Does anyone know where I can get the BNF or EBNF for the LOGO programming language?
A BNF grammar might not be too useful in certain circumstances...
Writing a LOGO that's accurately compatible with existing/historical implementation isn't an easy task (I worked on such a project). The problem is that the parser doesn't do the full job, and the evaluator (interpreter) has to work with partial data. Consider this example:
proc1 a b proc2 c
It could mean proc1(a, b, proc2(c)) or proc1(a, b, proc2(), c) according to the number of parameters for proc1 & proc2.
Furthermore the LOGO interpreters I know, for example Berkely LOGO, seem from a cursory glance not to write a traditional parser that additionally has access to each procedure and its arity; instead they run the procedures and the procedures 'eat up' the number of parameters that they need. This makes the parser a little naive and the main role is that of an interpreter, and thus parsing is kind of unusual.
There is no standard LOGO implementation.
Your best call is probably to look at the source of a popular implementation, such as UCBLogo
来源：https://stackoverflow.com/questions/6819505/ebnf-or-bnf-for-the-logo-programming-language