First attempt at .extract_expand() method, refs #239

sqlite_utils/db.py

@@ -1151,6 +1151,25 @@ class Table(Queryable):
         self.add_foreign_key(fk_column, table, "id")
         return self
 
+    def extract_expand(
+        self, column, expand, table=None, fk_column=None, fk_column_type=int, pk=None
+    ):
+        "Use expand function to transform values in column and extract them into a new table"
+        table = table or column
+        fk_column = fk_column or "{}_id".format(table)
+        self.add_column(fk_column, fk_column_type)
+        for row_pk, row in self.pks_and_rows_where():
+            value = row[column]
+            expanded = expand(value)
+            if isinstance(expanded, dict):
+                new_pk = self.db[table].insert(expanded, pk="id", replace=True).last_pk
+                self.update(row_pk, {fk_column: new_pk})
+        # Can drop the original column now
+        self.transform(drop=[column])
+        # And add that foreign key
+        self.add_foreign_key(fk_column, table, "id")
+        return self
+
     def create_index(self, columns, index_name=None, unique=False, if_not_exists=False):
         if index_name is None:
             index_name = "idx_{}_{}".format(

tests/test_extract.py

@@ -1,5 +1,6 @@
-from sqlite_utils.db import Index, InvalidColumns
+from sqlite_utils.db import Index, InvalidColumns, ForeignKey
 import itertools
+import json
 import pytest
 
 
@@ -171,3 +172,24 @@ def test_extract_error_on_incompatible_existing_lookup_table(fresh_db):
     fresh_db["species2"].insert({"id": 1, "common_name": 3.5})
     with pytest.raises(InvalidColumns):
         fresh_db["tree"].extract("common_name", table="species2")
+
+
+def test_extract_expand(fresh_db):
+    fresh_db["trees"].insert(
+        {"id": 1, "species": '{"id": 5, "name": "Tree 1", "common_name": "Palm"}'},
+        pk="id",
+    )
+    assert fresh_db.table_names() == ["trees"]
+    fresh_db["trees"].extract_expand(
+        "species", expand=json.loads, table="species", pk="id"
+    )
+    assert set(fresh_db.table_names()) == {"trees", "species"}
+    assert list(fresh_db["trees"].rows) == [{"id": 1, "species_id": 5}]
+    assert list(fresh_db["species"].rows) == [
+        {"id": 5, "name": "Tree 1", "common_name": "Palm"}
+    ]
+    assert fresh_db["trees"].foreign_keys == [
+        ForeignKey(
+            table="trees", column="species_id", other_table="species", other_column="id"
+        )
+    ]