Allow small rounding errors for bee_waggle_6

For this dataset we compute two of the features, and small rounding
errors can occur on different systems. This fix allows these
errors at least on Travis.

3 files changed, 27 insertions, 7 deletions

diff --git a/checksums.json b/checksums.json
index 11eeab9..a092d6b 100644
--- a/checksums.json
+++ b/checksums.json
@@ -3,7 +3,10 @@
 	"checksums": {
 		"apple.json": "22edb48471bd3711f7a6e15de6413643",
 		"bank.json": "5207135ea53fc6fa2a8119908da73abf",
-		"bee_waggle_6.json": "4f03feafecb3be0b069b3cb0d6b17d4f",
+		"bee_waggle_6.json": [
+			"4f03feafecb3be0b069b3cb0d6b17d4f",
+			"71311783488ee5f1122545d24c15429b"
+		],
 		"bitcoin.json": "f90ff14ed1fc0c3d47d4394d25cbce93",
 		"brent_spot.json": "79892116ef8a0aa16e2450123655b31d",
 		"businv.json": "d2ab178da17b2e659a10a102a4b9f332",
diff --git a/datasets/bee_waggle_6/get_bee_waggle_6.py b/datasets/bee_waggle_6/get_bee_waggle_6.py
index b68f378..6f80042 100644
--- a/datasets/bee_waggle_6/get_bee_waggle_6.py
+++ b/datasets/bee_waggle_6/get_bee_waggle_6.py
@@ -26,6 +26,8 @@ ZIP_URL = "https://web.archive.org/web/20191114130815if_/https://www.cc.gatech.e
 
 MD5_ZIP = "039843dc15c72fd5450eeb11c6e5599c"
 MD5_JSON = "4f03feafecb3be0b069b3cb0d6b17d4f"
+# alternative checksum for small rounding errors
+MD5_JSON_2 = "71311783488ee5f1122545d24c15429b"
 
 NAME_ZIP = "psslds.zip"
 NAME_JSON = "bee_waggle_6.json"
@@ -48,7 +50,7 @@ def check_md5sum(filename, checksum):
     return h == checksum
 
 
-def validate(checksum):
+def validate(checksum, alternative_checksum=None):
     """Decorator that validates the target file."""
 
     def validate_decorator(func):
@@ -57,10 +59,19 @@ def validate(checksum):
             target = kwargs.get("target_path", None)
             if os.path.exists(target) and check_md5sum(target, checksum):
                 return
+            if (
+                os.path.exists(target)
+                and alternative_checksum
+                and check_md5sum(target, alternative_checksum)
+            ):
+                return
             out = func(*args, **kwargs)
             if not os.path.exists(target):
                 raise FileNotFoundError("Target file expected at: %s" % target)
-            if not check_md5sum(target, checksum):
+            if not (check_md5sum(target, checksum) or (
+                alternative_checksum
+                and check_md5sum(target, alternative_checksum)
+            )):
                 raise ValidationError(target)
             return out
 
@@ -74,7 +85,7 @@ def download_zip(target_path=None):
     urlretrieve(ZIP_URL, target_path)
 
 
-@validate(MD5_JSON)
+@validate(MD5_JSON, MD5_JSON_2)
 def write_json(zip_path, target_path=None):
     with zipfile.ZipFile(zip_path) as thezip:
         with thezip.open("psslds/zips/data/sequence6/btf/ximage.btf") as fp:
@@ -138,7 +149,7 @@ def parse_args():
         choices=["collect", "clean"],
         help="Action to perform",
         default="collect",
-        nargs='?'
+        nargs="?",
     )
     return parser.parse_args()
 
diff --git a/utils/check_checksums.py b/utils/check_checksums.py
index 01dcd99..0d66400 100644
--- a/utils/check_checksums.py
+++ b/utils/check_checksums.py
@@ -73,8 +73,14 @@ def main():
         if not fname in data_files:
             raise FileNotFoundError("Missing data file: %s" % fname)
         md5 = md5sum(data_files[fname])
-        if not md5 == checksums[fname]:
-            raise ValueError(
+        if isinstance(checksums[fname], list):
+            if not md5 in checksums[fname]:
+                raise ValueError(
+                    "Checksums don't match for file: %s" % (data_files[fname])
+                    )
+        else:
+            if not md5 == checksums[fname]:
+                raise ValueError(
                 "Checksums don't match for file: %s" % (data_files[fname])
             )