X-Git-Url: http://matita.cs.unibo.it/gitweb/?a=blobdiff_plain;f=helm%2Fhttp_getter%2Fhttp_getter.pl.in;h=fd72b75094014106160cd3a8ecae90176e8eb2ce;hb=9632260a013dabd108d9965d6faffd99246e1f47;hp=f0392de2caa9acd9cbafb1b4805ba6f71e28e1c3;hpb=d71338dee9f65981e827bca3b4d6d79c0197b4d9;p=helm.git
diff --git a/helm/http_getter/http_getter.pl.in b/helm/http_getter/http_getter.pl.in
index f0392de2c..fd72b7509 100755
--- a/helm/http_getter/http_getter.pl.in
+++ b/helm/http_getter/http_getter.pl.in
@@ -27,12 +27,26 @@
use Env;
my $HELM_LIB_DIR = $ENV{"HELM_LIB_DIR"};
# this should be the only fixed constant
-my $DEFAULT_HELM_LIB_DIR = "@DEFAULT_HELM_LIB_DIR@";
+my $DEFAULT_HELM_LIB_DIR = "@HELM_LIB_DIR@";
if (defined ($HELM_LIB_DIR)) {
$HELM_LIB_PATH = $HELM_LIB_DIR."/configuration.pl";
} else {
$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 = $ENV{'HTTP_GETTER_CACHE_MODE'} || 'gzipped';
+if (($cachemode ne 'gzipped') and ($cachemode ne 'normal')) {
+ die "Invalid HTTP_GETTER_CACHE_MODE environment variable, must be".
+ "'normal' or 'gzipped'\n";
+}
+#
+
# next require defines: $helm_dir, $html_link, $dtd_dir, $uris_dbm
require $HELM_LIB_PATH;
@@ -48,11 +62,24 @@ use Compress::Zlib;
#CSC: altra roba da sistemare segnata con CSC
my $d = new HTTP::Daemon LocalPort => 8081;
+my $myownurl = $d->url;
+
+# Let's patch the returned URL
+$myownurl =~ s/\/$//; # chop the final slash
+my $myownport = $myownurl;
+$myownport =~ s/http:\/\/(.*):(.*)/$2/;
+$myownurl =~ s/http:\/\/(.*):(.*)/$1/;
+($myownurl) = gethostbyname($myownurl);
+$myownurl = "http://".$myownurl.":".$myownport;
+
tie(%map, 'DB_File', $uris_dbm.".db", O_RDONLY, 0664);
-print "Please contact me at: url, ">\n";
+print "Please contact me at: \n";
print "helm_dir: $helm_dir\n";
+print "style_dir: $style_dir\n";
print "dtd_dir: $dtd_dir\n";
print "urls_of_uris.db: $uris_dbm.db\n";
+print "cache mode: $cachemode\n";
+
$SIG{CHLD} = "IGNORE"; # do not accumulate defunct processes
$SIG{USR1} = \&update; # sent by the child to make the parent update
while (my $c = $d->accept) {
@@ -73,21 +100,19 @@ while (my $c = $d->accept) {
my $cicfilename = $cicuri;
$cicfilename =~ s/cic:(.*)/$1/;
$cicfilename =~ s/theory:(.*)/$1/;
-# $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'";
- }
+ 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)
@@ -127,45 +152,41 @@ 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 $cicurl = $map{$cicuri};
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: $annurl might be '.xml' or '.xml.gz'";
- }
- }
+ my ($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) ||
@@ -208,14 +229,54 @@ EOT
print "DTD: $inputuri ==> ($filename)\n";
if (stat($filename)) {
print "Using local copy\n";
- open(FD, $filename);
+ open(FD, $filename) or die "Cannot open $filename\n";
$cont = "";
- while() { $cont .= $_; }
+ while() {
+ # Vary bad heuristic for substituion of absolute URLs
+ # for relative ones
+ s/ENTITY (.*) SYSTEM\s+"/ENTITY $1 SYSTEM "$myownurl\/getdtd?uri=/g;
+ $cont .= $_;
+ }
close(FD);
answer($c,$cont);
} else {
die "Could not find DTD!";
}
+ } elsif ($http_method eq 'GET' and $http_path eq "/getstyleconf") {
+ my $filename = $inputuri;
+ $filename = $style_dir."/config/".$filename;
+ if (stat($filename)) {
+ open(FD, $filename) or die "Cannot open $filename\n";
+ $cont = "";
+ while() {
+ s/DOCTYPE (.*) SYSTEM\s+"/DOCTYPE $1 SYSTEM "$myownurl\/getstyleconf?uri=/g;
+ $cont .= $_;
+ }
+ close(FD);
+ answer($c,$cont);
+ } else {
+ die "Could not find Style Configuration File!";
+ }
+ } elsif ($http_method eq 'GET' and $http_path eq "/getxslt") {
+ my $filename = $inputuri;
+ $filename = $style_dir."/".$filename;
+ print "XSLT: $inputuri ==> ($filename)\n";
+ if (stat($filename)) {
+ print "Using local copy\n";
+ open(FD, $filename) or die "Cannot open $filename\n";
+ $cont = "";
+ while() {
+ # Vary bad heuristic for substituion of absolute URLs
+ # for relative ones
+ s/xsl:import\s+href="/xsl:import href="$myownurl\/getxslt?uri=/g ;
+ s/xsl:include\s+href="/xsl:include href="$myownurl\/getxslt?uri=/g ;
+ $cont .= $_;
+ }
+ close(FD);
+ answer($c,$cont);
+ } else {
+ die "Could not find XSLT!";
+ }
} elsif ($http_method eq 'GET' and $http_path eq "/conf") {
my $quoted_html_link = $html_link;
$quoted_html_link =~ s/&/&/g;
@@ -276,65 +337,112 @@ sub callback
$cont .= $data;
}
+sub gunzip { # gunzip a file and return the deflated content
+ my ($filename) = @_;
+
+ my ($gz, $buffer, $cont);
+
+ print "deflating $filename ...\n";
+ $gz = gzopen($filename, "r")
+ or die "Cannot open gzip'ed file $filename: $gzerrno";
+ $cont = "";
+ while ( $gz->gzread($buffer) > 0 ) {
+ $cont .= $buffer;
+ }
+ die "Error while reading : $gzerrno\n" if $gzerrno != Z_STREAM_END ;
+ $gz->gzclose();
+
+ return $cont;
+}
+
+sub gzip { # gzip the content argument and save it to filename argument
+ my ($cont, $filename) = @_;
+
+ my ($gz, $cont);
+
+ $gz = gzopen($filename, "w")
+ or die "Cannot gzopen for writing file $filename: $gzerrno";
+ $gz->gzwrite($cont) or die "error writing: $gzerrno\n" ;
+ $gz->gzclose();
+}
+
sub download
{
my ($remove_headers,$str,$url,$filename) = @_;
-#
my ($gz, $buffer);
- my $mode; # retrieve mode: "normal" (.xml) or "gzipped" (.xml.gz)
+ my $resourcetype; # retrieve mode: "normal" (.xml) or "gzipped" (.xml.gz)
if ($filename =~ /\.xml$/) { # set retrieve mode
- $mode = "normal";
+ $resourcetype = "normal";
} elsif ($filename =~ /\.xml\.gz$/) {
- $mode = "gzipped";
+ $resourcetype = "gzipped";
} else {
die "Unsupported download extension, might be '.gz' or '.xml'\n";
}
-#
+ my $basefname = $filename;
+ $basefname =~ s/\.gz$//; # get base resource name removing trailing .gz
$cont = ""; # modified by side-effect by the callback function
- if (stat($filename)) { # we already have local copy of requested file
- print "Using local copy for 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
+
+ my $localfname="";
+ if (stat($basefname)) {
+ $localfname=$basefname;
+ } elsif (stat($basefname.".gz")) {
+ $localfname=$basefname.".gz";
+ }
+ if ($localfname ne "") { # we already have local copy of requested file
+ # check both possible cache entry: gzipped or normal
+ print "Using local copy for the $str file\n";
+ if ($localfname =~ /\.xml\.gz$/) { # deflating cached file and return it
+ $cont = gunzip($localfname);
+ } elsif ($localfname =~ /\.xml$/) { # just return cached file
+ open(FD, $localfname) or die "Cannot open $localfname";
+ while() { $cont .= $_; }
+ close(FD);
+ } else { # error
+ die "Internal error: unexpected file name $localfname,"
+ ."must end with '.gz' or '.xml.gz'\n";
+ }
+ } 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);
- print "Storing the $str file\n";
- mkdirs($filename);
- 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
- }
-#
+ # cache retrieved file to disk
+# TODO: inefficent, I haven't yet undestood how to deflate
+# in memory gzipped file, without call "gzopen"
+ print "Storing the $str file\n";
+ mkdirs($filename);
+ open(FD, ">".$filename.".tmp") or die "Cannot open $filename.tmp\n";
+ print FD $cont;
+ close(FD);
+
+ # handle cache conversion normal->gzipped or gzipped->normal as user choice
+ if (($cachemode eq 'normal') and ($resourcetype eq 'normal')) {
+ # cache the file as is
+ rename "$filename.tmp", $filename;
+ } elsif (($cachemode eq 'gzipped') and ($resourcetype eq 'gzipped')) {
+ # cache the file as is
+ # and update the $cont variabile with deflated content
+ rename "$filename.tmp", $filename;
+ $cont = gunzip($filename);
+ } elsif (($cachemode eq 'normal') and ($resourcetype eq 'gzipped')) {
+ # deflate cache entry
+ # and update $cont
+ open(FD, "> $basefname") or die "cannot open $basefname\n";
+ $cont = gunzip($filename.".tmp");
+ print FD $cont;
+ close(FD);
+ unlink "$filename.tmp"; # delete old gzipped file
+ } elsif (($cachemode eq 'gzipped') and ($resourcetype eq 'normal')) {
+ # compress cache entry
+ gzip($cont, $basefname.".gz");
+ unlink "$filename.tmp"; # delete old uncompressed file
+ } else {
+ die "Internal error, unsopported cachemode, resourcetype couple\n";
+ }
+ # $cont now contained uncompressed data
+
}
if ($remove_headers) {
$cont =~ s/<\?xml [^?]*\?>//sg;