如何在数据库中存储“元”源代码

Question

I would like to store a computer program in a database instead of a number of text files. It should contain the structure and all objects, methods, dependencies etc. of the program. I do not want to store a specific language in the database but some kind of "meta" programming language. In a second step I would like to transform/export this structure in the database into either source code of a classic language (C#, Java, etc.) or compile it directly for CLR/JVM.

I think I am not the first person with this idea. I searched the internet and I think what I am looking for is called "source code in a database (SCID)" - unfortunately I could not find an implementation of this idea.

So my questions is:

Is there any program that stores "meta" program code inside of a database and let's you generate traditional text source code from it that can be compiled/executed?

Short remarks:
- It can also be a noSQL database
- I currently don't care how the program is imported/entered into the database

Answer 1

It sounds like you're looking for some kind of common markup language that adequately describes the common semantics of each target language - eg objects, functions, inputs, return values, etc.

This is less about storing in a database, and more about having a single (I imagine, XML-like) structure that can subsequently be parsed and eval'd by the target language to produce native source/bytecode. If there was such a thing, storing it in a database would be trivial -- that's not the hard part. Even a key/value database could handle that.

The hard part will be finding something that can abstract away the nuances of multiple languages and attempt to describe them in a common format.

Similar questions have already been asked, without satisfying solutions.

It may be that you don't need the full source, but instead just a description of the runtime data-- formats like XML and JSON are intended exactly for this purpose and provide a simplified description of Objects that can be parsed and mapped to native equivalents, with your source code built around the dynamic parsing of that data.

It may be possible to go further in certain languages. For example, if your language of choice converts to bytecode first, you might technically be able to store the binary bytecode in a BLOB and then run it directly. Languages that offer reflection and dynamic evaluation can probably handle this -- then your DB is simply a wrapper for storing that data on compilation, and retrieving it prior to running it. That'd depend on your target language and how compilation is handled.

Of course, if you're only working with interpreted languages, you can simply store the full source and eval it (in whatever manner is preferred by the target language).

If you give more info on your intended use case, I'm sure you'll get some decent suggestions on how to handle it without having to invent a sourcecode Babelfish.