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;
66 $getterport = $ENV{'HTTP_GETTER_PORT'} || $getterport;
68 # </move_to_conf_file>
70 # Let's override the configuration file
71 $style_dir = $ENV{"HELM_STYLE_DIR"} if (defined ($ENV{"HELM_STYLE_DIR"}));
72 $dtd_dir = $ENV{"HELM_DTD_DIR"} if (defined ($ENV{"HELM_DTD_DIR"}));
84 #CSC: mancano i controlli sulle condizioni di errore di molte funzioni
85 #CSC: ==> non e' robusto
86 #CSC: altra roba da sistemare segnata con CSC
88 my $d = new HTTP::Daemon LocalPort => $getterport
89 or die "Error: port $getterport not available.";
90 my $myownurl = $d->url;
92 # Let's patch the returned URL
93 $myownurl =~ s/\/$//; # chop the final slash
94 my $myownport = $myownurl;
95 $myownport =~ s/http:\/\/(.*):(.*)/$2/;
96 $myownurl =~ s/http:\/\/(.*):(.*)/$1/;
97 ($myownurl) = gethostbyname($myownurl);
98 $myownurl = "http://".$myownurl.":".$myownport;
100 tie(%xml_map, 'DB_File', $uris_dbm, O_RDWR, 0664); # open dbs
101 tie(%rdf_map, 'DB_File', $rdf_dbm, O_RDWR, 0664);
102 tie(%xslt_map, 'DB_File', $xslt_dbm, O_RDWR, 0664);
105 print "HTTP Getter $VERSION\n"; # print hello information
106 print "Please contact me at: <URL:", $myownurl."/", ">\n";
108 print "servers_file: $servers_file\n";
109 print "helm_dir: $helm_dir\n";
110 print "helm_rdf_dir: $helm_rdf_dir\n";
111 print "style_dir: $style_dir\n";
112 print "dtd_dir: $dtd_dir\n";
113 print "urls_of_uris db: $uris_dbm\n";
114 print "rdf db: $rdf_dbm\n";
115 print "xslt db: $xslt_dbm\n";
116 print "cache mode: $cachemode\n";
117 print "indexname: $indexname\n";
118 print "rdf_indexname: $rdf_indexname\n";
119 print "xslt_indexname: $xslt_indexname\n";
120 print "dtdbaseurl: $dtdbaseurl\n";
121 print "getterport: $getterport\n";
124 $SIG{CHLD} = "IGNORE"; # do not accumulate defunct processes
125 $SIG{USR1} = \&update; # sent by the child to make the parent update
127 while (my $c = $d->accept) {
129 while (my $r = $c->get_request) {
130 #CSC: mancano i controlli di sicurezza
132 my $http_method = $r->method;
133 my $http_path = $r->url->path;
134 my $http_query = uri_unescape($r->url->query);
135 my $cgi = new CGI("$http_query");
136 my $inputuri = $cgi->param('uri');
137 print "\nRequest: ".$r->url."\n\n";
139 print "\nUnescaped query: ".$http_query."\n";
141 # "getxml" works with rdf uris
142 if ($http_method eq 'GET' and $http_path eq "/getxml") {
143 # finds the uri, url and filename
145 my $answerformat = $cgi->param('format');
146 my $patch_dtd = $cgi->param('patch_dtd');
147 $answerformat = "" if (not defined($answerformat));
148 $patch_dtd = "yes" if (not defined($patch_dtd));
149 if (($answerformat ne "gz") and ($answerformat ne "normal")
150 and ($answerformat ne "")) {
151 die "Wrong output format: $answerformat, must be 'normal' ".
154 if (($patch_dtd ne "yes") and ($patch_dtd ne "no")
155 and ($patch_dtd ne "")) {
156 die "Wrong param, patch_dtd must be 'yes' or 'no'\n";
159 my $filename = $inputuri;
161 if (not isRdfUri($inputuri)) { # standad cic: or theory: uri
162 $filename =~ s/^cic:(.*)/$1/;
163 $filename =~ s/^theory:(.*)/$1/;
165 # The "helm:rdf/<<...>>//cic:" prefix is kept, but quoted
166 # (so that a "helm:rdf/<<...>>//cic:" dir is created
168 $filename =~ s/^(.*)\/\/cic:(.*)/$2/;
169 $filename =~ s/^(.*)\/\/theory:(.*)/$2/;
170 $prefix =~ s/\Q$filename\E//;
173 $filename = $prefix.$filename;
176 my $url = resolve ($inputuri); # resolve uri in url
177 if (not defined($url)) { # uri not found in uri2url map
178 die "NOT FOUND: uri \"$inputuri\" can't be resolved\n";
179 $c->send_error(RC_NOT_FOUND)
180 } else { # uri found and mapped to url
181 my $extension; # file extension
182 if ($url =~ /\.xml$/) { # non gzipped file
184 } elsif ($url =~ /\.xml\.gz$/) { # gzipped file
185 $extension = ".xml.gz";
186 } else { # error: unknown extension
187 die "unexpected extension in url: $url, might be '.xml'".
191 if (not isRdfUri ($inputuri)) { # save in uri std cache dir
192 $filename = $helm_dir.$filename.$extension;
193 } else { # save in rdf metadata cache dir
194 $filename = $helm_rdf_dir."/".$filename.$extension;
197 print_request($inputuri,$url,$filename);
200 my $ciccontent = download($patch_dtd,$url,$filename,$answerformat);
202 if ($answerformat eq "normal") { # answer the client in text/xml
203 answer($c,$ciccontent,"text/xml","");
204 } else { # answer the client in text/xml, gzip encoding
205 answer($c,$ciccontent,"text/xml","x-gzip");
209 # "/register" does not work with rdf uris
210 } elsif ($http_method eq 'GET' and $http_path eq "/register") {
211 my $inputurl = $cgi->param('url');
212 print "Register requested...\n";
213 $xml_map{$inputuri}=$inputurl;
215 # Now let's clean the cache
216 my $cicfilename = $inputuri;
217 $cicfilename =~ s/cic:(.*)/$1/;
218 $cicfilename =~ s/theory:(.*)/$1/;
220 print "Unlinking ".$helm_dir.$cicfilename.".xml[.gz]\n";
221 unlink ($helm_dir.$cicfilename.".xml");
222 unlink ($helm_dir.$cicfilename.".xml.gz");
224 kill(USR1,getppid()); # signal changes to parent
227 html_nice_answer($c,"Register done");
229 # "/resolve" works with rdf uri
230 } elsif ($http_method eq 'GET' and $http_path eq "/resolve") {
231 #my $outputurl = $xml_map{$inputuri};
232 my $outputurl = resolve($inputuri);
233 $outputurl = "" if (not defined($outputurl));
234 $cont = "<?xml version=\"1.0\" ?>\n\n";
235 if ($outputurl eq "") {
236 $cont .= "<unresolved />\n";
238 $cont .= "<url value=\"$outputurl\" />\n";
240 answer($c,$cont,"text/xml","");
242 } elsif ($http_method eq 'GET' and $http_path eq "/getdtd") {
244 my $filename = $inputuri;
245 $filename = $dtd_dir."/".$filename;
246 print "DTD: $inputuri ==> ($filename)\n";
247 if (stat($filename)) {
248 print "Using local copy\n";
249 open(FD, $filename) or die "Cannot open $filename\n";
252 # Vary bad heuristic for substituion of absolute URLs
254 s/ENTITY (.*) SYSTEM\s+"/ENTITY $1 SYSTEM "$myownurl\/getdtd?uri=/g;
258 answer($c,$cont,"text/xml","");
260 die "Could not find DTD!";
263 } elsif ($http_method eq 'GET' and $http_path eq "/getxslt") {
265 $url = $xslt_map{$inputuri};
266 print "Downloading XSLT ...\n";
267 helm_wget($tmp_dir, $url);
268 print "XSLT downloaded!\n";
269 $inputuri =~ s/.*\///;
270 my $filename = $tmp_dir . "/" . $inputuri;
272 die "Could not find XSLT!\n" unless stat($filename);
273 open(FD, $filename) or die "Can't open xslt local copy: '$filename'\n";
275 s/xsl:import\s+href="/xsl:import href="$myownurl\/getxslt?uri=/g;
276 s/xsl:include\s+href="/xsl:include href="$myownurl\/getxslt?uri=/g;
280 answer($c, $cont, "text/xml", "");
282 print "Can't unlink XSLT temp file '$filename'";
285 # my $filename = $inputuri;
286 # $filename = $style_dir."/".$filename;
287 # print "XSLT: $inputuri ==> ($filename)\n";
288 # if (stat($filename)) {
289 # print "Using local copy\n";
290 # open(FD, $filename) or die "Cannot open $filename\n";
293 # # Vary bad heuristic for substituion of absolute URLs
294 # # for relative ones
295 # s/xsl:import\s+href="/xsl:import href="$myownurl\/getxslt?uri=/g ;
296 # s/xsl:include\s+href="/xsl:include href="$myownurl\/getxslt?uri=/g ;
300 # answer($c,$cont,"text/xml","");
302 # die "Could not find XSLT!";
306 # "/update" works with rdf uri
307 } elsif ($http_method eq 'GET' and $http_path eq "/update") {
308 # rebuild urls_of_uris db and rdf uris db
309 print "Update requested...\n";
311 kill(USR1,getppid()); # signal changes to parent
313 html_nice_answer($c,"Update done");
315 # "/getalluris" does not work with rdf uris
316 } elsif ($http_method eq 'GET' and $http_path eq "/getalluris") {
317 # send back all the keys in xml
318 print "BASEURI $baseuri, FORMAT $outype\n";
319 $cont = getalluris();
320 answer($c,$cont,"text/xml","");
323 } elsif ($http_method eq 'GET' and $http_path eq "/getallrdfuris") {
324 # send back all the keys in xml
325 my $class = $cgi->param('class');
326 print "BASEURI $baseuri, FORMAT $outype, CLASS $class\n";
327 $cont = getallrdfuris($class);
328 answer($c,$cont,"text/xml","");
330 } elsif ($http_method eq 'GET' and $http_path eq "/ls") {
331 # send back keys that begin with a given uri
332 my ($uritype,$uripattern,$uriprefix);
333 my $baseuri = $cgi->param('baseuri');
334 chop $baseuri if ($baseuri =~ /.*\/$/); # remove trailing "/"
335 # output type, might be 'txt' or 'xml':
336 my $outype = $cgi->param('format');
337 $uripattern = $baseuri;
338 $uripattern =~ s/^.*:(.*)/$1/;
339 if ($baseuri =~ /^cic:/) {
341 } elsif ($baseuri =~ /^theory:/) {
344 $uritype = "invalid";
346 if ($uritype ne "invalid") { # uri is valid
347 if (($outype ne 'txt') and ($outype ne 'xml')) { # invalid out type
348 print "Invalid output type specified: $outype\n";
349 html_nice_answer($c,"Invalid output type, must be ".
351 } else { # valid output
352 print "BASEURI $baseuri, FORMAT $outype\n";
353 $cont = finduris($uritype,$uripattern,$outype);
354 if ($outype eq 'txt') {
355 answer($c,$cont,"text/plain","");
356 } elsif ($outype eq 'xml') {
357 answer($c,$cont,"text/xml","");
359 die "Internal error, exiting!";
362 } else { # invalid uri
363 print "Invalid uri: $baseuri, may begin with 'cic:', ".
364 "'theory:' or '*:'\n";
365 html_nice_answer($c,"Invalid uri , must begin with ".
366 "'cic:' or 'theory:'");
369 } elsif ($http_method eq 'GET' and $http_path eq "/help") {
371 print "Help requested!\n";
372 html_nice_answer($c,"HTTP Getter Version: $VERSION");
374 } elsif ($http_method eq 'GET' and $http_path eq "/getempty") {
375 # return an empty, but valid, xml document
376 my $emptyxml = <<EOXML;
377 <?xml version="1.0"?>
379 <!ELEMENT empty EMPTY>
383 answer($c, $emptyxml, "", "");
385 } elsif ($http_method eq 'GET' and $http_path =~ /\.cgi$/) {
387 print "CGI requested!\n";
388 if ($http_path !~ /^\/[^\/]*\.cgi$/) {
389 html_nice_answer($c,"Invalid CGI name: $http_path, ".
390 "you can't request CGI that contain a slash in the name\n");
391 } elsif (stat "$cgi_dir"."$http_path") {
392 if (not -x "$cgi_dir/$http_path") {
393 html_nice_answer($c,"CGI $http_path found but not ".
395 } else { # exec CGI and anser back its output
398 $ENV{'QUERY_STRING'} = $http_query;
399 my $cgi_output = `$cgi_dir/$http_path`;
400 answer($c,$cgi_output,"","");
404 html_nice_answer($c,"CGI '$http_path' not found ".
405 "in CGI dir '$cgi_dir'");
408 } else { # unsupported request
410 print "INVALID REQUEST!!!!!\n";
411 print "(PATH: ",$http_path,", ";
412 print "QUERY: ",$http_query,")\n";
413 $c->send_error(RC_FORBIDDEN)
416 print "\nRequest solved: ".$r->url."\n\n";
421 print "\nCONNECTION CLOSED\n\n";
426 #================================
428 sub isRdfUri { # return true if the uri is an rdf uri, false otherwise
430 # helm:rdf/cic:www.cs.unibo.it/helm/rdf/foo_schema//cic:\
431 # /Coq/Init/Logic/True_rec.con.types
433 # the format is "helm:rdf/<metadata_tree>:<metadata_scheme>//<xml_file_uri>"
436 if ($uri =~ /^helm:rdf(.*):(.*)\/\/(.*)/) {
443 sub resolve { # resolve an uri in a url, work both with standard cic: or theory:
445 print "RESOLVE subroutine\n";
447 print "GIVEN URI: \"$uri\"\n";
448 if (isRdfUri ($uri)) { # rdf uri, resolve using rdf db
449 print "IS A RDF URI\n";
450 print "I WILL RETURN '$rdf_map{$uri}'\n";
451 return ($rdf_map{$uri});
452 } else { # standard cic: or theory: uri, resolve using std uri db
453 print "IS NOT A RDF URI\n";
454 print "I WILL RETURN '$xml_map{$uri}'\n";
455 return ($xml_map{$uri});
459 sub getalluris { # get all the keys whose prefix is cic
462 my $debug=1; # for debug
464 $content .= '<?xml version="1.0" encoding="ISO-8859-1"?>' . "\n";
465 $content .= "<!DOCTYPE alluris SYSTEM ";
466 $content .= "\"$myownurl/getdtd?uri=alluris.dtd\">" . "\n\n";
467 $content .= "<alluris>\n";
468 foreach $uri (sort (keys(%xml_map))) { # select matching uris
469 if ($uri =~ /^cic:/ && not $uri =~ /.types$/) {
470 print "GETALLURI: $uri\n" if defined($debug);
471 $content .= "\t<uri value=\"$uri\"/>\n";
474 $content .= "</alluris>\n";
482 my $debug=1; # for debug
484 $content .= '<?xml version="1.0" encoding="ISO-8859-1"?>' . "\n";
485 $content .= "<!DOCTYPE allrdfuris SYSTEM ";
486 $content .= "\"$myownurl/getdtd?uri=alluris.dtd\">" . "\n\n";
487 $content .= "<allrdfuris>\n";
488 foreach $uri (sort (keys(%rdf_map))) {
489 if ($class eq "forward" &&
490 $uri =~ /^helm:rdf:www.cs.unibo.it\/helm\/rdf\/forward/
492 $class eq "backward" &&
493 $uri =~ /^helm:rdf:www.cs.unibo.it\/helm\/rdf\/backward/) {
494 print "GETALLRDFURI: $uri\n" if defined($debug);
495 $content .= "\t<uri value=\"$uri\"/>\n";
498 $content .= "</allrdfuris>\n";
502 sub finduris { # find uris for cic and theory trees generation
503 my ($uritype,$uripattern,$format) = @_;
505 my ($uri,$localpart,$basepart,$dirname,$suffix,$flags,$key);
506 my (@itemz,@already_pushed_dir);
507 my (%objects,%dirs); # map uris to suffixes' flags
508 my $debug=1; # for debug
510 print "FINDURIS, uritype: $uritype, uripattern: $uripattern, ".
511 "format: $format\n\n" if defined($debug);
513 if (($uritype eq "cic") or ($uritype eq "theory")) {
514 # get info only of one type: cic or theory
515 foreach (keys(%xml_map)) { # select matching uris
517 if ($uri =~ /^$uritype:$uripattern(\/|$|\.)/) {
518 if ($uri =~ /^$uritype:$uripattern\//) { # directory match
520 $localpart =~ s/^$uritype:$uripattern\/(.*)/$1/;
521 } elsif ($uri =~ /^$uritype:$uripattern($|\.)/) { # file match
523 $localpart =~ s/^.*\/([^\/]*)/$1/;
525 die "Internal error, seems that requested match is none of ".
526 "directory match or file match";
528 print "LOCALPART: $localpart\n" if defined($debug);
530 if ($localpart =~ /^[^\/]*$/) { # no slash, an OBJECT
531 $basepart = $localpart;
532 $basepart =~ s/^([^.]*\.[^.]*)((\.body)|(\.types))?(\.ann)?/$1/;
533 # remove exts .types, .body,
534 # .types.ann or .body.ann
535 $flags = $objects{$basepart}; # get old flags
536 if ($localpart =~ /\.ann$/) {
537 $flags = add_flag("ann","YES",$flags);
539 $flags = add_flag("ann","NO",$flags);
541 if ($localpart =~ /\.types$/) {
542 $flags = add_flag("types","YES",$flags);
543 } elsif ($localpart =~ /\.types\.ann$/) {
544 $flags = add_flag("types","ANN",$flags);
546 $flags = add_flag("types","NO",$flags);
548 if ($localpart =~ /\.body$/) {
549 $flags = add_flag("body","YES",$flags);
550 } elsif ($localpart =~ /\.body\.ann$/) {
551 $flags = add_flag("body","ANN",$flags);
553 $flags = add_flag("body","NO",$flags);
555 $objects{$basepart} = $flags; # save new flags
556 } else { # exists at least one slash, a DIR
557 ($dirname) = split (/\//, $localpart);
558 $dirs{$dirname} = ""; # no flags requirement for dir
563 die "getter internal error: unsupported uritype: \"$uritype\"";
565 # now generate OUTPUT:
566 # output will be saved in $content
567 if ($format eq "txt") { # TXT output
568 foreach $key (sort(keys %dirs)) {
569 $content .= "dir, " . $key . "\n";
571 foreach $key (sort(keys %objects)) {
572 $content .= "object, $key, " . $objects{$key} . "\n";
574 } elsif ($format eq "xml") { # XML output
575 $content .= '<?xml version="1.0" encoding="ISO-8859-1"?>' . "\n";
576 $content .= "<!DOCTYPE ls SYSTEM ";
577 $content .= "\"$myownurl/getdtd?uri=ls.dtd\">" . "\n\n";
578 $content .= "<ls>\n";
579 foreach $key (sort(keys %dirs)) {
580 $content .= "\t<section>$key</section>\n";
582 foreach $key (sort(keys %objects)) {
583 $content .= "\t<object name=\"$key\">\n";
584 $flags = $objects{$key};
585 $flags =~ s/^<(.*)>$/$1/;
586 my ($annflag,$typesflag,$bodyflag) = split /,/,$flags;
587 $content .= "\t\t<ann value=\"$annflag\" />\n";
588 $content .= "\t\t<types value=\"$typesflag\" />\n";
589 $content .= "\t\t<body value=\"$bodyflag\" />\n";
590 $content .= "\t</object>\n";
592 $content .= "</ls>\n";
593 } else { # may not enter this branch
594 die "Getter internal error: invalid format received by finduris sub";
600 # manage string like: "<ann_flag,type_flag,body_flag>"
601 # "ann_flag" may be one of "ann_YES", "ann_NO"
602 # "type_flag" may be one of "types_NO", "types_YES", "types_ANN"
603 # "body_flag" may be one of "body_NO", "body_YES", "body_ANN"
604 # when adding a flag the max between the current flag and the new flag
605 # is taken, the orders are ann_NO < ann_YES, types_NO < types_YES <
606 # types_ANN and body_NO < body_YES < body_ANN
607 my ($flagtype,$newflag,$str) = @_;
608 $str = "<,,>" if ($str eq "");
609 ($str =~ s/^<(.*,.*,.*)>$/$1/) or die "Internal error: ".
610 "wrong string format for flag adding in $str";
611 my ($annflag,$typeflag,$bodyflag) = split /,/,$str;
612 if ($flagtype eq "ann") { # ANN flag handling
613 if ($newflag eq "YES") {
615 } elsif ($newflag eq "NO") {
616 $annflag = "NO" unless ($annflag eq "YES");
618 die "Internal error: annflag must be \"YES\" or \"NO\"";
620 } elsif ($flagtype eq "types") { # TYPES flag handling
621 if ($newflag eq "ANN") {
623 } elsif ($newflag eq "YES") {
624 $typeflag = "YES" unless ($typeflag eq "ANN");
625 } elsif ($newflag eq "NO") {
627 unless (($typeflag eq "ANN") or ($typeflag eq "YES"));
629 die "Internal error: typeflag must be \"YES\", \"NO\" or \"ANN\"";
631 } elsif ($flagtype eq "body") { # BODY flag handling
632 if ($newflag eq "ANN") {
634 } elsif ($newflag eq "YES") {
635 $bodyflag = "YES" unless ($bodyflag eq "ANN");
636 } elsif ($newflag eq "NO") {
638 unless (($bodyflag eq "ANN") or ($bodyflag eq "YES"));
640 die "Internal error: typeflag must be \"YES\", \"NO\" or \"ANN\"";
643 die "Internal error: unsupported flagtype \"$flagtype\"";
645 $str = "<$annflag,$typeflag,$bodyflag>";
648 #CSC: Too much powerful: creates even /home, /home/users/, ...
649 #CSC: Does not raise errors if could not create dirs/files
653 my @dirs = split /\//,$pathname;
655 foreach $dir (@dirs) {
656 $tmp = ((defined($tmp)) ? $tmp."\/".$dir : "");
664 my ($uri,$url,$filename) = @_;
665 print "<download_request>\n";
668 print "filename: $filename\n\n";
669 print "</download_request>\n";
672 sub gunzip { # gunzip a file and return the deflated content
675 my ($gz, $buffer, $cont);
677 print "deflating $filename ...\n";
678 $gz = gzopen($filename, "r")
679 or die "Cannot open gzip'ed file $filename: $gzerrno";
681 while ( $gz->gzread($buffer) > 0 ) {
684 die "Error while reading : $gzerrno\n" if $gzerrno != Z_STREAM_END ;
690 sub gzip { # gzip the content argument and save it to filename argument
691 my ($cont, $filename) = @_;
692 #my $debug=1; # debug only
694 print "gzopening $filename ...\n" if (defined($debug));
695 my $gz = gzopen($filename, "w")
696 or die "Cannot gzopen for writing file $filename: $gzerrno";
697 print "gzwriting ...\n" if (defined($debug));
698 $gz->gzwrite($cont) or die "error writing: $gzerrno, exiting!\n";
699 print "gzclosing ...\n" if (defined($debug));
704 my ($patch_dtd,$url,$filename,$format) = @_;
707 # print "DOWNLOAD subs receives url: \"$url\"\n";
708 # print "DOWNLOAD subs receives filename: \"$filename\"\n";
710 my $resourcetype; # retrieve mode: "normal" (.xml) or "gzipped" (.xml.gz)
711 if ($filename =~ /\.xml$/) { # set retrieve mode
712 $resourcetype = "normal";
713 } elsif ($filename =~ /\.xml\.gz$/) {
714 $resourcetype = "gzipped";
716 die "Unsupported download extension, might be '.gz' or '.xml'\n";
718 my $basefname = $filename;
719 $basefname =~ s/\.gz$//; # get base resource name removing trailing .gz
720 $cont = ""; # modified by side-effect by the callback function
723 if (stat($basefname)) {
724 $localfname=$basefname;
725 } elsif (stat($basefname.".gz")) {
726 $localfname=$basefname.".gz";
728 if ($localfname ne "") { # we already have local copy of requested file
729 # check both possible cache entry: gzipped or normal
730 print "Using local copy.\n";
731 if ($localfname =~ /\.xml\.gz$/) { # deflating cached file and return it
732 $cont = gunzip($localfname);
733 } elsif ($localfname =~ /\.xml$/) { # just return cached file
734 open(FD, $localfname) or die "Cannot open $localfname";
735 while(<FD>) { $cont .= $_; }
738 die "Internal error: unexpected file name $localfname,"
739 ."must end with '.gz' or '.xml.gz'\n";
741 } else { # download file from net
742 print "Downloading file ...\n"; # download file
743 my $ua = LWP::UserAgent->new;
744 my $request = HTTP::Request->new(GET => "$url");
745 my $response = $ua->request($request, \&callback);
747 # cache retrieved file to disk
748 # <ZACK/> TODO: inefficent, I haven't yet undestood how to deflate
749 # in memory gzipped file, without call "gzopen"
750 print "Storing file ...\n";
751 # print "Making dirs ...\n";
753 # print "Opening tmp file for writing ...\n";
754 open(FD, ">".$filename.".tmp") or die "Cannot open $filename.tmp\n";
755 # print "Writing on tmp file ...\n";
757 # print "Closing tmp file ...\n";
760 # handle cache conversion normal->gzipped or gzipped->normal as user choice
761 print "cachemode:$cachemode, resourcetype:$resourcetype\n"
762 if (defined($debug));
763 if (($cachemode eq 'normal') and ($resourcetype eq 'normal')) {
764 # cache the file as is
765 rename "$filename.tmp", $filename;
766 } elsif (($cachemode eq 'gzipped') and ($resourcetype eq 'gzipped')) {
767 # cache the file as is
768 # and update the $cont variabile with deflated content
769 rename "$filename.tmp", $filename;
770 $cont = gunzip($filename);
771 } elsif (($cachemode eq 'normal') and ($resourcetype eq 'gzipped')) {
772 # deflate cache entry
774 open(FD, "> $basefname") or die "cannot open $basefname\n";
775 $cont = gunzip($filename.".tmp");
778 unlink "$filename.tmp"; # delete old gzipped file
779 } elsif (($cachemode eq 'gzipped') and ($resourcetype eq 'normal')) {
780 # compress cache entry
781 print "gzipping ...\n" if (defined($debug));
782 gzip($cont, $basefname.".gz");
783 unlink "$filename.tmp"; # delete old uncompressed file
785 die "Internal error, unsopported cachemode, resourcetype couple\n";
787 # $cont now contained uncompressed data
789 if ($patch_dtd eq "yes") {
790 $cont =~ s/DOCTYPE (.*) SYSTEM\s+"$dtdbaseurl\//DOCTYPE $1 SYSTEM "$myownurl\/getdtd?uri=/g;
792 if ($format eq "gz") {
793 gzip($cont,"$basefname.tmp");
794 open (TMP, "< $basefname.tmp")
795 or die "Can't open tempfile: $filename.tmp, exiting!\n";
801 unlink ($basefname . ".tmp") or
802 die "cannot unlink temporary file: $basefname.tmp\n";
809 # answer the client setting content, Content-Type and Content-Enconding
811 my ($c,$cont,$contype,$contenc) = @_;
812 my $res = new HTTP::Response;
813 $res->content($cont);
814 $res->push_header("Content-Type" => $contype)
815 unless ($contype eq "");
816 $res->push_header("Content-Encoding" => $contenc)
817 unless ($contenc eq "");
818 $res->push_header("Cache-Control" => "no-cache");
819 $res->push_header("Pragma" => "no-cache");
820 $res->push_header("Expires" => "0");
821 $c->send_response($res);
824 sub html_nice_answer {
825 # answer the client whith a nice html document
826 my ($c,$content) = @_;
827 $content = "<html><body><h1>$content</h1></body></html>";
828 answer($c,$content,"text/html","");
837 # retrieve a file from an url and write it to a temp dir
838 # used for retrieve resource index from servers
840 my ($prefix, $URL) = @_;
841 my $ua = LWP::UserAgent->new;
842 my $request = HTTP::Request->new(GET => "$URL");
843 my $response = $ua->request($request, \&callback);
844 my ($filename) = reverse (split "/", $URL); # get filename part of the URL
845 open (TEMP, "> $prefix/$filename")
846 || die "Cannot open temporary file: $prefix/$filename\n";
854 tie(%xml_map, 'DB_File', $uris_dbm, O_RDWR, 0664);
856 tie(%rdf_map, 'DB_File', $rdf_dbm, O_RDWR, 0664);
858 tie(%xslt_map, 'DB_File', $xslt_dbm, O_RDWR, 0664);
862 # rebuild dbs fetching resource indexes from servers.
863 # Rebuild urls_of_uris db (1), rdf db (2) and xslt db (3)
865 $server, $idxfile, $rdf_idxfile, $uri, $url, $comp, $line,
867 %urls_of_uris, %rdf, %xslt # local proxy of dbs
873 foreach my $file ($uris_dbm, $rdf_dbm, $xslt_dbm) { # remove old db file
874 if (stat $file) { # remove if exists
875 unlink ($file) or die "can't unlink old db file: $file\n";
878 tie(%urls_of_uris, 'DB_File', $uris_dbm, O_RDWR|O_CREAT, 0664);
879 tie(%rdf, 'DB_File', $rdf_dbm, O_RDWR|O_CREAT, 0664);
880 tie(%xslt, 'DB_File', $xslt_dbm, O_RDWR|O_CREAT, 0664);
882 open (SRVS, "< $servers_file") or
883 die "cannot open servers file: $servers_file\n";
884 @servers = <SRVS>; # read all servers
887 while ($server = pop @servers) { # cicle on servers in _reverse_ order
888 print "processing server: $server ...\n";
890 helm_wget($tmp_dir, $server."/".$indexname); # get index
891 helm_wget($tmp_dir, $server."/".$rdf_indexname); # get rdf index
892 helm_wget($tmp_dir, $server."/".$xslt_indexname); # get xslt index
893 $idxfile = $tmp_dir."/".$indexname;
894 $rdf_idxfile = $tmp_dir."/".$rdf_indexname;
895 $xslt_idxfile = $tmp_dir . "/" . $xslt_indexname;
897 # (1) REBUILD URLS_OF_URIS DB
898 open (INDEX, "< $idxfile") or
899 die "cannot open temporary index file: $idxfile\n";
900 while (my $line = <INDEX>) { # parse index and add entry to urls_of_uris db
902 ($uri,$comp) = split /[ \t]+/, $line;
905 $url = $uri . ".xml" . ".gz";
907 $url = $uri . ".xml";
909 $url =~ s/cic:/$server/;
910 $url =~ s/theory:/$server/;
911 $urls_of_uris{$uri} = $url;
915 open (RDF_INDEX, "< $rdf_idxfile") or
916 die "cannot open temporary RDF index file: $rdf_idxfile\n";
917 while (my $line = <RDF_INDEX>) { # parse index and add entry to rdf db
919 ($uri,$comp) = split /[ \t]+/, $line; # comp is "gz" or nothing
922 $url = $uri . ".xml" . ".gz";
924 $url = $uri . ".xml";
926 $url =~ s/^helm:rdf(.*)\/\/cic:/$server/;
927 $url =~ s/^helm:rdf(.*)\/\/theory:/$server/;
931 # (3) REBUILD XSLT DB
932 open (XSLT_INDEX, "< $xslt_idxfile") or
933 die "cannot open temporary XSLT index file: $xslt_idxfile\n";
934 while (my $line = <XSLT_INDEX>) { # parse index and add entry to xslt db
936 my $stylesheet = $line;
937 my $url = $server . "/" . $stylesheet;
938 $xslt{$stylesheet} = $url;
941 close INDEX; # close indexes
946 foreach my $file ($idxfile, $rdf_idxfile, $xslt_idxfile) {
947 print "cannot unlink temporary file: $file\n" if (unlink $file) != 1;
952 untie(%urls_of_uris); # untie local proxies
955 tie(%xml_map, 'DB_File', $uris_dbm, O_RDWR, 0664); # retie global ones
956 tie(%rdf_map, 'DB_File', $rdf_dbm, O_RDWR, 0664);
957 tie(%xslt_map, 'DB_File', $xslt_dbm, O_RDWR, 0664);
961 # vim modline: do not remove!