X-Git-Url: http://matita.cs.unibo.it/gitweb/?a=blobdiff_plain;f=helm%2Fhttp_getter%2Fhttp_getter.pl.in;h=4240edd806e78a9738946a8aa3b3c55ad8ac12be;hb=ef9e59ebe09298c1295ebdd52a9517f1eafcab63;hp=89f496ac7bc201e6cadc50bb09812b347eace3cb;hpb=083c36a4acd76127530b9af3d4c6c311f447d79b;p=helm.git diff --git a/helm/http_getter/http_getter.pl.in b/helm/http_getter/http_getter.pl.in index 89f496ac7..4240edd80 100755 --- a/helm/http_getter/http_getter.pl.in +++ b/helm/http_getter/http_getter.pl.in @@ -36,10 +36,6 @@ if (defined ($HELM_LIB_DIR)) { $HELM_LIB_PATH = $DEFAULT_HELM_LIB_DIR."/configuration.pl"; } -# Let's override the configuration file -$style_dir = $ENV{"HELM_STYLE_DIR"} if (defined ($ENV{"HELM_STYLE_DIR"})); -$dtd_dir = $ENV{"HELM_DTD_DIR"} if (defined ($ENV{"HELM_DTD_DIR"})); - # : TODO temporary, move this setting to configuration file # set the cache mode, may be "gzipped" or "normal" my $cachemode = $ENV{'HTTP_GETTER_CACHE_MODE'} || 'gzipped'; @@ -52,12 +48,18 @@ if (($cachemode ne 'gzipped') and ($cachemode ne 'normal')) { # next require defines: $helm_dir, $html_link, $dtd_dir, $uris_dbm require $HELM_LIB_PATH; +# Let's override the configuration file +$style_dir = $ENV{"HELM_STYLE_DIR"} if (defined ($ENV{"HELM_STYLE_DIR"})); +$dtd_dir = $ENV{"HELM_DTD_DIR"} if (defined ($ENV{"HELM_DTD_DIR"})); + use HTTP::Daemon; use HTTP::Status; use HTTP::Request; use LWP::UserAgent; use DB_File; use Compress::Zlib; +use CGI; +use URI::Escape; #CSC: mancano i controlli sulle condizioni di errore di molte funzioni #CSC: ==> non e' robusto @@ -89,11 +91,15 @@ while (my $c = $d->accept) { while (my $r = $c->get_request) { #CSC: mancano i controlli di sicurezza - my $inputuri = $r->url; + my $inputuri = uri_unescape($r->url); $inputuri =~ s/^[^?]*\?uri=(.*)/$1/; print "\nRequest: ".$r->url."\n\n"; my $http_method = $r->method; my $http_path = $r->url->path; + my $http_query = uri_unescape($r->url->query); + my $cgi = new CGI("$http_query"); + + print "\nUnescaped query: ".$http_query."\n"; if ($http_method eq 'GET' and $http_path eq "/getciconly") { # finds the uri, url and filename @@ -290,16 +296,52 @@ EOT $cont = "$quoted_html_link"; answer($c,$cont); } elsif ($http_method eq 'GET' and $http_path eq "/update") { - print "Update requested..."; - update(); - kill(USR1,getppid()); + # rebuild urls_of_uris.db + print "Update requested...\n"; + mk_urls_of_uris(); + kill(USR1,getppid()); # signal changes to parent print " done\n"; answer($c,"

Update done

"); - } elsif ($http_method eq 'GET' and $http_path eq "/version") { - print "Version requested!"; - answer($c,"

HTTP Getter Version ".$VERSION."

"); + } elsif ($http_method eq 'GET' and $http_path eq "/ls") { + # send back keys that begin with a given uri + my ($uritype,$uripattern,$uriprefix); + my $baseuri = $cgi->param('baseuri'); + chop $baseuri if ($baseuri =~ /.*\/$/); # remove trailing "/" + my $outype = $cgi->param('format'); # output type, might be 'txt' or 'xml' + $uripattern = $baseuri; + $uripattern =~ s/^.*:(.*)/$1/; + if ($baseuri =~ /^cic:/) { + $uritype = "cic"; + } elsif ($baseuri =~ /^theory:/) { + $uritype = "theory"; + } else { + $uritype = "invalid"; + } + if ($uritype ne "invalid") { # uri is valid + if (($outype ne 'txt') and ($outype ne 'xml')) { # invalid out type + print "Invalid output type specified: $outype\n"; + answer($c,"

Invalid output type, may be ". + "\"txt\" or \"xml\"

"); + } else { # valid output + print "BASEURI $baseuri, FORMAT $outype\n"; + $cont = finduris($uritype,$uripattern,$outype); + answer($c,$cont); + } + } else { # invalid uri + print "Invalid uri: $baseuri, may begin with 'cic:', ". + "'theory:' or '*:'\n"; + answer($c,"

Invalid uri , may begin with ". + "\"cic:\", \"theory:\" or \"*:\"

"); + } + } elsif ($http_method eq 'GET' and $http_path eq "/help") { + print "Help requested!"; + answer($c,"

HTTP Getter Version ". + $VERSION."

"); } else { - print "\nINVALID REQUEST!!!!!\n"; + print "\n"; + print "INVALID REQUEST!!!!!\n"; + print "(PATH: ",$http_path,", "; + print "QUERY: ",$http_query,")\n"; $c->send_error(RC_FORBIDDEN) } print "\nRequest solved: ".$r->url."\n\n"; @@ -313,6 +355,117 @@ EOT #================================ +sub finduris { # find uris for cic and theory trees generation + my ($uritype,$uripattern,$format) = @_; + my $content = ""; + my ($uri,$localpart,$basepart,$dirname,$suffix,$flags,$key); + my (@itemz,@already_pushed_dir); + my (%objects,%dirs); # map uris to suffixes' flags + + print "FINDURIS, uritype: $uritype, uripattern: $uripattern, ". + "format: $format\n\n"; + + if (($uritype eq "cic") or ($uritype eq "theory")) { + # get info only of one type: cic or theory + foreach (keys(%map)) { # select matching uris + $uri = $_; + if ($uri =~ /^$uritype:$uripattern\//) { + $localpart = $uri; + $localpart =~ s/^$uritype:$uripattern\/(.*)/$1/; + + if ($localpart =~ /^[^\/]*$/) { # no slash, an OBJECT + $basepart = $localpart; + $basepart =~ s/^(.*)\.types(\.ann)?/$1/; # remove exts .types or + # .types.ann + $flags = $objects{$basepart}; # get old flags + if ($localpart =~ /\.ann$/) { + $flags = add_flag("ann","YES",$flags); + } else { + $flags = add_flag("ann","NO",$flags); + } + if ($localpart =~ /\.types$/) { + $flags = add_flag("types","YES",$flags); + } elsif ($localpart =~ /\.types\.ann$/) { + $flags = add_flag("types","ANN",$flags); + } else { + $flags = add_flag("types","NO",$flags); + } + $objects{$basepart} = $flags; # save new flags + } else { # exists at least one slash, a DIR + ($dirname) = split (/\//, $localpart); + $dirs{$dirname} = ""; # no flags requirement for dir + } + } + } + } else { + die "getter internal error: unsupported uritype: \"$uritype\""; + } + # now generate OUTPUT: + # output will be saved in $content + if ($format eq "txt") { + foreach $key (sort(keys %dirs)) { + $content .= "dir, " . $key . "\n"; + } + foreach $key (sort(keys %objects)) { + $content .= "object, $key, " . $objects{$key} . "\n"; + } + } elsif ($format eq "xml") { + $content .= "\n"; + foreach $key (sort(keys %dirs)) { + $content .= "\t

$key

\n"; + } + foreach $key (sort(keys %objects)) { + $content .= "\t

\n"; + } + $content .= "\n"; + } else { # may not enter this branch + die "Getter internal error: invalid format received by finduris sub"; + } + return $content; +} + +sub add_flag { +# manage string like: "" +# "ann_flag" may be one of "ann_YES", "ann_NO" +# "type_flag" may be one of "types_NO", "types_YES", "types_ANN" +# when adding a flag the max between the current flag and the new flag +# is taken, the orders are ann_NO < ann_YES and types_NO < types_YES < +# types_ANN + my ($flagtype,$newflag,$str) = @_; + $str = "<,>" if ($str eq ""); + ($str =~ s/^<(.*,.*)>$/$1/) or die "Internal error: ". + "wrong string format for flag adding in $str"; + my ($annflag,$typeflag) = split /,/,$str; + if ($flagtype eq "ann") { # ANN flag handling + if ($newflag eq "YES") { + $annflag = "YES"; + } elsif ($newflag eq "NO") { + $annflag = "NO" unless ($annflag eq "YES"); + } else { + die "Internal error: annflag must be \"YES\" or \"NO\""; + } + } elsif ($flagtype eq "types") { # TYPES flag handling + if ($newflag eq "ANN") { + $typeflag = "ANN"; + } elsif ($newflag eq "YES") { + $typeflag = "YES" unless ($typeflag eq "ANN"); + } elsif ($newflag eq "NO") { + $typeflag = "NO" + unless (($typeflag eq "ANN") or ($typeflag eq "YES")); + } else { + die "Internal error: typeflag must be \"YES\", \"NO\" or \"ANN\""; + } + } else { + die "Internal error: unsupported flagtype \"$flagtype\""; + } + $str = "<$annflag,$typeflag>"; +} #CSC: Too much powerful: creates even /home, /home/users/, ... #CSC: Does not raise errors if could not create dirs/files @@ -409,9 +562,9 @@ sub download } } else { # download file from net print "Downloading the $str file\n"; # download file - $ua = LWP::UserAgent->new; - $request = HTTP::Request->new(GET => "$url"); - $response = $ua->request($request, \&callback); + my $ua = LWP::UserAgent->new; + my $request = HTTP::Request->new(GET => "$url"); + my $response = $ua->request($request, \&callback); # cache retrieved file to disk # TODO: inefficent, I haven't yet undestood how to deflate @@ -452,6 +605,8 @@ sub download if ($remove_headers) { $cont =~ s/<\?xml [^?]*\?>//sg; $cont =~ s/]*>//sg; + } else { + $cont =~ s/DOCTYPE (.*) SYSTEM\s+"http:\/\/www.cs.unibo.it\/helm\/dtd\//DOCTYPE $1 SYSTEM "$myownurl\/getdtd?uri=/g; } return $cont; } @@ -464,7 +619,71 @@ sub answer $c->send_response($res); } +sub helm_wget { +#retrieve a file from an url and write it to a temp dir +#used for retrieve resource index from servers + $cont = ""; + my ($prefix, $URL) = @_; + my $ua = LWP::UserAgent->new; + my $request = HTTP::Request->new(GET => "$URL"); + my $response = $ua->request($request, \&callback); + my ($filename) = reverse (split "/", $URL); # get filename part of the URL + open (TEMP, "> $prefix/$filename") + || die "Cannot open temporary file: $prefix/$filename\n"; + print TEMP $cont; + close TEMP; +} + sub update { untie %map; tie(%map, 'DB_File', $uris_dbm.".db", O_RDONLY, 0664); } + +sub mk_urls_of_uris { +#rebuild $uris_dbm.db fetching resource indexes from servers + my ( + $server, $idxfile, $uri, $url, $comp, $line, + @servers, + %urls_of_uris + ); + + untie %map; + if (stat $uris_dbm.".db") { # remove old db file + unlink($uris_dbm.".db") or + die "cannot unlink old db file: $uris_dbm.db\n"; + } + tie(%urls_of_uris, 'DB_File', $uris_dbm.".db", O_RDWR|O_CREAT, 0664); + + open (SRVS, "< $servers_file") or + die "cannot open servers file: $servers_file\n"; + @servers = ; + close (SRVS); + while ($server = pop @servers) { #cicle on servers in reverse order + print "processing server: $server ...\n"; + chomp $server; + helm_wget($tmp_dir, $server."/".$indexname); #get index + $idxfile = $tmp_dir."/".$indexname; + open (INDEX, "< $idxfile") or + die "cannot open temporary index file: $idxfile\n"; + while ($line = ) { #parse index and add entry to urls_of_uris + chomp $line; + ($uri,$comp) = split /[ \t]+/, $line; + # build url: + if ($comp =~ /gz/) { + $url = $uri . ".xml" . ".gz"; + } else { + $url = $uri . ".xml"; + } + $url =~ s/cic:/$server/; + $url =~ s/theory:/$server/; + $urls_of_uris{$uri} = $url; + } + close INDEX; + die "cannot unlink temporary file: $idxfile\n" + if (unlink $idxfile) != 1; + } + + untie(%urls_of_uris); + tie(%map, 'DB_File', $uris_dbm.".db", O_RDONLY, 0664); +} +