normalize_languages/

awk.rs

//! AWK language support.

use crate::external_packages::ResolvedPackage;
use crate::{
    Export, Import, Language, Symbol, SymbolKind, Visibility, VisibilityMechanism,
    simple_function_symbol,
};
use std::path::{Path, PathBuf};
use tree_sitter::Node;

/// AWK language support.
pub struct Awk;

impl Language for Awk {
    fn name(&self) -> &'static str {
        "AWK"
    }
    fn extensions(&self) -> &'static [&'static str] {
        &["awk", "gawk"]
    }
    fn grammar_name(&self) -> &'static str {
        "awk"
    }

    fn has_symbols(&self) -> bool {
        true
    }

    fn container_kinds(&self) -> &'static [&'static str] {
        &[]
    }

    fn function_kinds(&self) -> &'static [&'static str] {
        &["func_def"]
    }

    fn type_kinds(&self) -> &'static [&'static str] {
        &[]
    }
    fn import_kinds(&self) -> &'static [&'static str] {
        &[]
    }

    fn public_symbol_kinds(&self) -> &'static [&'static str] {
        &["func_def"]
    }

    fn visibility_mechanism(&self) -> VisibilityMechanism {
        VisibilityMechanism::AllPublic
    }

    fn extract_public_symbols(&self, node: &Node, content: &str) -> Vec<Export> {
        if node.kind() != "func_def" {
            return Vec::new();
        }

        let name = match self.node_name(node, content) {
            Some(n) => n.to_string(),
            None => return Vec::new(),
        };

        vec![Export {
            name,
            kind: SymbolKind::Function,
            line: node.start_position().row + 1,
        }]
    }

    fn scope_creating_kinds(&self) -> &'static [&'static str] {
        &["func_def", "block"]
    }

    fn control_flow_kinds(&self) -> &'static [&'static str] {
        &[
            "if_statement",
            "while_statement",
            "for_statement",
            "for_in_statement",
        ]
    }

    fn complexity_nodes(&self) -> &'static [&'static str] {
        &[
            "if_statement",
            "while_statement",
            "for_statement",
            "for_in_statement",
            "ternary_exp",
        ]
    }

    fn nesting_nodes(&self) -> &'static [&'static str] {
        &[
            "func_def",
            "if_statement",
            "while_statement",
            "for_statement",
        ]
    }

    fn signature_suffix(&self) -> &'static str {
        ""
    }

    fn extract_function(&self, node: &Node, content: &str, _in_container: bool) -> Option<Symbol> {
        let name = self.node_name(node, content)?;
        Some(simple_function_symbol(
            node,
            content,
            name,
            self.extract_docstring(node, content),
        ))
    }

    fn extract_container(&self, _node: &Node, _content: &str) -> Option<Symbol> {
        None
    }
    fn extract_type(&self, _node: &Node, _content: &str) -> Option<Symbol> {
        None
    }

    fn extract_docstring(&self, node: &Node, content: &str) -> Option<String> {
        let mut prev = node.prev_sibling();
        let mut doc_lines = Vec::new();

        while let Some(sibling) = prev {
            let text = &content[sibling.byte_range()];
            if sibling.kind() == "comment" && text.starts_with('#') {
                let line = text.strip_prefix('#').unwrap_or(text).trim();
                doc_lines.push(line.to_string());
                prev = sibling.prev_sibling();
            } else {
                break;
            }
        }

        if doc_lines.is_empty() {
            return None;
        }

        doc_lines.reverse();
        Some(doc_lines.join(" "))
    }

    fn extract_attributes(&self, _node: &Node, _content: &str) -> Vec<String> {
        Vec::new()
    }

    fn extract_imports(&self, _node: &Node, _content: &str) -> Vec<Import> {
        Vec::new()
    }

    fn format_import(&self, _import: &Import, _names: Option<&[&str]>) -> String {
        // AWK has no imports
        String::new()
    }

    fn is_public(&self, _node: &Node, _content: &str) -> bool {
        true
    }
    fn get_visibility(&self, _node: &Node, _content: &str) -> Visibility {
        Visibility::Public
    }

    fn is_test_symbol(&self, symbol: &crate::Symbol) -> bool {
        let name = symbol.name.as_str();
        match symbol.kind {
            crate::SymbolKind::Function | crate::SymbolKind::Method => name.starts_with("test_"),
            crate::SymbolKind::Module => name == "tests" || name == "test",
            _ => false,
        }
    }

    fn embedded_content(&self, _node: &Node, _content: &str) -> Option<crate::EmbeddedBlock> {
        None
    }

    fn container_body<'a>(&self, _node: &'a Node<'a>) -> Option<Node<'a>> {
        None
    }
    fn body_has_docstring(&self, _body: &Node, _content: &str) -> bool {
        false
    }

    fn node_name<'a>(&self, node: &Node, content: &'a str) -> Option<&'a str> {
        node.child_by_field_name("name")
            .map(|n| &content[n.byte_range()])
    }

    fn file_path_to_module_name(&self, path: &Path) -> Option<String> {
        let ext = path.extension()?.to_str()?;
        if ext != "awk" && ext != "gawk" {
            return None;
        }
        let stem = path.file_stem()?.to_str()?;
        Some(stem.to_string())
    }

    fn module_name_to_paths(&self, module: &str) -> Vec<String> {
        vec![format!("{}.awk", module)]
    }

    fn lang_key(&self) -> &'static str {
        "awk"
    }

    fn is_stdlib_import(&self, _import_name: &str, _project_root: &Path) -> bool {
        false
    }
    fn find_stdlib(&self, _project_root: &Path) -> Option<PathBuf> {
        None
    }
    fn resolve_local_import(&self, _: &str, _: &Path, _: &Path) -> Option<PathBuf> {
        None
    }
    fn resolve_external_import(&self, _: &str, _: &Path) -> Option<ResolvedPackage> {
        None
    }
    fn get_version(&self, _: &Path) -> Option<String> {
        None
    }
    fn find_package_cache(&self, _: &Path) -> Option<PathBuf> {
        None
    }
    fn indexable_extensions(&self) -> &'static [&'static str] {
        &["awk", "gawk"]
    }
    fn package_sources(&self, _: &Path) -> Vec<crate::PackageSource> {
        Vec::new()
    }

    fn should_skip_package_entry(&self, name: &str, is_dir: bool) -> bool {
        use crate::traits::{has_extension, skip_dotfiles};
        if skip_dotfiles(name) {
            return true;
        }
        !is_dir && !has_extension(name, self.indexable_extensions())
    }

    fn discover_packages(&self, _: &crate::PackageSource) -> Vec<(String, PathBuf)> {
        Vec::new()
    }

    fn package_module_name(&self, entry_name: &str) -> String {
        entry_name
            .strip_suffix(".awk")
            .or_else(|| entry_name.strip_suffix(".gawk"))
            .unwrap_or(entry_name)
            .to_string()
    }

    fn find_package_entry(&self, path: &Path) -> Option<PathBuf> {
        if path.is_file() {
            Some(path.to_path_buf())
        } else {
            None
        }
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::validate_unused_kinds_audit;

    #[test]
    fn unused_node_kinds_audit() {
        #[rustfmt::skip]
        let documented_unused: &[&str] = &[
            "break_statement", "continue_statement", "delete_statement", "do_while_statement",
            "else_clause", "exit_statement", "identifier", "next_statement", "nextfile_statement",
            "ns_qualified_name", "piped_io_statement", "print_statement", "printf_statement",
            "redirected_io_statement", "return_statement", "switch_body", "switch_case",
            "switch_statement",
        ];
        validate_unused_kinds_audit(&Awk, documented_unused)
            .expect("AWK unused node kinds audit failed");
    }
}