[backend] refactor url handling

- move urlencode to BSHTTP
- add urldecode
- add queryencode
- support a fragment

src/backend/BSHTTP.pm

@@ -53,6 +53,32 @@ This library contains functions to handle http requests in obs
 
 =cut
 
+sub urlencode {
+  my ($str, $iscgi) = @_;
+  if ($iscgi) {
+    $str =~ s/([\000-\037<>;\"#\?&\+=%[\177-\377])/sprintf("%%%02X",ord($1))/sge;
+    $str =~ s/ /+/sg;
+  } else {
+    $str =~ s/([\000-\040<>;\"#\?&\+=%[\177-\377])/sprintf("%%%02X",ord($1))/sge;
+  }
+  return $str;
+}
+
+sub urldecode {
+  my ($str, $iscgi) = @_;
+  $str =~ tr/+/ / if $iscgi;
+  $str =~ s/%([a-fA-F0-9]{2})/chr(hex($1))/sge;
+  return $str;
+}
+
+sub queryencode {
+  my (@args, $iscgi) = @_;
+  for (@args) {
+    $_ = urlencode($_, $iscgi);
+    s/%3D/=/;	# convert now escaped = back
+  }
+  return join('&', @args);
+}
 
 sub gethead {
   my ($h, $t) = @_;

src/backend/BSRPC.pm

@@ -51,9 +51,7 @@ sub import {
 my $tcpproto = getprotobyname('tcp');
 
 sub urlencode {
-  my $url = $_[0];
-  $url =~ s/([\000-\040<>;\"#\?&\+=%[\177-\377])/sprintf("%%%02X",ord($1))/sge;
-  return $url;
+  return BSHTTP::urlencode($_[0]);
 }
 
 sub createuri {
@@ -63,22 +61,13 @@ sub createuri {
     # encode uri, but do not encode the host part
     if ($uri =~ /^(https?:\/\/[^\/]*\/)(.*)$/s) {
       $uri = $1; 
-      $uri .= urlencode($2);
+      $uri .= BSHTTP::urlencode($2);
     } else {
-      $uri = urlencode($uri);
+      $uri = BSHTTP::urlencode($uri);
     }
   }
-  if (@args) {
-    for (@args) {
-      $_ = urlencode($_);
-      s/%3D/=/;	# convert first now escaped '=' back
-    }   
-    if ($uri =~ /\?/) {
-      $uri .= '&'.join('&', @args); 
-    } else {
-      $uri .= '?'.join('&', @args); 
-    }   
-  }
+  $uri .= (($uri =~ /\?/) ? '&' : '?') . BSHTTP::queryencode(@args) if @args;
+  $uri .= "#".BSHTTP::urlencode($param->{'fragment'}) if defined $param->{'fragment'};
   return $uri;
 }
 
@@ -251,11 +240,7 @@ sub rpc {
       push @xhdrs, 'Content-Type: application/x-www-form-urlencoded';
     }
     if ($formurlencode) {
-      for (@args) {
-        $_ = urlencode($_);
-        s/%3D/=/;	# convert now escaped = back
-      }
-      $data = join('&', @args);
+      $data = BSHTTP::queryencode(@args);
       @args = ();
     }
   }

src/backend/BSServer.pm

@@ -597,17 +597,18 @@ sub readrequest {
     $qu = '';
   }
   my $query_string = '';
-  if ($path =~ /^(.*?)\?(.*)$/) {
-    $path = $1;
-    $query_string = $2;
-  }
+  my $fragment;
+  ($path, $fragment) = ($1, $2) if $path =~ /^(.*?)\#(.*)$/;
+  ($path, $query_string) = ($1, $2) if $path =~ /^(.*?)\?(.*)$/;
   $path =~ s/%([a-fA-F0-9]{2})/chr(hex($1))/ge;	# unescape path
+  $fragment =~ s/%([a-fA-F0-9]{2})/chr(hex($1))/ge if defined $fragment;	# unescape fragment
   die("501 invalid path\n") unless $path =~ /^\//s; # forbid relative paths
   my %headers;
   BSHTTP::gethead(\%headers, $rawheaders);
   $req->{'action'} = $act;
   $req->{'path'} = $path;
   $req->{'query'} = $query_string;
+  $req->{'fragment'} = $fragment if defined $fragment;
   $req->{'headers'} = \%headers;
   $req->{'rawheaders'} = $rawheaders;
   if ($act eq 'POST' || $act eq 'PUT' || $act eq 'PATCH') {

src/backend/BSServerEvents.pm

@@ -327,13 +327,14 @@ sub getrequest {
       die("501 Bad method, must be GET\n") if $act ne 'GET';
     }
     my $query_string = '';
-    if ($path =~ /^(.*?)\?(.*)$/) {
-      $path = $1;
-      $query_string = $2;
-    }
+    my $fragment;
+    ($path, $fragment) = ($1, $2) if $path =~ /^(.*?)\#(.*)$/;
+    ($path, $query_string) = ($1, $2) if $path =~ /^(.*?)\?(.*)$/;
     $path =~ s/%([a-fA-F0-9]{2})/chr(hex($1))/ge;
+    $fragment =~ s/%([a-fA-F0-9]{2})/chr(hex($1))/ge if defined $fragment;
     die("501 invalid path\n") unless $path =~ /^\//;
     %$req = ( %$req, 'action' => $act, 'path' => $path, 'query' => $query_string, 'headers' => $headers, 'state' => 'processing' );
+    $req->{'fragment'} = $fragment if defined $fragment;
     # FIXME: should not use global
     local $BSServer::request = $req;
     my @r = $conf->{'dispatch'}->($conf, $req);

src/backend/BSUrlmapper.pm

@@ -23,7 +23,7 @@
 package BSUrlmapper;
 
 use BSConfiguration;
-use BSRPC;
+use BSHTTP;
 
 my $extrepodir = "$BSConfig::bsdir/repos";
 
@@ -101,13 +101,13 @@ sub get_downloadurl {
   # default to repodownload url
   return undef unless $BSConfig::repodownload;
   if ($extrep->[0] =~ /^\Q$BSConfig::bsdir\E\/repos\/(.*)$/) {
-    my $url = "$BSConfig::repodownload/".BSRPC::urlencode($1).'/';
+    my $url = "$BSConfig::repodownload/".BSHTTP::urlencode($1).'/';
     $url =~ s!//$!/!;
     return $url;
   }
   my $prp_ext = $prp;
   $prp_ext =~ s/:/:\//g;
-  return "$BSConfig::repodownload/".BSRPC::urlencode($prp_ext)."/";
+  return "$BSConfig::repodownload/".BSHTTP::urlencode($prp_ext)."/";
 }
 
 sub get_path_downloadurl {
@@ -124,12 +124,12 @@ sub get_path_downloadurl {
   $url = $extrep->[2] if !defined($url);
   if ((!defined($path) || !defined($url)) && $extrep->[0] =~ /^\Q$BSConfig::bsdir\E\/repos\/(.*)$/) {
     $path = $1 if !defined $path;
-    $url = "$BSConfig::repodownload/".BSRPC::urlencode($1) if $BSConfig::repodownload && !defined($url);
+    $url = "$BSConfig::repodownload/".BSHTTP::urlencode($1) if $BSConfig::repodownload && !defined($url);
   }
   if (!defined($url) && $BSConfig::repodownload)  {
     my $prp_ext = $prp;
     $prp_ext =~ s/:/:\//g;
-    $url = "$BSConfig::repodownload/".BSRPC::urlencode($prp_ext);
+    $url = "$BSConfig::repodownload/".BSHTTP::urlencode($prp_ext);
   }
   $url =~ s/\/?$/\// if defined $url;
   return ($path, $url);

src/backend/bs_repserver

@@ -2269,7 +2269,7 @@ sub putjob {
     # write jobstatus and free lock
     if (@{$kiwitree_tosign || []}) {
       my $c = '';
-      $c .= BSRPC::urlencode($_)."\n" for @$kiwitree_tosign;
+      $c .= BSHTTP::urlencode($_)."\n" for @$kiwitree_tosign;
       writestr("$dir/.kiwitree_tosign", undef, $c);
     } else {
       unlink("$dir/.kiwitree_tosign");

src/backend/bs_srcserver

@@ -42,6 +42,7 @@ use JSON::XS ();
 
 use BSConfiguration;
 use BSRPC ':https';
+use BSHTTP;
 use BSServer;
 use BSUtil;
 use BSFileDB;
@@ -5088,7 +5089,7 @@ sub published_path {
     }
     if (defined($p) && !$medium) {
       $path .= "/$p" if defined $path;
-      $url .= BSRPC::urlencode($p) if defined $url;
+      $url .= BSHTTP::urlencode($p) if defined $url;
     }
   }
 
@@ -5126,7 +5127,7 @@ sub published_path {
   }
   if ($medium) {
     $path .= "/$medium" if defined $path;
-    $url .= BSRPC::urlencode($medium) if defined $url;
+    $url .= BSHTTP::urlencode($medium) if defined $url;
   }
   $url =~ s/([^\/])\/$/$1/ if defined $url;		# compat
   $ret->{'path'} = $path if defined $path;

src/backend/bs_worker

@@ -3218,7 +3218,7 @@ sub buildkiwitree {
     my $sdir = shift @dirs;
     $sdir = '' unless defined $sdir;
     next unless -d $dir;
-    $tree{$1} = "d ". BSRPC::urlencode($1) if $sdir =~ /^(.+)\/$/;
+    $tree{$1} = "d ". BSHTTP::urlencode($1) if $sdir =~ /^(.+)\/$/;
     push @todo, $dir, $sdir;
   }
   while (@todo) {
@@ -3231,17 +3231,17 @@ sub buildkiwitree {
       if (-l $f) {
 	my $sl = readlink($f);
 	die("readlink $f: $!\n") unless defined $sl;
-	$tree{$sf} = "l ". BSRPC::urlencode($sf) . " " . BSRPC::urlencode($sl);
+	$tree{$sf} = "l ". BSHTTP::urlencode($sf) . " " . BSHTTP::urlencode($sl);
       } elsif (-d $f) {
-	$tree{$sf} = "d ". BSRPC::urlencode($sf);
+	$tree{$sf} = "d ". BSHTTP::urlencode($sf);
 	push @todo, $f, "$sf/";
       } elsif (-f $f) {
         my $leadsigmd5 = '';
 	Build::queryhdrmd5($f, \$leadsigmd5) if $f =~ /\.rpm$/;
 	if ($leadsigmd5) {
-	  $tree{$sf} = "f ". BSRPC::urlencode($sf) . " " . $leadsigmd5;
+	  $tree{$sf} = "f ". BSHTTP::urlencode($sf) . " " . $leadsigmd5;
 	} else {
-	  $tree{$sf} = "f ". BSRPC::urlencode($sf);
+	  $tree{$sf} = "f ". BSHTTP::urlencode($sf);
 	}
       }
     }