Metadata queries break on catalog, schema or table names containing a quote or backslash

オープン
  • 何か問題がありますか？
#1,747 コメント 0 件 リアクション 0 件 担当者 0 名 GitHub で見る

メンテナーはふだん 6 日以内に返信

@MsMansiDhruv がすでに取り組んでいます。

2026年10月29日 から。

  • #1758 @MsMansiDhruv による — オープン

評価

難易度
4/5
見積もり時間
3〜5日
初心者へのやさしさ
55/100
issue の種類
バグ
明瞭さ
明確に書かれている
活発さ
活発
技術スタック
python, sql
領域
backend, databases

調査の方向性

Start with the listed macros in relations/tags.sql, relations/components/, adapters/catalog.sql, adapters/metadata.sql, and materializations/clone/strategies.sql; compare their string literal handling with relations/components/comment.sql. Also inspect show_table_extended_sql and check_schema_exists_sql for the related LIKE pattern issue. Add unit cases for names containing both a quote and a backslash, and confirm the generated SQL escapes both correctly across the affected macros.

索引モデルが issue の本文から書いたものです。

説明

bug
Describe the bug

Several metadata macros paste relation names straight into SQL string literals without escaping them. Unity Catalog allows a ' or \ in a catalog, schema or table name when the name is backtick-quoted, for example `orders'eu`. For such a relation, the generated query is invalid:

... WHERE table_catalog = 'main' AND table_schema = 'prod' AND table_name = 'orders'eu'

The query fails to parse, so the materialization or command that issued it fails before running any DDL. Identifier quoting in these macros is fine; only the string literals are affected.

Affected macros (all build = '{{ relation.<part>|lower }}' predicates):

  • relations/tags.sql: fetch_tags_sql
  • relations/components/row_filter.sql: fetch_row_filters_sql
  • relations/components/column_mask.sql: fetch_column_masks_sql
  • relations/components/column_tags.sql: fetch_column_tags_sql
  • relations/components/constraints.sql: fetch_non_null_constraint_columns_sql, fetch_primary_key_constraints_sql, fetch_foreign_key_constraints_sql
  • adapters/catalog.sql: databricks__get_catalog_schemas_where_clause_sql, databricks__get_catalog_relations_where_clause_sql
  • adapters/metadata.sql: get_relation_last_modified_sql, get_view_description_sql, get_uc_tables_sql
  • materializations/clone/strategies.sql: clone_source_policies_sql, added in #1745, which runs on every dbt clone of a Unity Catalog table.

Related, but a different quoting problem: show_table_extended_sql and check_schema_exists_sql build LIKE '<name>' patterns from relation names. Those need both literal escaping and LIKE-pattern handling.

Steps To Reproduce
CREATE SCHEMA main.quote_probe;
CREATE TABLE main.quote_probe.`orders'eu` (id INT);

-- What fetch_row_filters_sql / clone_source_policies_sql generate today: fails to parse
SELECT 1 FROM main.information_schema.row_filters
WHERE table_catalog = 'main' AND table_schema = 'quote_probe' AND table_name = 'orders'eu';

-- Escaped: runs, and finds the filter once one is set
SELECT 1 FROM main.information_schema.row_filters
WHERE table_catalog = 'main' AND table_schema = 'quote_probe' AND table_name = 'orders\'eu';

In dbt, a model with alias: "orders'eu" hits this as soon as one of the macros above runs for it. Examples:

  • dbt clone once #1745 lands.
  • Tag or constraint diffs on a later dbt run.
  • dbt docs generate, which goes through the catalog macros.
Expected behavior

Every relation-name string literal in generated SQL is escaped, \ first and then ', the way relations/components/comment.sql already escapes comments: | replace("\\", "\\\\") | replace("'", "\\'"). A shared helper macro would keep the call sites consistent. Each fixed *_sql macro gets a unit case with a name containing ' and \.

System information

Reproduced on a Unity Catalog SQL warehouse against the 1.13.latest branch. The pattern is the same on main.

Additional context

Found while reviewing #1745. We agreed to fix all call sites together on the next .latest branch rather than in that PR.

主要言語
Python
スター
391
フォーク
231
平均マージ
9日 7時間
マージ済み PR（30日）
29

環境構築

はじめの一歩

  1. issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
  2. 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
  3. リポジトリをフォークし、ブランチを切って変更します。
  4. issue 番号を参照したプルリクエストを送ります。

databricks/dbt-databricks のほかの issue

databricks/dbt-databricks の issue をすべて見る

似ている issue

Python の issue をもっと見る

新しい issue をメールで受け取る

初心者向けの GitHub issue を短くまとめたダイジェスト。