AST, M-Expression, General List: same thing¶
The end-result of scanning and parsing is an M-expression, which is a
Python object of type
Expression. In compiler terminology the
M-expression is also called an abstract syntax tree or AST. The
first element of an
Expression is called the “head”.
Expression is to be evaluated, the head should a
Symbol for a Mathics3 Function or another
evaluated produces a function symbol.
In Mathics3, there are only very few different kinds of
Expression nodes, called “atoms” that can appear in the tree
initially from the parser:
With the exception of the addition of
Filename these are
almost the same atoms described in Basic Internal Architecture
for WL. However these are not all of the kinds of Atoms that exist when Mathics3 starts up.
There are other atoms like
We should note that
Symbol really has two distinct meanings. After
a parse, a
Symbol is just its name and this name is unique in the
same way that the integer 5 is unique or the string “foo”.
In a programming language you might think of a Symbol as it is born from the parser as akin to a variable identifier.
Later on in the evaluation process, symbols often have values bound to them. When this happens they more like variables in a programming language. There is not just the name but a value bound to that for some particular scope. Again, think about the difference between an identifier name in a programming language and the variable it represents.
The class definitions for these are given in mathics.core.parser.ast.
If you compare the above four AST types with other languages, you’ll find this is pretty sparse. For example, Python’s AST has well over 30 different types.
So what’s the difference? Python specializes AST types for different kinds of programming language constructs, such as for loops, conditional statements, exception blocks, different kinds of expressions, and so on. In WL and Mathics3, these are all simply built-in functions.
In the process of Evaluation, described in the next section, more kinds of objects over the above the four may get introduced as the M-expression is rewritten and transformed.
Here is an example of the transformation from an input string to the AST Form (an M-Expression)
We use the
--full-form option in the
mathics command-line to get this information.
Note that this shows the input before evaluation:
In:= 3 a - b System`Plus[System`Times[3, Global`a], System`Times[-1, Global`b]] Out= 3 a - b In:= 3 5 - 6 System`Plus[System`Times[3, 5], -6] Out= 9