Fixed forgotten missing: Wrequire 'My::Hash::Readonly';
[MyWeb.git] / 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                 );
41 our @ISA=qw(Tie::Handle Exporter);
42
43 my %packages_used_hash; # $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
44 my %packages_used_array;
45
46 BEGIN
47 {
48         use Carp qw(cluck confess);
49         $W->{"__My::Web_init"}=1;
50
51         sub Wrequire ($)
52         {
53         my($file)=@_;
54
55 #               print STDERR "Wrequire $file\n";
56                 $file=~s#/#::#g;
57                 $file=~s/[.]pm$//;
58                 my $class=$file;
59                 $file=~s#::#/#g;
60                 $file.=".pm";
61                 my %callers;
62                 for (my $depth=0;defined caller($depth);$depth++) {
63                         $callers{caller($depth)}=1;
64                         }
65                 my $selfpkg=__PACKAGE__;
66                 $callers{$selfpkg}=1;
67                 for my $target ($class,__PACKAGE__) {
68                         for my $caller (keys(%callers)) {
69                                 next if $caller eq $target;
70                                 next if $packages_used_hash{$caller}{$target}++;
71                                 push @{$packages_used_array{$caller}},$target;
72                                 }
73                         }
74                 eval { CORE::require "$file"; } or confess $@;
75                 1;      # Otherwise 'require' would already file above.
76         }
77
78         sub Wuse ($@)
79         {
80         my($file,@list)=@_;
81
82 #               print STDERR "Wuse $file\n";
83                 Wrequire $file;
84                 local $Exporter::ExportLevel=$Exporter::ExportLevel+1;
85                 $file->import(@list);
86                 1;
87         }
88
89         sub import
90         {
91         my($class,@rest)=@_;
92
93                 local $Exporter::ExportLevel=$Exporter::ExportLevel+1;
94                 Wrequire("$class");
95                 return $class->SUPER::import(@rest);
96         }
97 }
98
99 use WebConfig;  # see also below: Wuse 'WebConfig';
100 require CGI;
101 require Image::Size;    # for &imgsize
102 use File::Basename;     # &basename
103 use Carp qw(cluck confess);
104 use URI::Escape;
105 require HTTP::BrowserDetect;
106 require HTTP::Negotiate;
107 our $have_Geo_IP; BEGIN { $have_Geo_IP=eval { require Geo::IP; 1; }; }
108 # Do not: use ModPerl::Util qw(exit);
109 # to prevent in mod_perl2: "exit" is not exported by the ModPerl::Util module
110 # I do not know why.
111 use POSIX qw(strftime);
112 use Tie::Handle;
113 use Apache2::Const qw(HTTP_MOVED_TEMPORARILY OK HTTP_OK);
114 use URI;
115 use URI::QueryParam;
116 use Cwd;
117 require HTTP::Date;
118 require Storable;
119 require Digest::MD5;
120 require Data::Compare;
121 use Data::Dumper;
122 require Encode;
123 use Apache2::Filter;
124 use Apache2::Connection;
125
126
127 #our $W;
128
129 sub cleanup($)
130 {
131 my($apache_request)=@_;
132
133         $packages_used_hash{$W->{"__PACKAGE__"}}{"_done"}=1;
134         cache_finish();
135         # Sanity protection.
136         $W=undef();
137         return OK;
138 }
139
140 sub request_check(;$)
141 {
142 my($self)=@_;
143
144         # Use &eval to prevent: Global $r object is not available. Set:\n\tPerlOptions +GlobalRequest\nin ...
145         # CGI requires valid "r": check it beforehand here.
146         confess "Calling sensitive dynamic code from a static code" if !eval { Apache2::RequestUtil->request(); };
147         # Do not: confess "Calling sensitive dynamic code without My::Web::init" if !$W->{"__PACKAGE__"};
148         # as it is valid at least while preparing arguments to call: &project::Lib::init
149 }
150
151 sub init ($%)
152 {
153 my($class,%args)=@_;
154
155         print STDERR "$class->init ".Apache2::RequestUtil->request()->unparsed_uri()."\n";
156
157         # We need to track package dependencies, so we need to call it from &init.
158         # We cannot do it in BEGIN { } block
159         # as it would not be tracked for each of the toplevel users later.
160         Wuse 'WebConfig';
161         Wrequire 'My::Hash';
162
163         $W=My::Hash->new({
164                 "__PACKAGE__"=>scalar(caller()),
165                 %WebConfig,
166                 %args,  # override %WebConfig settings
167                 },"My::Hash::Sub","My::Hash::Push");
168
169         # {"__PACKAGE__"} is mandatory for mod_perl-2.0;
170         # $Apache2::Registry::curstash is no longer supported.
171         do { cluck "No $_" if !$W->{$_}; } for "__PACKAGE__";
172
173         # See: &escapeHTML
174         do { cluck "charset==$_, expecting ISO-8859-1" if $_ ne "ISO-8859-1"; } for CGI::charset();
175         CGI::charset("utf-8");
176
177         do { $W->{$_}=0  if !defined $W->{$_}; } for "detect_ent";
178         do { $W->{$_}=0  if !defined $W->{$_}; } for "detect_js";
179         do { $W->{$_}=1  if !defined $W->{$_}; } for "have_css";        # AFAIK it does not hurt anyone.
180         do { $W->{$_}=0  if !defined $W->{$_}; } for "css_inherit";
181         do { $W->{$_}=1  if !defined $W->{$_}; } for "footer";
182         do { $W->{$_}=1  if !defined $W->{$_}; } for "footer_delimit";
183         do { $W->{$_}=1  if !defined $W->{$_}; } for "footer_ids";
184         do { $W->{$_}=1  if !defined $W->{$_}; } for "indexme";
185         do { $W->{$_}="" if !defined $W->{$_}; } for "head";
186         do { $W->{$_}="" if !defined $W->{$_}; } for "body_attr";
187         do { $W->{$_}="en-US" if !defined $W->{$_}; } for "language";
188
189         my $footer_any=0;
190         for (qw(footer_ids)) {
191                 $W->{$_}=0 if !$W->{"footer"};
192                 $footer_any=1 if $W->{$_};
193                 }
194         $W->{"footer"}=0 if !$footer_any;
195         $W->{"footer_delimit"}=0 if !$W->{"footer"};
196
197         $W->{"r"}=Apache2::RequestUtil->request();
198
199         $W->{"r"}->push_handlers("PerlCleanupHandler"=>\&cleanup);
200
201         $W->{"web_hostname"}||=$W->{"r"}->hostname();
202
203         tie *STDOUT,$W->{"r"};
204         select *STDOUT;
205         $|=1;
206
207         $W->{"QUERY_STRING"}=$W->{"r"}->args() || "";
208         if ($W->{"detect_ent"}) {
209                          if ($W->{"QUERY_STRING"}=~/[&]amp;have_ent/)
210                         { $W->{"have_ent"}=0; }
211                 elsif ($W->{"QUERY_STRING"}=~    /[&]have_ent/)
212                         { $W->{"have_ent"}=1; }
213                 else
214                         { delete $W->{"have_ent"}; }
215                 if (!defined $W->{"have_ent"} && $W->{"r"}->method() eq "GET") {
216                         $W->{"head"}.='<meta http-equiv="Refresh" content="0; URL='
217                                         .escapeHTML("http://".$W->{"web_hostname"}."/".($W->{"r"}->uri()=~m#^/*(.*)$#)[0]
218                                                         ."?".($W->{"QUERY_STRING"} || "detect_ent_glue=1").'&have_ent=detect')
219                                         .'" />'."\n";
220                         }
221                 }
222         $W->{"QUERY_STRING"}=~s/([&])amp;/$1/g;
223         $W->{"r"}->args($W->{"QUERY_STRING"});
224         # Workaround: &CGI::Vars behaves weird if strings passed both as POST data and in: $QUERY_STRING
225         do { $W->{"r"}->args(""); delete $ENV{"QUERY_STRING"}; } if $W->{"r"}->method() eq "POST";
226         # Do not: $W->{"r"}->args()
227         # as it parses only QUERY_STRING (not POST data).
228         $W->{"args_orig_array"}=[ CGI->new($W->{"r"})->Vars() ];
229         $W->{"args"}={ @{$W->{"args_orig_array"}} };
230         for my $name (keys(%{$W->{"args"}})) {
231                 my @vals=split /\x00/,$W->{"args"}{$name};
232                 next if @vals<=1;
233                 $W->{"args"}{$name}=[@vals];
234                 }
235
236         $W->{"headers_in"}=$W->{"r"}->headers_in();
237         Wrequire 'My::Hash::Merge';
238         $W->{"headers_in"}=My::Hash::Merge->new(
239                         $W->{"headers_in"},
240                         My::Hash::Sub->new({
241                                 "_remote_ip"=>sub { return $W->{"r"}->connection()->remote_ip(); },
242                                 }),
243                         );
244         Wrequire 'My::Hash::Readonly';
245         $W->{"headers_in"}=My::Hash::Readonly->new($W->{"headers_in"});
246         
247         if ($W->{"r"}->method() eq "GET" || $W->{"r"}->method() eq "HEAD") {
248                 for (\$W->{"http_safe"}) {
249                         # Do not: # Extend the current ETag system instead if you would need it:
250                         #         cluck "Explicitely NOT HTTP-Safe for method \"".$W->{"r"}->method()."\"?!?"
251                         #                       if defined($$_) && !$$_;
252                         # as sometimes it just does not make sense to cache it.
253                         $$_=1 if !defined $$_;
254                         }
255                 }
256         else {
257                 for (\$W->{"http_safe"}) {
258                         cluck "Undefined HTTP-Safe-ty for method \"".$W->{"r"}->method()."\"!"
259                                         if !defined($$_);
260                         $$_=0 if !defined $$_;
261                         }
262                 }
263         if ($W->{"http_safe"}) {
264                 Wrequire 'My::Hash::RecordKeys';
265                 $W->{"headers_in_RecordKeys"}=My::Hash::RecordKeys->new($W->{"headers_in"});
266                 $W->{"headers_in"}=$W->{"headers_in_RecordKeys"};
267                 }
268
269         $W->{"browser"}=HTTP::BrowserDetect->new($W->{"headers_in"}{"User-Agent"});
270
271         if (!defined $W->{"have_style"}) {
272                 $W->{"have_style"}=(!$W->{"browser"}->netscape() || ($W->{"browser"}->major() && $W->{"browser"}->major()>4) ? 1 : 0);
273                 }
274
275         $W->{"have_js"}=($W->{"args"}{"have_js"} ? 1 : 0);
276         if ($W->{"detect_js"} && !$W->{"have_js"}) {
277                 $W->{"head"}.='<script type="text/javascript" src="'.uri_escaped(path_web('/My/HaveJS.pm')).'"></script>'."\n";
278                 }
279
280         do { _args_check(%$_) if $_; } for ($W->{"args_check"});
281
282         return bless $W,$class;
283 }
284
285 # Be aware other parts of code (non-My::Web) will NOT use this function!
286 # Do not: Wprint $W->{"heading"},"undef"=>1;
287 # as we would need to undef() it to turn it off and it would get defaulted in such case.
288 # Do not: exists $W->{"heading"}
289 # as we use a lot of 'for $W->{"heading"}' which instantiates it with the value: undef()
290 sub Wprint($%)
291 {
292 my($text,%args)=@_;
293
294         cluck "undef Wprint" if !defined $text && !$args{"undef"};
295         delete $args{"undef"};
296         cluck join(" ","Invalid arguments:",keys(%args)) if keys(%args);
297         return if !defined $text;
298         cluck "utf-8 untested" if Encode::is_utf8($text);
299         $W->{"r"}->puts($text);
300 }
301
302 sub escapeHTML($)
303 {
304 my($text)=@_;
305
306         # Prevent &CGI::escapeHTML breaking utf-8 strings like: \xC4\x9B eq \x{11B}
307         # Prevent case if we run under mod_perl but still just initializing:
308         request_check() if $ENV{"MOD_PERL"};
309         # Generally we are initialized from &init but we may be used without it without mod_perl
310         # and in such case check the change on all non-first invocations.
311         our $init;
312         if (!$ENV{"MOD_PERL"} && $init++) {
313                 do { cluck "charset==$_" if $_ ne "utf-8"; } for CGI::charset();
314                 }
315         CGI::charset("utf-8");
316
317         return CGI::escapeHTML($text);
318 }
319
320 # /home/user/www/webdir
321 sub dir_top_abs_disk()
322 {
323         our $dir_top_abs_disk;
324         if (!$dir_top_abs_disk) {
325                 my $selfpkg_relpath=__PACKAGE__;
326                 $selfpkg_relpath=~s{::}{/}g;
327                 $selfpkg_relpath.=".pm";
328                 my $selfpkg_abspath=$INC{$selfpkg_relpath} or do {
329                         cluck "Unable to find self package $selfpkg_relpath";
330                         return;
331                         };
332                 $selfpkg_abspath=~s{/*\Q$selfpkg_relpath\E$}{} or do {
333                         cluck "Unable to strip myself \"$selfpkg_relpath\" from the abspath: $selfpkg_abspath";
334                         return;
335                         };
336                 cluck "INC{myself} is relative?: $selfpkg_abspath" if $selfpkg_abspath!~m{^/};
337                 $dir_top_abs_disk=$selfpkg_abspath;
338                 }
339         return $dir_top_abs_disk;
340 }
341
342 sub unparsed_uri()
343 {
344         request_check();
345         if (!$W->{"unparsed_uri"}) {
346                 # Do not: $W->{"r"}
347                 # as we may be called before &init from: &My::Project::init
348                 my $r=Apache2::RequestUtil->request();
349                 cluck "Calling ".'&unparsed_uri'." from a static code, going to fail" if !$r;
350                 my $uri_string=$r->unparsed_uri() or cluck "Valid 'r' missing unparsed_uri()?";
351                 my $uri=URI->new_abs($uri_string,"http://".$W->{"web_hostname"}."/");
352                 $W->{"unparsed_uri"}=$uri;
353                 }
354         return $W->{"unparsed_uri"};
355 }
356
357 sub in_to_uri_abs($)
358 {
359 my($in)=@_;
360
361         # Otherwise we may have been already processed and thus legally relativized.
362         # FIXME data: Currently disabled, all the data are too violating such rule.
363         if (0 && !ref $in) {
364                 my $uri_check=URI->new($in);
365                 $uri_check->scheme() || $in=~m{^\Q./\E} || $in=~m{^/}
366                                 or cluck "Use './' or '/' prefix for all the local references: $in";
367                 }
368         my $uri=URI->new_abs($in,unparsed_uri());
369         $uri=$uri->canonical();
370         return $uri;
371 }
372
373 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
374 # $args{"abs"}=1;
375 sub path_web($%)
376 {
377 my($in,%args)=@_;
378
379         cluck if !$args{"uri_as_in"} && ref $in;
380         my $uri=in_to_uri_abs($in);
381         if (uri_is_local($uri)) {
382                 # Prefer the $uri values over "args_persistent" values.
383                 $uri->query_form_hash({
384                                 map({
385                                         my $key=$_;
386                                         my $val=$W->{"args"}{$key};
387                                         (!defined $val ? () : ($key=>$val));
388                                         } keys(%{$W->{"args_persistent"}})),
389                                 %{$uri->query_form_hash()},
390                                 });
391                 }
392         return $uri->abs(unparsed_uri()) if $W->{"args"}{"Wabs"} || $args{"abs"};
393         return $uri->rel(unparsed_uri());
394 }
395
396 sub path_abs_disk_register($)
397 {
398 my($path_abs_disk)=@_;
399
400         $W->{"path_abs_disk_register"}{$path_abs_disk}=1;
401 }
402
403 # $args{"uri_as_in"}=1 to permit passing URI objects as: $in
404 sub path_abs_disk($%)
405 {
406 my($in,%args)=@_;
407
408         cluck if !$args{"uri_as_in"} && ref $in;
409         my $uri=in_to_uri_abs($in);
410         cluck if !uri_is_local($uri);
411         my $path=$uri->path();
412         cluck "URI compatibility: ->path() not w/leading slash of URI \"$uri\"; path: $path" if $path!~m{^/};
413         my $r=dir_top_abs_disk().$path;
414         path_abs_disk_register $r if !defined $args{"register"} || $args{"register"};
415         return $r;
416 }
417
418 sub fatal (;$);
419
420 sub _args_check (%)
421 {
422 my(%tmpl)=@_;
423
424         while (my($name,$regex)=each(%tmpl)) {
425                 my $name_html="Parameter <span class=\"quote\">".escapeHTML($name)."</span>";
426                 $W->{"args"}{$name}="" if !defined $W->{"args"}{$name};
427                 $W->{"args"}{$name}=[ $W->{"args"}{$name} ] if !ref $W->{"args"}{$name} && ref $regex;
428                 fatal "$name_html passed as multivar although singlevar expected"
429                                 if ref $W->{"args"}{$name} && !ref $regex;
430                 $regex=$regex->[0] if ref $regex;
431                 for my $val (!ref $W->{"args"}{$name} ? $W->{"args"}{$name} : @{$W->{"args"}{$name}}) {
432                         $val="" if !defined $val;
433                         fatal "$name_html <span class=\"quote\">".escapeHTML($val)."</span>"
434                                                         ." does not match the required regex <span class=\"quote\">".escapeHTML($regex)."</span> "
435                                         if $regex ne "" && $val!~/$regex/;
436                         }
437                 }
438 }
439
440 sub vskip (;$)
441 {
442 my($height)=@_;
443
444         return '<p'.(!defined $height ? "" : ' style="height: '.$height.';"').'>&nbsp;</p>'."\n";
445 }
446
447 sub fatal (;$)
448 {
449 my($msg)=@_;
450
451         $msg="UNKNOWN" if !$msg;
452         cluck "FATAL: $msg";
453
454         # Do not send it unconditionally.
455         # The intial duplicated '<?xml...' crashes Gecko parser.
456         $W->{"heading_done"}=0 if $W->{"header_only"};
457         # Do not send it unconditionally.
458         # Prevents warn: Headers already sent
459         if (!$W->{"heading_done"}) {
460                 $W->{"indexme"}=0;      # For the case no heading was sent yet.
461                 $W->{"header_only"}=0;  # assurance for &heading
462                 My::Web->heading();
463                 }
464         Wprint "\n".vskip("3ex")."<hr /><h1 class=\"error\">FATAL ERROR: $msg!</h1>\n"
465                         ."<p>You can report this problem's details to"
466                         ." ".a_href("mailto:".$W->{"admin_mail"},"admin of this website").".</p>\n";
467         footer();
468 }
469
470 sub footer (;$)
471 {
472         exit 1 if $W->{"footer_passed"}++;      # deadlock prevention:
473
474         Wprint vskip if $W->{"footer_delimit"};
475
476         do { Wprint $_ if $_; } for $W->{"footing_delimit"};
477
478         Wprint "<hr />\n" if $W->{"footer"};
479
480         my $packages_used=$packages_used_array{$W->{"__PACKAGE__"}};
481
482         if ($W->{"footer_ids"}) {
483                 Wprint '<p class="cvs-id">';
484                 Wprint join("<br />\n",map({ my $package=$_;
485                         my $cvs_id=(eval('$'.$package."::CVS_ID")
486 #                                       || $package     # debug
487                                         );
488                         if (!$cvs_id) {
489                                 ();
490                                 }
491                         else {
492                                 $cvs_id='$'.$cvs_id.'$';        # Eaten by 'q' operator.
493                                 my @cvs_id_split=split / +/,$cvs_id;
494                                 if (@cvs_id_split==8) {
495                                         my $file=$package;
496                                         $file=~s#::#/#g;
497                                         my $ext;
498                                         my @tried;
499                                         for (qw(.pm)) {
500                                                 $ext=$_;
501                                                 my $path_abs_disk=path_abs_disk("/$file$ext");
502                                                 push @tried,$path_abs_disk;
503                                                 last if -r $path_abs_disk;
504                                                 cluck "Class file $file not found; tried: ".join(" ",@tried) if !$ext;
505                                                 }
506                                         $file.=$ext;
507                                         $cvs_id_split[2]=""
508                                                         .a_href((map({ my $s=$_; $s=~s#/viewcvs/#$&~checkout~/#; $s; } $W->{"viewcvs"}))[0]."$file?rev=".$cvs_id_split[2],
509                                                                         $cvs_id_split[2]);
510                                         $cvs_id_split[1]=a_href($W->{"viewcvs"}.$file,
511                                                         ($package!~/^Apache2::/ ? $package : $cvs_id_split[1]));
512                                         $cvs_id_split[5]=&{$W->{"cvs_id_author_sub"}}($cvs_id_split[5]);
513                                         }
514                                 join " ",@cvs_id_split;
515                                 }
516                         } @$packages_used));
517                 Wprint "</p>\n";
518                 }
519
520         for my $package (@$packages_used) {
521                 my $cvs_id=(eval('$'.$package."::CVS_ID")
522 #                               || $package     # debug
523                                 );
524                 Wprint '<!-- '.$package.' - $'.$cvs_id.'$ -->'."\n" if $cvs_id;
525                 }
526
527         do { Wprint $_ if $_; } for $W->{"footing"};
528
529         Wprint "</body></html>\n";
530         exit 0;
531 }
532
533 # Existing entries are overwritten.
534 sub header(%)
535 {
536 my(%pairs)=@_;
537
538         while (my($key,$val)=each(%pairs)) {
539                 do { cluck "Headers already sent"; next; } if $W->{"heading_done"};
540                 $W->{"r"}->headers_out()->set($key,$val);
541                 }
542 }
543
544 sub size_display ($)
545 {
546 my($size)=@_;
547
548            if ($size<4096)
549                 {}
550         elsif ($size<1024*1024)
551                 { $size=sprintf "%.1fK",$size/1024; }
552         else
553                 { $size=sprintf "%.1fM",$size/1024/1024; }
554         $size.="B";
555         return $size;
556 }
557
558 sub uri_is_local($)
559 {
560 my($in)=@_;
561
562         my $uri_rel=in_to_uri_abs($in)->rel(unparsed_uri());
563         # Do not: defined $uri_rel->("userinfo"|"host"|"port")();
564         # as they fail to be called for schemes not supporting them.
565         return 0 if $uri_rel->scheme();
566         return 0 if $uri_rel->authority();
567         return 1;
568 }
569
570 # &path_web still may be required for &uri_escaped !
571 sub uri_escaped($)
572 {
573 my($uri)=@_;
574
575         cluck if !ref $uri;
576         my $urient=escapeHTML($uri);
577         return $uri    if $uri eq $urient;
578         request_check();
579         return $urient if uri_is_local $uri;
580         return $uri    if defined $W->{"have_ent"} && !$W->{"have_ent"};        # non-ent client
581         return $urient if $W->{"have_ent"};     # ent client
582         # Unknown client, &escapeHTML should not be needed here:
583         return escapeHTML(path_web('/My/Redirect.pm?location='.uri_escape($uri->abs(unparsed_uri()))));
584 }
585
586 our $a_href_inhibited;
587 sub a_href($;$%)
588 {
589 my($in,$contents,%args)=@_;
590
591         request_check();
592         do { $$_=1 if !defined $$_; } for (\$args{"size"});
593         if (!defined $contents) {
594                 $contents=$in;
595                 $contents=File::Basename::basename($contents) if $args{"basename"};
596                 $contents=escapeHTML($contents);
597                 }
598         $contents=~s#<a\b[^>]*>##gi;
599         $contents=~s#</a>##gi;
600         return $contents if $a_href_inhibited;
601
602         my $path_web=path_web $in,%args;
603         my $r="";
604         $r.='<a href="';
605         $r.=uri_escaped $path_web;
606         $r.='"';
607         do { $r.=" $_" if $_; } for ($args{"attr"});
608         $r.='>'.$contents.'</a>';
609         if ($args{"size"} && uri_is_local($in) && ($args{"size"}>=2 || $in=~/[.](?:gz|Z|rpm|zip|deb|lha)/)) {   # Downloadable?
610                 my $path_abs_disk=path_abs_disk $in,%args;
611                 cluck "File not readable: $path_abs_disk" if !-r $path_abs_disk;
612                 $r.='&nbsp;('.size_display((stat($path_abs_disk))[7]).')';
613                 }
614         return $r;
615 }
616
617 sub a_href_inhibit($$;@)
618 {
619 my($self,$sub,@sub_args)=@_;
620
621         local $a_href_inhibited=1;
622         return &{$sub}(@sub_args);
623 }
624
625 sub input_hidden_persistents()
626 {
627         request_check();
628         return join("",map({
629                 my $key=$_;
630                 my $val=$W->{"args"}{$key};
631                 (!defined $val ? () : '<input type="hidden"'
632                                 .' name="'.escapeHTML($key).'"'
633                                 .' value="'.escapeHTML($val).'"'
634                                 .' />'."\n");
635                 } (keys(%{$W->{"args_persistent"}}))));
636 }
637
638 sub http_moved($$;$)
639 {
640 my($self,$url,$status)=@_;
641
642         $url=path_web($url,"abs"=>1);
643         $status||=HTTP_MOVED_TEMPORARILY;
644         $W->{"r"}->status($status);
645         $W->{"r"}->headers_out()->{"Location"}=$url;
646         $W->{"header_only"}=1;
647         My::Web->heading();
648         exit;
649         die "NOTREACHED";
650 }
651
652 sub remote_ip ()
653 {
654         # Do not: PerlModule                 Apache2::ForwardedFor
655         #         PerlPostReadRequestHandler Apache2::ForwardedFor
656         # As 'Apache2::ForwardedFor' takes the first of $ENV{"HTTP_X_FORWARDED_FOR"}
657         # while the contents is '127.0.0.1, 213.220.195.171' if client has its own proxy.
658         # We must take the last item ourselves.
659         # Be VERY sure you always retrieve all the headers unconditionally to hit: My::Hash::RecordKeys
660         my $x_forwarded_for=$W->{"headers_in"}{"X-Forwarded-For"};
661         $x_forwarded_for=~s/^.*,\s*// if $x_forwarded_for;
662         my $remote_ip=$W->{"headers_in"}{"_remote_ip"};
663         my $r;
664         $r||=$x_forwarded_for;
665         $r||=$remote_ip;
666         return $r;
667 }
668
669 # $url={"JP"=>"http://specific",...};
670 # $url={""=>"http://default",...};
671 sub a_href_cc($$;%)
672 {
673 my($url,$contents,%args)=@_;
674
675         # A bit ineffective but we must process all the possibilities to get stable 'headers_in' hits!
676         my %map=map(($_=>a_href($url->{$_},$contents,%args)),keys(%$url));
677         my $cc;
678         $cc||=Geo::IP->new()->country_code_by_addr(remote_ip()) if $have_Geo_IP;
679         $cc||="";
680         my $r=$map{$cc};
681         return $r if $r;
682         return $contents;
683 }
684
685 sub make ($)
686 {
687 my($cmd)=@_;
688
689         # FIXME: &alarm, --timeout is now infinite.
690         # FIXME: Try to remove bash(1).
691         # FIXME: Use: @PATH_FLOCK@
692         my @argv=('flock',dir_top_abs_disk(),'bash','-c',$cmd.' >&2');
693         print STDERR join(" ","SPAWN:",@argv)."\n";
694         system @argv;
695 }
696
697 sub make_file($$)
698 {
699 my($self,$file)=@_;
700
701         cluck "Pathname not absolute: $file" if $file!~m{^/};
702         return if -f $file;
703         # TODO: Somehow quickly check dependencies?
704         return make('make -s --no-print-directory'
705                                         .' -C '."'".File::Basename::dirname($file)."' '".File::Basename::basename($file)."'");
706 }
707
708 sub img_size ($$)
709 {
710 my($width,$height)=@_;
711
712         cluck if !defined $width || !defined $height;
713         return ($W->{"have_style"} ? "style=\"border:0;width:${width}px;height:${height}px\"" : "border=\"0\"")
714                         ." width=\"$width\" height=\"$height\"";
715 }
716
717 sub negotiate_variant (%)
718 {
719 my(%args)=@_;
720
721         my @fields=("id","qs","content-type","encoding","charset","lang","size");
722         return [ map(($args{$_}),@fields) ];
723 }
724
725 # Input: $self is required!
726 # Input: Put the fallback variant as the first one.
727 # Returns: always only scalar!
728 sub Negotiate_choose($$)
729 {
730 my($self,$variants)=@_;
731
732         # Limit these entries to generate proper 'Vary' header.
733         my %hash=(map(($_=>$W->{"headers_in"}{$_}),qw(
734                         Accept
735                         Accept-Charset
736                         Accept-Encoding
737                         Accept-Language
738                         )));
739         my $best=HTTP::Negotiate::choose($variants,
740                         # Do not: $W->{"r"}
741                         # to prevent: Can't locate object method "scan" via package "Apache2::RequestRec" at HTTP/Negotiate.pm line 84.
742                         # Do not: $W->{"r"}->headers_in()
743                         # to prevent: Can't locate object method "scan" via package "APR::Table" at HTTP/Negotiate.pm line 84.
744                         # Do not: HTTP::Headers->new($W->{"r"}->headers_in());
745                         # to prevent empty result or even: Odd number of elements in anonymous hash
746                         HTTP::Headers->new(%hash));
747         $best||=$variants->[0][0];      # $variants->[0]{"id"}; &HTTP::Negotiate::choose failed?
748         return $best;
749 }
750
751 my @img_variants=(
752                 { "id"=>"png","qs"=>0.9,"content-type"=>"image/png" },
753                 { "id"=>"gif","qs"=>0.7,"content-type"=>"image/gif" },
754                 );
755 my $img_variants_re='[.](?:'.join('|',"jpeg",map(($_->{"id"}),@img_variants)).')$';
756
757 # Returns: ($path_web,$path_abs_disk)
758 # URI path segments support ignored here. Where it is used? (';' path segment options)
759 sub _img_src($%)
760 {
761 my($in,%args)=@_;
762
763         cluck if !uri_is_local $in;
764         my $uri=in_to_uri_abs $in;
765         my $path_abs_disk=path_abs_disk $uri,%args,"uri_as_in"=>1,"register"=>0;
766
767         # Known image extension?
768         return path_web($uri,%args,"uri_as_in"=>1),$path_abs_disk if $uri->path()=~m#$img_variants_re#o;
769
770         my @nego_variants;
771         for my $var (@img_variants) {
772                 my $path_abs_disk_variant=$path_abs_disk.".".$var->{"id"};
773                 path_abs_disk_register($path_abs_disk_variant);
774                 __PACKAGE__->make_file($path_abs_disk_variant);
775                 push @nego_variants,negotiate_variant(
776                                 %$var,
777                                 "size"=>(stat $path_abs_disk_variant)[7],
778                                 );
779                 }
780         my $ext=__PACKAGE__->Negotiate_choose(\@nego_variants);
781
782         $uri->path($uri->path().".$ext");
783         return path_web($uri,%args,"uri_as_in"=>1),path_abs_disk($uri,%args,"uri_as_in"=>1);
784 }
785
786 # $args{"attr"}
787 sub img ($$%)
788 {
789 my($in,$alt,%args)=@_;
790
791         request_check();
792         my($path_web,$path_abs_disk)=_img_src($in,%args);
793         my($width,$height)=Image::Size::imgsize($path_abs_disk);
794         $alt=~s/<[^>]*>//g;
795         $alt=escapeHTML($alt);
796         my $content="<img src=\"".uri_escaped($path_web)."\" alt=\"$alt\" title=\"$alt\" ".img_size($width,$height)
797                         .(!$args{"attr"} ? "" : " ".$args{"attr"})." />";
798         do { return a_href((_img_src($_))[0],$content,"uri_as_in"=>1) if $_; } for $args{"a_href_img"};
799         do { return a_href $_,$content if $_; } for $args{"a_href"};
800         return $content;
801 }
802
803 sub centerimg
804 {
805         my $r="";
806         $r.='<table border="0" width="100%"><tr>'."\n";
807         @_=( [@_] ) if !ref $_[0];
808         for (@_) {
809                 $r.="\t".'<td align="center">'.&{\&img}(@$_).'</td>'."\n";
810                 }
811         $r.='</tr></table>'."\n";
812         return $r;
813 }
814
815 sub rightimg
816 {
817 my($text,@args_img)=@_;
818
819         # FIXME: Workaround bug of 'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)':
820         #        <col width="@{[ (!$W->{"browser"}->ie() ? "1*" : "90%" ) ]}" />
821         #        <col width="@{[ (!$W->{"browser"}->ie() ? "0*" : "10%" ) ]}" />
822         # causes whole invisible projects in: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.10) Gecko/20050719 Galeon/1.3.21
823         return <<"HERE";
824 <table border="0" width="100%">
825         <tr>
826                 <td align="left">
827                         @{[ $text ]}
828                 </td>
829                 <td align="right">
830                         @{[ &{\&img}(@args_img) ]}
831                 </td>
832         </tr>
833 </table>
834 HERE
835 }
836
837 sub readfile($$)
838 {
839 my($class,$filename)=@_;
840
841         local *F;
842         open F,$filename or cluck "Cannot open \"$filename\": $!";
843         my $F=do { local $/=undef(); <F>; };
844         close F or cluck "Cannot close \"$filename\": $!";
845         return $F;
846 }
847
848 sub _no_cache($)
849 {
850 my($self)=@_;
851
852         header("Expires"=>HTTP::Date::time2str(1000000000));    # date in the past
853         header("Last-Modified"=>HTTP::Date::time2str());        # always modified
854         header("Cache-Control"=>join(", ",
855                         "no-cache",
856                         "no-store",
857                         "must-revalidate",
858                         "max-age=0",
859                         "pre-check=0",  # MSIE
860                         "post-check=0", # MSIE
861                         ));     # HTTP/1.1
862         header("Pragma"=>"no-cache");   # HTTP/1.0
863         header("Vary"=>"*");    # content may ba based on unpredictable sources
864 }
865
866 sub headers_in_filtered(@)
867 {
868 my(@keys)=@_;
869
870         return map(($_=>$W->{"headers_in"}{$_}),@keys);
871 }
872
873 our %uri_args_frozen_to_headers_in_keys;
874 our %uri_args_headers_in_frozen_to_headers_out;
875
876 sub uri_args_headers_in_frozen_get($)
877 {
878 my($headers_in_keys_arrayref)=@_;
879
880         my %uri_args_headers_in_hash=(
881                 "uri_args_frozen"=>$W->{"uri_args_frozen"},
882                 "headers_in"=>{ headers_in_filtered(@$headers_in_keys_arrayref) },
883                 );
884         return do { local $Storable::canonical=1; Storable::freeze(\%uri_args_headers_in_hash); };
885 }
886
887 sub cache_output_filter($)
888 {
889 my($f)=@_;
890
891         while ($f->read(my $text,0x400)) {
892                 cluck "utf-8 untested" if Encode::is_utf8($text);       # Possible here at all?
893                 $f->print($text);
894                 $W->{"digest-md5"}->add($text);
895                 }
896         return OK;
897 }
898
899 sub cache_start()
900 {
901         if (!$W->{"http_safe"}) {
902                 __PACKAGE__->_no_cache();
903                 return;
904                 }
905
906         {
907                 # &Wrequire it here even if it will not be later used; to be stable!
908                 Wrequire 'My::Hash::RestrictTo';
909                 my %uri_args_hash=(
910                         "uri"=>"http://".$W->{"web_hostname"}."/".$W->{"r"}->uri(),
911                         "args"=>$W->{"args_orig_array"},
912                         );
913                 $W->{"uri_args_frozen"}=do { local $Storable::canonical=1; Storable::freeze(\%uri_args_hash); };
914                 last if !(my $headers_in_keys_arrayref=$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}});
915
916                 # Protection to be sure we are stable:
917                 $W->{"headers_in"}=My::Hash::RestrictTo->new($W->{"headers_in"},@$headers_in_keys_arrayref);
918
919                 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get($headers_in_keys_arrayref);
920                 last if !(my $headers_out_hashref=$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}});
921                 header(%$headers_out_hashref);
922                 my $status;
923                 {
924                         # &meets_conditions will always deny the attempt if !2xx status().
925                         # At least ap_read_request() sets: r->status=HTTP_REQUEST_TIME_OUT;     /* Until we get a request */
926                         my $status_old=$W->{"r"}->status();
927                         $W->{"r"}->status(HTTP_OK);
928                         # Update httpd's 'r->mtime' as the header "Last-Modified" is just not enough for ap_meets_conditions():
929                         # &update_mtime() argument is really in _secs_, not in _msecs_ as the docs claim.
930                         # Be aware '*1000000' would overflow Perl integer anyway.
931                         # &set_last_modified would also override the "Last-Modified" headers_out!
932                         # &mtime may exist but somehow does not work.
933                         $W->{"r"}->update_mtime(HTTP::Date::str2time($headers_out_hashref->{"Last-Modified"}));
934                         $status=$W->{"r"}->meets_conditions();
935                         $W->{"r"}->status($status_old);
936                         }
937                 last if OK==$status;
938                 $W->{"r"}->status($status);
939                 exit 0;
940                 die "NOTREACHED";
941                 }
942
943         $W->{"digest-md5"}=Digest::MD5->new();
944         $W->{"cache_active"}=1;
945         $W->{"r"}->add_output_filter(\&cache_output_filter);
946 }
947
948 sub cache_finish_last_modified()
949 {
950         cluck "Not yet done now? W __PACKAGE__: ".$W->{"__PACKAGE__"}
951                         if !$packages_used_hash{$W->{"__PACKAGE__"}}{"_done"};
952         for my $package_orig (@{$packages_used_array{$W->{"__PACKAGE__"}}}) {
953                 local $_=$package_orig.".pm";
954                 s{::}{/}g;
955                 path_abs_disk "/$_","register"=>1;
956                 }
957         my $mtime_newest;
958         for my $path_abs_disk (keys(%{$W->{"path_abs_disk_register"}})) {
959                 my $mtime=(stat $path_abs_disk)[9];
960                 do { cluck "No mtime for: $path_abs_disk"; next; } if !$mtime;
961                 $mtime_newest=$mtime if !$mtime_newest || $mtime_newest<$mtime;
962                 }
963         cluck "No mtime_newest found for the current W __PACKAGE__: ".$W->{"__PACKAGE__"}
964                         if !$mtime_newest;
965         return HTTP::Date::time2str($mtime_newest);
966 }
967
968
969 sub cache_finish()
970 {
971         # Do not: return if !$W->{"uri_args_frozen"};
972         # as we may have just gave 304 and 'exit 0;' without starting the caching.
973         return if !$W->{"cache_active"};
974
975         # Fill-in/check: %uri_args_frozen_to_headers_in_keys
976         my $headers_in_keys_stored_arrayref_ref=\$uri_args_frozen_to_headers_in_keys{$W->{"uri_args_frozen"}};
977         my @headers_in_keys=tied(%{$W->{"headers_in_RecordKeys"}})->accessed();
978         if (!$$headers_in_keys_stored_arrayref_ref
979                         || !Data::Compare::Compare(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)) {
980                 cluck "Non-matching generated 'headers_in_keys' per 'uri_args_frozen' key:\n"
981                                                 .Dumper(\@headers_in_keys,$$headers_in_keys_stored_arrayref_ref)
982                                 if $$headers_in_keys_stored_arrayref_ref;
983                 # Build or possibly prevent such further warn dupes:
984                 $$headers_in_keys_stored_arrayref_ref=\@headers_in_keys;
985                 # Build or regenerate as obsoleted now:
986                 $W->{"uri_args_headers_in_frozen"}=uri_args_headers_in_frozen_get(\@headers_in_keys);
987                 }
988
989         # Prepare 'headers_out' for the future reusal:
990         my %headers_out;
991         $headers_out{"Content-MD5"}=$W->{"digest-md5"}->b64digest();
992         # In fact we could also use MD5 for ETag as if we know ETag we also know MD5.
993         # But this way we do not need to calculate MD5 and we still can provide such ETag. So.
994         # $W->{"r"}->set_etag() ?
995         $headers_out{"ETag"}='"'.Digest::MD5::md5_base64($W->{"uri_args_headers_in_frozen"}).'"';
996         # $W->{"r"}->set_content_length() ?
997         $headers_out{"Content-Length"}=$W->{"r"}->bytes_sent();
998         my %Vary=map(($_=>1),(@headers_in_keys));
999         for (keys(%Vary)) {
1000                 next if !/^_/;
1001                 $Vary{"*"}=1;
1002                 delete $Vary{$_};
1003                 }
1004         %Vary=("*"=>1) if $Vary{"*"};
1005         $headers_out{"Vary"}=join(", ",sort keys(%Vary));
1006         # $W->{"r"}->set_last_modified() ?
1007         $headers_out{"Last-Modified"}=cache_finish_last_modified();
1008
1009         # Fill-in/check: %uri_args_headers_in_frozen_to_headers_out
1010         my $headers_out_stored_hashref_ref=\$uri_args_headers_in_frozen_to_headers_out{$W->{"uri_args_headers_in_frozen"}};
1011         if (!$$headers_out_stored_hashref_ref
1012                         || !Data::Compare::Compare(\%headers_out,$$headers_out_stored_hashref_ref)) {
1013                 cluck "Non-matching generated 'headers_out' per 'uri_args_headers_in_frozen' key:\n"
1014                                                 .Dumper(\%headers_out,$$headers_out_stored_hashref_ref)
1015                                 if $$headers_out_stored_hashref_ref;
1016                 # Build or possibly prevent such further warn dupes:
1017                 $$headers_out_stored_hashref_ref=\%headers_out;
1018                 }
1019
1020 ###print STDERR Dumper(\%uri_args_frozen_to_headers_in_keys,\%uri_args_headers_in_frozen_to_headers_out);
1021 }
1022
1023 sub heading()
1024 {
1025 my($class)=@_;
1026
1027         if (!$W->{"header_only"}) {
1028                 header("Content-Style-Type"=>"text/css");
1029                 header("Content-Script-Type"=>"text/javascript");
1030                 # $W->{"r"}->content_languages() ?
1031                 do { header("Content-Language"=>$_) if $_; } for $W->{"language"};
1032                 }
1033         # TODO: Support also: private
1034         header("Cache-Control"=>"public");      # HTTP/1.1
1035
1036         # $ENV{"CLIENT_CHARSET"} ignored (mod_czech support dropped!)
1037         my $client_charset=$W->{"force_charset"} || "us-ascii";
1038
1039         # Workaround bug
1040         #   https://bugzilla.mozilla.org/show_bug.cgi?id=120556
1041         # of at least
1042         #   Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1043         my $mime;
1044         # http://validator.w3.org/ does not send ANY "Accept" headers!
1045         $mime||="application/xhtml+xml" if 1
1046                         && !$W->{"headers_in"}{"Accept"}
1047                         && ($W->{"headers_in"}{"User-Agent"}||"")=~m{^W3C_Validator/}i;
1048         $mime||=$class->Negotiate_choose([
1049                         # Put the fallback variant as the first one.
1050                         # Rate both variants the same to prefer "text/html" for undecided clients.
1051                         # At least
1052                         #   Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b) Gecko/20050217
1053                         # prefers "application/xhtml+xml" over "text/html" itself:
1054                         #   text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5
1055                         negotiate_variant(
1056                                         "id"=>"text/html",
1057                                         "content-type"=>"text/html",
1058                                         "qs"=>0.6,
1059                                         "charset"=>$client_charset,
1060                                         "lang"=>$W->{"language"},
1061                                         ),
1062                         negotiate_variant(
1063                                         "id"=>"application/xhtml+xml",
1064                                         "content-type"=>"application/xhtml+xml",
1065                                         "qs"=>0.6,
1066                                         "charset"=>$client_charset,
1067                                         "lang"=>$W->{"language"},
1068                                         ),
1069                         # application/xml ?
1070                         # text/xml ?
1071                         ]);
1072         # mod_perl doc: If you set this header via the headers_out table directly, it
1073         #               will be ignored by Apache. So do not do that.
1074         $W->{"r"}->content_type("$mime; charset=$client_charset");
1075
1076         cache_start();
1077         return if $W->{"header_only"};
1078         # We still can append headers before we put out some text.
1079         # FIXME: It is not clean to still append them without overwriting.
1080         return if $W->{"heading_done"}++;
1081
1082         Wprint '<?xml version="1.0" encoding="'.$client_charset.'"?>'."\n" if $mime=~m{^application/\w+[+]xml$};
1083         return if $W->{"xml_header_only"};
1084         Wprint '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">'."\n";
1085         Wprint '<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="'.$W->{"language"}.'">'."\n";
1086         my $title=$W->{"title_prefix"}.join("",map({ ': '.$_; } ($W->{"title"} || ())));
1087         # Do not: cluck if $title=~/[<>]/;
1088         # as it is not solved just by: &a_href_inhibit
1089         # as sometimes titles use also: <i>...</i>
1090         $title=~s#<[^>]*>##g;
1091         Wprint "<head>";
1092         Wprint "<title>$title</title>\n";
1093         if ($W->{"have_css"}) {
1094                 # Everything can get overriden later.
1095                 for my $css ("/My/Web.css",map((!$_ ? () : ("ARRAY" ne ref($_) ? $_ : @$_)),$W->{"css_push"})) {
1096                         Wprint <<"HERE";
1097 <link rel="stylesheet" type="text/css" href="@{[ uri_escaped(path_web $css) ]}" />
1098 HERE
1099                         }
1100                 if ($W->{"css_inherit"}) {
1101                         Wprint <<"HERE";
1102 <script type="text/javascript" src="@{[ uri_escaped(path_web('/My/css_inherit.js')) ]}" />
1103 HERE
1104                         }
1105                 }
1106         Wprint '<meta name="robots" content="'.($W->{"indexme"} ? "" : "no" ).'index,follow" />'."\n";
1107         Wprint $W->{"head"};
1108         for my $type (qw(prev next index contents start up)) {
1109                 do { Wprint '<link rel="'.$type.'" href="'.uri_escaped(path_web $_).'" />'."\n" if $_; }
1110                                 for ($W->{"rel_$type"});
1111                 }
1112         Wprint "</head><body";
1113 #       Wprint ' bgcolor="black" text="white" link="aqua" vlink="teal"'
1114 #                       if $W->{"browser"}->netscape() && (!$W->{"browser"}->major() || $W->{"browser"}->major()<=4);
1115         Wprint $W->{"body_attr"};
1116         Wprint ">\n";
1117
1118         do { Wprint $_ if $_; } for $W->{"heading"};
1119 }
1120
1121 BEGIN {
1122         delete $W->{"__My::Web_init"};
1123         }
1124
1125 1;