X-Git-Url: http://matita.cs.unibo.it/gitweb/?a=blobdiff_plain;ds=sidebyside;f=helm%2Fhttp_getter%2Fhttp_getter.pl.in;h=bee3315e7583cb9e1a98700559c36fc9eafb34b4;hb=4e8b137b58ec6763785b8ee490ee8cd6cb77b5bf;hp=e8adb3bc02f321d0c3ca1adc05b4f894749c0758;hpb=8e1a7e55cbc7750446f0a7ab3d071190594243fb;p=helm.git
diff --git a/helm/http_getter/http_getter.pl.in b/helm/http_getter/http_getter.pl.in
index e8adb3bc0..bee3315e7 100755
--- a/helm/http_getter/http_getter.pl.in
+++ b/helm/http_getter/http_getter.pl.in
@@ -25,22 +25,23 @@
# First of all, let's load HELM configuration
use Env;
-my $HELM_LIBRARY_DIR = $ENV{"HELM_LIBRARY_DIR"};
+my $HELM_LIB_DIR = $ENV{"HELM_LIB_DIR"};
# this should be the only fixed constant
-my $DEFAULT_HELM_LIBRARY_DIR = "@DEFAULT_HELM_LIBRARY_DIR@";
-if (defined ($HELM_LIBRARY_DIR)) {
- $HELM_LIBRARY_PATH = $HELM_LIBRARY_DIR."/configuration.pl";
+my $DEFAULT_HELM_LIB_DIR = "@HELM_LIB_DIR@";
+if (defined ($HELM_LIB_DIR)) {
+ $HELM_LIB_PATH = $HELM_LIB_DIR."/configuration.pl";
} else {
- $HELM_LIBRARY_PATH = $DEFAULT_HELM_LIBRARY_DIR."/configuration.pl";
+ $HELM_LIB_PATH = $DEFAULT_HELM_LIB_DIR."/configuration.pl";
}
# next require defines: $helm_dir, $html_link, $dtd_dir, $uris_dbm
-require $HELM_LIBRARY_PATH;
+require $HELM_LIB_PATH;
use HTTP::Daemon;
use HTTP::Status;
use HTTP::Request;
use LWP::UserAgent;
use DB_File;
+use Compress::Zlib;
#CSC: mancano i controlli sulle condizioni di errore di molte funzioni
#CSC: ==> non e' robusto
@@ -72,9 +73,21 @@ while (my $c = $d->accept) {
my $cicfilename = $cicuri;
$cicfilename =~ s/cic:(.*)/$1/;
$cicfilename =~ s/theory:(.*)/$1/;
- $cicfilename = $helm_dir.$cicfilename.".xml";
-
+# $cicfilename = $helm_dir.$cicfilename.".xml";
+#
my $cicurl = $map{$cicuri};
+ my $extension;
+ if ($cicurl =~ /\.xml$/) { # non gzipped file
+ $extension = ".xml";
+ } elsif ($cicurl =~ /\.xml\.gz$/) { # gzipped file
+ $extension = ".xml.gz";
+ } else { # error: unknown extension
+ die "unexpected extension in url: $cicurl, might be '.xml' or '.xml.gz'";
+ }
+ $cicfilename = $helm_dir.$cicfilename.$extension;
+
+ #my $cicurl = $map{$cicuri};
+#
if (!defined($cicurl)) {
print "\nNOT FOUND!!!!!\n";
$c->send_error(RC_NOT_FOUND)
@@ -114,13 +127,45 @@ while (my $c = $d->accept) {
$cicfilename =~ s/theory:(.*)/$1/;
$cicfilename = $helm_dir.$cicfilename;
- my $typesfilename = $cicfilename.".types.xml" if $typesuri;
- my $annfilename = $cicfilename.$annsuffix.".xml" if $annuri;
- $cicfilename .= ".xml";
+# my $typesfilename = $cicfilename.".types.xml" if $typesuri;
+# my $annfilename = $cicfilename.$annsuffix.".xml" if $annuri;
+# $cicfilename .= ".xml";
+#
my $cicurl = $map{$cicuri};
- my $typesurl = $map{$typesuri} if $typesuri;
- my $annurl = $map{$annuri} if $annuri;
+ my $typesurl = $map{$typesuri} if (defined($typesuri));
+ my $annurl = $map{$annuri} if (defined($annuri));
+ my ($cicext, $typesext, $annext);
+ if ($cicurl =~ /\.xml$/) { # normal file
+ $cicext = ".xml";
+ } elsif ($cicurl =~ /\.xml\.gz$/) { # gzipped file
+ $cicext = ".xml.gz";
+ } else {
+ die "unexpected extension in url: $cicurl; might be '.xml' or '.xml.gz'";
+ }
+ if (defined($typesuri)) { # extension selection for types file
+ if ($typesurl =~ /\.xml$/) { # normal file
+ $typesext = ".types.xml";
+ } elsif ($typesurl =~ /\.xml\.gz$/) { # gzipped file
+ $typesext = ".types.xml.gz";
+ } else {
+ die "unexpected extension in url: $typesurl; might be '.xml' or '.xml.gz'";
+ }
+ }
+ if (defined($annuri)) { # extension selection for annotation file
+ if ($annurl =~ /\.xml$/) { # normal file
+ $annext = ".xml";
+ } elsif ($annurl =~ /\.xml\.gz$/) { # gzipped file
+ $annext = ".xml.gz";
+ } else {
+ die "unexpected extension in url: $annurl might be '.xml' or '.xml.gz'";
+ }
+ }
+ my $typesfilename = $cicfilename.$typesext if $typesuri;
+ my $annfilename = $cicfilename.$annsuffix.$annext if $annuri;
+ $cicfilename .= $cicext;
+#
+
if (!defined($cicurl) ||
(!defined($typesurl) && $typesuri) ||
@@ -234,14 +279,40 @@ sub callback
sub download
{
my ($remove_headers,$str,$url,$filename) = @_;
+#
+ my ($gz, $buffer);
+
+ my $mode; # retrieve mode: "normal" (.xml) or "gzipped" (.xml.gz)
+ if ($filename =~ /\.xml$/) { # set retrieve mode
+ $mode = "normal";
+ } elsif ($filename =~ /\.xml\.gz$/) {
+ $mode = "gzipped";
+ } else {
+ die "Unsupported download extension, might be '.gz' or '.xml'\n";
+ }
+#
$cont = ""; # modified by side-effect by the callback function
- if (stat($filename)) {
+ if (stat($filename)) { # we already have local copy of requested file
print "Using local copy for the $str file\n";
- open(FD, $filename);
- while() { $cont .= $_; }
- close(FD);
- } else {
- print "Downloading the $str file\n";
+#
+ if ($mode eq "gzipped") { # deflating cached file
+ print "deflating local file ...\n";
+ $gz = gzopen($filename, "r") or die "Cannot open gzip'ed file $filename: $gzerrno";
+ while ( $gz->gzread($buffer) > 0 ) {
+ $cont .= $buffer;
+ }
+ die "Error while reading : $gzerrno\n" if $gzerrno != Z_STREAM_END ;
+ $gz->gzclose();
+ } elsif ($mode eq "normal") { # return cached file
+ open(FD, $filename);
+ while() { $cont .= $_; }
+ close(FD);
+ } else { # error
+ die "Internal error: unexpected mode: $mode, might be 'normal' or 'gzipped'";
+ }
+#
+ } 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);
@@ -251,6 +322,19 @@ sub download
open(FD, ">".$filename);
print FD $cont;
close(FD);
+#
+ if ($mode eq "gzipped") { # deflate gzipped retrieved file
+ print "deflating just retrieved file ...\n";
+ $cont = ""; # reset $cont, cause $cont actually contain gzipped data
+ $gz = gzopen($filename, "r") or die "Cannot open gzip'ed file $filename: $gzerrno";
+ while ( $gz->gzread($buffer) > 0 ) {
+ $cont .= $buffer;
+ }
+ die "Error while reading : $gzerrno\n" if $gzerrno != Z_STREAM_END ;
+ $gz->gzclose();
+ # now $cont contain deflated, clear text data
+ }
+#
}
if ($remove_headers) {
$cont =~ s/<\?xml [^?]*\?>//sg;