SQLGlot ASTs as Portable Lineage Evidence

Migration teams do not fail because they forgot to count SQL files. They fail because the SQL files hide meaning.

SQL text hides migration risk

SQLGlot is a SQL parser, transpiler, optimizer, and engine. Its documentation exposes expression classes and APIs for parsing and working with SQL structure. That makes it useful for migration work because SQL can be inspected as a tree rather than only as text.

Lineage derived from parsed SQL is not perfect. It can miss runtime behavior, dynamic SQL, permissions, data quality assumptions, and business meaning. It is still valuable because it creates evidence before migration work reaches production.

ASTs make structure inspectable

An abstract syntax tree can show which tables are referenced, which columns are selected, which expressions transform them, and where dialect-specific behavior appears. That helps teams build a migration risk register with more discipline than search-and-replace scripts.

The important part is humility. Parser-derived evidence should feed catalog lineage and validation tests. It should not pretend to be the whole truth about the data product.

The ODI migration pattern

Open Data Infrastructure migration should preserve meaning, policy, lineage, and consumer trust while platforms change. SQLGlot can help identify SQL structure and dialect drift. Catalogs and lineage systems can connect that evidence to owners, datasets, and downstream contracts.

For adjacent context, read SQLGlot for ODI migration, SQLGlot lineage for portable data models, and dialect drift budgets.

What breaks first

• The parser identifies columns, but no one checks whether semantic meaning changed.
• Dialect rewrites pass syntax tests but fail business validation.
• Temporary migration mappings become permanent undocumented lineage.
• Teams treat parser output as authoritative instead of evidence that needs review.

Questions to ask

Ask which SQL dialects are in scope, which expressions are unsupported, and which outputs need human review. Ask how AST findings are tied to catalog metadata, tests, and migration sign-off.

Portable SQL starts with structure, but it earns trust through validation.

Sources to start with

These primary sources anchor the technical claims in this guide.

• SQLGlot documentation
• SQLGlot expressions API documentation
• SQLGlot AST primer
• OpenLineage object model documentation

The AST is not the destination. It is the first map that does not lie about structure.