From 7ee74cd1dd28ee902582e71cf8d3ebddd06c0599 Mon Sep 17 00:00:00 2001
From: Arne Georg Gleditsch <argggh@pelargir.(none)>
Date: Tue, 8 Jul 2008 15:24:07 +0200
Subject: Better support for whitespace in filenames.

---
 lib/LXRng/Context.pm            | 1 +
 lib/LXRng/Repo/Git.pm           | 2 +-
 lib/LXRng/Repo/Git/Directory.pm | 2 +-
 lib/LXRng/Repo/Git/Iterator.pm  | 2 +-
 lib/LXRng/Web.pm                | 5 +++++
 lxr-genxref                     | 9 +++++++++
 6 files changed, 18 insertions(+), 3 deletions(-)

diff --git a/lib/LXRng/Context.pm b/lib/LXRng/Context.pm
index d316ede..bf717ab 100644
--- a/lib/LXRng/Context.pm
+++ b/lib/LXRng/Context.pm
@@ -43,6 +43,7 @@ sub new {
 	my $path = $ENV{'REQUEST_URI'};
 	$path =~ s/\?.*//;
 	$path =~ s,/+,/,g;
+	$path =~ s/%([0-9a-f][0-9a-f])/hex($1) >= 32 ? pack("C", hex($1)) : '_'/ge;
 	$$self{'req_base'} = $host.$ENV{'SCRIPT_NAME'};
 
 	foreach my $p ($args{'query'}->param) {
diff --git a/lib/LXRng/Repo/Git.pm b/lib/LXRng/Repo/Git.pm
index 9e03f83..7407ca0 100644
--- a/lib/LXRng/Repo/Git.pm
+++ b/lib/LXRng/Repo/Git.pm
@@ -127,7 +127,7 @@ sub node {
     else {
 	my $git = $self->_git_cmd('ls-tree', $release, $path);
 	my ($mode, $gitpath);
-	($mode, $type, $rev, $gitpath) = split(" ", <$git>);
+	($mode, $type, $rev, $gitpath) = split(" ", <$git>, 4);
     }
 
     if ($type eq 'tree') {
diff --git a/lib/LXRng/Repo/Git/Directory.pm b/lib/LXRng/Repo/Git/Directory.pm
index 6b24214..cf46915 100644
--- a/lib/LXRng/Repo/Git/Directory.pm
+++ b/lib/LXRng/Repo/Git/Directory.pm
@@ -60,7 +60,7 @@ sub contents {
     my (@dirs, @files);
     while (<$git>) {
 	chomp;
-	my ($mode, $type, $ref, $node) = split(" ", $_);
+	my ($mode, $type, $ref, $node) = split(" ", $_, 4);
 	if ($type eq 'tree') {
 	    push(@dirs, LXRng::Repo::Git::Directory->new($$self{'repo'},
 							    $prefix.$node,
diff --git a/lib/LXRng/Repo/Git/Iterator.pm b/lib/LXRng/Repo/Git/Iterator.pm
index 702b4ca..418142c 100644
--- a/lib/LXRng/Repo/Git/Iterator.pm
+++ b/lib/LXRng/Repo/Git/Iterator.pm
@@ -28,7 +28,7 @@ sub new {
     my @refs;
     my $git = $repo->_git_cmd('ls-tree', '-r', $release);
     while (<$git>) {
-	if (/\S+\s+blob\s+(\S+)\s+(\S+)/) {
+	if (/\S+\s+blob\s+(\S+)\s+(.+)/) {
 	    push(@refs, [$2, $1]);
 	}
     }
diff --git a/lib/LXRng/Web.pm b/lib/LXRng/Web.pm
index c9acc71..df2422e 100644
--- a/lib/LXRng/Web.pm
+++ b/lib/LXRng/Web.pm
@@ -125,6 +125,11 @@ sub print_markedup_file {
 	mkpath($cfile, 0, 0777);
 	open($cache, '>', "$cfile/0");
     }
+    else {
+	# Don't chunk file transfers unless cache is enabled.
+	$focus = 1;
+    }
+
     my $handle = $node->handle();
     LXRng::Lang->init($context);
     my $lang   = LXRng::Lang->new($node);
diff --git a/lxr-genxref b/lxr-genxref
index 3173bdd..d61526f 100755
--- a/lxr-genxref
+++ b/lxr-genxref
@@ -230,10 +230,18 @@ sub reference_file($$$) {
 sub hash_file($$$) {
     my ($file, $fileid, $rels) = @_;
 
+    return 0 if $context->config->{'search_size_limit'} > 0 and
+	$file->size > $context->config->{'search_size_limit'};
+
     my $docid;
     if ($index->to_hash($fileid)) {
 	my $handle;
 	sysopen($handle, $file->phys_path, 0) || die($!);
+	unless (-T $handle) {
+	    $handle->close();
+	    return 0;
+	}
+
 	warn("--- hashing     ".$file->name." [".$file->revision."]\n");
 	my $doc = $hash->new_document($file->name);
 	while (<$handle>) {
@@ -255,6 +263,7 @@ sub hash_file($$$) {
     }
     else {
 	$docid = $index->get_hashed_document($fileid);
+	return 0 unless $docid;
 	my $doc = $hash->get_document($docid);
 	if (reference_file($file, $fileid, $doc)) {
 	    $hash->save_document($docid, $doc);
-- 
cgit v1.2.3