[PATCH] mid: compress Message-IDs with '%' in them

[PATCH] mid: compress Message-IDs with '%' in them

[Eric Wong]

Some HTTP servers (apache2 2.2.22-13+deb7u5) on my system
apparently do not handle "%25" correctly.  I'm not yet sure if
it's something weird with my rewrite rules or what....
---
 lib/PublicInbox/MID.pm | 8 ++++++++
 t/view.t               | 9 +++++++++
 2 files changed, 17 insertions(+)

diff --git a/lib/PublicInbox/MID.pm b/lib/PublicInbox/MID.pm
index d097011..c75aa0e 100644
--- a/lib/PublicInbox/MID.pm
+++ b/lib/PublicInbox/MID.pm
@@ -20,6 +20,14 @@ sub mid_clean {
 # this is idempotent
 sub mid_compressed {
 	my ($mid) = @_;
+
+	# XXX dirty hack! FIXME!
+	# Some HTTP servers (apache2 2.2.22-13+deb7u5 on my system)
+	# apparently do not handle "%25" in the URL path component correctly.
+	# I'm not yet sure if it's something weird with my rewrite rules
+	# or what; will need to debug...
+	return sha1_hex($mid) if (index($mid, '%') >= 0);
+
 	return $mid if (length($mid) <= MID_MAX);
 	sha1_hex($mid);
 }
diff --git a/t/view.t b/t/view.t
index 3107285..463fc07 100644
--- a/t/view.t
+++ b/t/view.t
@@ -144,4 +144,13 @@ EOF
 	like($html, qr/\bhi = bye\b/, "HTML output decoded QP");
 }
 
+
+{	# XXX dirty hack
+	use PublicInbox::MID qw/mid_compressed/;
+	like(mid_compressed('foo%bar@wtf'), qr/\A[a-f0-9]{40}\z/,
+		"percent always converted to sha1 to workaround buggy httpds");
+	is(mid_compressed('foobar@wtf'), 'foobar@wtf',
+		'regular MID not compressed');
+}
+
 done_testing();
-- 
EW

[PATCH] search: bump schema version for '%' compression change

[Eric Wong]

commit 0fea7793b22efd2596983283947ee43687e0cfac
("mid: compress Message-IDs with '%' in them")
requires re-indexing of repositories with '%' in Message-IDs :<
---
 lib/PublicInbox/Search.pm | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/lib/PublicInbox/Search.pm b/lib/PublicInbox/Search.pm
index 862ed6d..2608a58 100644
--- a/lib/PublicInbox/Search.pm
+++ b/lib/PublicInbox/Search.pm
@@ -16,7 +16,8 @@ use constant {
 	# 0 - initial
 	# 1 - subject_path is lower-cased
 	# 2 - subject_path is mid_compressed in the index, only
-	SCHEMA_VERSION => 2,
+	# 3 - message-ID is compressed if it includes '%' (hack!)
+	SCHEMA_VERSION => 3,
 	LANG => 'english',
 	QP_FLAGS => FLAG_PHRASE|FLAG_BOOLEAN|FLAG_LOVEHATE|FLAG_WILDCARD,
 };
-- 
EW