Fix bad timestamps (#85)

Handle bad timestamps that have `00` for the day in CDX records.

wayback/_utils.py

@@ -47,8 +47,14 @@ def parse_timestamp(time_string):
     """
     Given a Wayback-style timestamp string, return an equivalent ``datetime``.
     """
+    # Fix bad timestamps
+    timestamp_chars = list(time_string)
+    # If the timestamp has a day of "00"
+    if timestamp_chars[6:8] == ['0', '0']:
+        del timestamp_chars[6:8]
+        timestamp_chars.extend(['0', '0'])
     return (datetime
-            .strptime(time_string, URL_DATE_FORMAT)
+            .strptime(''.join(timestamp_chars), URL_DATE_FORMAT)
             .replace(tzinfo=timezone.utc))
 
 

wayback/tests/test_client.py

@@ -218,6 +218,32 @@ def test_search_handles_no_length_cdx_records(requests_mock):
         assert record_list[-1].length is None
 
 
+def test_search_handles_bad_timestamp_cdx_records(requests_mock):
+    """
+    The CDX index can contain a timestamp with an invalid day "00", which can't be
+    parsed into an timestamp. We should handle this.
+
+    Because these are rare and hard to get all in a single CDX query that isn't
+    *huge*, we use a made-up mock for this one instead of a VCR recording.
+    """
+    with open(Path(__file__).parent / 'test_files' / 'bad_timestamp_cdx.txt') as f:
+        bad_cdx_data = f.read()
+
+    with WaybackClient() as client:
+        requests_mock.get('http://web.archive.org/cdx/search/cdx'
+                          '?url=www.usatoday.com%2F%2A'
+                          '&matchType=domain&filter=statuscode%3A200'
+                          '&showResumeKey=true&resolveRevisits=true',
+                          [{'status_code': 200, 'text': bad_cdx_data}])
+        records = client.search('www.usatoday.com/*',
+                                matchType="domain",
+                                filter_field="statuscode:200")
+
+        record_list = list(records)
+        assert 5 == len(record_list)
+        assert record_list[-1].timestamp.day == 24
+
+
 @ia_vcr.use_cassette()
 def test_get_memento():
     with WaybackClient() as client:

wayback/tests/test_files/bad_timestamp_cdx.txt

@@ -0,0 +1,5 @@
+com,usatoday)/2000/century/tech/003d.htm 20011120210446 http://www.usatoday.com:80/2000/century/tech/003d.htm text/html 200 EJTUZEVOPPFGLXXQK2KV4DPFRSOULYVN 3823
+com,usatoday)/2000/century/tech/004.htm 20000125210430 http://www.usatoday.com:80/2000/century/tech/004.htm text/html 200 EBWZW6DNCJK2PU2DYX2JX2SWD6NQMUXK 6822
+com,usatoday)/2000/century/tech/004.htm 20000311052312 http://www.usatoday.com:80/2000/century/tech/004.htm text/html 200 BTVE5SD57GD4HZHWISTWPLXRH7XONXW6 6214
+com,usatoday)/2000/century/tech/004.htm 20000613174049 http://www.usatoday.com:80/2000/century/tech/004.htm text/html 200 RT4WYWDBYOFDEIJ2ZI2HD5UMT7UH7LRC 6566
+com,usatoday)/2000/century/tech/004.htm 20000800241623 http://www.usatoday.com:80/2000/century/tech/004.htm text/html 200 PAJWSPCRQMVBTYWV4NPJPNDQHKWJC3OO 6177