A proprietary SQL parsing engine. 500x faster than ANTLR. A fraction of the memory.
Built from the ground up for one job: parse SQL correctly, at speed.
The usual approach
SQL is one of the hardest languages to parse correctly. Every database engine ships its own dialect. Every dialect extends the standard differently. Off-the-shelf parser generators produce subpar code, and hand-writing a parser for even one SQL dialect takes years.
The SQL Tran approach
We built Hummingbird because nothing on the market met our standard for speed, correctness, and completeness. It is the foundation everything in SQL Tran is built on.
How it works
Hummingbird is a hand-written parser generator designed exclusively for SQL. It transforms raw SQL source text into a strongly-typed, richly structured syntax tree, capturing every statement, every expression, every keyword, every function, and every type in your codebase.
That syntax tree is the input for everything that follows: static analysis, dependency mapping, translation, and test generation. The quality of the parse determines the quality of every downstream step.
500x
Faster than ANTLR
0.5%
Memory footprint vs. competitors
What makes it different
Not a general-purpose parser adapted for SQL. Hummingbird handles every statement, every construct, every dialect-specific extension natively.
Hummingbird allocates between 0.5% and 3% of the memory consumed by competing parsers. Enterprise-scale codebases parse on modest hardware without memory pressure.
A parser that is fast but wrong is useless. Hummingbird was designed for correctness across the full surface area of every supported SQL dialect, then optimized for speed on top of that foundation.
Every supported dialect is parsed in its entirety: all statements, all constructs, all keywords, all types, all built-in functions. No shortcuts. No partial grammars.
Key benefits
Start a free assessment and get a complete picture of your migration project: translation statistics, success percentage, and every dependency mapped.