Add comprehensive LookML adapter fixes and kitchen sink tests#27
Add comprehensive LookML adapter fixes and kitchen sink tests#27
Conversation
…ests
Implement 4 major fixes to sidemantic's LookML adapter: (1) recursive dimension reference resolution for ${dimension_name} SQL expansions, (2) proper LookML filter syntax conversion to SQL CASE WHEN expressions, (3) derived measure support (type: number) referencing other measures, and (4) multi-hop join detection with adjacency graph building. Add 86 comprehensive integration tests with real DuckDB execution across 8-entity e-commerce model covering 36 distinct measure types and join patterns.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 3310f909e2
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| # Numeric value | ||
| filters.append(f"{{model}}.{field} = {value}") | ||
| else: | ||
| # String value | ||
| filters.append(f"{{model}}.{field} = '{value}'") |
There was a problem hiding this comment.
LookML filter values can encode comparison operators (e.g. ">1000", "<=100" as used in tests/fixtures/lookml/advanced_measures.lkml), but this conversion always emits equality, so those filters become sale_amount = '>1000' instead of sale_amount > 1000. That changes the meaning of filtered measures like large_sales_count and will typically return 0 or incorrect results for numeric columns. The conversion needs to parse operator-prefixed values rather than treating them as literal strings.
Useful? React with 👍 / 👎.
Parse operator-prefixed filter values (">1000", "<=100", ">=5", "<10", "!=0")
as proper SQL comparisons instead of string equality. Previously these were
incorrectly converted to `field = '>1000'` instead of `field > 1000`.
1 participant
Implement 4 major LookML adapter fixes: (1) recursive dimension reference resolution for ${dimension_name} SQL expansions, (2) filter syntax conversion to SQL CASE WHEN, (3) derived measure support (type: number), and (4) multi-hop join detection with adjacency graph building. Add 86 integration tests with real DuckDB execution across 8-entity e-commerce model covering 36 distinct measure types and join patterns.