refactor, validate, and utility functions

seqcol/__init__.py

@@ -2,18 +2,12 @@
 
 import logging
 
-from ._version import __version__
-from .seqcol import SeqColClient
-from .seqcol import (
-    parse_fasta,
-    explain_flag,
-    fasta_to_seqcol,
-    build_sorted_name_length_pairs,
-)
-
 from .const import *
+from .seqcol import *
+from .seqcol_client import *
 from .utilities import *
+from ._version import __version__
 
 
 __classes__ = ["SeqColClient"]
-__all__ = __classes__ + ["build_sorted_name_length_pairs"]
+__all__ = (__classes__ + ["build_sorted_name_length_pairs", "compare", "validate_seqcol"],)

seqcol/const.py

@@ -46,3 +46,6 @@ def _schema_path(name):
 SCHEMA_NAMES = [ASL_NAME + ".yaml"]
 SCHEMA_FILEPATH = os.path.join(os.path.dirname(__file__), "schemas")
 INTERNAL_SCHEMAS = [_schema_path(s) for s in SCHEMA_NAMES]
+
+# Alias dict to make typehinting clearer
+SeqCol = dict

seqcol/exceptions.py

@@ -0,0 +1,10 @@
+class InvalidSeqColError(Exception):
+    """Object was not validated successfully according to schema."""
+
+    def __init__(self, message, errors):
+        super().__init__(message)
+        self.message = message
+        self.errors = errors
+
+    def __str__(self):
+        return f"InvalidSeqColError ({self.message}): {self.errors}"

seqcol/schemas/seqcol.yaml

@@ -21,6 +21,7 @@ properties:
       description: "Digests of sequences computed using the GA4GH digest algorithm (sha512t24u)."
 required:
   - lengths
+  - names
 inherent:
   - lengths
   - names