Language Convergence Infrastructure

Overall setup. The methodology for grammar convergence has been presented in [1] and elaborated in a large case study [2]. In short, it is a method of establishing relationships between language grammars by extracting them from available grammar artifacts and transforming until they become equivalent. The relationship is represented by the transformation chain properties: its length, the type of steps it consists of, the correspondence with the properties expected a priori from documentation, etc. Grammar convergence is a complicated process that can only be automated partially and therefore requires expert knowledge to use successfully. In order to make things as simple as possible for the grammar engineer, we need a solid transformation operators suite and a powerful tool support. The former, called XBGF, where BGF stands for BNF-like Grammar Format—the XML variant that we use to store grammar knowledge, was sketched in [1] and is far too complex and voluminous to be explained in this abstract. Thus, the latter becomes the main topic of our interest. Overall tool support for grammar convergence has been developed and released as a part of SLPS, Software Language Processing Suite, slps.sf.net. It comprises several stand-alone scripts targeting comparison, transformation, benchmarking, validation, extraction—most of those scripts written in Python, Prolog and Shell. The centre that ties it all together is called LCI, or Language Convergence Infrastructure. It operates on a DSL we call LCF (F for Format) in which the input configuration must be expressed. Core convergence tools. There are three top-level tools that are used universally on all grammars: comparison, transformation and validation. Comparison tool references an external program that takes two BGF grammars and returns the verdict on their equivalence. Since the premise of grammar convergence method was to document grammar relationships, the comparator is not expected to do any sophisticated matching. Transformation tool takes a BGF grammar and an XBGF script and applies the latter on the former, resulting in a transformed BGF grammar (or an error return code). Validation is an optional tool that is asked to check the XML validity of every grammar produced in the convergence process. Convergence sources. In LCF one