schema: use a more flexible hash schema (iterative#3333)

Fixes iterative#3331

dvc/output/__init__.py

@@ -1,5 +1,5 @@
 from urllib.parse import urlparse
-from voluptuous import Any, Required
+from voluptuous import Any, Required, Lower, Length, Coerce, And, SetTo
 
 from dvc.output.base import OutputBase
 from dvc.output.gs import OutputGS
@@ -29,6 +29,12 @@
     Schemes.LOCAL: OutputLOCAL,
 }
 
+CHECKSUM_SCHEMA = Any(
+    None,
+    And(str, Length(max=0), SetTo(None)),
+    And(Any(str, And(int, Coerce(str))), Length(min=3), Lower),
+)
+
 # NOTE: currently there are only 3 possible checksum names:
 #
 #    1) md5 (LOCAL, SSH, GS);
@@ -37,15 +43,15 @@
 #
 # so when a few types of outputs share the same name, we only need
 # specify it once.
-CHECKSUM_SCHEMA = {
-    RemoteLOCAL.PARAM_CHECKSUM: Any(str, None),
-    RemoteS3.PARAM_CHECKSUM: Any(str, None),
-    RemoteHDFS.PARAM_CHECKSUM: Any(str, None),
+CHECKSUMS_SCHEMA = {
+    RemoteLOCAL.PARAM_CHECKSUM: CHECKSUM_SCHEMA,
+    RemoteS3.PARAM_CHECKSUM: CHECKSUM_SCHEMA,
+    RemoteHDFS.PARAM_CHECKSUM: CHECKSUM_SCHEMA,
 }
 
-TAGS_SCHEMA = {str: CHECKSUM_SCHEMA}
+TAGS_SCHEMA = {str: CHECKSUMS_SCHEMA}
 
-SCHEMA = CHECKSUM_SCHEMA.copy()
+SCHEMA = CHECKSUMS_SCHEMA.copy()
 SCHEMA[Required(OutputBase.PARAM_PATH)] = str
 SCHEMA[OutputBase.PARAM_CACHE] = bool
 SCHEMA[OutputBase.PARAM_METRIC] = OutputBase.METRIC_SCHEMA

dvc/stage.py

@@ -191,7 +191,7 @@ class Stage(object):
     PARAM_ALWAYS_CHANGED = "always_changed"
 
     SCHEMA = {
-        PARAM_MD5: Any(str, None),
+        PARAM_MD5: output.CHECKSUM_SCHEMA,
         PARAM_CMD: Any(str, None),
         PARAM_WDIR: Any(str, None),
         PARAM_DEPS: Any([dependency.SCHEMA], None),

tests/unit/output/test_output.py

@@ -0,0 +1,42 @@
+import pytest
+
+from voluptuous import Schema, MultipleInvalid
+
+from dvc.output import CHECKSUM_SCHEMA
+
+
+@pytest.mark.parametrize(
+    "value,expected",
+    [
+        ("", None),
+        (None, None),
+        (11111, "11111"),
+        ("11111", "11111"),
+        ("aAaBa", "aaaba"),
+        (
+            "3cc286c534a71504476da009ed174423",
+            "3cc286c534a71504476da009ed174423",
+        ),  # md5
+        (
+            "d41d8cd98f00b204e9800998ecf8427e-38",
+            "d41d8cd98f00b204e9800998ecf8427e-38",
+        ),  # etag
+        (
+            "000002000000000000000000c16859d1d071c6b1ffc9c8557d4909f1",
+            "000002000000000000000000c16859d1d071c6b1ffc9c8557d4909f1",
+        ),  # hdfs checksum
+        # Not much we can do about hex and oct values without writing our own
+        # parser. So listing these test cases just to acknowledge this.
+        # See https://github.com/iterative/dvc/issues/3331.
+        (0x3451, "13393"),
+        (0o1244, "676"),
+    ],
+)
+def test_checksum_schema(value, expected):
+    assert Schema(CHECKSUM_SCHEMA)(value) == expected
+
+
+@pytest.mark.parametrize("value", ["1", "11", {}, {"a": "b"}, [], [1, 2]])
+def test_checksum_schema_fail(value):
+    with pytest.raises(MultipleInvalid):
+        Schema(CHECKSUM_SCHEMA)(value)["md5"]