From 95125e250b9e4a3023a6b769c372990530d1aa87 Mon Sep 17 00:00:00 2001
From: Helmut Grohne <helmut@subdivi.de>
Date: Tue, 22 Jul 2014 21:16:14 +0200
Subject: store filenames as blobs

To facilitate this importpkg.py decodes filenames using iso-8859-1.
While this is not the encoding used for most filenames, decoding will
never fail. Conversely, readyaml.py encodes to iso-8859-1 to undo the
above effect.

webapp.py cannot currently cope with the above change and is broken by
this commit.
---
 readyaml.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

(limited to 'readyaml.py')

diff --git a/readyaml.py b/readyaml.py
index 2ef9a3b..3e5ba87 100755
--- a/readyaml.py
+++ b/readyaml.py
@@ -45,8 +45,11 @@ def readyaml(db, stream):
             db.commit()
             return
 
+        # iso-8859-1 is used as a safe representation of binary data within
+        # utf-8.
+        filename = buffer(entry["name"].encode("iso-8859-1"))
         cur.execute("INSERT INTO content (pid, filename, size) VALUES (?, ?, ?);",
-                    (pid, entry["name"], entry["size"]))
+                    (pid, filename, entry["size"]))
         cid = cur.lastrowid
         cur.executemany("INSERT INTO hash (cid, fid, hash) VALUES (?, ?, ?);",
                         ((cid, funcmapping[func], hexhash)
-- 
cgit v1.2.3