tests: zipnum: better test coverage for incorrect idx or loc files, add invalid sample files zipnum-bad{.idx, .loc}, #112

pywb/cdx/test/test_zipnum.py

@@ -142,6 +142,22 @@ org,iana)/time-zones 20140126200737    zipnum    9631    166    38
 Traceback (most recent call last):
 NotFoundException: No Captures found for: http://aaa.zz/ (domain query)
 
+# Invalid .idx filesor or missing loc
+
+>>> zip_test_err(url='http://example.com/', matchType='exact')
+Traceback (most recent call last):
+Exception: No Locations Found for: foo
+
+
+>>> zip_test_err(url='http://iana.org/x', matchType='exact')
+Traceback (most recent call last):
+IOError: [Errno 2] No such file or directory: './sample_archive/invalid'
+
+
+>>> zip_test_err(url='http://example.zz/x', matchType='exact')
+Traceback (most recent call last):
+Exception: No Locations Found for: foo2
+
 """
 
 from test_cdxops import cdx_ops_test
@@ -161,6 +177,10 @@ def zip_ops_test(url, **kwargs):
     sources = test_zipnum
     cdx_ops_test(url, sources, **kwargs)
 
+def zip_test_err(url, **kwargs):
+    sources = get_test_dir() + 'zipcdx/zipnum-bad.idx'
+    cdx_ops_test(url, sources, **kwargs)
+
 
 def test_zip_prefix_load():
 

pywb/cdx/zipnum.py

@@ -290,6 +290,11 @@ class ZipNumCluster(CDXSource):
         last_exc = None
         last_traceback = None
 
+        try:
+            locations = self.loc_resolver(blocks.part, query)
+        except:
+            raise Exception('No Locations Found for: ' + blocks.part)
+
         for location in self.loc_resolver(blocks.part, query):
             try:
                 return self.load_blocks(location, blocks, ranges, query)
@@ -299,9 +304,9 @@ class ZipNumCluster(CDXSource):
                 last_traceback = sys.exc_info()[2]
 
         if last_exc:
-            raise exc, None, last_traceback
+            raise last_exc, None, last_traceback
         else:
-            raise Exception('No Locations Found for: ' + block.part)
+            raise Exception('No Locations Found for: ' + blocks.part)
 
     def load_blocks(self, location, blocks, ranges, query):
         """ Load one or more blocks of compressed cdx lines, return

sample_archive/zipcdx/zipnum-bad.idx

@@ -0,0 +1,3 @@
+com,example)/ 20140127171200	foo	0	275	1
+org,iana)/ 20140127171238	bar	275	328	2
+zz,example)/ 20140127171238	foo2	275	328	2

sample_archive/zipcdx/zipnum-bad.loc

@@ -0,0 +1,3 @@
+bar	./sample_archive/invalid
+foo2
+zipnum	./sample_archive/x-bad-path-to-ignore-x	./sample_archive/zipcdx/zipnum-sample.cdx.gz