indexing: declare 'record' and bail if no record was loaded, add test for empty file indexing, fixes #168

pywb/warc/archiveiterator.py

@@ -80,6 +80,7 @@ class ArchiveIterator(object):
 
         raise_invalid_gzip = False
         empty_record = False
+        record = None
 
         while True:
             try:
@@ -93,7 +94,8 @@ class ArchiveIterator(object):
             except EOFError:
                 empty_record = True
 
-            self.read_to_end(record)
+            if record:
+                self.read_to_end(record)
 
             if self.reader.decompressor:
                 # if another gzip member, continue

pywb/warc/test/test_indexing.py

@@ -329,6 +329,17 @@ com,example)/ 20140216050221 {"url": "http://example.com/", "digest": "PEWDX5GTH
 """)
 
 
+def test_cdxj_empty():
+    options = dict(cdxj=True)
+
+    buff = BytesIO()
+
+    empty = BytesIO()
+
+    write_cdx_index(buff, empty,  'empty.warc.gz', **options)
+
+    assert buff.getvalue() == b''
+
 
 
 if __name__ == "__main__":