big_code_analysis_ast/lang_helpers/

elixir.rs

//! Elixir: keyword `Call` identity (`def`, `defmodule`, `quote`, …),
//! and applied versus named operators.

use crate::Elixir;
use crate::node::{Ancestors, Node};

/// Reads the text of the `target` field of an Elixir `Call` node.
///
/// Most of Elixir's control-flow constructs (`if`, `unless`, `for`,
/// `while`, `case`, `cond`, `with`, `try`) and method-defining macros
/// (`def`, `defp`, `defmacro`, …) parse as `Call` nodes whose `target`
/// is an `Identifier` whose source text spells the keyword. The
/// `Cyclomatic` and `Exit` impls already follow this pattern; this
/// helper centralises the byte-text lookup so `Cognitive` and `Abc`
/// can share it.
///
/// Returns `None` for Calls whose target is not a simple identifier
/// (e.g. `Module.func(…)` parses as `RemoteCallWithParentheses` with
/// the dotted name as target) or when the bytes are not valid UTF-8.
#[inline]
#[must_use]
pub fn elixir_call_keyword<'a>(node: &'a Node<'a>, code: &'a [u8]) -> Option<&'a str> {
    if node.kind_id() != Elixir::Call as u16 {
        return None;
    }
    let target = node.child_by_field_name("target")?;
    if target.kind_id() != Elixir::Identifier as u16 {
        return None;
    }
    target.utf8_text(code)
}

/// Method-defining macros (`def`, `defp`, `defmacro`, `defmacrop`). The set
/// is duplicated across checker, getter, and several metric impls
/// because each consults it from a different trait surface; centralising
/// the literal here keeps future additions (e.g. `defguard`) consistent.
#[inline]
#[must_use]
pub fn elixir_is_method_macro(kw: &str) -> bool {
    matches!(kw, "def" | "defp" | "defmacro" | "defmacrop")
}

/// Class-defining macro (`defmodule`). Paired with [`elixir_is_method_macro`]
/// where a caller needs both ("any space-opening declaration").
#[inline]
#[must_use]
pub fn elixir_is_class_macro(kw: &str) -> bool {
    kw == "defmodule"
}

/// Returns true when `node` is lexically nested inside the `do_block` of a
/// `quote do … end` Call (Elixir's metaprogramming template). A `def` /
/// `defp` / `defmacro` / `defmacrop` inside `quote` does not define a
/// method of any enclosing module — the syntax tree is a code template
/// emitted later, when the surrounding macro is invoked. Treating those
/// quoted Calls as methods inflates `Wmc` and disagrees with `Npm`'s
/// direct-children classification (#310).
///
/// Walks the ancestor chain looking for a `quote` Call ancestor. Stops at
/// the first match (true) or at the root (false). Each step is a single
/// `child_by_field_name("target")` + identifier byte compare, so the cost
/// is O(steps) when `ancestors` is known — with `Ancestors::unknown` each
/// step additionally pays `Node::parent`'s O(depth) (#1084).
#[inline]
#[must_use]
pub fn elixir_is_inside_quote_block<'a>(
    node: &Node<'a>,
    code: &[u8],
    ancestors: Ancestors<'a, '_>,
) -> bool {
    ancestors
        .iter(node)
        .any(|(n, _)| elixir_call_keyword(&n, code) == Some("quote"))
}

/// Iterates the direct-child `Call` nodes inside the `do_block` of an
/// Elixir Call (typically a `defmodule`). Used by `Npm` / `Npa` to scan
/// a module body for method-defining macros / `defstruct` without
/// descending into nested modules. Yields no items when the Call has
/// no `do_block`.
#[inline]
pub fn elixir_do_block_call_children<'a>(
    node: &'a Node<'a>,
) -> impl Iterator<Item = Node<'a>> + 'a {
    node.children()
        .filter(|child| child.kind_id() == Elixir::DoBlock as u16)
        .flat_map(|do_block| do_block.children())
        .filter(|stmt| stmt.kind_id() == Elixir::Call as u16)
}

/// The `binary_operator` that applies `op`, an operator token, or `None`
/// when the token is not applied there.
///
/// An operator token also *names* the operator: `&==/2`, `&and/2` and
/// `Kernel.||(a, b)` hold it inside an `operator_identifier`, and the
/// single-token `atom` / `keyword` spell it too. None of those is a
/// `binary_operator` with operands to compare or chain, and under error
/// recovery the parent can be an `ERROR` with no operands, so every
/// decision metric keys on this parent rather than on the token (#1531,
/// #1534). A remote call that names the operator still *applies* it,
/// though; see [`elixir_applying_remote_call`].
///
/// Matched by rule name rather than by `kind_id`: the grammar aliases
/// `binary_operator` to three ids (`Elixir::BinaryOperator` through
/// `BinaryOperator3`), and a name comparison stays correct when a bump
/// adds a fourth (grammar-dispatch §1).
#[must_use]
pub fn elixir_applying_operator<'a>(
    op: &Node<'a>,
    ancestors: Ancestors<'a, '_>,
) -> Option<Node<'a>> {
    ancestors
        .parent(op)
        .filter(|parent| parent.kind() == "binary_operator")
}

/// The name of the short-circuit operator a remote call on `Kernel` with
/// arguments applies, or `None` for any other node: the `||` of
/// `Kernel.||(a, b)`, or of `x |> Kernel.||(y)`, which the pipe rewrites
/// to it. Such a call expands the `Kernel` macro exactly as `a || b`
/// does, so it short-circuits and decides as the operator does, where a
/// capture (`&Kernel.||/2`, a call with no arguments) only names it. It
/// has no `binary_operator`, so its operands are the call's arguments,
/// and under a pipe the piped value, rather than `left` / `right` fields.
///
/// The receiver must spell `Kernel` (or `Elixir.Kernel`): a user module
/// may define `||/2` as an ordinary function, and `Foo.||(a, b)` neither
/// short-circuits nor decides.
///
/// The grammar lexes a symbolic name as a token inside the call's
/// `operator_identifier`, which is what is returned. `Kernel.and(a, b)`
/// and `Kernel.or(a, b)` spell their names as a plain `identifier`, with
/// no `And` / `Or` token to find, so that identifier is returned instead
/// (#1582). Any other name — `Kernel.+`, `Kernel.!`, `Kernel.not`,
/// `Kernel.in` — is a function or decides nothing, and is not found; nor
/// is the childless `..`. [`elixir_short_circuit_name`] says which
/// operation the name applies.
///
/// Every identifier a walk meets asks this of its call, so the receiver,
/// which turns nearly all of them away without a cursor, is read before
/// the call's children are scanned for its `arguments`.
///
/// Matched by rule name: `dot` and `arguments` are aliased to several
/// ids each (grammar-dispatch §1).
#[must_use]
pub fn elixir_kernel_short_circuit_name<'a>(call: &Node<'a>, code: &[u8]) -> Option<Node<'a>> {
    kernel_short_circuit_parts(call, code).map(|(name, _)| name)
}

fn kernel_short_circuit_parts<'a>(call: &Node<'a>, code: &[u8]) -> Option<(Node<'a>, Node<'a>)> {
    if call.kind() != "call" {
        return None;
    }
    let dot = call.child_by_field_name("target").filter(|target| {
        target.kind() == "dot"
            && target.child_by_field_name("left").is_some_and(|receiver| {
                matches!(
                    code.get(receiver.start_byte()..receiver.end_byte()),
                    Some(b"Kernel" | b"Elixir.Kernel")
                )
            })
    })?;
    let name = dot.child_by_field_name("right")?;
    let name = if name.kind() == "operator_identifier" {
        name.child(0)?
    } else {
        name
    };
    elixir_short_circuit_name(&name, code)
        .and_then(|_| call.children().find(|child| child.kind() == "arguments"))
        .map(|arguments| (name, arguments))
}

/// The short-circuit operation an operator token kind applies, keyed by
/// its symbol: `&&` for `&&` and `and`, `||` for `||` and `or`. A word
/// form demands a boolean left operand, but it is the same conjunction or
/// disjunction to read, so one sequence key serves both spellings.
/// `None` for any other kind.
#[must_use]
pub fn elixir_short_circuit_kind(kind_id: u16) -> Option<Elixir> {
    match kind_id.into() {
        Elixir::AMPAMP | Elixir::And => Some(Elixir::AMPAMP),
        Elixir::PIPEPIPE | Elixir::Or => Some(Elixir::PIPEPIPE),
        _ => None,
    }
}

/// [`elixir_short_circuit_kind`] for an operator's name as
/// [`elixir_kernel_short_circuit_name`] returns it: a token, or the
/// `identifier` spelling `Kernel.and` / `Kernel.or` (#1582), whose
/// bytes decide. Ask it only of such a name — an identifier is an `and`
/// operator nowhere else.
#[must_use]
pub fn elixir_short_circuit_name(name: &Node, code: &[u8]) -> Option<Elixir> {
    if name.kind_id() != Elixir::Identifier as u16 {
        return elixir_short_circuit_kind(name.kind_id());
    }
    match code.get(name.start_byte()..name.end_byte())? {
        b"and" => Some(Elixir::AMPAMP),
        b"or" => Some(Elixir::PIPEPIPE),
        _ => None,
    }
}

/// A recognized short-circuit `Kernel` call and the context its operands need.
pub struct ElixirRemoteCall<'tree, 'chain> {
    /// The call that applies the operator.
    pub call: Node<'tree>,
    /// The argument list found while recognizing the call.
    pub arguments: Node<'tree>,
    /// The call's own ancestry.
    pub ancestors: Ancestors<'tree, 'chain>,
}

/// The `Kernel` remote call that applies `op`, a short-circuit operator's
/// name — the call [`elixir_kernel_short_circuit_name`] answers `op` for
/// — with its argument list and ancestry, or `None` when `op` sits
/// anywhere else. A token's parents are the `operator_identifier`, then
/// the `dot`, then the call; the `identifier` naming `Kernel.and` sits
/// on the `dot` directly. Any other node fails on the first one or two
/// parents' kinds, so the check is cheap enough to ask of every
/// identifier a walk meets.
#[must_use]
pub fn elixir_applying_remote_call<'a, 'c>(
    op: &Node<'a>,
    code: &[u8],
    ancestors: Ancestors<'a, 'c>,
) -> Option<ElixirRemoteCall<'a, 'c>> {
    let mut up = ancestors.iter(op);
    let (mut dot, _) = up.next()?;
    if dot.kind() == "operator_identifier" {
        dot = up.next()?.0;
    }
    if dot.kind() != "dot" {
        return None;
    }
    let (call, call_ancestors) = up.next()?;
    kernel_short_circuit_parts(&call, code)
        .filter(|(name, _)| name.id() == op.id())
        .map(|(_, arguments)| ElixirRemoteCall {
            call,
            arguments,
            ancestors: call_ancestors,
        })
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::ElixirCode;
    use crate::node::Tree;
    use crate::test_support::for_each_node_with_chain;

    #[cfg(feature = "elixir")]
    #[test]
    fn short_circuit_name_requires_the_identifiers_source_span() {
        let source = b"Kernel.and(a, b)";
        let tree = Tree::new::<ElixirCode>(source);
        let name = tree
            .get_root()
            .preorder()
            .find(|node| {
                node.kind_id() == Elixir::Identifier as u16 && node.utf8_text(source) == Some("and")
            })
            .expect("fixture contains the operator name");
        assert_eq!(
            elixir_short_circuit_name(&name, source),
            Some(Elixir::AMPAMP)
        );
        assert_eq!(
            elixir_short_circuit_name(&name, &source[..name.end_byte() - 1]),
            None
        );
    }

    #[cfg(feature = "elixir")]
    #[test]
    fn remote_call_lookup_rejects_a_root_and_keeps_its_argument_list() {
        let source = b"Kernel.and(a, b)\nKernel.||(c, d)\n";
        let mut recognized = Vec::new();
        for_each_node_with_chain::<ElixirCode>(source, |node, chain| {
            let known = elixir_applying_remote_call(node, source, Ancestors::known(chain));
            let unknown = elixir_applying_remote_call(node, source, Ancestors::unknown());
            assert_eq!(
                known.as_ref().map(|r| (r.call.id(), r.arguments.id())),
                unknown.map(|r| (r.call.id(), r.arguments.id()))
            );
            if chain.is_empty() {
                assert!(known.is_none());
            }
            if let Some(remote) = known {
                assert_eq!(remote.arguments.kind(), "arguments");
                assert_eq!(
                    remote.ancestors.parent(&remote.call).map(|p| p.kind()),
                    Some("source")
                );
                recognized.push((
                    node.utf8_text(source)
                        .expect("operator is UTF-8")
                        .to_owned(),
                    remote
                        .arguments
                        .utf8_text(source)
                        .expect("arguments are UTF-8")
                        .to_owned(),
                ));
            }
        });
        assert_eq!(
            recognized,
            [
                ("and".to_owned(), "(a, b)".to_owned()),
                ("||".to_owned(), "(c, d)".to_owned())
            ]
        );
    }
}