Improved column type introspection, closes #92

sqlite_utils/db.py

@@ -1,4 +1,4 @@
-from .utils import sqlite3, OperationalError, suggest_column_types
+from .utils import sqlite3, OperationalError, suggest_column_types, column_affinity
 from collections import namedtuple, OrderedDict
 import datetime
 import hashlib
@@ -65,15 +65,6 @@ if np:
     )
 
 
-REVERSE_COLUMN_TYPE_MAPPING = {
-    "": str,  # Columns in views sometimes have type = ''
-    "TEXT": str,
-    "BLOB": bytes,
-    "INTEGER": int,
-    "FLOAT": float,
-}
-
-
 class AlterError(Exception):
     pass
 
@@ -457,10 +448,7 @@ class Queryable:
     @property
     def columns_dict(self):
         "Returns {column: python-type} dictionary"
-        return {
-            column.name: REVERSE_COLUMN_TYPE_MAPPING[column.type]
-            for column in self.columns
-        }
+        return {column.name: column_affinity(column.type) for column in self.columns}
 
     @property
     def schema(self):

sqlite_utils/utils.py

@@ -33,3 +33,22 @@ def suggest_column_types(records):
             t = str
         column_types[key] = t
     return column_types
+
+
+def column_affinity(column_type):
+    # Implementation of SQLite affinity rules from
+    # https://www.sqlite.org/datatype3.html#determination_of_column_affinity
+    assert isinstance(column_type, str)
+    column_type = column_type.upper().strip()
+    if column_type == "":
+        return str  # We differ from spec, which says it should be BLOB
+    if "INT" in column_type:
+        return int
+    if "CHAR" in column_type or "CLOB" in column_type or "TEXT" in column_type:
+        return str
+    if "BLOB" in column_type:
+        return bytes
+    if "REAL" in column_type or "FLOA" in column_type or "DOUB" in column_type:
+        return float
+    # Default is 'NUMERIC', which we currently also treat as float
+    return float