On development of static analysis tools for string-embedded languages

Abstract

Some programs can produce string expressions with embedded code in other programming languages while running. This embedded code should be syntactically correct as it is typically executed by some subsystem. A program in Java language that builds and sends SQL queries to the database it works with can be considered as an example. In such scenarios, languages like SQL are called string-embedded and ones like Java -- host languages.In spite of the fact such an approach of programs building is being replaced by alternative ones, for example by ORM and LINQ, string-embedding is still used in practice. Development and reengineering of the programs with string-embedded languages is complicated because the IDE and similar tools process the code embedded in strings as host language string literals and cannot provide the functionality to work with this code. To facilitate the development process, string-embedded code highlighting, completion, navigation and static errors checking would be useful. For the purposes of reengineering, embedded code metrics computation would be helpful.Currently existing tools to string-embedded languages support only operate with one host language and a fixed set of string-embedded ones. Their functionality is often limited. Moreover, it is almost impossible or requires a substantial amount of work to add a support for both new host and string-embedded language. Attempts to extend their functionality often result in the same problem.In this paper we present the platform which can be used for relatively fast and easy building of endpoint tools that provide a support for different string-embedded languages inside different host languages. The tools built for T-SQL and arithmetic expressions language embedding in C# are demonstrated as the examples of how the platform can be used.