1 files changed, 50 insertions, 0 deletions
diff --git a/uploader/datautils.py b/uploader/datautils.py
new file mode 100644
index 0000000..d132c42
--- /dev/null
+++ b/uploader/datautils.py
@@ -0,0 +1,50 @@
+"""Generic data utilities: Rename module."""
+import math
+import json
+import base64
+from functools import reduce
+from typing import Union, Sequence
+
+def enumerate_sequence(seq: Sequence[dict], start:int = 1) -> Sequence[dict]:
+    """Enumerate sequence beginning at 1"""
+    return tuple({**item, "sequence_number": seqno}
+                 for seqno, item in enumerate(seq, start=start))
+
+
+def order_by_family(items: tuple[dict, ...],
+                    family_key: str = "Family",
+                    order_key: str = "FamilyOrderId") -> list:
+    """Order the populations by their families."""
+    def __family_order__(item):
+        orderval = item[order_key]
+        return math.inf if orderval is None else orderval
+
+    def __order__(ordered, current):
+        _key = (__family_order__(current), current[family_key])
+        return {
+            **ordered,
+            _key: ordered.get(_key, tuple()) + (current,)
+        }
+
+    return sorted(tuple(reduce(__order__, items, {}).items()),
+                  key=lambda item: item[0][0])
+
+
+def safe_int(val: Union[str, int, float]) -> int:
+    """
+    Convert val into an integer: if val cannot be converted, return a zero.
+    """
+    try:
+        return int(val)
+    except ValueError:
+        return 0
+
+
+def base64_encode_dict(dct: dict, **kwargs) -> bytes:
+    """Base64 encode a dictionary. Takes the same keywords as `json.dumps` function."""
+    return base64.urlsafe_b64encode(json.dumps(dct, **kwargs).encode("utf-8"))
+
+
+def base64_decode_to_dict(value: str, **kwargs) -> dict:
+    """Base64 encode a dictionary. Takes the same keywords as `json.loads` function."""
+    return json.loads(base64.urlsafe_b64decode(value), **kwargs)