fix(bigquery): strip whitespace from bigquery field names (#9160)

Removes extra whitespace from the generated field name.

xref #9112 but doesn't strictly solve the issue

ibis/backends/bigquery/compiler.py

@@ -673,7 +673,7 @@ def visit_HashBytes(self, op, *, arg, how):
 
     @staticmethod
     def _gen_valid_name(name: str) -> str:
-        return "_".join(_NAME_REGEX.findall(name)) or "tmp"
+        return "_".join(map(str.strip, _NAME_REGEX.findall(name))) or "tmp"
 
     def visit_CountStar(self, op, *, arg, where):
         if where is not None:

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_approx/filter-approx_median/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  approx_quantiles(IF(`t0`.`month` > 0, `t0`.`double_col`, NULL), IF(`t0`.`month` > 0, 2, NULL))[offset(1)] AS `ApproxMedian_double_col_ Greater_month_ 0`
+  approx_quantiles(IF(`t0`.`month` > 0, `t0`.`double_col`, NULL), IF(`t0`.`month` > 0, 2, NULL))[offset(1)] AS `ApproxMedian_double_col_Greater_month_0`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_approx/filter-approx_nunique/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  APPROX_COUNT_DISTINCT(IF(`t0`.`month` > 0, `t0`.`double_col`, NULL)) AS `ApproxCountDistinct_double_col_ Greater_month_ 0`
+  APPROX_COUNT_DISTINCT(IF(`t0`.`month` > 0, `t0`.`double_col`, NULL)) AS `ApproxCountDistinct_double_col_Greater_month_0`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_binary/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  CAST(`t0`.`value` AS BYTES) AS `Cast_value_ binary`
+  CAST(`t0`.`value` AS BYTES) AS `Cast_value_binary`
 FROM `t` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_bit/filter-bit_and/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  bit_and(IF(`t0`.`bigint_col` > 0, `t0`.`int_col`, NULL)) AS `BitAnd_int_col_ Greater_bigint_col_ 0`
+  bit_and(IF(`t0`.`bigint_col` > 0, `t0`.`int_col`, NULL)) AS `BitAnd_int_col_Greater_bigint_col_0`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_bit/filter-bit_or/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  bit_or(IF(`t0`.`bigint_col` > 0, `t0`.`int_col`, NULL)) AS `BitOr_int_col_ Greater_bigint_col_ 0`
+  bit_or(IF(`t0`.`bigint_col` > 0, `t0`.`int_col`, NULL)) AS `BitOr_int_col_Greater_bigint_col_0`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_bit/filter-bit_xor/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  bit_xor(IF(`t0`.`bigint_col` > 0, `t0`.`int_col`, NULL)) AS `BitXor_int_col_ Greater_bigint_col_ 0`
+  bit_xor(IF(`t0`.`bigint_col` > 0, `t0`.`int_col`, NULL)) AS `BitXor_int_col_Greater_bigint_col_0`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_bool_reducers_where_conj/out.sql

@@ -9,5 +9,5 @@ SELECT
       CAST(`t0`.`bool_col` AS INT64),
       NULL
     )
-  ) AS `Sum_bool_col_ And_Greater_month_ 6_ Less_month_ 10`
+  ) AS `Sum_bool_col_And_Greater_month_6_Less_month_10`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_bool_reducers_where_simple/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  AVG(IF(`t0`.`month` > 6, CAST(`t0`.`bool_col` AS INT64), NULL)) AS `Mean_bool_col_ Greater_month_ 6`
+  AVG(IF(`t0`.`month` > 6, CAST(`t0`.`bool_col` AS INT64), NULL)) AS `Mean_bool_col_Greater_month_6`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_cast_float_to_int/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  CAST(trunc(`t0`.`double_col`) AS INT64) AS `Cast_double_col_ int64`
+  CAST(trunc(`t0`.`double_col`) AS INT64) AS `Cast_double_col_int64`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_cov/pop/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  COVAR_POP(`t0`.`double_col`, `t0`.`double_col`) AS `Covariance_double_col_ double_col`
+  COVAR_POP(`t0`.`double_col`, `t0`.`double_col`) AS `Covariance_double_col_double_col`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_cov/sample/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  COVAR_SAMP(`t0`.`double_col`, `t0`.`double_col`) AS `Covariance_double_col_ double_col`
+  COVAR_SAMP(`t0`.`double_col`, `t0`.`double_col`) AS `Covariance_double_col_double_col`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_day_of_week/date/index.sql

@@ -1,2 +1,2 @@
 SELECT
-  MOD(EXTRACT(dayofweek FROM DATE(2017, 1, 1)) + 5, 7) AS `DayOfWeekIndex_datetime_date_2017_ 1_ 1`
+  MOD(EXTRACT(dayofweek FROM DATE(2017, 1, 1)) + 5, 7) AS `DayOfWeekIndex_datetime_date_2017_1_1`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_day_of_week/date/name.sql

@@ -1,2 +1,2 @@
 SELECT
-  INITCAP(CAST(DATE(2017, 1, 1) AS STRING FORMAT 'DAY')) AS `DayOfWeekName_datetime_date_2017_ 1_ 1`
+  INITCAP(CAST(DATE(2017, 1, 1) AS STRING FORMAT 'DAY')) AS `DayOfWeekName_datetime_date_2017_1_1`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_day_of_week/datetime/index.sql

@@ -1,2 +1,2 @@
 SELECT
-  MOD(EXTRACT(dayofweek FROM datetime('2017-01-01T04:55:59')) + 5, 7) AS `DayOfWeekIndex_datetime_datetime_2017_ 1_ 1_ 4_ 55_ 59`
+  MOD(EXTRACT(dayofweek FROM datetime('2017-01-01T04:55:59')) + 5, 7) AS `DayOfWeekIndex_datetime_datetime_2017_1_1_4_55_59`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_day_of_week/datetime/name.sql

@@ -1,2 +1,2 @@
 SELECT
-  INITCAP(CAST(datetime('2017-01-01T04:55:59') AS STRING FORMAT 'DAY')) AS `DayOfWeekName_datetime_datetime_2017_ 1_ 1_ 4_ 55_ 59`
+  INITCAP(CAST(datetime('2017-01-01T04:55:59') AS STRING FORMAT 'DAY')) AS `DayOfWeekName_datetime_datetime_2017_1_1_4_55_59`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_day_of_week/string_date/index.sql

@@ -1,2 +1,2 @@
 SELECT
-  MOD(EXTRACT(dayofweek FROM DATE(2017, 1, 1)) + 5, 7) AS `DayOfWeekIndex_datetime_date_2017_ 1_ 1`
+  MOD(EXTRACT(dayofweek FROM DATE(2017, 1, 1)) + 5, 7) AS `DayOfWeekIndex_datetime_date_2017_1_1`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_day_of_week/string_date/name.sql

@@ -1,2 +1,2 @@
 SELECT
-  INITCAP(CAST(DATE(2017, 1, 1) AS STRING FORMAT 'DAY')) AS `DayOfWeekName_datetime_date_2017_ 1_ 1`
+  INITCAP(CAST(DATE(2017, 1, 1) AS STRING FORMAT 'DAY')) AS `DayOfWeekName_datetime_date_2017_1_1`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_day_of_week/string_timestamp/index.sql

@@ -1,2 +1,2 @@
 SELECT
-  MOD(EXTRACT(dayofweek FROM datetime('2017-01-01T04:55:59')) + 5, 7) AS `DayOfWeekIndex_datetime_datetime_2017_ 1_ 1_ 4_ 55_ 59`
+  MOD(EXTRACT(dayofweek FROM datetime('2017-01-01T04:55:59')) + 5, 7) AS `DayOfWeekIndex_datetime_datetime_2017_1_1_4_55_59`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_day_of_week/string_timestamp/name.sql

@@ -1,2 +1,2 @@
 SELECT
-  INITCAP(CAST(datetime('2017-01-01T04:55:59') AS STRING FORMAT 'DAY')) AS `DayOfWeekName_datetime_datetime_2017_ 1_ 1_ 4_ 55_ 59`
+  INITCAP(CAST(datetime('2017-01-01T04:55:59') AS STRING FORMAT 'DAY')) AS `DayOfWeekName_datetime_datetime_2017_1_1_4_55_59`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_day_of_week/timestamp/index.sql

@@ -1,2 +1,2 @@
 SELECT
-  MOD(EXTRACT(dayofweek FROM datetime('2017-01-01T04:55:59')) + 5, 7) AS `DayOfWeekIndex_datetime_datetime_2017_ 1_ 1_ 4_ 55_ 59`
+  MOD(EXTRACT(dayofweek FROM datetime('2017-01-01T04:55:59')) + 5, 7) AS `DayOfWeekIndex_datetime_datetime_2017_1_1_4_55_59`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_day_of_week/timestamp/name.sql

@@ -1,2 +1,2 @@
 SELECT
-  INITCAP(CAST(datetime('2017-01-01T04:55:59') AS STRING FORMAT 'DAY')) AS `DayOfWeekName_datetime_datetime_2017_ 1_ 1_ 4_ 55_ 59`
+  INITCAP(CAST(datetime('2017-01-01T04:55:59') AS STRING FORMAT 'DAY')) AS `DayOfWeekName_datetime_datetime_2017_1_1_4_55_59`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_day_of_week/timestamp_date/index.sql

@@ -1,2 +1,2 @@
 SELECT
-  MOD(EXTRACT(dayofweek FROM DATE(2017, 1, 1)) + 5, 7) AS `DayOfWeekIndex_datetime_date_2017_ 1_ 1`
+  MOD(EXTRACT(dayofweek FROM DATE(2017, 1, 1)) + 5, 7) AS `DayOfWeekIndex_datetime_date_2017_1_1`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_day_of_week/timestamp_date/name.sql

@@ -1,2 +1,2 @@
 SELECT
-  INITCAP(CAST(DATE(2017, 1, 1) AS STRING FORMAT 'DAY')) AS `DayOfWeekName_datetime_date_2017_ 1_ 1`
+  INITCAP(CAST(DATE(2017, 1, 1) AS STRING FORMAT 'DAY')) AS `DayOfWeekName_datetime_date_2017_1_1`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_divide_by_zero/floordiv/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  CAST(FLOOR(ieee_divide(`t0`.`double_col`, 0)) AS INT64) AS `FloorDivide_double_col_ 0`
+  CAST(FLOOR(ieee_divide(`t0`.`double_col`, 0)) AS INT64) AS `FloorDivide_double_col_0`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_divide_by_zero/truediv/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  ieee_divide(`t0`.`double_col`, 0) AS `Divide_double_col_ 0`
+  ieee_divide(`t0`.`double_col`, 0) AS `Divide_double_col_0`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_literal_year/date/out.sql

@@ -1,2 +1,2 @@
 SELECT
-  EXTRACT(year FROM DATE(2017, 1, 1)) AS `ExtractYear_datetime_date_2017_ 1_ 1`
+  EXTRACT(year FROM DATE(2017, 1, 1)) AS `ExtractYear_datetime_date_2017_1_1`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_literal_year/datetime/out.sql

@@ -1,2 +1,2 @@
 SELECT
-  EXTRACT(year FROM datetime('2017-01-01T04:55:59')) AS `ExtractYear_datetime_datetime_2017_ 1_ 1_ 4_ 55_ 59`
+  EXTRACT(year FROM datetime('2017-01-01T04:55:59')) AS `ExtractYear_datetime_datetime_2017_1_1_4_55_59`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_literal_year/string_date/out.sql

@@ -1,2 +1,2 @@
 SELECT
-  EXTRACT(year FROM DATE(2017, 1, 1)) AS `ExtractYear_datetime_date_2017_ 1_ 1`
+  EXTRACT(year FROM DATE(2017, 1, 1)) AS `ExtractYear_datetime_date_2017_1_1`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_literal_year/string_timestamp/out.sql

@@ -1,2 +1,2 @@
 SELECT
-  EXTRACT(year FROM datetime('2017-01-01T04:55:59')) AS `ExtractYear_datetime_datetime_2017_ 1_ 1_ 4_ 55_ 59`
+  EXTRACT(year FROM datetime('2017-01-01T04:55:59')) AS `ExtractYear_datetime_datetime_2017_1_1_4_55_59`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_literal_year/timestamp/out.sql

@@ -1,2 +1,2 @@
 SELECT
-  EXTRACT(year FROM datetime('2017-01-01T04:55:59')) AS `ExtractYear_datetime_datetime_2017_ 1_ 1_ 4_ 55_ 59`
+  EXTRACT(year FROM datetime('2017-01-01T04:55:59')) AS `ExtractYear_datetime_datetime_2017_1_1_4_55_59`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_literal_year/timestamp_date/out.sql

@@ -1,2 +1,2 @@
 SELECT
-  EXTRACT(year FROM DATE(2017, 1, 1)) AS `ExtractYear_datetime_date_2017_ 1_ 1`
+  EXTRACT(year FROM DATE(2017, 1, 1)) AS `ExtractYear_datetime_date_2017_1_1`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_to_timestamp_no_timezone/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  parse_timestamp('%F', `t0`.`date_string_col`, 'UTC') AS `StringToTimestamp_date_string_col_ '%F'`
+  parse_timestamp('%F', `t0`.`date_string_col`, 'UTC') AS `StringToTimestamp_date_string_col_'%F'`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/snapshots/test_compiler/test_to_timestamp_timezone/out.sql

@@ -1,3 +1,3 @@
 SELECT
-  parse_timestamp('%F %Z', CONCAT(`t0`.`date_string_col`, ' America/New_York'), 'UTC') AS `StringToTimestamp_StringConcat_ '%F %Z'`
+  parse_timestamp('%F %Z', CONCAT(`t0`.`date_string_col`, ' America/New_York'), 'UTC') AS `StringToTimestamp_StringConcat_'%F %Z'`
 FROM `functional_alltypes` AS `t0`

ibis/backends/bigquery/tests/unit/test_compiler.py

@@ -13,6 +13,7 @@
 import ibis.expr.datatypes as dt
 import ibis.expr.operations as ops
 from ibis import _
+from ibis.backends.bigquery.compiler import BigQueryCompiler
 from ibis.common.annotations import ValidationError
 
 to_sql = ibis.bigquery.compile
@@ -633,3 +634,15 @@ def test_unnest(snapshot):
         ).select(level_two=lambda t: t.level_one.unnest())
     )
     snapshot.assert_match(result, "out_two_unnests.sql")
+
+
+@pytest.mark.parametrize(
+    "fieldname, expected",
+    [
+        ("TryCast(b, Float64)", "TryCast_b_Float64"),
+        ("Cast(b, Int64)", "Cast_b_Int64"),
+        ("that, is, a, lot, of, spaces", "that_is_a_lot_of_spaces"),
+    ],
+)
+def test_field_names_strip_whitespace(fieldname, expected):
+    assert BigQueryCompiler._gen_valid_name(fieldname) == expected

ibis/backends/bigquery/tests/unit/udf/snapshots/test_builtin/test_farm_fingerprint/out.sql

@@ -1,2 +1,2 @@
 SELECT
-  farm_fingerprint(CAST('48656c6c6f2c20576f726c6421' AS BYTES FORMAT 'HEX')) AS `farm_fingerprint_0_b'Hello_ World_'`
+  farm_fingerprint(CAST('48656c6c6f2c20576f726c6421' AS BYTES FORMAT 'HEX')) AS `farm_fingerprint_0_b'Hello_World_'`