Parsing sh/bash scripts: Get all executable/function calls out of it

Question

I'm doing some analysing of some code written in bash.
Now I must know all the "things" that are executed (or can be executed) from inside the script, without executing the script.
So basically the script must be parsed and all the occurrences of executions of programs and sh functions must be listed.
Does anyone know if there is a tool/script for this somewhere?
Thanks.
Last edited by Dieter@be (2008-11-08 15:12:07)

Answer 1

First you should know that it's impossible to get a guaranteed-correct answer without executing the script. Scripts can call programs/functions whose names are found/generated at runtime. The most common example is opening a text file using the program whose name is stored in $EDITOR.
That said, you can get an approximation by writing a parser. Parsers are hard. Smart programmers use a parser generator like bison or ANTLR to abstract away much of the complexity, but it's still sort of a pain in the butt the first time or two.
If you don't want to write a parser, a very approximate answer can be found by splitting each line by whitespace and returning the first word. Then finding other places commands can hide, like after semicolons, backticks, "$(", etc, but not when quoted. At this point, you will have written a very shoddy parser and taken longer at it than you would have if you'd bothered to learn bison or ANTLR.
Last edited by skymt (2008-11-08 15:34:36)