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@";
33 my (%xml_map, %rdf_map, %xslt_map);
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";
58 $helm_rdf_dir = $ENV{'HTTP_GETTER_RDF_DIR'} || $helm_rdf_dir;
59 $rdf_dbm = $ENV{'HTTP_GETTER_RDF_DBM'} || $rdf_dbm;
60 $xslt_dbm = $ENV{'HTTP_GETTER_XSLT_DBM'} || $xslt_dbm;
61 $rdf_indexname = $ENV{'HTTP_GETTER_RDF_INDEXNAME'} || $rdf_indexname;
62 $xslt_indexname = $ENV{'HTTP_GETTER_XSLT_INDEXNAME'} || $xslt_indexname;
63 $servers_file = $ENV{'HTTP_GETTER_SERVERS_FILE'} || $servers_file;
64 $uris_dbm = $ENV{'HTTP_GETTER_URIS_DBM'} || $uris_dbm;
65 $dtdbaseurl = $ENV{'HTTP_GETTER_DTD_BASE_URL'} || $dtdbaseurl;
67 # </move_to_conf_file>
69 # Let's override the configuration file
70 $style_dir = $ENV{"HELM_STYLE_DIR"} if (defined ($ENV{"HELM_STYLE_DIR"}));
71 $dtd_dir = $ENV{"HELM_DTD_DIR"} if (defined ($ENV{"HELM_DTD_DIR"}));
83 #CSC: mancano i controlli sulle condizioni di errore di molte funzioni
84 #CSC: ==> non e' robusto
85 #CSC: altra roba da sistemare segnata con CSC
87 my $d = new HTTP::Daemon LocalPort => 8081
88 or die "Error: port 8081 not available.";
89 my $myownurl = $d->url;
91 # Let's patch the returned URL
92 $myownurl =~ s/\/$//; # chop the final slash
93 my $myownport = $myownurl;
94 $myownport =~ s/http:\/\/(.*):(.*)/$2/;
95 $myownurl =~ s/http:\/\/(.*):(.*)/$1/;
96 ($myownurl) = gethostbyname($myownurl);
97 $myownurl = "http://".$myownurl.":".$myownport;
99 tie(%xml_map, 'DB_File', $uris_dbm, O_RDWR, 0664); # open dbs
100 tie(%rdf_map, 'DB_File', $rdf_dbm, O_RDWR, 0664);
101 tie(%xslt_map, 'DB_File', $xslt_dbm, O_RDWR, 0664);
104 print "HTTP Getter $VERSION\n"; # print hello information
105 print "Please contact me at: <URL:", $myownurl."/", ">\n";
107 print "servers_file: $servers_file\n";
108 print "helm_dir: $helm_dir\n";
109 print "helm_rdf_dir: $helm_rdf_dir\n";
110 print "style_dir: $style_dir\n";
111 print "dtd_dir: $dtd_dir\n";
112 print "urls_of_uris db: $uris_dbm\n";
113 print "rdf db: $rdf_dbm\n";
114 print "xslt db: $xslt_dbm\n";
115 print "cache mode: $cachemode\n";
116 print "indexname: $indexname\n";
117 print "rdf_indexname: $rdf_indexname\n";
118 print "xslt_indexname: $xslt_indexname\n";
119 print "dtdbaseurl: $dtdbaseurl\n";
122 $SIG{CHLD} = "IGNORE"; # do not accumulate defunct processes
123 $SIG{USR1} = \&update; # sent by the child to make the parent update
125 while (my $c = $d->accept) {
127 while (my $r = $c->get_request) {
128 #CSC: mancano i controlli di sicurezza
130 my $http_method = $r->method;
131 my $http_path = $r->url->path;
132 my $http_query = uri_unescape($r->url->query);
133 my $cgi = new CGI("$http_query");
134 my $inputuri = $cgi->param('uri');
135 print "\nRequest: ".$r->url."\n\n";
137 print "\nUnescaped query: ".$http_query."\n";
139 # "getxml" works with rdf uris
140 if ($http_method eq 'GET' and $http_path eq "/getxml") {
141 # finds the uri, url and filename
143 my $answerformat = $cgi->param('format');
144 my $patch_dtd = $cgi->param('patch_dtd');
145 $answerformat = "" if (not defined($answerformat));
146 $patch_dtd = "yes" if (not defined($patch_dtd));
147 if (($answerformat ne "gz") and ($answerformat ne "normal")
148 and ($answerformat ne "")) {
149 die "Wrong output format: $answerformat, must be 'normal' ".
152 if (($patch_dtd ne "yes") and ($patch_dtd ne "no")
153 and ($patch_dtd ne "")) {
154 die "Wrong param, patch_dtd must be 'yes' or 'no'\n";
157 my $filename = $inputuri;
159 if (not isRdfUri($inputuri)) { # standad cic: or theory: uri
160 $filename =~ s/^cic:(.*)/$1/;
161 $filename =~ s/^theory:(.*)/$1/;
163 # The "helm:rdf/<<...>>//cic:" prefix is kept, but quoted
164 # (so that a "helm:rdf/<<...>>//cic:" dir is created
166 $filename =~ s/^(.*)\/\/cic:(.*)/$2/;
167 $filename =~ s/^(.*)\/\/theory:(.*)/$2/;
168 $prefix =~ s/\Q$filename\E//;
171 $filename = $prefix.$filename;
174 my $url = resolve ($inputuri); # resolve uri in url
175 if (not defined($url)) { # uri not found in uri2url map
176 die "NOT FOUND: uri \"$inputuri\" can't be resolved\n";
177 $c->send_error(RC_NOT_FOUND)
178 } else { # uri found and mapped to url
179 my $extension; # file extension
180 if ($url =~ /\.xml$/) { # non gzipped file
182 } elsif ($url =~ /\.xml\.gz$/) { # gzipped file
183 $extension = ".xml.gz";
184 } else { # error: unknown extension
185 die "unexpected extension in url: $url, might be '.xml'".
189 if (not isRdfUri ($inputuri)) { # save in uri std cache dir
190 $filename = $helm_dir.$filename.$extension;
191 } else { # save in rdf metadata cache dir
192 $filename = $helm_rdf_dir."/".$filename.$extension;
195 print_request($inputuri,$url,$filename);
198 my $ciccontent = download($patch_dtd,$url,$filename,$answerformat);
200 if ($answerformat eq "normal") { # answer the client in text/xml
201 answer($c,$ciccontent,"text/xml","");
202 } else { # answer the client in text/xml, gzip encoding
203 answer($c,$ciccontent,"text/xml","x-gzip");
207 # "/register" does not work with rdf uris
208 } elsif ($http_method eq 'GET' and $http_path eq "/register") {
209 my $inputurl = $cgi->param('url');
210 print "Register requested...\n";
211 $xml_map{$inputuri}=$inputurl;
213 # Now let's clean the cache
214 my $cicfilename = $inputuri;
215 $cicfilename =~ s/cic:(.*)/$1/;
216 $cicfilename =~ s/theory:(.*)/$1/;
218 print "Unlinking ".$helm_dir.$cicfilename.".xml[.gz]\n";
219 unlink ($helm_dir.$cicfilename.".xml");
220 unlink ($helm_dir.$cicfilename.".xml.gz");
222 kill(USR1,getppid()); # signal changes to parent
225 html_nice_answer($c,"Register done");
227 # "/resolve" works with rdf uri
228 } elsif ($http_method eq 'GET' and $http_path eq "/resolve") {
229 #my $outputurl = $xml_map{$inputuri};
230 my $outputurl = resolve($inputuri);
231 $outputurl = "" if (not defined($outputurl));
232 $cont = "<?xml version=\"1.0\" ?>\n\n";
233 if ($outputurl eq "") {
234 $cont .= "<unresolved />\n";
236 $cont .= "<url value=\"$outputurl\" />\n";
238 answer($c,$cont,"text/xml","");
240 } elsif ($http_method eq 'GET' and $http_path eq "/getdtd") {
242 my $filename = $inputuri;
243 $filename = $dtd_dir."/".$filename;
244 print "DTD: $inputuri ==> ($filename)\n";
245 if (stat($filename)) {
246 print "Using local copy\n";
247 open(FD, $filename) or die "Cannot open $filename\n";
250 # Vary bad heuristic for substituion of absolute URLs
252 s/ENTITY (.*) SYSTEM\s+"/ENTITY $1 SYSTEM "$myownurl\/getdtd?uri=/g;
256 answer($c,$cont,"text/xml","");
258 die "Could not find DTD!";
261 } elsif ($http_method eq 'GET' and $http_path eq "/getxslt") {
263 $url = $xslt_map{$inputuri};
264 print "Downloading XSLT ...\n";
265 helm_wget($tmp_dir, $url);
266 print "XSLT downloaded!\n";
267 $inputuri =~ s/.*\///;
268 my $filename = $tmp_dir . "/" . $inputuri;
270 die "Could not find XSLT!\n" unless stat($filename);
271 open(FD, $filename) or die "Can't open xslt local copy: '$filename'\n";
273 s/xsl:import\s+href="/xsl:import href="$myownurl\/getxslt?uri=/g;
274 s/xsl:include\s+href="/xsl:include href="$myownurl\/getxslt?uri=/g;
278 answer($c, $cont, "text/xml", "");
280 print "Can't unlink XSLT temp file '$filename'";
283 # my $filename = $inputuri;
284 # $filename = $style_dir."/".$filename;
285 # print "XSLT: $inputuri ==> ($filename)\n";
286 # if (stat($filename)) {
287 # print "Using local copy\n";
288 # open(FD, $filename) or die "Cannot open $filename\n";
291 # # Vary bad heuristic for substituion of absolute URLs
292 # # for relative ones
293 # s/xsl:import\s+href="/xsl:import href="$myownurl\/getxslt?uri=/g ;
294 # s/xsl:include\s+href="/xsl:include href="$myownurl\/getxslt?uri=/g ;
298 # answer($c,$cont,"text/xml","");
300 # die "Could not find XSLT!";
304 # "/update" works with rdf uri
305 } elsif ($http_method eq 'GET' and $http_path eq "/update") {
306 # rebuild urls_of_uris db and rdf uris db
307 print "Update requested...\n";
309 kill(USR1,getppid()); # signal changes to parent
311 html_nice_answer($c,"Update done");
313 # "/getalluris" does not work with rdf uris
314 } elsif ($http_method eq 'GET' and $http_path eq "/getalluris") {
315 # send back all the keys in xml
316 print "BASEURI $baseuri, FORMAT $outype\n";
317 $cont = getalluris();
318 answer($c,$cont,"text/xml","");
321 } elsif ($http_method eq 'GET' and $http_path eq "/getallrdfuris") {
322 # send back all the keys in xml
323 my $class = $cgi->param('class');
324 print "BASEURI $baseuri, FORMAT $outype, CLASS $class\n";
325 $cont = getallrdfuris($class);
326 answer($c,$cont,"text/xml","");
328 } elsif ($http_method eq 'GET' and $http_path eq "/ls") {
329 # send back keys that begin with a given uri
330 my ($uritype,$uripattern,$uriprefix);
331 my $baseuri = $cgi->param('baseuri');
332 chop $baseuri if ($baseuri =~ /.*\/$/); # remove trailing "/"
333 # output type, might be 'txt' or 'xml':
334 my $outype = $cgi->param('format');
335 $uripattern = $baseuri;
336 $uripattern =~ s/^.*:(.*)/$1/;
337 if ($baseuri =~ /^cic:/) {
339 } elsif ($baseuri =~ /^theory:/) {
342 $uritype = "invalid";
344 if ($uritype ne "invalid") { # uri is valid
345 if (($outype ne 'txt') and ($outype ne 'xml')) { # invalid out type
346 print "Invalid output type specified: $outype\n";
347 html_nice_answer($c,"Invalid output type, must be ".
349 } else { # valid output
350 print "BASEURI $baseuri, FORMAT $outype\n";
351 $cont = finduris($uritype,$uripattern,$outype);
352 if ($outype eq 'txt') {
353 answer($c,$cont,"text/plain","");
354 } elsif ($outype eq 'xml') {
355 answer($c,$cont,"text/xml","");
357 die "Internal error, exiting!";
360 } else { # invalid uri
361 print "Invalid uri: $baseuri, may begin with 'cic:', ".
362 "'theory:' or '*:'\n";
363 html_nice_answer($c,"Invalid uri , must begin with ".
364 "'cic:' or 'theory:'");
367 } elsif ($http_method eq 'GET' and $http_path eq "/help") {
369 print "Help requested!\n";
370 html_nice_answer($c,"HTTP Getter Version: $VERSION");
372 } elsif ($http_method eq 'GET' and $http_path eq "/getempty") {
373 # return an empty, but valid, xml document
374 my $emptyxml = <<EOXML;
375 <?xml version="1.0"?>
377 <!ELEMENT empty EMPTY>
381 answer($c, $emptyxml, "", "");
383 } elsif ($http_method eq 'GET' and $http_path =~ /\.cgi$/) {
385 print "CGI requested!\n";
386 if ($http_path !~ /^\/[^\/]*\.cgi$/) {
387 html_nice_answer($c,"Invalid CGI name: $http_path, ".
388 "you can't request CGI that contain a slash in the name\n");
389 } elsif (stat "$cgi_dir"."$http_path") {
390 if (not -x "$cgi_dir/$http_path") {
391 html_nice_answer($c,"CGI $http_path found but not ".
393 } else { # exec CGI and anser back its output
396 $ENV{'QUERY_STRING'} = $http_query;
397 my $cgi_output = `$cgi_dir/$http_path`;
398 answer($c,$cgi_output,"","");
402 html_nice_answer($c,"CGI '$http_path' not found ".
403 "in CGI dir '$cgi_dir'");
406 } else { # unsupported request
408 print "INVALID REQUEST!!!!!\n";
409 print "(PATH: ",$http_path,", ";
410 print "QUERY: ",$http_query,")\n";
411 $c->send_error(RC_FORBIDDEN)
414 print "\nRequest solved: ".$r->url."\n\n";
419 print "\nCONNECTION CLOSED\n\n";
424 #================================
426 sub isRdfUri { # return true if the uri is an rdf uri, false otherwise
428 # helm:rdf/cic:www.cs.unibo.it/helm/rdf/foo_schema//cic:\
429 # /Coq/Init/Logic/True_rec.con.types
431 # the format is "helm:rdf/<metadata_tree>:<metadata_scheme>//<xml_file_uri>"
434 if ($uri =~ /^helm:rdf(.*):(.*)\/\/(.*)/) {
441 sub resolve { # resolve an uri in a url, work both with standard cic: or theory:
443 print "RESOLVE subroutine\n";
445 print "GIVEN URI: \"$uri\"\n";
446 if (isRdfUri ($uri)) { # rdf uri, resolve using rdf db
447 print "IS A RDF URI\n";
448 print "I WILL RETURN '$rdf_map{$uri}'\n";
449 return ($rdf_map{$uri});
450 } else { # standard cic: or theory: uri, resolve using std uri db
451 print "IS NOT A RDF URI\n";
452 print "I WILL RETURN '$xml_map{$uri}'\n";
453 return ($xml_map{$uri});
457 sub getalluris { # get all the keys whose prefix is cic
460 my $debug=1; # for debug
462 $content .= '<?xml version="1.0" encoding="ISO-8859-1"?>' . "\n";
463 $content .= "<!DOCTYPE alluris SYSTEM ";
464 $content .= "\"$myownurl/getdtd?uri=alluris.dtd\">" . "\n\n";
465 $content .= "<alluris>\n";
466 foreach $uri (sort (keys(%xml_map))) { # select matching uris
467 if ($uri =~ /^cic:/ && not $uri =~ /.types$/) {
468 print "GETALLURI: $uri\n" if defined($debug);
469 $content .= "\t<uri value=\"$uri\"/>\n";
472 $content .= "</alluris>\n";
480 my $debug=1; # for debug
482 $content .= '<?xml version="1.0" encoding="ISO-8859-1"?>' . "\n";
483 $content .= "<!DOCTYPE allrdfuris SYSTEM ";
484 $content .= "\"$myownurl/getdtd?uri=alluris.dtd\">" . "\n\n";
485 $content .= "<allrdfuris>\n";
486 foreach $uri (sort (keys(%rdf_map))) {
487 if ($class eq "forward" &&
488 $uri =~ /^helm:rdf:www.cs.unibo.it\/helm\/rdf\/forward/
490 $class eq "backward" &&
491 $uri =~ /^helm:rdf:www.cs.unibo.it\/helm\/rdf\/backward/) {
492 print "GETALLRDFURI: $uri\n" if defined($debug);
493 $content .= "\t<uri value=\"$uri\"/>\n";
496 $content .= "</allrdfuris>\n";
500 sub finduris { # find uris for cic and theory trees generation
501 my ($uritype,$uripattern,$format) = @_;
503 my ($uri,$localpart,$basepart,$dirname,$suffix,$flags,$key);
504 my (@itemz,@already_pushed_dir);
505 my (%objects,%dirs); # map uris to suffixes' flags
506 my $debug=1; # for debug
508 print "FINDURIS, uritype: $uritype, uripattern: $uripattern, ".
509 "format: $format\n\n" if defined($debug);
511 if (($uritype eq "cic") or ($uritype eq "theory")) {
512 # get info only of one type: cic or theory
513 foreach (keys(%xml_map)) { # select matching uris
515 if ($uri =~ /^$uritype:$uripattern(\/|$|\.)/) {
516 if ($uri =~ /^$uritype:$uripattern\//) { # directory match
518 $localpart =~ s/^$uritype:$uripattern\/(.*)/$1/;
519 } elsif ($uri =~ /^$uritype:$uripattern($|\.)/) { # file match
521 $localpart =~ s/^.*\/([^\/]*)/$1/;
523 die "Internal error, seems that requested match is none of ".
524 "directory match or file match";
526 print "LOCALPART: $localpart\n" if defined($debug);
528 if ($localpart =~ /^[^\/]*$/) { # no slash, an OBJECT
529 $basepart = $localpart;
530 $basepart =~ s/^([^.]*\.[^.]*)(\.types)?(\.ann)?/$1/;
531 # remove exts .types or
533 $flags = $objects{$basepart}; # get old flags
534 if ($localpart =~ /\.ann$/) {
535 $flags = add_flag("ann","YES",$flags);
537 $flags = add_flag("ann","NO",$flags);
539 if ($localpart =~ /\.types$/) {
540 $flags = add_flag("types","YES",$flags);
541 } elsif ($localpart =~ /\.types\.ann$/) {
542 $flags = add_flag("types","ANN",$flags);
544 $flags = add_flag("types","NO",$flags);
546 $objects{$basepart} = $flags; # save new flags
547 } else { # exists at least one slash, a DIR
548 ($dirname) = split (/\//, $localpart);
549 $dirs{$dirname} = ""; # no flags requirement for dir
554 die "getter internal error: unsupported uritype: \"$uritype\"";
556 # now generate OUTPUT:
557 # output will be saved in $content
558 if ($format eq "txt") { # TXT output
559 foreach $key (sort(keys %dirs)) {
560 $content .= "dir, " . $key . "\n";
562 foreach $key (sort(keys %objects)) {
563 $content .= "object, $key, " . $objects{$key} . "\n";
565 } elsif ($format eq "xml") { # XML output
566 $content .= '<?xml version="1.0" encoding="ISO-8859-1"?>' . "\n";
567 $content .= "<!DOCTYPE ls SYSTEM ";
568 $content .= "\"$myownurl/getdtd?uri=ls.dtd\">" . "\n\n";
569 $content .= "<ls>\n";
570 foreach $key (sort(keys %dirs)) {
571 $content .= "\t<section>$key</section>\n";
573 foreach $key (sort(keys %objects)) {
574 $content .= "\t<object name=\"$key\">\n";
575 $flags = $objects{$key};
576 $flags =~ s/^<(.*)>$/$1/;
577 my ($annflag,$typesflag) = split /,/,$flags;
578 $content .= "\t\t<ann value=\"$annflag\" />\n";
579 $content .= "\t\t<types value=\"$typesflag\" />\n";
580 $content .= "\t</object>\n";
582 $content .= "</ls>\n";
583 } else { # may not enter this branch
584 die "Getter internal error: invalid format received by finduris sub";
590 # manage string like: "<ann_flag,type_flag>"
591 # "ann_flag" may be one of "ann_YES", "ann_NO"
592 # "type_flag" may be one of "types_NO", "types_YES", "types_ANN"
593 # when adding a flag the max between the current flag and the new flag
594 # is taken, the orders are ann_NO < ann_YES and types_NO < types_YES <
596 my ($flagtype,$newflag,$str) = @_;
597 $str = "<,>" if ($str eq "");
598 ($str =~ s/^<(.*,.*)>$/$1/) or die "Internal error: ".
599 "wrong string format for flag adding in $str";
600 my ($annflag,$typeflag) = split /,/,$str;
601 if ($flagtype eq "ann") { # ANN flag handling
602 if ($newflag eq "YES") {
604 } elsif ($newflag eq "NO") {
605 $annflag = "NO" unless ($annflag eq "YES");
607 die "Internal error: annflag must be \"YES\" or \"NO\"";
609 } elsif ($flagtype eq "types") { # TYPES flag handling
610 if ($newflag eq "ANN") {
612 } elsif ($newflag eq "YES") {
613 $typeflag = "YES" unless ($typeflag eq "ANN");
614 } elsif ($newflag eq "NO") {
616 unless (($typeflag eq "ANN") or ($typeflag eq "YES"));
618 die "Internal error: typeflag must be \"YES\", \"NO\" or \"ANN\"";
621 die "Internal error: unsupported flagtype \"$flagtype\"";
623 $str = "<$annflag,$typeflag>";
626 #CSC: Too much powerful: creates even /home, /home/users/, ...
627 #CSC: Does not raise errors if could not create dirs/files
631 my @dirs = split /\//,$pathname;
633 foreach $dir (@dirs) {
634 $tmp = ((defined($tmp)) ? $tmp."\/".$dir : "");
642 my ($uri,$url,$filename) = @_;
643 print "<download_request>\n";
646 print "filename: $filename\n\n";
647 print "</download_request>\n";
650 sub gunzip { # gunzip a file and return the deflated content
653 my ($gz, $buffer, $cont);
655 print "deflating $filename ...\n";
656 $gz = gzopen($filename, "r")
657 or die "Cannot open gzip'ed file $filename: $gzerrno";
659 while ( $gz->gzread($buffer) > 0 ) {
662 die "Error while reading : $gzerrno\n" if $gzerrno != Z_STREAM_END ;
668 sub gzip { # gzip the content argument and save it to filename argument
669 my ($cont, $filename) = @_;
670 #my $debug=1; # debug only
672 print "gzopening $filename ...\n" if (defined($debug));
673 my $gz = gzopen($filename, "w")
674 or die "Cannot gzopen for writing file $filename: $gzerrno";
675 print "gzwriting ...\n" if (defined($debug));
676 $gz->gzwrite($cont) or die "error writing: $gzerrno, exiting!\n";
677 print "gzclosing ...\n" if (defined($debug));
682 my ($patch_dtd,$url,$filename,$format) = @_;
685 # print "DOWNLOAD subs receives url: \"$url\"\n";
686 # print "DOWNLOAD subs receives filename: \"$filename\"\n";
688 my $resourcetype; # retrieve mode: "normal" (.xml) or "gzipped" (.xml.gz)
689 if ($filename =~ /\.xml$/) { # set retrieve mode
690 $resourcetype = "normal";
691 } elsif ($filename =~ /\.xml\.gz$/) {
692 $resourcetype = "gzipped";
694 die "Unsupported download extension, might be '.gz' or '.xml'\n";
696 my $basefname = $filename;
697 $basefname =~ s/\.gz$//; # get base resource name removing trailing .gz
698 $cont = ""; # modified by side-effect by the callback function
701 if (stat($basefname)) {
702 $localfname=$basefname;
703 } elsif (stat($basefname.".gz")) {
704 $localfname=$basefname.".gz";
706 if ($localfname ne "") { # we already have local copy of requested file
707 # check both possible cache entry: gzipped or normal
708 print "Using local copy.\n";
709 if ($localfname =~ /\.xml\.gz$/) { # deflating cached file and return it
710 $cont = gunzip($localfname);
711 } elsif ($localfname =~ /\.xml$/) { # just return cached file
712 open(FD, $localfname) or die "Cannot open $localfname";
713 while(<FD>) { $cont .= $_; }
716 die "Internal error: unexpected file name $localfname,"
717 ."must end with '.gz' or '.xml.gz'\n";
719 } else { # download file from net
720 print "Downloading file ...\n"; # download file
721 my $ua = LWP::UserAgent->new;
722 my $request = HTTP::Request->new(GET => "$url");
723 my $response = $ua->request($request, \&callback);
725 # cache retrieved file to disk
726 # <ZACK/> TODO: inefficent, I haven't yet undestood how to deflate
727 # in memory gzipped file, without call "gzopen"
728 print "Storing file ...\n";
729 # print "Making dirs ...\n";
731 # print "Opening tmp file for writing ...\n";
732 open(FD, ">".$filename.".tmp") or die "Cannot open $filename.tmp\n";
733 # print "Writing on tmp file ...\n";
735 # print "Closing tmp file ...\n";
738 # handle cache conversion normal->gzipped or gzipped->normal as user choice
739 print "cachemode:$cachemode, resourcetype:$resourcetype\n"
740 if (defined($debug));
741 if (($cachemode eq 'normal') and ($resourcetype eq 'normal')) {
742 # cache the file as is
743 rename "$filename.tmp", $filename;
744 } elsif (($cachemode eq 'gzipped') and ($resourcetype eq 'gzipped')) {
745 # cache the file as is
746 # and update the $cont variabile with deflated content
747 rename "$filename.tmp", $filename;
748 $cont = gunzip($filename);
749 } elsif (($cachemode eq 'normal') and ($resourcetype eq 'gzipped')) {
750 # deflate cache entry
752 open(FD, "> $basefname") or die "cannot open $basefname\n";
753 $cont = gunzip($filename.".tmp");
756 unlink "$filename.tmp"; # delete old gzipped file
757 } elsif (($cachemode eq 'gzipped') and ($resourcetype eq 'normal')) {
758 # compress cache entry
759 print "gzipping ...\n" if (defined($debug));
760 gzip($cont, $basefname.".gz");
761 unlink "$filename.tmp"; # delete old uncompressed file
763 die "Internal error, unsopported cachemode, resourcetype couple\n";
765 # $cont now contained uncompressed data
767 if ($patch_dtd eq "yes") {
768 $cont =~ s/DOCTYPE (.*) SYSTEM\s+"$dtdbaseurl\//DOCTYPE $1 SYSTEM "$myownurl\/getdtd?uri=/g;
770 if ($format eq "gz") {
771 gzip($cont,"$basefname.tmp");
772 open (TMP, "< $basefname.tmp")
773 or die "Can't open tempfile: $filename.tmp, exiting!\n";
779 unlink ($basefname . ".tmp") or
780 die "cannot unlink temporary file: $basefname.tmp\n";
787 # answer the client setting content, Content-Type and Content-Enconding
789 my ($c,$cont,$contype,$contenc) = @_;
790 my $res = new HTTP::Response;
791 $res->content($cont);
792 $res->push_header("Content-Type" => $contype)
793 unless ($contype eq "");
794 $res->push_header("Content-Encoding" => $contenc)
795 unless ($contenc eq "");
796 $res->push_header("Cache-Control" => "no-cache");
797 $res->push_header("Pragma" => "no-cache");
798 $res->push_header("Expires" => "0");
799 $c->send_response($res);
802 sub html_nice_answer {
803 # answer the client whith a nice html document
804 my ($c,$content) = @_;
805 $content = "<html><body><h1>$content</h1></body></html>";
806 answer($c,$content,"text/html","");
815 # retrieve a file from an url and write it to a temp dir
816 # used for retrieve resource index from servers
818 my ($prefix, $URL) = @_;
819 my $ua = LWP::UserAgent->new;
820 my $request = HTTP::Request->new(GET => "$URL");
821 my $response = $ua->request($request, \&callback);
822 my ($filename) = reverse (split "/", $URL); # get filename part of the URL
823 open (TEMP, "> $prefix/$filename")
824 || die "Cannot open temporary file: $prefix/$filename\n";
832 tie(%xml_map, 'DB_File', $uris_dbm, O_RDWR, 0664);
834 tie(%rdf_map, 'DB_File', $rdf_dbm, O_RDWR, 0664);
836 tie(%xslt_map, 'DB_File', $xslt_dbm, O_RDWR, 0664);
840 # rebuild dbs fetching resource indexes from servers.
841 # Rebuild urls_of_uris db (1), rdf db (2) and xslt db (3)
843 $server, $idxfile, $rdf_idxfile, $uri, $url, $comp, $line,
845 %urls_of_uris, %rdf, %xslt # local proxy of dbs
851 foreach my $file ($uris_dbm, $rdf_dbm, $xslt_dbm) { # remove old db file
852 if (stat $file) { # remove if exists
853 unlink ($file) or die "can't unlink old db file: $file\n";
856 tie(%urls_of_uris, 'DB_File', $uris_dbm, O_RDWR|O_CREAT, 0664);
857 tie(%rdf, 'DB_File', $rdf_dbm, O_RDWR|O_CREAT, 0664);
858 tie(%xslt, 'DB_File', $xslt_dbm, O_RDWR|O_CREAT, 0664);
860 open (SRVS, "< $servers_file") or
861 die "cannot open servers file: $servers_file\n";
862 @servers = <SRVS>; # read all servers
865 while ($server = pop @servers) { # cicle on servers in _reverse_ order
866 print "processing server: $server ...\n";
868 helm_wget($tmp_dir, $server."/".$indexname); # get index
869 helm_wget($tmp_dir, $server."/".$rdf_indexname); # get rdf index
870 helm_wget($tmp_dir, $server."/".$xslt_indexname); # get xslt index
871 $idxfile = $tmp_dir."/".$indexname;
872 $rdf_idxfile = $tmp_dir."/".$rdf_indexname;
873 $xslt_idxfile = $tmp_dir . "/" . $xslt_indexname;
875 # (1) REBUILD URLS_OF_URIS DB
876 open (INDEX, "< $idxfile") or
877 die "cannot open temporary index file: $idxfile\n";
878 while (my $line = <INDEX>) { # parse index and add entry to urls_of_uris db
880 ($uri,$comp) = split /[ \t]+/, $line;
883 $url = $uri . ".xml" . ".gz";
885 $url = $uri . ".xml";
887 $url =~ s/cic:/$server/;
888 $url =~ s/theory:/$server/;
889 $urls_of_uris{$uri} = $url;
893 open (RDF_INDEX, "< $rdf_idxfile") or
894 die "cannot open temporary RDF index file: $rdf_idxfile\n";
895 while (my $line = <RDF_INDEX>) { # parse index and add entry to rdf db
897 ($uri,$comp) = split /[ \t]+/, $line; # comp is "gz" or nothing
900 $url = $uri . ".xml" . ".gz";
902 $url = $uri . ".xml";
904 $url =~ s/^helm:rdf(.*)\/\/cic:/$server/;
905 $url =~ s/^helm:rdf(.*)\/\/theory:/$server/;
909 # (3) REBUILD XSLT DB
910 open (XSLT_INDEX, "< $xslt_idxfile") or
911 die "cannot open temporary XSLT index file: $xslt_idxfile\n";
912 while (my $line = <XSLT_INDEX>) { # parse index and add entry to xslt db
914 my $stylesheet = $line;
915 my $url = $server . "/" . $stylesheet;
916 $xslt{$stylesheet} = $url;
919 close INDEX; # close indexes
924 foreach my $file ($idxfile, $rdf_idxfile, $xslt_idxfile) {
925 print "cannot unlink temporary file: $file\n" if (unlink $file) != 1;
930 untie(%urls_of_uris); # untie local proxies
933 tie(%xml_map, 'DB_File', $uris_dbm, O_RDWR, 0664); # retie global ones
934 tie(%rdf_map, 'DB_File', $rdf_dbm, O_RDWR, 0664);
935 tie(%xslt_map, 'DB_File', $xslt_dbm, O_RDWR, 0664);
939 # vim modline: do not remove!