[Feature] Supports non half width alphanumeric for generic test (#8203)

* Supports non half width alphanumeric for generic test * Unify conditional statements * add CHANGELOG entries * added test for Japanese * Move the fix further upstream * Remove the changes in core/dbt/task/runnable.py * Fix accidental removal of `_` substitution character --------- Co-authored-by: Doug Beatty <[email protected]>
dbt-labs · Nov 9, 2023 · f45b013 · f45b013
1 parent e547c0e
commit f45b013
Show file tree

Hide file tree

Showing 4 changed files with 61 additions and 5 deletions.
diff --git a/.changes/unreleased/Features-20230725-122847.yaml b/.changes/unreleased/Features-20230725-122847.yaml
@@ -0,0 +1,6 @@
+kind: Features
+body: Supports non half width alphanumeric for generic test
+time: 2023-07-25T12:28:47.88711+09:00
+custom:
+  Author: d-kaneshiro
+  Issue: "8204"
diff --git a/core/dbt/parser/generic_test_builders.py b/core/dbt/parser/generic_test_builders.py
@@ -56,7 +56,7 @@ def synthesize_generic_test_names(
 
         flat_args.extend([str(part) for part in parts])
 
-    clean_flat_args = [re.sub("[^0-9a-zA-Z_]+", "_", arg) for arg in flat_args]
+    clean_flat_args = [re.sub(r"\W+", "_", arg) for arg in flat_args]
     unique = "__".join(clean_flat_args)
 
     # for the file path + alias, the name must be <64 characters

diff --git a/tests/adapter/dbt/tests/adapter/basic/files.py b/tests/adapter/dbt/tests/adapter/basic/files.py
@@ -12,6 +12,20 @@
 10,Nora,1976-03-01T16:51:39
 """.lstrip()
 
+seeds_jp_base_csv = """
+id,名前,ふりがな,部署ｺｰﾄﾞ,イニシャル
+1,比嘉,ひが,10000,H
+2,金城,きんじょう,10000,K
+3,大城,おおしろ,10001,O
+4,宮城,みやぎ,10001,M
+5,新垣,あらかき,10002,A
+6,玉城,たましろ,10002,T
+7,上原,うえはら,10002,U
+8,島袋,しまぶくろ,10003,S
+9,平良,たいら,10003,T
+10,山城,やましろ,10003,Y
+""".lstrip()
+
 
 seeds_added_csv = (
     seeds_base_csv
@@ -61,6 +75,18 @@
      - name: id
        tests:
          - not_null
+     - name: 名前
+       tests:
+         - not_null
+     - name: ふりがな
+       tests:
+         - not_null
+     - name: 部署ｺｰﾄﾞ
+       tests:
+         - not_null
+     - name: イニシャル
+       tests:
+         - not_null
 """
 
 generic_test_view_yml = """
@@ -71,6 +97,18 @@
      - name: id
        tests:
          - not_null
+     - name: 名前
+       tests:
+         - not_null
+     - name: ふりがな
+       tests:
+         - not_null
+     - name: 部署ｺｰﾄﾞ
+       tests:
+         - not_null
+     - name: イニシャル
+       tests:
+         - not_null
 """
 
 generic_test_table_yml = """
@@ -81,6 +119,18 @@
      - name: id
        tests:
          - not_null
+     - name: 名前
+       tests:
+         - not_null
+     - name: ふりがな
+       tests:
+         - not_null
+     - name: 部署ｺｰﾄﾞ
+       tests:
+         - not_null
+     - name: イニシャル
+       tests:
+         - not_null
 """
 
 test_passing_sql = """

diff --git a/tests/adapter/dbt/tests/adapter/basic/test_generic_tests.py b/tests/adapter/dbt/tests/adapter/basic/test_generic_tests.py
@@ -1,7 +1,7 @@
 import pytest
 from dbt.tests.util import run_dbt
 from dbt.tests.adapter.basic.files import (
-    seeds_base_csv,
+    seeds_jp_base_csv,
     generic_test_seed_yml,
     base_view_sql,
     base_table_sql,
@@ -19,7 +19,7 @@ def project_config_update(self):
     @pytest.fixture(scope="class")
     def seeds(self):
         return {
-            "base.csv": seeds_base_csv,
+            "base.csv": seeds_jp_base_csv,
             "schema.yml": generic_test_seed_yml,
         }
 
@@ -39,15 +39,15 @@ def test_generic_tests(self, project):
 
         # test command selecting base model
         results = run_dbt(["test", "-m", "base"])
-        assert len(results) == 1
+        assert len(results) == 5
 
         # run command
         results = run_dbt(["run"])
         assert len(results) == 2
 
         # test command, all tests
         results = run_dbt(["test"])
-        assert len(results) == 3
+        assert len(results) == 15
 
 
 class TestGenericTests(BaseGenericTests):