Abstract <p>This paper introduces ‘‘Regular Table Language’’ (RTL), a noveldomain-specific language designed for extracting recordsets fromarbitrary tables presented in machine-readable documents, such asspreadsheets, rich text, web pages (HTML), etc. The foundation ofRTL rests on the hypothesis that any table can be aligned with apattern that defines its structure. Furthermore, a whole class oftables designed by one template can be specified by one pattern.RTL offers a formalized way to express such patterns in a clearand concise manner. It is also proposed the ‘‘Interpretable TableModel’’ (ITM) as an intermediary between source tables and targetrecordsets. This model extends the conventional table structuresfound in widely used formats by adding semantics that provide theautomatic inference of recordsets from tables. This semantics isrecovered through matching appropriate RTL-patterns withITM-instances generated from source tables. Both RTL and ITM havebeen implemented as core components of <span>RegTab</span>, anopen-source software library designed to streamline thedevelopment of custom applications where data sources are varioustables presented in electronic documents.</p>