Struct difft::parse::tree_sitter_parser::TreeSitterConfig

pub(crate) struct TreeSitterConfig {
    pub(crate) language: Language,
    atom_nodes: FxHashSet<&'static str>,
    delimiter_tokens: Vec<(&'static str, &'static str)>,
    highlight_query: Query,
    sub_languages: Vec<TreeSitterSubLanguage>,
}

Configuration for a tree-sitter parser.

Fields

language: Language

The tree-sitter language parser.

atom_nodes: FxHashSet<&'static str>

Force these tree-sitter nodes to be difftastic atoms and ignore their children.

Difftastic only cares about list delimiters and atom contents. This ensures that “x” and “ x“ are different, but [x] and [ x] are not.

This causes problems for tree-sitter grammars that have more complex structure for literals. If string interpolation produces an AST with a separate interpolation node, difftastic will think that “$x” and “ $x“ are the same, because the atom is just $x and the delimiter is “.

This problem also occurs when the tree-sitter AST is missing some children. This is known limitation of tree-sitter, and occurs more often for complex string syntax. https://github.com/tree-sitter/tree-sitter/issues/1156

By forcing the tree-sitter subtree to be a difftastic atom, we guarantee a correct diff, at the cost of losing some structure in the tree-sitter AST.

delimiter_tokens: Vec<(&'static str, &'static str)>

We want to consider delimiter tokens as part of lists, not standalone atoms. Tree-sitter includes delimiter tokens, so mark which token pairs we consider to be delimiters.

highlight_query: Query

The tree-sitter query used for syntax highlighting this language.

sub_languages: Vec<TreeSitterSubLanguage>

Sub-languages in use, if any.