Skip to content

Commit 55f376f

Browse files
committed
Fix(table_diff): Widen the data types that --decimals applies to
1 parent c62cac1 commit 55f376f

File tree

2 files changed

+48
-5
lines changed

2 files changed

+48
-5
lines changed

sqlmesh/core/table_diff.py

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -367,8 +367,8 @@ def _column_expr(name: str, table: str) -> exp.Expression:
367367
column_type = matched_columns[name]
368368
qualified_column = exp.column(name, table)
369369

370-
if column_type.is_type(*exp.DataType.FLOAT_TYPES):
371-
return exp.func("ROUND", qualified_column, exp.Literal.number(self.decimals))
370+
if column_type.is_type(*exp.DataType.REAL_TYPES):
371+
return self.adapter._normalize_decimal_value(qualified_column, self.decimals)
372372
if column_type.is_type(*exp.DataType.NESTED_TYPES):
373373
return self.adapter._normalize_nested_value(qualified_column)
374374

tests/core/test_table_diff.py

Lines changed: 46 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -144,7 +144,7 @@ def test_data_diff(sushi_context_fixed_date, capsys, caplog):
144144
assert row_diff.t_sample.shape == (1, 6)
145145

146146

147-
def test_data_diff_decimals(sushi_context_fixed_date):
147+
def test_data_diff_decimals_on_float(sushi_context_fixed_date):
148148
engine_adapter = sushi_context_fixed_date.engine_adapter
149149

150150
engine_adapter.ctas(
@@ -230,6 +230,49 @@ def test_data_diff_decimals(sushi_context_fixed_date):
230230
assert stripped_output == stripped_expected
231231

232232

233+
def test_data_diff_decimals_on_numeric():
234+
engine_adapter = DuckDBConnectionConfig().create_engine_adapter()
235+
236+
columns_to_types = {
237+
"key": exp.DataType.build("int"),
238+
"value": exp.DataType.build("decimal(10,5)"),
239+
}
240+
241+
engine_adapter.create_table("src", columns_to_types)
242+
engine_adapter.create_table("target", columns_to_types)
243+
244+
src_records = [
245+
(1, "25.12344"),
246+
(2, "25.1234"),
247+
(3, "25.124"),
248+
(4, "25.14"),
249+
(5, "25.4"),
250+
]
251+
252+
target_records = [
253+
(1, "25.12343"),
254+
(2, "25.1233"),
255+
(3, "25.123"),
256+
(4, "25.13"),
257+
(5, "25.3"),
258+
]
259+
260+
src_df = pd.DataFrame(data=src_records, columns=columns_to_types.keys())
261+
target_df = pd.DataFrame(data=target_records, columns=columns_to_types.keys())
262+
263+
engine_adapter.insert_append("src", src_df)
264+
engine_adapter.insert_append("target", target_df)
265+
266+
for decimals in range(5, 0, -1):
267+
table_diff = TableDiff(
268+
adapter=engine_adapter, source="src", target="target", on=["key"], decimals=decimals
269+
)
270+
diff = table_diff.row_diff()
271+
272+
assert diff.full_match_count == 5 - decimals
273+
assert diff.partial_match_count + diff.full_match_count == 5
274+
275+
233276
def test_grain_check(sushi_context_fixed_date):
234277
expressions = d.parse(
235278
"""
@@ -344,7 +387,7 @@ def test_generated_sql(sushi_context_fixed_date: Context, mocker: MockerFixture)
344387
),
345388
)
346389

347-
query_sql = 'CREATE TABLE IF NOT EXISTS "memory"."sqlmesh_temp_test"."__temp_diff_abcdefgh" AS WITH "__source" AS (SELECT "s"."key", "s"."value", "s"."key" AS "__sqlmesh_join_key" FROM "table_diff_source" AS "s"), "__target" AS (SELECT "t"."key", "t"."value", "t"."key" AS "__sqlmesh_join_key" FROM "table_diff_target" AS "t"), "__stats" AS (SELECT "s"."key" AS "s__key", "s"."value" AS "s__value", "s"."__sqlmesh_join_key" AS "s____sqlmesh_join_key", "t"."key" AS "t__key", "t"."value" AS "t__value", "t"."__sqlmesh_join_key" AS "t____sqlmesh_join_key", CASE WHEN NOT "s"."__sqlmesh_join_key" IS NULL THEN 1 ELSE 0 END AS "s_exists", CASE WHEN NOT "t"."__sqlmesh_join_key" IS NULL THEN 1 ELSE 0 END AS "t_exists", CASE WHEN "s"."__sqlmesh_join_key" = "t"."__sqlmesh_join_key" THEN 1 ELSE 0 END AS "row_joined", CASE WHEN "s"."key" IS NULL AND "t"."key" IS NULL THEN 1 ELSE 0 END AS "null_grain", CASE WHEN "s"."key" = "t"."key" THEN 1 WHEN ("s"."key" IS NULL) AND ("t"."key" IS NULL) THEN 1 WHEN ("s"."key" IS NULL) OR ("t"."key" IS NULL) THEN 0 ELSE 0 END AS "key_matches", CASE WHEN ROUND("s"."value", 3) = ROUND("t"."value", 3) THEN 1 WHEN ("s"."value" IS NULL) AND ("t"."value" IS NULL) THEN 1 WHEN ("s"."value" IS NULL) OR ("t"."value" IS NULL) THEN 0 ELSE 0 END AS "value_matches" FROM "__source" AS "s" FULL JOIN "__target" AS "t" ON "s"."__sqlmesh_join_key" = "t"."__sqlmesh_join_key") SELECT *, CASE WHEN "key_matches" = 1 AND "value_matches" = 1 THEN 1 ELSE 0 END AS "row_full_match" FROM "__stats"'
390+
query_sql = 'CREATE TABLE IF NOT EXISTS "memory"."sqlmesh_temp_test"."__temp_diff_abcdefgh" AS WITH "__source" AS (SELECT "s"."key", "s"."value", "s"."key" AS "__sqlmesh_join_key" FROM "table_diff_source" AS "s"), "__target" AS (SELECT "t"."key", "t"."value", "t"."key" AS "__sqlmesh_join_key" FROM "table_diff_target" AS "t"), "__stats" AS (SELECT "s"."key" AS "s__key", "s"."value" AS "s__value", "s"."__sqlmesh_join_key" AS "s____sqlmesh_join_key", "t"."key" AS "t__key", "t"."value" AS "t__value", "t"."__sqlmesh_join_key" AS "t____sqlmesh_join_key", CASE WHEN NOT "s"."__sqlmesh_join_key" IS NULL THEN 1 ELSE 0 END AS "s_exists", CASE WHEN NOT "t"."__sqlmesh_join_key" IS NULL THEN 1 ELSE 0 END AS "t_exists", CASE WHEN "s"."__sqlmesh_join_key" = "t"."__sqlmesh_join_key" THEN 1 ELSE 0 END AS "row_joined", CASE WHEN "s"."key" IS NULL AND "t"."key" IS NULL THEN 1 ELSE 0 END AS "null_grain", CASE WHEN "s"."key" = "t"."key" THEN 1 WHEN ("s"."key" IS NULL) AND ("t"."key" IS NULL) THEN 1 WHEN ("s"."key" IS NULL) OR ("t"."key" IS NULL) THEN 0 ELSE 0 END AS "key_matches", CASE WHEN CAST(CAST("s"."value" AS DOUBLE) AS DECIMAL(38, 3)) = CAST(CAST("t"."value" AS DOUBLE) AS DECIMAL(38, 3)) THEN 1 WHEN ("s"."value" IS NULL) AND ("t"."value" IS NULL) THEN 1 WHEN ("s"."value" IS NULL) OR ("t"."value" IS NULL) THEN 0 ELSE 0 END AS "value_matches" FROM "__source" AS "s" FULL JOIN "__target" AS "t" ON "s"."__sqlmesh_join_key" = "t"."__sqlmesh_join_key") SELECT *, CASE WHEN "key_matches" = 1 AND "value_matches" = 1 THEN 1 ELSE 0 END AS "row_full_match" FROM "__stats"'
348391
summary_query_sql = 'SELECT SUM("s_exists") AS "s_count", SUM("t_exists") AS "t_count", SUM("row_joined") AS "join_count", SUM("null_grain") AS "null_grain_count", SUM("row_full_match") AS "full_match_count", SUM("key_matches") AS "key_matches", SUM("value_matches") AS "value_matches", COUNT(DISTINCT ("s____sqlmesh_join_key")) AS "distinct_count_s", COUNT(DISTINCT ("t____sqlmesh_join_key")) AS "distinct_count_t" FROM "memory"."sqlmesh_temp_test"."__temp_diff_abcdefgh"'
349392
compare_sql = 'SELECT ROUND(100 * (CAST(SUM("key_matches") AS DECIMAL) / COUNT("key_matches")), 9) AS "key_matches", ROUND(100 * (CAST(SUM("value_matches") AS DECIMAL) / COUNT("value_matches")), 9) AS "value_matches" FROM "memory"."sqlmesh_temp_test"."__temp_diff_abcdefgh" WHERE "row_joined" = 1'
350393
sample_query_sql = 'WITH "source_only" AS (SELECT \'source_only\' AS "__sqlmesh_sample_type", "s__key", "s__value", "s____sqlmesh_join_key", "t__key", "t__value", "t____sqlmesh_join_key" FROM "memory"."sqlmesh_temp_test"."__temp_diff_abcdefgh" WHERE "s_exists" = 1 AND "row_joined" = 0 ORDER BY "s__key" NULLS FIRST LIMIT 20), "target_only" AS (SELECT \'target_only\' AS "__sqlmesh_sample_type", "s__key", "s__value", "s____sqlmesh_join_key", "t__key", "t__value", "t____sqlmesh_join_key" FROM "memory"."sqlmesh_temp_test"."__temp_diff_abcdefgh" WHERE "t_exists" = 1 AND "row_joined" = 0 ORDER BY "t__key" NULLS FIRST LIMIT 20), "common_rows" AS (SELECT \'common_rows\' AS "__sqlmesh_sample_type", "s__key", "s__value", "s____sqlmesh_join_key", "t__key", "t__value", "t____sqlmesh_join_key" FROM "memory"."sqlmesh_temp_test"."__temp_diff_abcdefgh" WHERE "row_joined" = 1 AND "row_full_match" = 0 ORDER BY "s__key" NULLS FIRST, "t__key" NULLS FIRST LIMIT 20) SELECT "__sqlmesh_sample_type", "s__key", "s__value", "s____sqlmesh_join_key", "t__key", "t__value", "t____sqlmesh_join_key" FROM "source_only" UNION ALL SELECT "__sqlmesh_sample_type", "s__key", "s__value", "s____sqlmesh_join_key", "t__key", "t__value", "t____sqlmesh_join_key" FROM "target_only" UNION ALL SELECT "__sqlmesh_sample_type", "s__key", "s__value", "s____sqlmesh_join_key", "t__key", "t__value", "t____sqlmesh_join_key" FROM "common_rows"'
@@ -384,7 +427,7 @@ def test_generated_sql(sushi_context_fixed_date: Context, mocker: MockerFixture)
384427
where="key = 2",
385428
)
386429

387-
query_sql_where = 'CREATE TABLE IF NOT EXISTS "memory"."sqlmesh_temp"."__temp_diff_abcdefgh" AS WITH "__source" AS (SELECT "s"."key", "s"."value", "s"."key" AS "__sqlmesh_join_key" FROM "table_diff_source" AS "s" WHERE "s"."key" = 2), "__target" AS (SELECT "t"."key", "t"."value", "t"."key" AS "__sqlmesh_join_key" FROM "table_diff_target" AS "t" WHERE "t"."key" = 2), "__stats" AS (SELECT "s"."key" AS "s__key", "s"."value" AS "s__value", "s"."__sqlmesh_join_key" AS "s____sqlmesh_join_key", "t"."key" AS "t__key", "t"."value" AS "t__value", "t"."__sqlmesh_join_key" AS "t____sqlmesh_join_key", CASE WHEN NOT "s"."__sqlmesh_join_key" IS NULL THEN 1 ELSE 0 END AS "s_exists", CASE WHEN NOT "t"."__sqlmesh_join_key" IS NULL THEN 1 ELSE 0 END AS "t_exists", CASE WHEN "s"."__sqlmesh_join_key" = "t"."__sqlmesh_join_key" THEN 1 ELSE 0 END AS "row_joined", CASE WHEN "s"."key" IS NULL AND "t"."key" IS NULL THEN 1 ELSE 0 END AS "null_grain", CASE WHEN "s"."key" = "t"."key" THEN 1 WHEN ("s"."key" IS NULL) AND ("t"."key" IS NULL) THEN 1 WHEN ("s"."key" IS NULL) OR ("t"."key" IS NULL) THEN 0 ELSE 0 END AS "key_matches", CASE WHEN ROUND("s"."value", 3) = ROUND("t"."value", 3) THEN 1 WHEN ("s"."value" IS NULL) AND ("t"."value" IS NULL) THEN 1 WHEN ("s"."value" IS NULL) OR ("t"."value" IS NULL) THEN 0 ELSE 0 END AS "value_matches" FROM "__source" AS "s" FULL JOIN "__target" AS "t" ON "s"."__sqlmesh_join_key" = "t"."__sqlmesh_join_key") SELECT *, CASE WHEN "key_matches" = 1 AND "value_matches" = 1 THEN 1 ELSE 0 END AS "row_full_match" FROM "__stats"'
430+
query_sql_where = 'CREATE TABLE IF NOT EXISTS "memory"."sqlmesh_temp"."__temp_diff_abcdefgh" AS WITH "__source" AS (SELECT "s"."key", "s"."value", "s"."key" AS "__sqlmesh_join_key" FROM "table_diff_source" AS "s" WHERE "s"."key" = 2), "__target" AS (SELECT "t"."key", "t"."value", "t"."key" AS "__sqlmesh_join_key" FROM "table_diff_target" AS "t" WHERE "t"."key" = 2), "__stats" AS (SELECT "s"."key" AS "s__key", "s"."value" AS "s__value", "s"."__sqlmesh_join_key" AS "s____sqlmesh_join_key", "t"."key" AS "t__key", "t"."value" AS "t__value", "t"."__sqlmesh_join_key" AS "t____sqlmesh_join_key", CASE WHEN NOT "s"."__sqlmesh_join_key" IS NULL THEN 1 ELSE 0 END AS "s_exists", CASE WHEN NOT "t"."__sqlmesh_join_key" IS NULL THEN 1 ELSE 0 END AS "t_exists", CASE WHEN "s"."__sqlmesh_join_key" = "t"."__sqlmesh_join_key" THEN 1 ELSE 0 END AS "row_joined", CASE WHEN "s"."key" IS NULL AND "t"."key" IS NULL THEN 1 ELSE 0 END AS "null_grain", CASE WHEN "s"."key" = "t"."key" THEN 1 WHEN ("s"."key" IS NULL) AND ("t"."key" IS NULL) THEN 1 WHEN ("s"."key" IS NULL) OR ("t"."key" IS NULL) THEN 0 ELSE 0 END AS "key_matches", CASE WHEN CAST(CAST("s"."value" AS DOUBLE) AS DECIMAL(38, 3)) = CAST(CAST("t"."value" AS DOUBLE) AS DECIMAL(38, 3)) THEN 1 WHEN ("s"."value" IS NULL) AND ("t"."value" IS NULL) THEN 1 WHEN ("s"."value" IS NULL) OR ("t"."value" IS NULL) THEN 0 ELSE 0 END AS "value_matches" FROM "__source" AS "s" FULL JOIN "__target" AS "t" ON "s"."__sqlmesh_join_key" = "t"."__sqlmesh_join_key") SELECT *, CASE WHEN "key_matches" = 1 AND "value_matches" = 1 THEN 1 ELSE 0 END AS "row_full_match" FROM "__stats"'
388431
spy_execute.assert_any_call(query_sql_where, False)
389432

390433

0 commit comments

Comments
 (0)