Small additions

Author: erezsh

data_diff/diff_tables.py

@@ -18,6 +18,7 @@
 from .databases.base import Database
 from .databases.database_types import (
     ArithUUID,
+    IKey,
     NumericType,
     PrecisionType,
     StringType,
@@ -212,6 +213,8 @@ def count_and_checksum(self) -> Tuple[int, int]:
                 "We recommend increasing --bisection-factor or decreasing --threads."
             )
 
+        if count:
+            assert checksum, (count, checksum)
         return count or 0, checksum if checksum is None else int(checksum)
 
     def query_key_range(self) -> Tuple[int, int]:
@@ -306,6 +309,10 @@ def diff_tables(self, table1: TableSegment, table2: TableSegment) -> DiffResult:
 
         key_type = table1._schema[table1.key_column]
         key_type2 = table2._schema[table2.key_column]
+        if not isinstance(key_type, IKey):
+            raise NotImplementedError(f"Cannot use column of type {key_type} as a key")
+        if not isinstance(key_type2, IKey):
+            raise NotImplementedError(f"Cannot use column of type {key_type2} as a key")
         assert key_type.python_type is key_type2.python_type
 
         # We add 1 because our ranges are exclusive of the end (like in Python)

data_diff/sql.py

@@ -121,9 +121,10 @@ class Checksum(Sql):
 
     def compile(self, c: Compiler):
         if len(self.exprs) > 1:
-            compiled_exprs = ", ".join(f"coalesce({c.compile(expr)}, 'null')" for expr in self.exprs)
+            compiled_exprs = ", ".join(f"coalesce({c.compile(expr)}, '<null>')" for expr in self.exprs)
             expr = f"concat({compiled_exprs})"
         else:
+            # No need to coalesce - safe to assume that key cannot be null
             (expr,) = self.exprs
             expr = c.compile(expr)
         md5 = c.database.md5_to_int(expr)