1 Quote and unquote
An Elixir program can be represented by its own data structures. In this chapter, we will learn what those structures look like and how to compose them. The concepts learned in this chapter are the building blocks for macros, which we are going to take a deeper look at in the next chapter.
1.1 Quoting
The building block of an Elixir program is a tuple with three elements. For example, the function call sum(1,2,3) is represented internally as:
{:sum, [], [1, 2, 3]}
You can get the representation of any expression by using the quote macro:
iex> quote do: sum(1, 2, 3)
{:sum, [], [1, 2, 3]}
The first element is the function name, the second is a keyword list containing metadata and the third is the arguments list.
Operators are also represented as such tuples:
iex> quote do: 1 + 2
{:+, [context: Elixir, import: Kernel], [1, 2]}
Even a map is represented as a call to %{}:
iex> quote do: %{1 => 2}
{:%{}, [], [{1, 2}]}
Variables are also represented using such triplets, except the last element is an atom, instead of a list:
iex> quote do: x
{ :x, [], Elixir }
When quoting more complex expressions, we can see that the code is represented in such tuples, which are often nested inside each other in a structure resembling a tree. Many languages would call such representations an Abstract Syntax Tree (AST). Elixir calls them quoted expressions:
iex> quote do: sum(1, 2 + 3, 4)
{:sum, [], [1, {:+, [context: Elixir, import: Kernel], [2, 3]}, 4]}
Sometimes when working with quoted expressions, it may be useful to get the textual code representation back. This can be done with Macro.to_string/1:
iex> Macro.to_string(quote do: sum(1, 2 + 3, 4))
"sum(1, 2 + 3, 4)"
In general, the tuples above are structured according to the following format:
{ tuple | atom, list, list | atom }
- The first element is an atom or another tuple in the same representation;
- The second element is a keyword list containing metadata, like numbers and contexts;
- The third element is either a list of arguments for the function call or an atom. When this element is an atom, it means the tuple represents a variable.
Besides the tuple defined above, there are five Elixir literals that, when quoted, return themselves (and not a tuple). They are:
:sum #=> Atoms
1.0 #=> Numbers
[1, 2] #=> Lists
"strings" #=> Strings
{key, value} #=> Tuples with two elements
Most Elixir code has a straight-forward translation to its underlying quoted expression. We recommend you try out different code samples and see what the results are. For example, what does String.upcase("foo") expand to? We have also learned that if(true, do: :this, else: :that) is the same as if true do :this else :that end. How does this affirmation hold with quoted expressions?
1.2 Unquoting
Quote is about retrieving the inner representation of some particular chunk of code. However, sometimes it may be necessary to inject some other particular chunk of code inside the representation we want to retrieve.
For example, imagine you have a variable number which contains the number you want to inject inside a quoted expression. The number can be injected into the quoted representation by using unquote:
iex> number = 13
iex> Macro.to_string(quote do: 11 + unquote(number))
"11 + 13"
unquote can even be used to inject function names:
iex> fun = :hello
iex> Macro.to_string(quote do: unquote(fun)(:world))
"hello(:world)"
In some cases, it may be necessary to inject many values inside a list. For example, imagine you have a list containing [1, 2, 6] and we want to inject [3, 4, 5] into it. Using unquote won't yield the desired result:
iex> inner = [3, 4, 5]
iex> Macro.to_string(quote do: [1, 2, unquote(inner), 6])
"[1, 2, [3, 4, 5], 6]"
That's when unquote_splicing becomes handy:
iex> inner = [3, 4, 5]
iex> Macro.to_string(quote do: [1, 2, unquote_splicing(inner), 6])
"[1, 2, 3, 4, 5, 6]"
Unquoting is very useful when working with macros. When writing macros, developers are able to receive code chunks and inject them inside other code chunks, which can be used to transform code or write code that generates code during compilation.
1.3 Escaping
As we saw at the beginning of this chapter, only some values are valid quoted expressions in Elixir. For example, a map is not a valid quoted expression. Neither is a tuple with four elements. However, such values can be expressed as a quoted expression:
iex> quote do: %{1 => 2}
{:%{}, [], [{1, 2}]}
In some cases, you may need to inject such values into quoted expressions. To do that, we need to first escape those values into quoted expressions with the help of Macro.escape/1:
iex> map = %{hello: :world}
iex> Macro.escape(map)
{:%{}, [], [hello: :world]}
Macros receive quoted expressions and must return quoted expressions. However, sometimes during the execution of a macro, you may need to work with values and making a distinction between values and quoted expressions will be required.
In other words, it is important to make a distinction between a regular Elixir value (like a list, a map, a process, a reference, etc) and a quoted expression. Some values, such as integers, atoms and strings, have a quoted expression equal to the value itself. Other values, like maps, need to be explicitly converted. Finally, values like functions and references cannot be converted to a quoted expression at all.
You can read more about quote and unquote in the Kernel.SpecialForms module. Documentation for Macro.escape/1 and other functions related to quoted expressions can be found in the Macro module.
In this introduction we have laid the groundwork to finally write our first macro, so let's move to the next chapter.