3 # Copyright (C) 2000, HELM Team.
5 # This file is part of HELM, an Hypertextual, Electronic
6 # Library of Mathematics, developed at the Computer Science
7 # Department, University of Bologna, Italy.
9 # HELM is free software; you can redistribute it and/or
10 # modify it under the terms of the GNU General Public License
11 # as published by the Free Software Foundation; either version 2
12 # of the License, or (at your option) any later version.
14 # HELM is distributed in the hope that it will be useful,
15 # but WITHOUT ANY WARRANTY; without even the implied warranty of
16 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 # GNU General Public License for more details.
19 # You should have received a copy of the GNU General Public License
20 # along with HELM; if not, write to the Free Software
21 # Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
23 # For details, see the HELM World-Wide-Web page,
24 # http://cs.unibo.it/helm/.
28 my $VERSION = "@VERSION@";
32 my $cgi_dir = "@HELM_CGI_DIR@";
35 # First of all, let's load HELM configuration
37 my $HELM_LIB_DIR = $ENV{"HELM_LIB_DIR"};
38 # this should be the only fixed constant
39 my $DEFAULT_HELM_LIB_DIR = "@HELM_LIB_DIR@";
40 if (defined ($HELM_LIB_DIR)) {
41 $HELM_LIB_PATH = $HELM_LIB_DIR."/configuration.pl";
43 $HELM_LIB_PATH = $DEFAULT_HELM_LIB_DIR."/configuration.pl";
46 # next require defines: $helm_dir, $html_link, $dtd_dir, $uris_dbm, $indexname
47 require $HELM_LIB_PATH;
49 # TEMP: TODO put these vars in configuration file configuration.xml
51 my $cachemode = $ENV{'HTTP_GETTER_CACHE_MODE'} ||
53 if (($cachemode ne 'gzipped') and ($cachemode ne 'normal')) {
54 die "Invalid HTTP_GETTER_CACHE_MODE environment variable, must be".
55 "'normal' or 'gzipped'\n";
57 my $helm_rdf_dir = $ENV{'HTTP_GETTER_RDF_DIR'} ||
58 "/usr/local/helm/rdf_library";
59 my $rdf_dbm = $ENV{'HTTP_GETTER_RDF_DBM'} ||
60 "/usr/local/helm/rdf_urls_of_uris";
61 my $rdf_indexname = $ENV{'HTTP_GETTER_RDF_INDEXNAME'} ||
63 # </move_to_conf_file>
65 # Let's override the configuration file
66 $style_dir = $ENV{"HELM_STYLE_DIR"} if (defined ($ENV{"HELM_STYLE_DIR"}));
67 $dtd_dir = $ENV{"HELM_DTD_DIR"} if (defined ($ENV{"HELM_DTD_DIR"}));
78 #CSC: mancano i controlli sulle condizioni di errore di molte funzioni
79 #CSC: ==> non e' robusto
80 #CSC: altra roba da sistemare segnata con CSC
82 my $d = new HTTP::Daemon LocalPort => 8081
83 or die "Error: port 8081 not available.";
84 my $myownurl = $d->url;
86 # Let's patch the returned URL
87 $myownurl =~ s/\/$//; # chop the final slash
88 my $myownport = $myownurl;
89 $myownport =~ s/http:\/\/(.*):(.*)/$2/;
90 $myownurl =~ s/http:\/\/(.*):(.*)/$1/;
91 ($myownurl) = gethostbyname($myownurl);
92 $myownurl = "http://".$myownurl.":".$myownport;
94 tie(%map, 'DB_File', $uris_dbm.".db", O_RDWR, 0664); # open dbs
95 tie(%rdf_map, 'DB_File', $rdf_dbm.".db", O_RDWR, 0664);
98 print "HTTP Getter $VERSION\n"; # print hello information
99 print "Please contact me at: <URL:", $myownurl."/", ">\n";
101 print "helm_dir: $helm_dir\n";
102 print "helm_rdf_dir: $helm_rdf_dir\n";
103 print "style_dir: $style_dir\n";
104 print "dtd_dir: $dtd_dir\n";
105 print "urls_of_uris db: $uris_dbm.db\n";
106 print "rdf db: $rdf_dbm.db\n";
107 print "cache mode: $cachemode\n";
108 print "indexname: $indexname\n";
109 print "rdf_indexname: $rdf_indexname\n";
112 $SIG{CHLD} = "IGNORE"; # do not accumulate defunct processes
113 $SIG{USR1} = \&update; # sent by the child to make the parent update
115 while (my $c = $d->accept) {
117 while (my $r = $c->get_request) {
118 #CSC: mancano i controlli di sicurezza
120 my $http_method = $r->method;
121 my $http_path = $r->url->path;
122 my $http_query = uri_unescape($r->url->query);
123 my $cgi = new CGI("$http_query");
124 my $inputuri = $cgi->param('uri');
125 print "\nRequest: ".$r->url."\n\n";
127 print "\nUnescaped query: ".$http_query."\n";
129 # "getxml" works with rdf uris
130 if ($http_method eq 'GET' and $http_path eq "/getxml") {
131 # finds the uri, url and filename
133 my $answerformat = $cgi->param('format');
134 my $patch_dtd = $cgi->param('patch_dtd');
135 $answerformat = "" if (not defined($answerformat));
136 $patch_dtd = "yes" if (not defined($patch_dtd));
137 if (($answerformat ne "gz") and ($answerformat ne "normal")
138 and ($answerformat ne "")) {
139 die "Wrong output format: $answerformat, must be 'normal' ".
142 if (($patch_dtd ne "yes") and ($patch_dtd ne "no")
143 and ($patch_dtd ne "")) {
144 die "Wrong param, patch_dtd must be 'yes' or 'no'\n";
147 my $filename = $inputuri;
149 if (not isRdfUri($inputuri)) { # standad cic: or theory: uri
150 $filename =~ s/^cic:(.*)/$1/;
151 $filename =~ s/^theory:(.*)/$1/;
153 # The "helm:rdf/<<...>>//cic:" prefix is kept, but quoted
154 # (so that a "helm:rdf/<<...>>//cic:" dir is created
156 $filename =~ s/^(.*)\/\/cic:(.*)/$2/;
157 $filename =~ s/^(.*)\/\/theory:(.*)/$2/;
158 $prefix =~ s/\Q$filename\E//;
161 $filename = $prefix.$filename;
164 my $url = resolve ($inputuri); # resolve uri in url
165 if (not defined($url)) { # uri not found in uri2url map
166 die "NOT FOUND: uri \"$inputuri\" can't be resolved\n";
167 $c->send_error(RC_NOT_FOUND)
168 } else { # uri found and mapped to url
169 my $extension; # file extension
170 if ($url =~ /\.xml$/) { # non gzipped file
172 } elsif ($url =~ /\.xml\.gz$/) { # gzipped file
173 $extension = ".xml.gz";
174 } else { # error: unknown extension
175 die "unexpected extension in url: $url, might be '.xml'".
179 if (not isRdfUri ($inputuri)) { # save in uri std cache dir
180 $filename = $helm_dir.$filename.$extension;
181 } else { # save in rdf metadata cache dir
182 $filename = $helm_rdf_dir."/".$filename.$extension;
185 print_request($inputuri,$url,$filename);
188 my $ciccontent = download($patch_dtd,$url,$filename,$answerformat);
190 if ($answerformat eq "normal") { # answer the client in text/xml
191 answer($c,$ciccontent,"text/xml","");
192 } else { # answer the client in text/xml, gzip encoding
193 answer($c,$ciccontent,"text/xml","x-gzip");
197 # "/register" does not work with rdf uris
198 } elsif ($http_method eq 'GET' and $http_path eq "/register") {
199 my $inputurl = $cgi->param('url');
200 print "Register requested...\n";
201 $map{$inputuri}=$inputurl;
203 # Now let's clean the cache
204 my $cicfilename = $inputuri;
205 $cicfilename =~ s/cic:(.*)/$1/;
206 $cicfilename =~ s/theory:(.*)/$1/;
208 print "Unlinking ".$helm_dir.$cicfilename.".xml[.gz]\n";
209 unlink ($helm_dir.$cicfilename.".xml");
210 unlink ($helm_dir.$cicfilename.".xml.gz");
212 kill(USR1,getppid()); # signal changes to parent
215 html_nice_answer($c,"Register done");
217 # "/resolve" works with rdf uri
218 } elsif ($http_method eq 'GET' and $http_path eq "/resolve") {
219 #my $outputurl = $map{$inputuri};
220 my $outputurl = resolve($inputuri);
221 $outputurl = "" if (not defined($outputurl));
222 $cont = "<?xml version=\"1.0\" ?>\n\n";
223 if ($outputurl eq "") {
224 $cont .= "<unresolved />\n";
226 $cont .= "<url value=\"$outputurl\" />\n";
228 answer($c,$cont,"text/xml","");
230 } elsif ($http_method eq 'GET' and $http_path eq "/getdtd") {
232 # my $filename = $inputuri;
233 # $filename = $dtd_dir."/".$filename;
234 # print "DTD: $inputuri ==> ($filename)\n";
235 # if (stat($filename)) {
236 # print "Using local copy\n";
237 # open(FD, $filename) or die "Cannot open $filename\n";
240 # # Vary bad heuristic for substituion of absolute URLs
241 # # for relative ones
242 # s/ENTITY (.*) SYSTEM\s+"/ENTITY $1 SYSTEM "$myownurl\/getdtd?uri=/g;
246 # answer($c,$cont,"text/xml","");
248 # die "Could not find DTD!";
251 } elsif ($http_method eq 'GET' and $http_path eq "/getstyleconf") {
252 my $filename = $inputuri;
253 $filename = $style_dir."/config/".$filename;
254 if (stat($filename)) {
255 open(FD, $filename) or die "Cannot open $filename\n";
258 s/DOCTYPE (.*) SYSTEM\s+"/DOCTYPE $1 SYSTEM "$myownurl\/getstyleconf?uri=/g;
262 answer($c,$cont,"text/plain","");
264 die "Could not find Style Configuration File!";
267 } elsif ($http_method eq 'GET' and $http_path eq "/getxslt") {
268 my $filename = $inputuri;
269 $filename = $style_dir."/".$filename;
270 print "XSLT: $inputuri ==> ($filename)\n";
271 if (stat($filename)) {
272 print "Using local copy\n";
273 open(FD, $filename) or die "Cannot open $filename\n";
276 # Vary bad heuristic for substituion of absolute URLs
278 s/xsl:import\s+href="/xsl:import href="$myownurl\/getxslt?uri=/g ;
279 s/xsl:include\s+href="/xsl:include href="$myownurl\/getxslt?uri=/g ;
283 answer($c,$cont,"text/xml","");
285 die "Could not find XSLT!";
288 # "/update" works with rdf uri
289 } elsif ($http_method eq 'GET' and $http_path eq "/update") {
290 # rebuild urls_of_uris db and rdf uris db
291 print "Update requested...\n";
293 kill(USR1,getppid()); # signal changes to parent
295 html_nice_answer($c,"Update done");
297 # "/ls" does not work with rdf uris
298 } elsif ($http_method eq 'GET' and $http_path eq "/ls") {
299 # send back keys that begin with a given uri
300 my ($uritype,$uripattern,$uriprefix);
301 my $baseuri = $cgi->param('baseuri');
302 chop $baseuri if ($baseuri =~ /.*\/$/); # remove trailing "/"
303 # output type, might be 'txt' or 'xml':
304 my $outype = $cgi->param('format');
305 $uripattern = $baseuri;
306 $uripattern =~ s/^.*:(.*)/$1/;
307 if ($baseuri =~ /^cic:/) {
309 } elsif ($baseuri =~ /^theory:/) {
312 $uritype = "invalid";
314 if ($uritype ne "invalid") { # uri is valid
315 if (($outype ne 'txt') and ($outype ne 'xml')) { # invalid out type
316 print "Invalid output type specified: $outype\n";
317 html_nice_answer($c,"Invalid output type, must be ".
319 } else { # valid output
320 print "BASEURI $baseuri, FORMAT $outype\n";
321 $cont = finduris($uritype,$uripattern,$outype);
322 if ($outype eq 'txt') {
323 answer($c,$cont,"text/plain","");
324 } elsif ($outype eq 'xml') {
325 answer($c,$cont,"text/xml","");
327 die "Internal error, exiting!";
330 } else { # invalid uri
331 print "Invalid uri: $baseuri, may begin with 'cic:', ".
332 "'theory:' or '*:'\n";
333 html_nice_answer($c,"Invalid uri , must begin with ".
334 "'cic:' or 'theory:'");
337 } elsif ($http_method eq 'GET' and $http_path eq "/help") {
339 print "Help requested!\n";
340 html_nice_answer($c,"HTTP Getter Version: $VERSION");
342 } elsif ($http_method eq 'GET' and $http_path eq "/getempty") {
343 # return an empty, but valid, xml document
344 my $emptyxml = <<EOXML;
345 <?xml version="1.0"?>
347 <!ELEMENT empty EMPTY>
351 answer($c, $emptyxml, "", "");
353 } elsif ($http_method eq 'GET' and $http_path =~ /\.cgi$/) {
355 print "CGI requested!\n";
356 if ($http_path !~ /^\/[^\/]*\.cgi$/) {
357 html_nice_answer($c,"Invalid CGI name: $http_path, ".
358 "you can't request CGI that contain a slash in the name\n");
359 } elsif (stat "$cgi_dir"."$http_path") {
360 if (not -x "$cgi_dir/$http_path") {
361 html_nice_answer($c,"CGI $http_path found but not ".
363 } else { # exec CGI and anser back its output
366 $ENV{'QUERY_STRING'} = $http_query;
367 my $cgi_output = `$cgi_dir/$http_path`;
368 answer($c,$cgi_output,"","");
372 html_nice_answer($c,"CGI '$http_path' not found ".
373 "in CGI dir '$cgi_dir'");
376 } else { # unsupported request
378 print "INVALID REQUEST!!!!!\n";
379 print "(PATH: ",$http_path,", ";
380 print "QUERY: ",$http_query,")\n";
381 $c->send_error(RC_FORBIDDEN)
384 print "\nRequest solved: ".$r->url."\n\n";
389 print "\nCONNECTION CLOSED\n\n";
394 #================================
396 sub isRdfUri { # return true if the uri is an rdf uri, false otherwise
398 # helm:rdf/cic:www.cs.unibo.it/helm/rdf/foo_schema//cic:\
399 # /Coq/Init/Logic/True_rec.con.types
401 # the format is "helm:rdf/<metadata_tree>:<metadata_scheme>//<xml_file_uri>"
404 if ($uri =~ /^helm:rdf(.*):(.*)\/\/(.*)/) {
411 sub resolve { # resolve an uri in a url, work both with standard cic: or theory:
413 print "RESOLVE subroutine\n";
415 print "GIVEN URI: \"$uri\"\n";
416 if (isRdfUri ($uri)) { # rdf uri, resolve using rdf db
417 print "IS A RDF URI\n";
418 print "I WILL RETURN '$rdf_map{$uri}'\n";
419 return ($rdf_map{$uri});
420 } else { # standard cic: or theory: uri, resolve using std uri db
421 print "IS NOT A RDF URI\n";
422 print "I WILL RETURN '$map{$uri}'\n";
427 sub finduris { # find uris for cic and theory trees generation
428 my ($uritype,$uripattern,$format) = @_;
430 my ($uri,$localpart,$basepart,$dirname,$suffix,$flags,$key);
431 my (@itemz,@already_pushed_dir);
432 my (%objects,%dirs); # map uris to suffixes' flags
433 my $debug=1; # for debug
435 print "FINDURIS, uritype: $uritype, uripattern: $uripattern, ".
436 "format: $format\n\n" if defined($debug);
438 if (($uritype eq "cic") or ($uritype eq "theory")) {
439 # get info only of one type: cic or theory
440 foreach (keys(%map)) { # select matching uris
442 if ($uri =~ /^$uritype:$uripattern(\/|$|\.)/) {
443 if ($uri =~ /^$uritype:$uripattern\//) { # directory match
445 $localpart =~ s/^$uritype:$uripattern\/(.*)/$1/;
446 } elsif ($uri =~ /^$uritype:$uripattern($|\.)/) { # file match
448 $localpart =~ s/^.*\/([^\/]*)/$1/;
450 die "Internal error, seems that requested match is none of ".
451 "directory match or file match";
453 print "LOCALPART: $localpart\n" if defined($debug);
455 if ($localpart =~ /^[^\/]*$/) { # no slash, an OBJECT
456 $basepart = $localpart;
457 $basepart =~ s/^([^.]*\.[^.]*)(\.types)?(\.ann)?/$1/;
458 # remove exts .types or
460 $flags = $objects{$basepart}; # get old flags
461 if ($localpart =~ /\.ann$/) {
462 $flags = add_flag("ann","YES",$flags);
464 $flags = add_flag("ann","NO",$flags);
466 if ($localpart =~ /\.types$/) {
467 $flags = add_flag("types","YES",$flags);
468 } elsif ($localpart =~ /\.types\.ann$/) {
469 $flags = add_flag("types","ANN",$flags);
471 $flags = add_flag("types","NO",$flags);
473 $objects{$basepart} = $flags; # save new flags
474 } else { # exists at least one slash, a DIR
475 ($dirname) = split (/\//, $localpart);
476 $dirs{$dirname} = ""; # no flags requirement for dir
481 die "getter internal error: unsupported uritype: \"$uritype\"";
483 # now generate OUTPUT:
484 # output will be saved in $content
485 if ($format eq "txt") { # TXT output
486 foreach $key (sort(keys %dirs)) {
487 $content .= "dir, " . $key . "\n";
489 foreach $key (sort(keys %objects)) {
490 $content .= "object, $key, " . $objects{$key} . "\n";
492 } elsif ($format eq "xml") { # XML output
493 $content .= '<?xml version="1.0" encoding="ISO-8859-1"?>' . "\n";
494 $content .= "<!DOCTYPE ls SYSTEM ";
495 $content .= "\"$myownurl/getdtd?uri=ls.dtd\">" . "\n\n";
496 $content .= "<ls>\n";
497 foreach $key (sort(keys %dirs)) {
498 $content .= "\t<section>$key</section>\n";
500 foreach $key (sort(keys %objects)) {
501 $content .= "\t<object name=\"$key\">\n";
502 $flags = $objects{$key};
503 $flags =~ s/^<(.*)>$/$1/;
504 my ($annflag,$typesflag) = split /,/,$flags;
505 $content .= "\t\t<ann value=\"$annflag\" />\n";
506 $content .= "\t\t<types value=\"$typesflag\" />\n";
507 $content .= "\t</object>\n";
509 $content .= "</ls>\n";
510 } else { # may not enter this branch
511 die "Getter internal error: invalid format received by finduris sub";
517 # manage string like: "<ann_flag,type_flag>"
518 # "ann_flag" may be one of "ann_YES", "ann_NO"
519 # "type_flag" may be one of "types_NO", "types_YES", "types_ANN"
520 # when adding a flag the max between the current flag and the new flag
521 # is taken, the orders are ann_NO < ann_YES and types_NO < types_YES <
523 my ($flagtype,$newflag,$str) = @_;
524 $str = "<,>" if ($str eq "");
525 ($str =~ s/^<(.*,.*)>$/$1/) or die "Internal error: ".
526 "wrong string format for flag adding in $str";
527 my ($annflag,$typeflag) = split /,/,$str;
528 if ($flagtype eq "ann") { # ANN flag handling
529 if ($newflag eq "YES") {
531 } elsif ($newflag eq "NO") {
532 $annflag = "NO" unless ($annflag eq "YES");
534 die "Internal error: annflag must be \"YES\" or \"NO\"";
536 } elsif ($flagtype eq "types") { # TYPES flag handling
537 if ($newflag eq "ANN") {
539 } elsif ($newflag eq "YES") {
540 $typeflag = "YES" unless ($typeflag eq "ANN");
541 } elsif ($newflag eq "NO") {
543 unless (($typeflag eq "ANN") or ($typeflag eq "YES"));
545 die "Internal error: typeflag must be \"YES\", \"NO\" or \"ANN\"";
548 die "Internal error: unsupported flagtype \"$flagtype\"";
550 $str = "<$annflag,$typeflag>";
553 #CSC: Too much powerful: creates even /home, /home/users/, ...
554 #CSC: Does not raise errors if could not create dirs/files
558 my @dirs = split /\//,$pathname;
560 foreach $dir (@dirs) {
561 $tmp = ((defined($tmp)) ? $tmp."\/".$dir : "");
569 my ($uri,$url,$filename) = @_;
570 print "<download_request>\n";
573 print "filename: $filename\n\n";
574 print "</download_request>\n";
583 sub gunzip { # gunzip a file and return the deflated content
586 my ($gz, $buffer, $cont);
588 print "deflating $filename ...\n";
589 $gz = gzopen($filename, "r")
590 or die "Cannot open gzip'ed file $filename: $gzerrno";
592 while ( $gz->gzread($buffer) > 0 ) {
595 die "Error while reading : $gzerrno\n" if $gzerrno != Z_STREAM_END ;
601 sub gzip { # gzip the content argument and save it to filename argument
602 my ($cont, $filename) = @_;
603 #my $debug=1; # debug only
605 print "gzopening $filename ...\n" if (defined($debug));
606 my $gz = gzopen($filename, "w")
607 or die "Cannot gzopen for writing file $filename: $gzerrno";
608 print "gzwriting ...\n" if (defined($debug));
609 $gz->gzwrite($cont) or die "error writing: $gzerrno, exiting!\n";
610 print "gzclosing ...\n" if (defined($debug));
615 my ($patch_dtd,$url,$filename,$format) = @_;
618 # print "DOWNLOAD subs receives url: \"$url\"\n";
619 # print "DOWNLOAD subs receives filename: \"$filename\"\n";
621 my $resourcetype; # retrieve mode: "normal" (.xml) or "gzipped" (.xml.gz)
622 if ($filename =~ /\.xml$/) { # set retrieve mode
623 $resourcetype = "normal";
624 } elsif ($filename =~ /\.xml\.gz$/) {
625 $resourcetype = "gzipped";
627 die "Unsupported download extension, might be '.gz' or '.xml'\n";
629 my $basefname = $filename;
630 $basefname =~ s/\.gz$//; # get base resource name removing trailing .gz
631 $cont = ""; # modified by side-effect by the callback function
634 if (stat($basefname)) {
635 $localfname=$basefname;
636 } elsif (stat($basefname.".gz")) {
637 $localfname=$basefname.".gz";
639 if ($localfname ne "") { # we already have local copy of requested file
640 # check both possible cache entry: gzipped or normal
641 print "Using local copy.\n";
642 if ($localfname =~ /\.xml\.gz$/) { # deflating cached file and return it
643 $cont = gunzip($localfname);
644 } elsif ($localfname =~ /\.xml$/) { # just return cached file
645 open(FD, $localfname) or die "Cannot open $localfname";
646 while(<FD>) { $cont .= $_; }
649 die "Internal error: unexpected file name $localfname,"
650 ."must end with '.gz' or '.xml.gz'\n";
652 } else { # download file from net
653 print "Downloading file ...\n"; # download file
654 my $ua = LWP::UserAgent->new;
655 my $request = HTTP::Request->new(GET => "$url");
656 my $response = $ua->request($request, \&callback);
658 # cache retrieved file to disk
659 # <ZACK/> TODO: inefficent, I haven't yet undestood how to deflate
660 # in memory gzipped file, without call "gzopen"
661 print "Storing file ...\n";
662 # print "Making dirs ...\n";
664 # print "Opening tmp file for writing ...\n";
665 open(FD, ">".$filename.".tmp") or die "Cannot open $filename.tmp\n";
666 # print "Writing on tmp file ...\n";
668 # print "Closing tmp file ...\n";
671 # handle cache conversion normal->gzipped or gzipped->normal as user choice
672 print "cachemode:$cachemode, resourcetype:$resourcetype\n"
673 if (defined($debug));
674 if (($cachemode eq 'normal') and ($resourcetype eq 'normal')) {
675 # cache the file as is
676 rename "$filename.tmp", $filename;
677 } elsif (($cachemode eq 'gzipped') and ($resourcetype eq 'gzipped')) {
678 # cache the file as is
679 # and update the $cont variabile with deflated content
680 rename "$filename.tmp", $filename;
681 $cont = gunzip($filename);
682 } elsif (($cachemode eq 'normal') and ($resourcetype eq 'gzipped')) {
683 # deflate cache entry
685 open(FD, "> $basefname") or die "cannot open $basefname\n";
686 $cont = gunzip($filename.".tmp");
689 unlink "$filename.tmp"; # delete old gzipped file
690 } elsif (($cachemode eq 'gzipped') and ($resourcetype eq 'normal')) {
691 # compress cache entry
692 print "gzipping ...\n" if (defined($debug));
693 gzip($cont, $basefname.".gz");
694 unlink "$filename.tmp"; # delete old uncompressed file
696 die "Internal error, unsopported cachemode, resourcetype couple\n";
698 # $cont now contained uncompressed data
700 if ($patch_dtd eq "yes") {
701 $cont =~ s/DOCTYPE (.*) SYSTEM\s+"http:\/\/www.cs.unibo.it\/helm\/dtd\//DOCTYPE $1 SYSTEM "$myownurl\/getdtd?uri=/g;
703 if ($format eq "gz") {
704 gzip($cont,"$basefname.tmp");
705 open (TMP, "< $basefname.tmp")
706 or die "Can't open tempfile: $filename.tmp, exiting!\n";
712 unlink ($basefname . ".tmp") or
713 die "cannot unlink temporary file: $basefname.tmp\n";
720 # answer the client setting content, Content-Type and Content-Enconding
722 my ($c,$cont,$contype,$contenc) = @_;
723 my $res = new HTTP::Response;
724 $res->content($cont);
725 $res->push_header("Content-Type" => $contype)
726 unless ($contype eq "");
727 $res->push_header("Content-Encoding" => $contenc)
728 unless ($contenc eq "");
729 $res->push_header("Cache-Control" => "no-cache");
730 $res->push_header("Pragma" => "no-cache");
731 $res->push_header("Expires" => "0");
732 $c->send_response($res);
735 sub html_nice_answer {
736 # answer the client whith a nice html document
737 my ($c,$content) = @_;
738 $content = "<html><body><h1>$content</h1></body></html>";
739 answer($c,$content,"text/html","");
743 #retrieve a file from an url and write it to a temp dir
744 #used for retrieve resource index from servers
746 my ($prefix, $URL) = @_;
747 my $ua = LWP::UserAgent->new;
748 my $request = HTTP::Request->new(GET => "$URL");
749 my $response = $ua->request($request, \&callback);
750 my ($filename) = reverse (split "/", $URL); # get filename part of the URL
751 open (TEMP, "> $prefix/$filename")
752 || die "Cannot open temporary file: $prefix/$filename\n";
760 tie(%map, 'DB_File', $uris_dbm.".db", O_RDWR, 0664);
762 tie(%rdf_map, 'DB_File', $rdf_dbm.".db", O_RDWR, 0664);
766 # rebuild dbs fetching resource indexes from servers.
767 # Rebuild urls_of_uris db (1) and rdf db (2)
769 $server, $idxfile, $rdf_idxfile, $uri, $url, $comp, $line,
771 %urls_of_uris, %rdf # local proxy of $map and $rdf_map
776 foreach my $file ($uris_dbm, $rdf_dbm) { # remove old db file
777 if (stat $file.".db") { # remove if exists
778 unlink ($file.".db") or die "can't unlink old db file: $file.db\n";
781 tie(%urls_of_uris, 'DB_File', $uris_dbm.".db", O_RDWR|O_CREAT, 0664);
782 tie(%rdf, 'DB_File', $rdf_dbm.".db", O_RDWR|O_CREAT, 0664);
784 open (SRVS, "< $servers_file") or
785 die "cannot open servers file: $servers_file\n";
786 @servers = <SRVS>; # read all servers
789 while ($server = pop @servers) { # cicle on servers in _reverse_ order
790 print "processing server: $server ...\n";
792 helm_wget($tmp_dir, $server."/".$indexname); # get index
793 helm_wget($tmp_dir, $server."/".$rdf_indexname); # get rdf index
794 $idxfile = $tmp_dir."/".$indexname;
795 $rdf_idxfile = $tmp_dir."/".$rdf_indexname;
797 open (INDEX, "< $idxfile") or # (1) REBUILD URLS_OF_URIS DB
798 die "cannot open temporary index file: $idxfile\n";
799 while ($line = <INDEX>) { # parse index and add entry to urls_of_uris db
801 ($uri,$comp) = split /[ \t]+/, $line;
804 $url = $uri . ".xml" . ".gz";
806 $url = $uri . ".xml";
808 $url =~ s/cic:/$server/;
809 $url =~ s/theory:/$server/;
810 $urls_of_uris{$uri} = $url;
813 open (RDF_INDEX, "< $rdf_idxfile") or # (2) REBUILD RDF DB
814 die "cannot open temporary rdf index file: $rdf_idxfile\n";
815 while ($line = <RDF_INDEX>) { # parse index and add entry to rdf db
817 ($uri,$comp) = split /[ \t]+/, $line; # comp is "gz" or nothing
820 $url = $uri . ".xml" . ".gz";
822 $url = $uri . ".xml";
824 $url =~ s/^helm:rdf(.*)\/\/cic:/$server/;
825 $url =~ s/^helm:rdf(.*)\/\/theory:/$server/;
829 close INDEX; # close indexes
831 die "cannot unlink temporary file: $idxfile\n" # remove temp files
832 if (unlink $idxfile) != 1;
833 die "cannot unlink temporary file: $rdf_idxfile\n"
834 if (unlink $rdf_idxfile) != 1;
838 untie(%urls_of_uris); # untie local proxies
840 tie(%map, 'DB_File', $uris_dbm.".db", O_RDWR, 0664); # retie global ones
841 tie(%rdf_map, 'DB_File', $rdf_dbm.".db", O_RDWR, 0664);
844 # vim modline: do not remove!