From 46b93836752a6fb221df44111de43c435174926d Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Tue, 23 Apr 2019 17:19:29 -0700
Subject: ignore empty files on backwards import

---
 arabesque.py | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/arabesque.py b/arabesque.py
index 5b8e209..1c5663b 100755
--- a/arabesque.py
+++ b/arabesque.py
@@ -416,7 +416,11 @@ def backward(log_file, map_db, output_db, hit_mimetypes=FULLTEXT_MIMETYPES):
             continue
 
         if line.mimetype == "application/octet-stream" and int(line.size_bytes) < 1000:
-            counts['skip-tiny-octetstream-'] += 1
+            counts['skip-tiny-octetstream'] += 1
+            continue
+
+        if int(line.size_bytes) == 0 or line.sha1 == "3I42H3S6NNFQ2MSVX7XZKYAYSCX5QBYJ":
+            counts['skip-empty-file'] += 1
             continue
 
         #print(time.time())
-- 
cgit v1.2.3