Merge branch 'dev' of github.com:refgenie/seqcol into dev

seqcol/__init__.py

@@ -6,4 +6,4 @@
 
 
 __classes__ = ["SeqColClient"]
-__all__ = (__classes__ + ["build_sorted_name_length_pairs", "compare", "validate_seqcol"],)
+__all__ = (__classes__ + ["build_sorted_name_length_pairs", "compare", "validate_seqcol", "fasta_to_digest"],)

seqcol/seqcol.py

@@ -19,6 +19,11 @@ def explain_flag(flag):
         if flag & 2**e:
             print(FLAGS[2**e])
 
+def fasta_to_digest(fa_file_path: str) -> str:
+    """Given a fasta, return a digest"""
+    seqcol_obj = fasta_to_seqcol(fa_file_path)
+    return seqcol_digest(seqcol_obj)
+
 
 def parse_fasta(fa_file) -> pyfaidx.Fasta:
     """

seqcol/seqcol_client.py

@@ -119,7 +119,7 @@ def compare_digests(self, digestA, digestB):
         B = self.retrieve(digestB, reclimit=1)
         # _LOGGER.info(A)
         # _LOGGER.info(B)
-        return compare(A, B)
+        return compare_seqcols(A, B)
 
     def retrieve(self, druid, reclimit=None, raw=False):
         try:

seqcol/utilities.py

@@ -56,3 +56,17 @@ def validate_seqcol(seqcol_obj: SeqCol, schema=None) -> Optional[dict]:
         errors = sorted(validator.iter_errors(seqcol_obj), key=lambda e: e.path)
         raise InvalidSeqColError("Validation failed", errors)
     return True
+
+def format_itemwise(csc: SeqCol) -> list:
+    """ 
+    Format a SeqCol object into a list of dicts, one per sequence.
+    """
+    list_of_dicts = []
+    for i in range(len(csc["lengths"])):
+        list_of_dicts.append(
+            {
+                "name": csc["names"][i],
+                "length": csc["lengths"][i],
+                "sequence": csc["sequences"][i],
+            })
+    return list_of_dicts