unable to find self package $selfpkg_relpath";
 529                         return;
 530                         };
 531                 $selfpkg_abspath=~s{/*\Q$selfpkg_relpath\E$}{} or do {
 532                         cluck "Unable to strip myself \"$selfpkg_relpath\" from the abspath: $selfpkg_abspath";
 533                         return;
 534                         };
 535                 cluck "INC{myself} is relative?: $selfpkg_abspath" if $selfpkg_abspath!~m{^/};
 536                 $dir_top_abs_disk=$selfpkg_abspath;
 537                 }
 538         return $dir_top_abs_disk;
 539 }
 540
 541 sub unparsed_uri()
 542 {
 543         request_check();
 544         if (!$W->{"unparsed_uri"}) {
 545                 # Do not: $W->{"r"}
 546                 # as we may be called before &init from: &My::Project::init
 547                 my $r=Apache2::RequestUtil->request();
 548                 cluck "Calling ".'&unparsed_uri'." from a static code, going to fail" if !$r;
 549                 my $uri_string=$r->unparsed_uri() or cluck "Valid 'r' missing unparsed_uri()?";
 550                 my $uri=URI->new_abs($uri_string,"http://".$W->{"web_hostname"}."/");
 551                 $W->{"unparsed_uri"}=$uri;
 552                 }
 553         return $W->{"unparsed_uri"};
 554 }
 555
 556 sub in_to_uri_abs($)
 557 {
 558 my($in)=@_;
 559
 560         # Otherwise we may have been already processed and thus legally relativized.
 561         # FIXME data: Currently disabled, all the data are too violating such rule.
 562         if (0 && !ref $in) {
 563                 my $uri_check=URI->new($in);
 564                 $uri_check->scheme() || $in=~m{^\Q./\E} || $in=~m{^/}
 565                                 or cluck "Use './' or '/' prefix for all the local references: $in";
 566                 }
 567         my $uri=URI->new_abs($in,unparsed_uri());
 568         $uri=$uri->canonical();
 569         return $uri;
 570 }
 571
 572 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
 573 # $args{"abs"}=0 || 1;  # overrides: $W->{"args"}{"Wabs"}
 574 sub path_web($%)
 575 {
 576 my($in,%args)=@_;
 577
 578         cluck if !$args{"uri_as_in"} && ref $in;
 579         my $uri=in_to_uri_abs($in);
 580         if (uri_is_local($uri)) {
 581                 # Prefer the $uri values over "args_persistent" values.
 582                 # &query_form_hash comes from: URI::QueryParam
 583                 $uri->query_form_hash({
 584                                 map({
 585                                         my $key=$_;
 586                                         my $val=$W->{"args"}{$key};
 587                                         (!defined $val ? () : ($key=>$val));
 588                                         } keys(%{$W->{"args_persistent"}})),
 589                                 %{$uri->query_form_hash()},
 590                                 });
 591                 }
 592         my $abs;
 593         do { $abs=$_ if defined; } for $W->{"args"}{"Wabs"},$args{"abs"};
 594         return $uri->abs(unparsed_uri()) if $abs;
 595         return $uri->rel(unparsed_uri());
 596 }
 597
 598 sub path_abs_disk_register($)
 599 {
 600 my($path_abs_disk)=@_;
 601
 602         $W->{"path_abs_disk_register"}{$path_abs_disk}=1;
 603 }
 604
 605 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
 606 sub path_abs_disk($%)
 607 {
 608 my($in,%args)=@_;
 609
 610         cluck if !$args{"uri_as_in"} && ref $in;
 611         my $uri=in_to_uri_abs($in);
 612         cluck if !uri_is_local($uri);
 613         my $path=$uri->path();
 614         cluck "URI compatibility: ->path() not w/leading slash of URI \"$uri\"; path: $path" if $path!~m{^/};
 615         my $r=dir_top_abs_disk().$path;
 616         path_abs_disk_register $r if !defined $args{"register"} || $args{"register"};
 617         return $r;
 618 }
 619
 620 sub fatal (;$);
 621
 622 sub _args_check (%)
 623 {
 624 my(%tmpl)=@_;
 625
 626         while (my($name,$regex)=each(%tmpl)) {
 627                 my $name_html="Parameter <span class=\"quote\">".escapeHTML($name)."</span>";
 628                 $W->{"args"}{$name}="" if !defined $W->{"args"}{$name};
 629                 $W->{"args"}{$name}=[ $W->{"args"}{$name} ] if !ref $W->{"args"}{$name} && ref $regex;
 630                 fatal "$name_html passed as multivar although singlevar expected"
 631                                 if ref $W->{"args"}{$name} && !ref $regex;
 632                 $regex=$regex->[0] if ref $regex;
 633                 for my $val (!ref $W->{"args"}{$name} ? $W->{"args"}{$name} : @{$W->{"args"}{$name}}) {
 634                         $val="" if !defined $val;
 635                         fatal "$name_html <span class=\"quote\">".escapeHTML($val)."</span>"
 636                                                         ." does not match the required regex <span class=\"quote\">".escapeHTML($regex)."</span> "
 637                                         if $regex ne "" && $val!~/$regex/;
 638                         }
 639                 }
 640 }
 641
 642 sub vskip (;$)
 643 {
 644 my($height)=@_;
 645
 646         return '<p'.(!defined $height ? "" : ' style="height: '.$height.';"').'>&nbsp;</p>'."\n";
 647 }
 648
 649 sub fatal (;$)
 650 {
 651 my($msg)=@_;
 652
 653         $msg="UNKNOWN" if !$msg;
 654         cluck "FATAL: $msg";
 655
 656         # Do not send it unconditionally.
 657         # The intial duplicated '<?xml...' crashes Gecko parser.
 658         $W->{"heading_done"}=0 if $W->{"header_only"};
 659         # Do not send it unconditionally.
 660         # Prevents warn: Headers already sent
 661         if (!$W->{"heading_done"}) {
 662                 $W->{"indexme"}=0;      # For the case no heading was sent yet.
 663                 $W->{"header_only"}=0;  # assurance for &heading
 664                 $W->{"content_type"}="text/html";       # Force HTML and avoid strictly checked XHTML.
 665                 My::Web->heading();
 666                 }
 667         Wprint "\n".vskip("3ex")."<hr /><h1 class=\"error\">FATAL ERROR: $msg!</h1>\n"
 668                         ."<p>You can report this problem's details to"
 669                         ." ".a_href("mailto:".$W->{"admin_mail"},"admin of this website").".</p>\n";
 670         exit;
 671 }
 672
 673 sub footer_packages_used_comments()
 674 {
 675         my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
 676         for my $package (@$packages_used) {
 677                 my $cvs_id=(eval('$'.$package."::CVS_ID")
 678 #                               || $package     # debug
 679                                 );
 680                 Wprint '<!-- '.$package.' - $'.$cvs_id.'$ -->'."\n" if $cvs_id;
 681                 }
 682 }
 683
 684 sub footer()
 685 {
 686         cluck 'Explicit &footer call is deprecated, !_exit_dne' if !$W->{"_exit_done"};
 687         exit if $W->{"footer_done"}++;  # deadlock prevention:
 688         &{$_}() for reverse @{$W->{"footer_sub_push"}};
 689         if ($W->{"header_only"}) {
 690                 $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
 691                 exit;
 692                 }
 693
 694         Wprint vskip if $W->{"footer_delimit"};
 695         &{$_}() for reverse @{$W->{"footing_delimit_sub_push"}};
 696         Wprint "<hr />\n" if $W->{"footer"};
 697
 698         # Never update the package list while we examine it!
 699         $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
 700
 701         # Dump the whole packages stack or just the primary one?
 702         #my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
 703         my $packages_used=[$W->{"__PACKAGE__"}];
 704
 705         if ($W->{"footer_ids"}) {
 706                 Wprint '<p class="cvs-id">';
 707                 Wprint join("<br />\n",map({ my $package=$_;
 708                         my $cvs_id=(eval('$'.$package."::CVS_ID")
 709 #                                       || $package     # debug
 710                                         );
 711                         if (!$cvs_id) {
 712                                 ();
 713                                 }
 714                         else {
 715                                 $cvs_id='$'.$cvs_id.'$';        # Eaten by 'q' operator.
 716                                 my @cvs_id_split=split / +/,$cvs_id;
 717                                 if (@cvs_id_split==8) {
 718                                         my $file=$package;
 719                                         $file=~s#::#/#g;
 720                                         my $ext;
 721                                         my @tried;
 722                                         for (qw(.pm)) {
 723                                                 $ext=$_;
 724                                                 my $path_abs_disk=path_abs_disk("/$file$ext");
 725                                                 push @tried,$path_abs_disk;
 726                                                 last if -r $path_abs_disk;
 727                                                 cluck "Class file $file not found; tried: ".join(" ",@tried) if !$ext;
 728                                                 }
 729                                         $file.=$ext;
 730                                         my $viewcvs;
 731                                         if ((my $file_cvs=$file)=~s{^My/}{}) {
 732                                                 $viewcvs=$W->{"viewcvs_My"}.$file_cvs;
 733                                                 }
 734                                         else {
 735                                                 $viewcvs=$W->{"viewcvs"}.$file;
 736                                                 }
 737                                         $cvs_id_split[2]=""
 738                                                         .a_href((map({ my $s=$_; $s=~s#/viewcvs/#$&~checkout~/#; $s; } $viewcvs))[0]."?rev=".$cvs_id_split[2],
 739                                                                         $cvs_id_split[2]);
 740                                         $cvs_id_split[1]=a_href($viewcvs,($package!~/^Apache2::/ ? $package : $cvs_id_split[1]));
 741                                         $cvs_id_split[5]=&{$W->{"cvs_id_author_sub"}}($cvs_id_split[5]);
 742                                         }
 743                                 join " ",@cvs_id_split;
 744                                 }
 745                         } @$packages_used));
 746                 Wprint "</p>\n";
 747                 }
 748
 749         footer_packages_used_comments();
 750
 751         do { Wprint $_ if $_; } for $W->{"footing"};
 752
 753         Wprint "</body></html>\n";
 754         exit;
 755 }
 756
 757 # Existing entries are overwritten.
 758 sub header(%)
 759 {
 760 my(%pairs)=@_;
 761
 762         while (my($key,$val)=each(%pairs)) {
 763                 do { cluck "Headers already sent"; next; } if $W->{"heading_done"};
 764                 $W->{"r"}->headers_out()->set($key,$val);
 765                 }
 766 }
 767
 768 sub size_display ($)
 769 {
 770 my($size)=@_;
 771
 772            if ($size<4096)
 773                 {}
 774         elsif ($size<1024*1024)
 775                 { $size=sprintf "%.1fK",$size/1024; }
 776         else
 777                 { $size=sprintf "%.1fM",$size/1024/1024; }
 778         $size.="B";
 779         return $size;
 780 }
 781
 782 sub uri_is_local($)
 783 {
 784 my($in)=@_;
 785
 786         my $uri_rel=in_to_uri_abs($in)->rel(unparsed_uri());
 787         # Do not: defined $uri_rel->("userinfo"|"host"|"port")();
 788         # as they fail to be called for schemes not supporting them.
 789         return 0 if $uri_rel->scheme();
 790         return 0 if $uri_rel->authority();
 791         return 1;
 792 }
 793
 794 # &path_web still may be required for &uri_escaped !
 795 sub uri_escaped($)
 796 {
 797 my($uri)=@_;
 798
 799         cluck if !ref $uri;
 800         my $urient=escapeHTML($uri);
 801         return $uri    if $uri eq $urient;
 802         request_check();
 803         return $urient if uri_is_local $uri;
 804         return $uri    if defined $W->{"have_ent"} && !$W->{"have_ent"};        # non-ent client
 805         return $urient if $W->{"have_ent"};     # ent client
 806         # Unknown client, &escapeHTML should not be needed here:
 807         return escapeHTML(path_web('/My/Redirect.pm?location='.uri_escape($uri->abs(unparsed_uri()))));
 808 }
 809
 810 our $a_href_inhibited;
 811 sub a_href($;$%)
 812 {
 813 my($in,$contents,%args)=@_;
 814
 815         request_check();
 816         do { $$_=1 if !defined $$_; } for (\$args{"size"});
 817         if (!defined $contents) {
 818                 $contents=$in;
 819                 $contents=File::Basename::basename($contents) if $args{"basename"};
 820                 $contents=~s/^mailto:([-.\w]+@[-.\w]+)$/$1/;
 821                 $contents=escapeHTML($contents);
 822                 }
 823         $contents=~s#<a\b[^>]*>##gi;
 824         $contents=~s#</a>##gi;
 825         return $contents if $a_href_inhibited;
 826
 827         my $path_web=path_web $in,%args;
 828         my $r="";
 829         $r.='<a href="';
 830         $r.=uri_escaped $path_web;
 831         $r.='"';
 832         do { $r.=" $_" if $_; } for ($args{"attr"});
 833         $r.='>'.$contents.'</a>';
 834         if ($args{"size"} && uri_is_local($in) && ($args{"size"}>=2 || $in=~/[.](?:gz|Z|rpm|zip|deb|lha)/)) {   # Downloadable?
 835                 my $path_abs_disk=path_abs_disk $in,%args;
 836                 cluck "File not readable: $path_abs_disk" if !-r $path_abs_disk;
 837                 $r.='&nbsp;('.size_display((stat($path_abs_disk))[7]).')';
 838                 }
 839         return $r;
 840 }
 841
 842 sub a_href_inhibit($$;@)
 843 {
 844 my($self,$sub,@sub_args)=@_;
 845
 846         local $a_href_inhibited=1;
 847         return &{$sub}(@sub_args);
 848 }
 849
 850 sub input_hidden_persistents()
 851 {
 852         request_check();
 853         return join("",map({
 854                 my $key=$_;
 855                 my $val=$W->{"args"}{$key};
 856                 (!defined $val ? () : '<input type="hidden"'
 857                                 .' name="'.escapeHTML($key).'"'
 858                                 .' value="'.escapeHTML($val).'"'
 859                                 .' />'."\n");
 860                 } (keys(%{$W->{"args_persistent"}}))));
 861 }
 862
 863 sub http_moved($$;$)
 864 {
 865 my($self,$url,$status)=@_;
 866
 867         $url=path_web($url,"abs"=>1);
 868         $status||=HTTP_MOVED_TEMPORARILY;
 869         $W->{"r"}->status($status);
 870         $W->{"r"}->headers_out()->{"Location"}=$url;
 871         $W->{"header_only"}=1;
 872         $W->{"content_type"}=0;
 873         $W->{"charset"}=0;
 874         My::Web->heading();
 875         exit;
 876         die "NOTREACHED";
 877 }
 878
 879 sub remote_ip ()
 880 {
 881         # Do not: PerlModule                 Apache2::ForwardedFor
 882         #         PerlPostReadRequestHandler Apache2::ForwardedFor
 883         # As 'Apache2::ForwardedFor' takes the first of $ENV{"HTTP_X_FORWARDED_FOR"}
 884         # while the contents is '127.0.0.1, 213.220.195.171' if client has its own proxy.
 885         # We must take the last item ourselves.
 886         # Be VERY sure you always retrieve all the headers unconditionally to hit: My::Hash::RecordKeys
 887         my $x_forwarded_for=$W->{"headers_in"}{"X-Forwarded-For"};
 888         $x_forwarded_for=~s/^.*,\s*// if $x_forwarded_for;
 889         my $remote_ip=$W->{"headers_in"}{"_remote_ip"};
 890         my $r;
 891         $r||=$x_forwarded_for;
 892         $r||=$remote_ip;
 893         return $r;
 894 }
 895
 896 sub _cc()
 897 {
 898         my $r;
 899         $r||=Geo::IP->new()->country_code_by_addr(remote_ip()) if $have_Geo_IP;
 900         $r||="";
 901         return $r;
 902 }
 903
 904 # $url={"JP"=>"http://specific",...};
 905 # $url={""=>"http://default",...};
 906 sub a_href_cc($$;%)
 907 {
 908 my($url,$contents,%args)=@_;
 909
 910         # A bit ineffective but we must process all the possibilities to get stable 'headers_in' hits!
 911         my %map=map(($_=>a_href($url->{$_},$contents,%args)),keys(%$url));
 912         my $r;
 913         $r||=$map{_cc()};
 914         $r||=$map{""};
 915         return $r if $r;
 916         return $contents;
 917 }
 918
 919 # $tree={"JP"=>"specific",...};
 920 # $tree={""=>"default",...};
 921 sub text_cc($)
 922 {
 923 my($tree)=@_;
 924
 925         cluck if !$tree->{""};
 926         my $r;
 927         $r||=$tree->{_cc()};
 928         $r||=$tree->{""};
 929         return $r;
 930 }
 931
 932 sub make ($)
 933 {
 934 my($cmd)=@_;
 935
 936         # FIXME: &alarm, --timeout is now infinite.
 937         # FIXME: Try to remove bash(1).
 938         # FIXME: Use: @PATH_FLOCK@
 939         my @argv=('flock',dir_top_abs_disk(),'bash','-c',$cmd.' >&2');
 940         print STDERR join(" ","SPAWN:",@argv)."\n";
 941         system @argv;
 942 }
 943
 944 sub make_file($$)
 945 {
 946 my($self,$file)=@_;
 947
 948         cluck "Pathname not absolute: $file" if $file!~m{^/};
 949         return if -f $file;
 950         # TODO: Somehow quickly check dependencies?
 951         return make('make -s --no-print-directory'
 952                                         .' -C '."'".File::Basename::dirname($file)."' '".File::Basename::basename($file)."'");
 953 }
 954
 955 sub img_size ($$)
 956 {
 957 my($width,$height)=@_;
 958
 959         cluck if !defined $width || !defined $height;
 960         return ($W->{"have_style"} ? "style=\"border:0;width:${width}px;height:${height}px\"" : "border=\"0\"")
 961                         ." width=\"$width\" height=\"$height\"";
 962 }
 963
 964 sub negotiate_variant (%)
 965 {
 966 my(%args)=@_;
 967
 968         my @fields=("id","qs","content-type","encoding","charset","lang","size");
 969         return [ map(($args{$_}),@fields) ];
 970 }
 971
 972 # Returns: 'HTTP::Headers' instance.
 973 sub http_headers_in_for($@)
 974 {
 975 my($self,@headers)=@_;
 976
 977         # Limit these entries to generate proper 'Vary' header.
 978         return HTTP::Headers->new(map(($_=>$W->{"headers_in"}{$_}),@headers));
 979 }
 980
 981 # Input: $self is required!
 982 # Input: Put the fallback variant as the first one.
 983 # Returns: always only scalar!
 984 sub Negotiate_choose($$)
 985 {
 986 my($self,$variants)=@_;
 987
 988         my $best=HTTP::Negotiate::choose($variants,
 989                         # Do not: $W->{"r"}
 990                         # to prevent: Can't locate object method "scan" via package "Apache2::RequestRec" at HTTP/Negotiate.pm line 84.
 991                         # Do not: $W->{"r"}->headers_in()
 992                         # to prevent: Can't locate object method "scan" via package "APR::Table" at HTTP/Negotiate.pm line 84.
 993                         # Do not: HTTP::Headers->new($W->{"r"}->headers_in());
 994                         # to prevent empty result or even: Odd number of elements in anonymous hash
 995                         $self->http_headers_in_for(qw(
 996                                         Accept
 997                                         Accept-Charset
 998                                         Accept-Encoding
 999                                         Accept-Language
1000                                         )));
1001         $best||=$variants->[0][0];      # $variants->[0]{"id"}; &HTTP::Negotiate::choose failed?
1002         return $best;
1003 }
1004
1005 # FIXME: Use for "content-type": MIME::Types
1006 my @img_variants=(
1007         { "id"=>"svg","qs"=>0.9,"content-type"=>"image/svg+xml" },
1008         { "id"=>"png","qs"=>0.8,"content-type"=>"image/png" },
1009         { "id"=>"gif","qs"=>0.7,"content-type"=>"image/gif" },
1010         # Do not prefer these to avoid fedding them for '*/*' browsers.
1011         { "id"=>"dia","qs"=>0.6,"content-type"=>"application/x-dia-diagram" },
1012         { "id"=>"fig","qs"=>0.6,"content-type"=>"image/x-xfig" },
1013         { "id"=>"fig","qs"=>0.6,"content-type"=>"application/x-xfig" },
1014         { "id"=>"sxd","qs"=>0.6,"content-type"=>"application/vnd.sun.xml.draw" },
1015         { "id"=>"sxi","qs"=>0.6,"content-type"=>"application/vnd.sun.xml.impress" },
1016         );
1017 # Unsupported by Image::Size 2.992: sxd sxi dia
1018 #   Supported by Image::Size 2.992: fig png gif jpeg
1019 #   Expensive by Image::Size 2.992: svg
1020 my %img_map=(
1021         "svg" =>[qw(png gif)],  # svg
1022         "png" =>[qw(png gif)],
1023         "jpeg"=>[qw(jpeg)],
1024         "dia" =>[qw(png gif)],  # svg
1025         "fig" =>[qw(fig png gif)],
1026         "sxd" =>[qw(png gif)],
1027         "sxi" =>[qw(png gif)],
1028         );
1029
1030 # Returns: ($path_web,$path_abs_disk)
1031 # URI path segments support ignored here. Where it is used? (';' path segment options)
1032 sub _img_src($%)
1033 {
1034 my($in,%args)=@_;
1035
1036         cluck if !uri_is_local $in;
1037         my $uri=in_to_uri_abs $in;
1038         my $path_abs_disk=path_abs_disk $uri,%args,"uri_as_in"=>1,"register"=>0;
1039
1040         cluck "Not exists image path_abs_disk: $path_abs_disk" if !-r $path_abs_disk;
1041         my($path_abs_disk_base,$ext_orig)=($path_abs_disk=~/^(.*)[.](\w+)$/) or cluck;
1042         my $map_arrayref=$img_map{$ext_orig} or cluck;
1043
1044         my $ext=$ext_orig;
1045         if (1!=@$map_arrayref) {
1046                 my @nego_variants;
1047                 for my $ext (@$map_arrayref) {
1048                         my $path_abs_disk_variant=$path_abs_disk_base.".".$ext;
1049                         path_abs_disk_register($path_abs_disk_variant);
1050                         __PACKAGE__->make_file($path_abs_disk_variant);
1051                         for my $var (@img_variants) {
1052                                 next if $var->{"id"} ne $ext;
1053                                 push @nego_variants,negotiate_variant(
1054                                                 %$var,
1055                                                 "size"=>(stat $path_abs_disk_variant)[7],
1056                                                 );
1057                                 }
1058                         }
1059                 $ext=__PACKAGE__->Negotiate_choose(\@nego_variants);
1060                 }
1061
1062         my $uri_path=$uri->path();
1063         $uri_path=~s/\Q.$ext_orig\E$/.$ext/ or cluck;
1064         $uri->path($uri_path);
1065         return path_web($uri,%args,"uri_as_in"=>1),path_abs_disk($uri,%args,"uri_as_in"=>1);
1066 }
1067
1068 # $args{"attr"}
1069 sub img ($$%)
1070 {
1071 my($in,$alt,%args)=@_;
1072
1073         request_check();
1074         my($path_web,$path_abs_disk)=_img_src($in,%args);
1075         my($width,$height)=Image::Size::imgsize($path_abs_disk);
1076         $alt=~s/<[^>]*>//g;
1077         $alt=escapeHTML($alt);
1078         my $content="<img src=\"".uri_escaped($path_web)."\" alt=\"$alt\" title=\"$alt\" ".img_size($width,$height)
1079                         .(!$args{"attr"} ? "" : " ".$args{"attr"})." />";
1080         do { return a_href((_img_src($_))[0],$content,"uri_as_in"=>1) if $_; } for $args{"a_href_img"};
1081         do { return a_href $_,$content if $_; } for $args{"a_href"};
1082         return $content;
1083 }
1084
1085 sub centerimg
1086 {
1087         my $r="";
1088         $r.='<table border="0" width="100%"><tr>'."\n";
1089         @_=( [@_] ) if !ref $_[0];
1090         for (@_) {
1091                 $r.="\t".'<td align="center">'.&{\&img}(@$_).'</td>'."\n";
1092                 }
1093         $r.='</tr></table>'."\n";
1094         return $r;
1095 }
1096
1097 # Optional: Provide 'text' as 1==@args_img item.
1098 sub rightimg
1099 {
1100 my($text,@args_img)=@_;
1101
1102         # FIXME: Workaround bug of 'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)':
1103         #        <col width="@{[ (!$W->{"browser"}->ie() ? "1*" : "90%" ) ]}" />
1104         #        <col width="@{[ (!$W->{"browser"}->ie() ? "0*" : "10%" ) ]}" />
1105         # causes whole invisible projects in: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.10) Gecko/20050719 Galeon/1.3.21
1106         return <<"HERE";
1107 <table border="0" width="100%" class="textimg">
1108         <tr>
1109                 <td align="left">
1110                         @{[ $text ]}
1111                 </td>
1112                 <td>&nbsp;&nbsp;</td>
1113                 <td align="right">
1114                         @{[ 1==@args_img ? $args_img[0] : &{\&img}(@args_img) ]}
1115                 </td>
1116         </tr>
1117 </table>
1118 HERE
1119 }
1120
1121 # Optional: Provide 'text' as 1==@args_img item.
1122 sub leftimg
1123 {
1124 my($text,@args_img)=@_;
1125
1126         # FIXME: Workaround bug of 'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)':
1127         #        <col width="@{[ (!$W->{"browser"}->ie() ? "0*" : "10%" ) ]}" />
1128         #        <col width="@{[ (!$W->{"browser"}->ie() ? "1*" : "90%" ) ]}" />
1129         # causes whole invisible projects in: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.10) Gecko/20050719 Galeon/1.3.21
1130         return <<"HERE";
1131 <table border="0" width="100%">
1132         <tr>
1133                 <td align="center">
1134                         @{[ 1==@args_img ? $args_img[0] : &{\&img}(@args_img) ]}
1135                 </td>
1136                 <td>&nbsp;&nbsp;</td>
1137                 <td align="left">
1138                         @{[ $text ]}
1139                 </td>
1140         </tr>
1141 </table>
1142 HERE
1143 }
1144
1145 sub readfile($$)
1146 {
1147 my($class,$filename)=@_;
1148
1149         local *F;
1150         open F,$filename or cluck "Cannot open \"$filename\": $!";
1151         my $F=do { local $/=undef(); <F>; };
1152         close F or cluck "Cannot close \"$filename\": $!";
1153         return $F;
1154 }
1155
1156 sub _no_cache($)
1157 {
1158 my($self)=@_;
1159
1160         header("Expires"=>HTTP::Date::time2str(1000000000));    # date in the past
1161         header("Last-Modified"=>HTTP::Date::time2str());        # always modified
1162         header("Cache-Control"=>join(", ",
1163                         "no-cache",
1164                         "no-store",
1165                         "must-revalidate",
1166                         "max-age=0",
1167                         "pre-check=0",  # MSIE
1168                         "post-check=0", # MSIE
1169                         ));     # HTTP/1.1
1170         header("Pragma"=>"no-cache");   # HTTP/1.0
1171         header("Vary"=>"*");    # content may ba based on unpredictable sources
1172 }
1173
1174 sub headers_in_filtered(@)
1175 {
1176 my(@keys)=@_;
1177
1178         return map(($_=>$W->{"headers_in"}{$_}),@keys);
1179 }
1180
1181 our %uri_args_frozen_to_headers_in_keys;
1182 our %uri_args_headers_in_frozen_to_headers_out;
1183
1184 sub uri_args_headers_in_frozen_get($)
1185 {
1186 my($headers_in_keys_arrayref)=@_;
1187
1188         my %uri_args_headers_in_hash=(
1189                 "uri_args_frozen"=>$W->{"uri_args_frozen"},
1190                 "headers_in"=>{ headers_in_filtered(@$headers_in_keys_arrayref) },
1191                 );
1192         return do { local $Storable::canonical=1; Storable::freeze(\%uri_args_headers_in_hash); };
1193 }
1194
1195 sub cache_output_filter($)
1196 {
1197 my($f)=@_;
1198
1199         while ($f->read(my $text,0x400)) {
1200                 cluck "utf-8 untested" if Encode::is_utf8($text);       # Possible here at all?
1201                 $f->print($text);
1202                 $W->{"digest-md5"}->add($text);
1203                 }
1204         return OK;
1205 }
1206
1207 sub cache_start()
1208 {
1209         # Used only if: !$W->{"http_safe"}
1210         # but we would cause on different method(): Appending to the '_done' package list
1211         # &Wrequire it here even if it will not be later used; to be stable!
1212         Wrequire 'My::Hash::RestrictTo';
1213         if (!$W->{"http_safe"}) {
1214                 __PACKAGE__->_no_cache();
1215                 return;
1216                 }
1217
1218         {
1219                 my %uri_args_hash=(
1220                         "method"=>$W->{"r"}->method(),
1221                         "uri"=>"http://".$W->{"web_hostname"}."/".$W->{"r"}->uri(),
1222                         "args"=>$W->{"args_orig"},
1223                         );
1224                 $W->{"uri_args_frozen"}=do { local $Storable::canonical=1; Storable::freeze(\%uri_args_hash); };
1225                 last if !(my $headers_in_keys_arrayref=$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}});
1226
1227                 # Protection to be sure we are stable:
1228                 $W->{"headers_in"}=My::Hash::RestrictTo->new($W->{"headers_in"},@$headers_in_keys_arrayref);
1229
1230                 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get($headers_in_keys_arrayref);
1231                 last if !(my $headers_out_hashref=$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}});
1232                 header(%$headers_out_hashref);
1233                 my $status;
1234                 {
1235                         # &meets_conditions will always deny the attempt if !2xx status().
1236                         # At least ap_read_request() sets: r->status=HTTP_REQUEST_TIME_OUT;     /* Until we get a request */
1237                         my $status_old=$W->{"r"}->status();
1238                         $W->{"r"}->status(HTTP_OK);
1239                         # Update httpd's 'r->mtime' as the header "Last-Modified" is just not enough for ap_meets_conditions():
1240                         # &update_mtime() argument is really in _secs_, not in _msecs_ as the docs claim.
1241                         # Be aware '*1000000' would overflow Perl integer anyway.
1242                         # &set_last_modified would also override the "Last-Modified" headers_out!
1243                         # &mtime may exist but somehow does not work.
1244                         $W->{"r"}->update_mtime(HTTP::Date::str2time($headers_out_hashref->{"Last-Modified"}));
1245                         $status=$W->{"r"}->meets_conditions();
1246                         $W->{"r"}->status($status_old);
1247                         }
1248                 last if OK==$status;
1249                 $W->{"r"}->status($status);
1250                 $W->{"header_only"}=1;  # Inhibit &footer output.
1251                 exit;
1252                 die "NOTREACHED";
1253                 }
1254
1255         $W->{"digest-md5"}=Digest::MD5->new();
1256         $W->{"cache_active"}=1;
1257         $W->{"r"}->add_output_filter(\&cache_output_filter);
1258 }
1259
1260 sub cache_finish_last_modified()
1261 {
1262         cluck "Not yet done now? W __PACKAGE__: ".$W->{"__PACKAGE__"}
1263                         if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
1264         for my $package_orig (@{$packages_used_array{$W->{"__PACKAGE__"}}}) {
1265                 local $_=$package_orig.".pm";
1266                 s{::}{/}g;
1267                 path_abs_disk "/$_","register"=>1;
1268                 }
1269         my $mtime_newest;
1270         for my $path_abs_disk (keys(%{$W->{"path_abs_disk_register"}})) {
1271                 my $mtime=(stat $path_abs_disk)[9];
1272                 do { cluck "No mtime for: $path_abs_disk"; next; } if !$mtime;
1273                 $mtime_newest=$mtime if !$mtime_newest || $mtime_newest<$mtime;
1274                 }
1275         cluck "No mtime_newest found for the current W __PACKAGE__: ".$W->{"__PACKAGE__"}
1276                         if !$mtime_newest;
1277         return HTTP::Date::time2str($mtime_newest);
1278 }
1279
1280
1281 sub cache_finish()
1282 {
1283         # Do not: return if !$W->{"uri_args_frozen"};
1284         # as we may have just gave 304 and 'exit;' without starting the caching.
1285         return if !$W->{"cache_active"};
1286
1287         # Headers may not be complete in this case; not sure, just trying.
1288         return if $W->{"r"}->connection()->aborted();
1289
1290         # Fill-in/check: %uri_args_frozen_to_headers_in_keys
1291         my $headers_in_keys_stored_arrayref_ref=\$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}};
1292         my @headers_in_keys=tied(%{$W->{"headers_in_RecordKeys"}})->accessed();
1293         if (!$$headers_in_keys_stored_arrayref_ref
1294                         || !Data::Compare::Compare(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)) {
1295                 cluck "Non-matching generated 'headers_in_keys' per 'uri_args_frozen' key:\n"
1296                                                 .Dumper(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)
1297                                 if $$headers_in_keys_stored_arrayref_ref;
1298                 # Build or possibly prevent such further warn dupes:
1299                 $$headers_in_keys_stored_arrayref_ref=\@headers_in_keys;
1300                 # Build or regenerate as obsoleted now:
1301                 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get(\@headers_in_keys);
1302                 }
1303
1304         # Prepare 'headers_out' for the future reusal:
1305         my %headers_out;
1306         # Do not: $W->{"digest-md5"}->b64digest();
1307         # as it will not provide the trailing filling '='s.
1308         # RFC 1864 is not clear if they should be there but its sample provides them.
1309         # Do not try to provide canonical "\r\n" form of newlines as is said by RFC 1864.
1310         # RFC 2068 (HTTP/1.1) section 14.16 says the newlines should NOT be converted for HTTP.
1311         # ',""' to avoid breaking the headers by its default "\n".
1312         $headers_out{"Content-MD5"}=MIME::Base64::encode_base64($W->{"digest-md5"}->digest(),"");
1313         # In fact we could also use MD5 for ETag as if we know ETag we also know MD5.
1314         # But this way we do not need to calculate MD5 and we still can provide such ETag. So.
1315         # $W->{"r"}->set_etag() ?
1316         $headers_out{"ETag"}='"'.Digest::MD5::md5_base64($W->{"uri_args_headers_in_frozen"}).'"';
1317         # $W->{"r"}->set_content_length() ?
1318         $headers_out{"Content-Length"}=$W->{"r"}->bytes_sent();
1319         my %Vary=map(($_=>1),(@headers_in_keys));
1320         for (keys(%Vary)) {
1321                 next if !/^_/;
1322                 $Vary{"*"}=1;
1323                 delete $Vary{$_};
1324                 }
1325         %Vary=("*"=>1) if $Vary{"*"};
1326         $headers_out{"Vary"}=join(", ",sort keys(%Vary)) if keys(%Vary);
1327         # $W->{"r"}->set_last_modified() ?
1328         $headers_out{"Last-Modified"}=cache_finish_last_modified();
1329
1330         # Fill-in/check: %uri_args_headers_in_frozen_to_headers_out
1331         my $headers_out_stored_hashref_ref=\$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}};
1332         if (!$$headers_out_stored_hashref_ref
1333                         || !Data::Compare::Compare(\%headers_out,$$headers_out_stored_hashref_ref)) {
1334                 cluck "Non-matching generated 'headers_out' per 'uri_args_headers_in_frozen' key:\n"
1335                                                 .Dumper(\%headers_out,$$headers_out_stored_hashref_ref)
1336                                 if $$headers_out_stored_hashref_ref;
1337                 # Build or possibly prevent such further warn dupes:
1338                 $$headers_out_stored_hashref_ref=\%headers_out;
1339                 }
1340
1341 ###print STDERR Dumper(\%uri_args_frozen_to_headers_in_keys,\%uri_args_headers_in_frozen_to_headers_out);
1342 }
1343
1344 sub heading()
1345 {
1346 my($class)=@_;
1347
1348         if (!$W->{"header_only"}) {
1349                 header("Content-Style-Type"=>"text/css");
1350                 # Do not: text/javascript
1351                 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
1352                 # "application/javascript" so far standardized till 2005-12-08 by:
1353                 #       http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
1354                 header("Content-Script-Type"=>"application/javascript");
1355                 # $W->{"r"}->content_languages() ?
1356                 do { header("Content-Language"=>$_) if $_; } for $W->{"language"};
1357                 }
1358         # TODO: Support also: private
1359         header("Cache-Control"=>"public");      # HTTP/1.1
1360
1361         # Use $W->{"charset"}=0 to disable charset.
1362         $W->{"charset"}="us-ascii"
1363                         if !defined $W->{"charset"} && (!defined($W->{"content_type"}) || $W->{"content_type"});
1364
1365         # Workaround bug
1366         #   https://bugzilla.mozilla.org/show_bug.cgi?id=120556
1367         # of at least
1368         #   Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1369         # http://validator.w3.org/ does not send ANY "Accept" headers!
1370         if (!defined $W->{"content_type"}) {
1371                 # Be _stable_ for "headers_in".
1372                 my $accept=$W->{"headers_in"}{"Accept"};
1373                 my $user_agent=$W->{"headers_in"}{"User-Agent"}||"";
1374                 $W->{"content_type"}="application/xhtml+xml"
1375                                 if !$accept && $user_agent=~m{^W3C_Validator/}i;
1376                 # Be _stable_:
1377                 my $negotiated=$class->Negotiate_choose([
1378                                 # Put the fallback variant as the first one.
1379                                 # Rate both variants the same to prefer "text/html" for undecided clients.
1380                                 # At least
1381                                 #   Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1382                                 # prefers "application/xhtml+xml" over "text/html" itself:
1383                                 #   text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5
1384                                 negotiate_variant(
1385                                                 "id"=>"text/html",
1386                                                 "content-type"=>"text/html",
1387                                                 "qs"=>0.6,
1388                                                 (!$W->{"charset"} ? () : "charset"=>$W->{"charset"}),
1389                                                 "lang"=>$W->{"language"},
1390                                                 ),
1391                                 negotiate_variant(
1392                                                 "id"=>"application/xhtml+xml",
1393                                                 "content-type"=>"application/xhtml+xml",
1394                                                 "qs"=>0.6,
1395                                                 (!$W->{"charset"} ? () : "charset"=>$W->{"charset"}),
1396                                                 "lang"=>$W->{"language"},
1397                                                 ),
1398                                 # application/xml ?
1399                                 # text/xml ?
1400                                 ]);
1401                 $W->{"content_type"}=$negotiated if !defined $W->{"content_type"};
1402                 }
1403         # mod_perl doc: If you set this header via the headers_out table directly, it
1404         #               will be ignored by Apache. So do not do that.
1405         my $type;
1406         if ($W->{"content_type"}) {
1407                 $type=MIME::Types->new()->type($W->{"content_type"});
1408                 cluck "MIME::Types type '".$W->{"content_type"}."' not known" if !$type;
1409                 }
1410         cluck "charset='".$W->{"charset"}."' does not match content-type='".$W->{"content_type"}."'"
1411                         if ($W->{"charset"} ? 1 : 0) != (!$type ? 0 : $type->isAscii());
1412         $W->{"r"}->content_type($W->{"content_type"}.(!$W->{"charset"} ? "" : "; charset=".$W->{"charset"}))
1413                         if $W->{"content_type"};
1414
1415         cache_start();
1416         # We still can append headers before we put out some text.
1417         # FIXME: It is not clean to still append them without overwriting.
1418         return if $W->{"heading_done"};
1419         Wprint '<?xml version="1.0" encoding="'.$W->{"charset"}.'"?>'."\n"
1420                         if (!$W->{"header_only"} || $W->{"header_only"} eq "xml") && (0
1421                                         || $W->{"content_type"}=~m{^application/\w+[+]xml$}
1422                                         || $W->{"content_type"} eq "text/vnd.wap.wml");
1423         return if $W->{"header_only"};
1424         # Split 'heading_done' for the proper handling of: /project/Rel.pm
1425         $W->{"heading_done"}++;
1426
1427         Wprint '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">'."\n";
1428         Wprint '<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="'.$W->{"language"}.'">'."\n";
1429         my $title=$W->{"title_prefix"}.join("",map({ ': '.$_; } ($W->{"title"} || ())));
1430         # Do not: cluck if $title=~/[<>]/;
1431         # as it is not solved just by: &a_href_inhibit
1432         # as sometimes titles use also: <i>...</i>
1433         $title=~s#<[^>]*>##g;
1434         Wprint "<head>";
1435         Wprint "<title>$title</title>\n";
1436         if ($W->{"have_css"}) {
1437                 # Everything can get overriden later.
1438                 for my $css ("/My/Web.css",@{$W->{"css_push"}}) {
1439                         Wprint <<"HERE";
1440 <link rel="stylesheet" type="text/css" href="@{[ uri_escaped(path_web $css) ]}" />
1441 HERE
1442                         }
1443                 if ($W->{"css_inherit"}) {
1444                         $W->{"js_push"}="/My/css_inherit.js";
1445                         }
1446                 }
1447         for my $js (@{$W->{"js_push"}}) {
1448                 # Do not: <script />
1449                 # as at least Lynx inhibits any further HTML output.
1450                 # Do not: text/javascript
1451                 # as it does not look as registered, at least according to: MIME::Types $VERSION 1.15
1452                 # "application/javascript" so far standardized till 2005-12-08 by:
1453                 #       http://www.ietf.org/internet-drafts/draft-hoehrmann-script-types-03.txt
1454                 Wprint <<"HERE";
1455 <script type="application/javascript" src="@{[ uri_escaped(path_web $js) ]}"></script>
1456 <script type="text/javascript"        src="@{[ uri_escaped(path_web $js) ]}"></script>
1457 HERE
1458                 }
1459         Wprint '<meta name="robots" content="'.($W->{"indexme"} ? "" : "no" ).'index,follow" />'."\n";
1460         for my $head (@{$W->{"head_push"}}) {
1461                 $head=&{$head}() if "CODE" eq ref $head;
1462                 Wprint $head;
1463                 }
1464         for my $type (qw(prev next index contents start up)) {
1465                 do { Wprint '<link rel="'.$type.'" href="'.uri_escaped(path_web $_).'" />'."\n" if $_; }
1466                                 for ($W->{"rel_$type"});
1467                 }
1468         Wprint "</head><body";
1469 #       Wprint ' bgcolor="black" text="white" link="aqua" vlink="teal"'
1470 #                       if $W->{"browser"}->netscape() && (!$W->{"browser"}->major() || $W->{"browser"}->major()<=4);
1471         Wprint $W->{"body_attr"};
1472         Wprint ">\n";
1473
1474         do { Wprint $_ if $_; } for $W->{"heading"};
1475 }
1476
1477 BEGIN {
1478         delete $W->{"__My::Web_init"};
1479         }
1480
1481 1;