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 # <ZACK>: TODO temporary, move this setting to configuration file
47 # set the cache mode, may be "gzipped" or "normal"
48 my $cachemode = $ENV{'HTTP_GETTER_CACHE_MODE'} || 'gzipped';
49 if (($cachemode ne 'gzipped') and ($cachemode ne 'normal')) {
50 die "Invalid HTTP_GETTER_CACHE_MODE environment variable, must be".
51 "'normal' or 'gzipped'\n";
55 # next require defines: $helm_dir, $html_link, $dtd_dir, $uris_dbm, $indexname
56 require $HELM_LIB_PATH;
58 # TEMP: TODO put these vars in configuration file configuration.xml
60 $helm_rdf_dir = "/projects/helm/shared/V7/rdf_library";
61 $rdf_dbm = "/projects/helm/shared/V7/rdf_urls_of_uris";
62 $rdf_indexname = "rdf_index.txt";
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 my $myownurl = $d->url;
85 # Let's patch the returned URL
86 $myownurl =~ s/\/$//; # chop the final slash
87 my $myownport = $myownurl;
88 $myownport =~ s/http:\/\/(.*):(.*)/$2/;
89 $myownurl =~ s/http:\/\/(.*):(.*)/$1/;
90 ($myownurl) = gethostbyname($myownurl);
91 $myownurl = "http://".$myownurl.":".$myownport;
93 tie(%map, 'DB_File', $uris_dbm.".db", O_RDWR, 0664); # open dbs
94 tie(%rdf_map, 'DB_File', $rdf_dbm.".db", O_RDWR, 0664);
97 print "HTTP Getter $VERSION\n"; # print hello information
98 print "Please contact me at: <URL:", $myownurl."/", ">\n";
100 print "helm_dir: $helm_dir\n";
101 print "helm_rdf_dir: $helm_rdf_dir\n";
102 print "style_dir: $style_dir\n";
103 print "dtd_dir: $dtd_dir\n";
104 print "urls_of_uris.db: $uris_dbm.db\n";
105 print "cache mode: $cachemode\n";
106 print "indexname: $indexname\n";
107 print "rdf_indexname: $rdf_indexname\n";
110 $SIG{CHLD} = "IGNORE"; # do not accumulate defunct processes
111 $SIG{USR1} = \&update; # sent by the child to make the parent update
112 while (my $c = $d->accept) {
114 while (my $r = $c->get_request) {
115 #CSC: mancano i controlli di sicurezza
117 my $http_method = $r->method;
118 my $http_path = $r->url->path;
119 my $http_query = uri_unescape($r->url->query);
120 my $cgi = new CGI("$http_query");
121 my $inputuri = $cgi->param('uri');
122 print "\nRequest: ".$r->url."\n\n";
124 print "\nUnescaped query: ".$http_query."\n";
126 # "getxml" works with rdf uris
127 if ($http_method eq 'GET' and $http_path eq "/getxml") {
128 # finds the uri, url and filename
130 my $answerformat = $cgi->param('format');
131 my $patch_dtd = $cgi->param('patch_dtd');
132 $answerformat = "" if (not defined($answerformat));
133 $patch_dtd = "yes" if (not defined($patch_dtd));
134 if (($answerformat ne "gz") and ($answerformat ne "normal")
135 and ($answerformat ne "")) {
136 die "Wrong output format: $answerformat, must be 'normal' ".
139 if (($patch_dtd ne "yes") and ($patch_dtd ne "no")
140 and ($patch_dtd ne "")) {
141 die "Wrong param, patch_dtd must be 'yes' or 'no'\n";
144 my $filename = $inputuri;
145 if (not isRdfUri($inputuri)) { # standad cic: or theory: uri
146 $filename =~ s/^cic:(.*)/$1/;
147 $filename =~ s/^theory:(.*)/$1/;
149 $filename =~ s/^(.*)\/\/cic:(.*)/$2/;
150 $filename =~ s/^(.*)\/\/theory:(.*)/$2/;
153 my $url = resolve ($inputuri); # resolve uri in url
154 if (not defined($url)) {
155 die "uri \"$inputuri\" can't be resolved\n";
158 my $extension; # file extension
159 if ($url =~ /\.xml$/) { # non gzipped file
161 } elsif ($url =~ /\.xml\.gz$/) { # gzipped file
162 $extension = ".xml.gz";
163 } else { # error: unknown extension
164 die "unexpected extension in url: $url, might be '.xml'".
168 if (not isRdfUri ($inputuri)) {
169 $filename = $helm_dir.$filename.$extension;
171 $filename = $helm_rdf_dir.$filename.$extension;
174 if (!defined($url)) {
175 print "\nNOT FOUND!!!!!\n";
176 $c->send_error(RC_NOT_FOUND)
178 print_request("cic",$inputuri,$url,$filename);
181 my $ciccontent = download($patch_dtd,"cic",$url,$filename,$answerformat);
183 # Answering the client
184 if ($answerformat eq "normal") {
185 answer($c,$ciccontent,"text/xml","");
187 answer($c,$ciccontent,"text/xml","x-gzip");
190 # "/register" does not work with rdf uris
191 } elsif ($http_method eq 'GET' and $http_path eq "/register") {
192 my $inputurl = $cgi->param('url');
193 print "Register requested...\n";
194 $map{$inputuri}=$inputurl;
196 # Now let's clean the cache
197 my $cicfilename = $inputuri;
198 $cicfilename =~ s/cic:(.*)/$1/;
199 $cicfilename =~ s/theory:(.*)/$1/;
201 print "Unlinking ".$helm_dir.$cicfilename.".xml[.gz]\n";
202 unlink ($helm_dir.$cicfilename.".xml");
203 unlink ($helm_dir.$cicfilename.".xml.gz");
205 kill(USR1,getppid()); # signal changes to parent
208 html_nice_answer($c,"Register done");
209 # "/resolve" works with rdf uri
210 } elsif ($http_method eq 'GET' and $http_path eq "/resolve") {
211 #my $outputurl = $map{$inputuri};
212 my $outputurl = resolve($inputuri);
213 $outputurl = "" if (not defined($outputurl));
214 $cont = "<?xml version=\"1.0\" ?>\n\n";
215 if ($outputurl eq "") {
216 $cont .= "<unresolved />\n";
218 $cont .= "<url value=\"$outputurl\" />\n";
220 answer($c,$cont,"text/xml","");
221 } elsif ($http_method eq 'GET' and $http_path eq "/getdtd") {
222 my $filename = $inputuri;
223 $filename = $dtd_dir."/".$filename;
224 print "DTD: $inputuri ==> ($filename)\n";
225 if (stat($filename)) {
226 print "Using local copy\n";
227 open(FD, $filename) or die "Cannot open $filename\n";
230 # Vary bad heuristic for substituion of absolute URLs
232 s/ENTITY (.*) SYSTEM\s+"/ENTITY $1 SYSTEM "$myownurl\/getdtd?uri=/g;
236 answer($c,$cont,"text/xml","");
238 die "Could not find DTD!";
240 } elsif ($http_method eq 'GET' and $http_path eq "/getstyleconf") {
241 my $filename = $inputuri;
242 $filename = $style_dir."/config/".$filename;
243 if (stat($filename)) {
244 open(FD, $filename) or die "Cannot open $filename\n";
247 s/DOCTYPE (.*) SYSTEM\s+"/DOCTYPE $1 SYSTEM "$myownurl\/getstyleconf?uri=/g;
251 answer($c,$cont,"text/plain","");
253 die "Could not find Style Configuration File!";
255 } elsif ($http_method eq 'GET' and $http_path eq "/getxslt") {
256 my $filename = $inputuri;
257 $filename = $style_dir."/".$filename;
258 print "XSLT: $inputuri ==> ($filename)\n";
259 if (stat($filename)) {
260 print "Using local copy\n";
261 open(FD, $filename) or die "Cannot open $filename\n";
264 # Vary bad heuristic for substituion of absolute URLs
266 s/xsl:import\s+href="/xsl:import href="$myownurl\/getxslt?uri=/g ;
267 s/xsl:include\s+href="/xsl:include href="$myownurl\/getxslt?uri=/g ;
271 answer($c,$cont,"text/xml","");
273 die "Could not find XSLT!";
275 } elsif ($http_method eq 'GET' and $http_path eq "/update") {
276 # rebuild urls_of_uris db and rdf uris db
277 print "Update requested...\n";
279 kill(USR1,getppid()); # signal changes to parent
281 html_nice_answer($c,"Update done");
282 # "/ls" does not work with rdf uris
283 } elsif ($http_method eq 'GET' and $http_path eq "/ls") {
284 # send back keys that begin with a given uri
285 my ($uritype,$uripattern,$uriprefix);
286 my $baseuri = $cgi->param('baseuri');
287 chop $baseuri if ($baseuri =~ /.*\/$/); # remove trailing "/"
288 my $outype = $cgi->param('format'); # output type, might be 'txt' or 'xml'
289 $uripattern = $baseuri;
290 $uripattern =~ s/^.*:(.*)/$1/;
291 if ($baseuri =~ /^cic:/) {
293 } elsif ($baseuri =~ /^theory:/) {
296 $uritype = "invalid";
298 if ($uritype ne "invalid") { # uri is valid
299 if (($outype ne 'txt') and ($outype ne 'xml')) { # invalid out type
300 print "Invalid output type specified: $outype\n";
301 html_nice_answer($c,"Invalid output type, must be ".
303 } else { # valid output
304 print "BASEURI $baseuri, FORMAT $outype\n";
305 $cont = finduris($uritype,$uripattern,$outype);
306 if ($outype eq 'txt') {
307 answer($c,$cont,"text/plain","");
308 } elsif ($outype eq 'xml') {
309 answer($c,$cont,"text/xml","");
311 die "Internal error, exiting!";
314 } else { # invalid uri
315 print "Invalid uri: $baseuri, may begin with 'cic:', ".
316 "'theory:' or '*:'\n";
317 html_nice_answer($c,"Invalid uri , must begin with ".
318 "'cic:' or 'theory:'");
320 } elsif ($http_method eq 'GET' and $http_path eq "/help") {
322 print "Help requested!\n";
323 html_nice_answer($c,"HTTP Getter Version: $VERSION");
324 } elsif ($http_method eq 'GET' and $http_path =~ /\.cgi$/) {
326 print "CGI requested!\n";
327 if ($http_path !~ /^\/[^\/]*\.cgi$/) {
328 html_nice_answer($c,"Invalid CGI name: $http_path, ".
329 "you can't request CGI that contain a slash in the name\n");
330 } elsif (stat "$cgi_dir"."$http_path") {
331 if (not -x "$cgi_dir/$http_path") {
332 html_nice_answer($c,"CGI $http_path found but not ".
334 } else { # exec CGI and anser back its output
337 $ENV{'QUERY_STRING'} = $http_query;
338 my $cgi_output = `$cgi_dir/$http_path`;
339 answer($c,$cgi_output,"","");
343 html_nice_answer($c,"CGI '$http_path' not found ".
344 "in CGI dir '$cgi_dir'");
346 } else { # unsupported request
348 print "INVALID REQUEST!!!!!\n";
349 print "(PATH: ",$http_path,", ";
350 print "QUERY: ",$http_query,")\n";
351 $c->send_error(RC_FORBIDDEN)
353 print "\nRequest solved: ".$r->url."\n\n";
357 print "\nCONNECTION CLOSED\n\n";
362 #================================
364 sub isRdfUri { # return true if the uri is an rdf uri, false otherwise
366 # helm:rdf/cic:www.cs.unibo.it/helm/rdf/foo_schema//cic:\
367 # /Coq/Init/Logic/True_rec.con.types
369 # the format is "helm:rdf/<metadata_tree>:<metadata_scheme>//<xml_file_uri>"
372 if ($uri =~ /^helm:rdf(.*):(.*)\/\/(.*)/) {
379 sub resolve { # resolve an uri in a url, work both with standard cic: or theory:
382 if (isRdfUri ($uri)) { # rdf uri, resolve using rdf db
383 return ($rdf_map{$uri});
384 } else { # standard cic: or theory: uri, resolve using std uri db
389 sub finduris { # find uris for cic and theory trees generation
390 my ($uritype,$uripattern,$format) = @_;
392 my ($uri,$localpart,$basepart,$dirname,$suffix,$flags,$key);
393 my (@itemz,@already_pushed_dir);
394 my (%objects,%dirs); # map uris to suffixes' flags
395 #my $debug=1; # for debug
397 print "FINDURIS, uritype: $uritype, uripattern: $uripattern, ".
398 "format: $format\n\n" if defined($debug);
400 if (($uritype eq "cic") or ($uritype eq "theory")) {
401 # get info only of one type: cic or theory
402 foreach (keys(%map)) { # select matching uris
404 if ($uri =~ /^$uritype:$uripattern(\/|$|\.)/) {
405 if ($uri =~ /^$uritype:$uripattern\//) { # directory match
407 $localpart =~ s/^$uritype:$uripattern\/(.*)/$1/;
408 } elsif ($uri =~ /^$uritype:$uripattern($|\.)/) { # file match
410 $localpart =~ s/^.*\/([^\/]*)/$1/;
412 die "Internal error, seems that requested match is none of ".
413 "directory match or file match";
415 print "LOCALPART: $localpart\n" if defined($debug);
417 if ($localpart =~ /^[^\/]*$/) { # no slash, an OBJECT
418 $basepart = $localpart;
419 $basepart =~ s/^([^.]*\.[^.]*)(\.types)?(\.ann)?/$1/;
420 # remove exts .types or
422 $flags = $objects{$basepart}; # get old flags
423 if ($localpart =~ /\.ann$/) {
424 $flags = add_flag("ann","YES",$flags);
426 $flags = add_flag("ann","NO",$flags);
428 if ($localpart =~ /\.types$/) {
429 $flags = add_flag("types","YES",$flags);
430 } elsif ($localpart =~ /\.types\.ann$/) {
431 $flags = add_flag("types","ANN",$flags);
433 $flags = add_flag("types","NO",$flags);
435 $objects{$basepart} = $flags; # save new flags
436 } else { # exists at least one slash, a DIR
437 ($dirname) = split (/\//, $localpart);
438 $dirs{$dirname} = ""; # no flags requirement for dir
443 die "getter internal error: unsupported uritype: \"$uritype\"";
445 # now generate OUTPUT:
446 # output will be saved in $content
447 if ($format eq "txt") { # TXT output
448 foreach $key (sort(keys %dirs)) {
449 $content .= "dir, " . $key . "\n";
451 foreach $key (sort(keys %objects)) {
452 $content .= "object, $key, " . $objects{$key} . "\n";
454 } elsif ($format eq "xml") { # XML output
455 $content .= '<?xml version="1.0" encoding="ISO-8859-1"?>' . "\n";
456 $content .= "<!DOCTYPE ls SYSTEM ";
457 $content .= "\"$myownurl/getdtd?uri=ls.dtd\">" . "\n\n";
458 $content .= "<ls>\n";
459 foreach $key (sort(keys %dirs)) {
460 $content .= "\t<section>$key</section>\n";
462 foreach $key (sort(keys %objects)) {
463 $content .= "\t<object name=\"$key\">\n";
464 $flags = $objects{$key};
465 $flags =~ s/^<(.*)>$/$1/;
466 my ($annflag,$typesflag) = split /,/,$flags;
467 $content .= "\t\t<ann value=\"$annflag\" />\n";
468 $content .= "\t\t<types value=\"$typesflag\" />\n";
469 $content .= "\t</object>\n";
471 $content .= "</ls>\n";
472 } else { # may not enter this branch
473 die "Getter internal error: invalid format received by finduris sub";
479 # manage string like: "<ann_flag,type_flag>"
480 # "ann_flag" may be one of "ann_YES", "ann_NO"
481 # "type_flag" may be one of "types_NO", "types_YES", "types_ANN"
482 # when adding a flag the max between the current flag and the new flag
483 # is taken, the orders are ann_NO < ann_YES and types_NO < types_YES <
485 my ($flagtype,$newflag,$str) = @_;
486 $str = "<,>" if ($str eq "");
487 ($str =~ s/^<(.*,.*)>$/$1/) or die "Internal error: ".
488 "wrong string format for flag adding in $str";
489 my ($annflag,$typeflag) = split /,/,$str;
490 if ($flagtype eq "ann") { # ANN flag handling
491 if ($newflag eq "YES") {
493 } elsif ($newflag eq "NO") {
494 $annflag = "NO" unless ($annflag eq "YES");
496 die "Internal error: annflag must be \"YES\" or \"NO\"";
498 } elsif ($flagtype eq "types") { # TYPES flag handling
499 if ($newflag eq "ANN") {
501 } elsif ($newflag eq "YES") {
502 $typeflag = "YES" unless ($typeflag eq "ANN");
503 } elsif ($newflag eq "NO") {
505 unless (($typeflag eq "ANN") or ($typeflag eq "YES"));
507 die "Internal error: typeflag must be \"YES\", \"NO\" or \"ANN\"";
510 die "Internal error: unsupported flagtype \"$flagtype\"";
512 $str = "<$annflag,$typeflag>";
515 #CSC: Too much powerful: creates even /home, /home/users/, ...
516 #CSC: Does not raise errors if could not create dirs/files
520 my @dirs = split /\//,$pathname;
522 foreach $dir (@dirs) {
523 $tmp = ((defined($tmp)) ? $tmp."\/".$dir : "");
531 my ($str,$uri,$url,$filename) = @_;
532 print $str."uri: $uri\n";
533 print $str."url: $url\n";
534 print $str."filename: $filename\n\n";
543 sub gunzip { # gunzip a file and return the deflated content
546 my ($gz, $buffer, $cont);
548 print "deflating $filename ...\n";
549 $gz = gzopen($filename, "r")
550 or die "Cannot open gzip'ed file $filename: $gzerrno";
552 while ( $gz->gzread($buffer) > 0 ) {
555 die "Error while reading : $gzerrno\n" if $gzerrno != Z_STREAM_END ;
561 sub gzip { # gzip the content argument and save it to filename argument
562 my ($cont, $filename) = @_;
563 #my $debug=1; # debug only
565 print "gzopening $filename ...\n" if (defined($debug));
566 my $gz = gzopen($filename, "w")
567 or die "Cannot gzopen for writing file $filename: $gzerrno";
568 print "gzwriting ...\n" if (defined($debug));
569 $gz->gzwrite($cont) or die "error writing: $gzerrno, exiting!\n";
570 print "gzclosing ...\n" if (defined($debug));
575 my ($patch_dtd,$str,$url,$filename,$format) = @_;
578 #my $debug=1; # for DEBUG only
580 my $resourcetype; # retrieve mode: "normal" (.xml) or "gzipped" (.xml.gz)
581 if ($filename =~ /\.xml$/) { # set retrieve mode
582 $resourcetype = "normal";
583 } elsif ($filename =~ /\.xml\.gz$/) {
584 $resourcetype = "gzipped";
586 die "Unsupported download extension, might be '.gz' or '.xml'\n";
588 my $basefname = $filename;
589 $basefname =~ s/\.gz$//; # get base resource name removing trailing .gz
590 $cont = ""; # modified by side-effect by the callback function
593 if (stat($basefname)) {
594 $localfname=$basefname;
595 } elsif (stat($basefname.".gz")) {
596 $localfname=$basefname.".gz";
598 if ($localfname ne "") { # we already have local copy of requested file
599 # check both possible cache entry: gzipped or normal
600 print "Using local copy for the $str file\n";
601 if ($localfname =~ /\.xml\.gz$/) { # deflating cached file and return it
602 $cont = gunzip($localfname);
603 } elsif ($localfname =~ /\.xml$/) { # just return cached file
604 open(FD, $localfname) or die "Cannot open $localfname";
605 while(<FD>) { $cont .= $_; }
608 die "Internal error: unexpected file name $localfname,"
609 ."must end with '.gz' or '.xml.gz'\n";
611 } else { # download file from net
612 print "Downloading the $str file\n"; # download file
613 my $ua = LWP::UserAgent->new;
614 my $request = HTTP::Request->new(GET => "$url");
615 my $response = $ua->request($request, \&callback);
617 # cache retrieved file to disk
618 # <ZACK/> TODO: inefficent, I haven't yet undestood how to deflate
619 # in memory gzipped file, without call "gzopen"
620 print "Storing the $str file\n";
621 print "Making dirs ...\n" if (defined($debug));
623 print "Opening tmp file for writing ...\n" if (defined($debug));
624 open(FD, ">".$filename.".tmp") or die "Cannot open $filename.tmp\n";
625 print "Writing on tmp file ...\n" if (defined($debug));
627 print "Closing tmp file ...\n" if (defined($debug));
630 # handle cache conversion normal->gzipped or gzipped->normal as user choice
631 print "cachemode:$cachemode, resourcetype:$resourcetype\n" if (defined($debug));
632 if (($cachemode eq 'normal') and ($resourcetype eq 'normal')) {
633 # cache the file as is
634 rename "$filename.tmp", $filename;
635 } elsif (($cachemode eq 'gzipped') and ($resourcetype eq 'gzipped')) {
636 # cache the file as is
637 # and update the $cont variabile with deflated content
638 rename "$filename.tmp", $filename;
639 $cont = gunzip($filename);
640 } elsif (($cachemode eq 'normal') and ($resourcetype eq 'gzipped')) {
641 # deflate cache entry
643 open(FD, "> $basefname") or die "cannot open $basefname\n";
644 $cont = gunzip($filename.".tmp");
647 unlink "$filename.tmp"; # delete old gzipped file
648 } elsif (($cachemode eq 'gzipped') and ($resourcetype eq 'normal')) {
649 # compress cache entry
650 print "gzipping ...\n" if (defined($debug));
651 gzip($cont, $basefname.".gz");
652 unlink "$filename.tmp"; # delete old uncompressed file
654 die "Internal error, unsopported cachemode, resourcetype couple\n";
656 # $cont now contained uncompressed data
658 if ($patch_dtd eq "yes") {
659 $cont =~ s/DOCTYPE (.*) SYSTEM\s+"http:\/\/www.cs.unibo.it\/helm\/dtd\//DOCTYPE $1 SYSTEM "$myownurl\/getdtd?uri=/g;
661 if ($format eq "gz") {
662 gzip($cont,"$basefname.tmp");
663 open (TMP, "< $basefname.tmp")
664 or die "Can't open tempfile: $filename.tmp, exiting!\n";
670 unlink ($basefname . ".tmp") or
671 die "cannot unlink temporary file: $basefname.tmp\n";
678 # answer the client setting content, Content-Type and Content-Enconding
680 my ($c,$cont,$contype,$contenc) = @_;
681 my $res = new HTTP::Response;
682 $res->content($cont);
683 $res->push_header("Content-Type" => $contype)
684 unless ($contype eq "");
685 $res->push_header("Content-Encoding" => $contenc)
686 unless ($contenc eq "");
687 $res->push_header("Cache-Control" => "no-cache");
688 $res->push_header("Pragma" => "no-cache");
689 $res->push_header("Expires" => "0");
690 $c->send_response($res);
693 sub html_nice_answer {
694 # answer the client whith a nice html document
695 my ($c,$content) = @_;
696 $content = "<html><body><h1>$content</h1></body></html>";
697 answer($c,$content,"text/html","");
701 #retrieve a file from an url and write it to a temp dir
702 #used for retrieve resource index from servers
704 my ($prefix, $URL) = @_;
705 my $ua = LWP::UserAgent->new;
706 my $request = HTTP::Request->new(GET => "$URL");
707 my $response = $ua->request($request, \&callback);
708 my ($filename) = reverse (split "/", $URL); # get filename part of the URL
709 open (TEMP, "> $prefix/$filename")
710 || die "Cannot open temporary file: $prefix/$filename\n";
717 tie(%map, 'DB_File', $uris_dbm.".db", O_RDWR, 0664);
719 tie(%rdf_map, 'DB_File', $uris_dbm.".db", O_RDWR, 0664);
723 # rebuild dbs fetching resource indexes from servers.
724 # Rebuild urls_of_uris db (1) and rdf db (2)
726 $server, $idxfile, $rdf_idxfile, $uri, $url, $comp, $line,
728 %urls_of_uris, %rdf # local proxy of $map and $rdf_map
733 foreach my $file ($uris_dbm, $rdf_dbm) { # remove old db file
734 if (stat $file.".db") { # remove if exists
735 unlink ($file.".db") or die "can't unlink old db file: $file.db\n";
738 tie(%urls_of_uris, 'DB_File', $uris_dbm.".db", O_RDWR|O_CREAT, 0664);
739 tie(%rdf, 'DB_File', $rdf_dbm.".db", O_RDWR|O_CREAT, 0664);
741 open (SRVS, "< $servers_file") or
742 die "cannot open servers file: $servers_file\n";
743 @servers = <SRVS>; # read all servers
746 while ($server = pop @servers) { # cicle on servers in _reverse_ order
747 print "processing server: $server ...\n";
749 helm_wget($tmp_dir, $server."/".$indexname); # get index
750 helm_wget($tmp_dir, $server."/".$rdf_indexname); # get rdf index
751 $idxfile = $tmp_dir."/".$indexname;
752 $rdf_idxfile = $tmp_dir."/".$rdf_indexname;
754 open (INDEX, "< $idxfile") or # (1) REBUILD URLS_OF_URIS DB
755 die "cannot open temporary index file: $idxfile\n";
756 while ($line = <INDEX>) { # parse index and add entry to urls_of_uris db
758 ($uri,$comp) = split /[ \t]+/, $line;
761 $url = $uri . ".xml" . ".gz";
763 $url = $uri . ".xml";
765 $url =~ s/cic:/$server/;
766 $url =~ s/theory:/$server/;
767 $urls_of_uris{$uri} = $url;
770 open (RDF_INDEX, "< $rdf_idxfile") or # (2) REBUILD RDF DB
771 die "cannot open temporary rdf index file: $rdf_idxfile\n";
772 while ($line = <RDF_INDEX>) { # parse index and add entry to rdf db
774 ($uri,$comp) = split /[ \t]+/, $line; # comp is "gz" or nothing
777 $url = $uri . ".xml" . ".gz";
779 $url = $uri . ".xml";
781 $url =~ s/^helm:rdf(.*)\/\/cic:/$server/;
782 $url =~ s/^helm:rdf(.*)\/\/theory:/$server/;
786 close INDEX; # close indexes
788 die "cannot unlink temporary file: $idxfile\n" # remove temp files
789 if (unlink $idxfile) != 1;
790 die "cannot unlink temporary file: $rdf_idxfile\n"
791 if (unlink $rdf_idxfile) != 1;
795 untie(%urls_of_uris); # untie local proxies
797 tie(%map, 'DB_File', $uris_dbm.".db", O_RDWR, 0664); # retie global ones
798 tie(%rdf_map, 'DB_File', $rdf_dbm.".db", O_RDWR, 0664);