car-ast 0.15.0

use super::{extract_doc_comment, field_text, node_text};
use crate::types::*;

pub fn extract(tree: &tree_sitter::Tree, source: &[u8]) -> (Vec<Symbol>, Vec<Import>) {
    let root = tree.root_node();
    let mut symbols = Vec::new();
    let mut imports = Vec::new();

    extract_children(&root, source, &mut symbols, &mut imports, None);

    (symbols, imports)
}

fn extract_children(
    node: &tree_sitter::Node,
    source: &[u8],
    symbols: &mut Vec<Symbol>,
    imports: &mut Vec<Import>,
    parent_name: Option<&str>,
) {
    let mut cursor = node.walk();
    for child in node.children(&mut cursor) {
        match child.kind() {
            "function_definition" => {
                if let Some(sym) = extract_function(&child, source, parent_name) {
                    symbols.push(sym);
                }
            }
            "method_declaration" => {
                if let Some(sym) = extract_method(&child, source, parent_name) {
                    symbols.push(sym);
                }
            }
            "class_declaration" => {
                if let Some(sym) = extract_type_decl(&child, source, SymbolKind::Class, parent_name)
                {
                    symbols.push(sym);
                }
            }
            "interface_declaration" => {
                if let Some(sym) =
                    extract_type_decl(&child, source, SymbolKind::Interface, parent_name)
                {
                    symbols.push(sym);
                }
            }
            "trait_declaration" => {
                if let Some(sym) = extract_type_decl(&child, source, SymbolKind::Trait, parent_name)
                {
                    symbols.push(sym);
                }
            }
            "enum_declaration" => {
                if let Some(sym) = extract_type_decl(&child, source, SymbolKind::Enum, parent_name)
                {
                    symbols.push(sym);
                }
            }
            "namespace_definition" => {
                extract_namespace(&child, source, symbols, imports);
            }
            "namespace_use_declaration" => {
                let text = node_text(&child, source).trim().to_string();
                imports.push(Import {
                    path: text,
                    alias: None,
                    span: Span::from_node(&child),
                });
            }
            "const_declaration" => {
                extract_const(&child, source, symbols, parent_name);
            }
            "property_declaration" => {
                extract_property(&child, source, symbols, parent_name);
            }
            // Recurse into container nodes
            "declaration_list" | "program" => {
                extract_children(&child, source, symbols, imports, parent_name);
            }
            _ => {}
        }
    }
}

fn extract_function(
    node: &tree_sitter::Node,
    source: &[u8],
    parent_name: Option<&str>,
) -> Option<Symbol> {
    let name = field_text(node, "name", source)?;

    let signature = if let Some(body) = node.child_by_field_name("body") {
        let sig = &source[node.start_byte()..body.start_byte()];
        std::str::from_utf8(sig).unwrap_or("").trim().to_string()
    } else {
        node_text(node, source).to_string()
    };

    Some(Symbol {
        name: name.to_string(),
        kind: SymbolKind::Function,
        span: Span::from_node(node),
        signature,
        doc_comment: extract_doc_comment(node, source),
        parent: parent_name.map(|s| s.to_string()),
        children: Vec::new(),
    })
}

fn extract_method(
    node: &tree_sitter::Node,
    source: &[u8],
    parent_name: Option<&str>,
) -> Option<Symbol> {
    let name = field_text(node, "name", source)?;

    let kind = if parent_name.is_some() {
        SymbolKind::Method
    } else {
        SymbolKind::Function
    };

    let signature = if let Some(body) = node.child_by_field_name("body") {
        let sig = &source[node.start_byte()..body.start_byte()];
        std::str::from_utf8(sig).unwrap_or("").trim().to_string()
    } else {
        // Abstract methods or interface method signatures
        node_text(node, source)
            .trim_end_matches(';')
            .trim()
            .to_string()
    };

    Some(Symbol {
        name: name.to_string(),
        kind,
        span: Span::from_node(node),
        signature,
        doc_comment: extract_doc_comment(node, source),
        parent: parent_name.map(|s| s.to_string()),
        children: Vec::new(),
    })
}

fn extract_type_decl(
    node: &tree_sitter::Node,
    source: &[u8],
    kind: SymbolKind,
    parent_name: Option<&str>,
) -> Option<Symbol> {
    let name = field_text(node, "name", source)?;

    // Find the declaration_list child for the body
    let body = find_child_of_kind(node, "declaration_list");
    let signature = if let Some(ref body) = body {
        let sig = &source[node.start_byte()..body.start_byte()];
        std::str::from_utf8(sig).unwrap_or("").trim().to_string()
    } else {
        node_text(node, source)
            .lines()
            .next()
            .unwrap_or("")
            .trim()
            .to_string()
    };

    let mut children = Vec::new();
    let mut child_imports = Vec::new();
    if let Some(body) = body {
        extract_children(&body, source, &mut children, &mut child_imports, Some(name));
    }

    Some(Symbol {
        name: name.to_string(),
        kind,
        span: Span::from_node(node),
        signature,
        doc_comment: extract_doc_comment(node, source),
        parent: parent_name.map(|s| s.to_string()),
        children,
    })
}

/// PHP namespace_definition is flat -- the namespace statement is followed
/// by sibling declarations (not child declarations). We collect all sibling
/// declarations that follow the namespace_definition until the next namespace
/// or end of file.
fn extract_namespace(
    node: &tree_sitter::Node,
    source: &[u8],
    symbols: &mut Vec<Symbol>,
    imports: &mut Vec<Import>,
) {
    let name = field_text(node, "name", source).unwrap_or("(anonymous)");

    // Check if this namespace has a body (braced form)
    let body = find_child_of_kind(node, "compound_statement");
    if let Some(body) = body {
        let mut children = Vec::new();
        extract_children(&body, source, &mut children, imports, Some(name));
        symbols.push(Symbol {
            name: name.to_string(),
            kind: SymbolKind::Module,
            span: Span::from_node(node),
            signature: format!("namespace {}", name),
            doc_comment: extract_doc_comment(node, source),
            parent: None,
            children,
        });
        return;
    }

    // Flat namespace: collect subsequent siblings
    let mut children = Vec::new();
    let mut sibling = node.next_sibling();
    while let Some(s) = sibling {
        if s.kind() == "namespace_definition" {
            break;
        }
        match s.kind() {
            "class_declaration"
            | "interface_declaration"
            | "trait_declaration"
            | "enum_declaration"
            | "function_definition"
            | "const_declaration" => {
                let mut child_syms = Vec::new();
                // Process the sibling declaration
                match s.kind() {
                    "class_declaration" => {
                        if let Some(sym) =
                            extract_type_decl(&s, source, SymbolKind::Class, Some(name))
                        {
                            child_syms.push(sym);
                        }
                    }
                    "interface_declaration" => {
                        if let Some(sym) =
                            extract_type_decl(&s, source, SymbolKind::Interface, Some(name))
                        {
                            child_syms.push(sym);
                        }
                    }
                    "trait_declaration" => {
                        if let Some(sym) =
                            extract_type_decl(&s, source, SymbolKind::Trait, Some(name))
                        {
                            child_syms.push(sym);
                        }
                    }
                    "enum_declaration" => {
                        if let Some(sym) =
                            extract_type_decl(&s, source, SymbolKind::Enum, Some(name))
                        {
                            child_syms.push(sym);
                        }
                    }
                    "function_definition" => {
                        if let Some(sym) = extract_function(&s, source, Some(name)) {
                            child_syms.push(sym);
                        }
                    }
                    "const_declaration" => {
                        extract_const(&s, source, &mut child_syms, Some(name));
                    }
                    _ => {}
                }
                children.extend(child_syms);
            }
            "namespace_use_declaration" => {
                let text = node_text(&s, source).trim().to_string();
                imports.push(Import {
                    path: text,
                    alias: None,
                    span: Span::from_node(&s),
                });
            }
            _ => {}
        }
        sibling = s.next_sibling();
    }

    symbols.push(Symbol {
        name: name.to_string(),
        kind: SymbolKind::Module,
        span: Span::from_node(node),
        signature: format!("namespace {}", name),
        doc_comment: extract_doc_comment(node, source),
        parent: None,
        children,
    });
}

fn extract_const(
    node: &tree_sitter::Node,
    source: &[u8],
    symbols: &mut Vec<Symbol>,
    parent_name: Option<&str>,
) {
    let mut cursor = node.walk();
    for child in node.children(&mut cursor) {
        if child.kind() == "const_element" {
            if let Some(name) = field_text(&child, "name", source) {
                symbols.push(Symbol {
                    name: name.to_string(),
                    kind: SymbolKind::Const,
                    span: Span::from_node(&child),
                    signature: node_text(node, source)
                        .trim_end_matches(';')
                        .trim()
                        .to_string(),
                    doc_comment: extract_doc_comment(node, source),
                    parent: parent_name.map(|s| s.to_string()),
                    children: Vec::new(),
                });
            }
        }
    }
}

fn extract_property(
    node: &tree_sitter::Node,
    source: &[u8],
    symbols: &mut Vec<Symbol>,
    parent_name: Option<&str>,
) {
    let mut cursor = node.walk();
    for child in node.children(&mut cursor) {
        if child.kind() == "property_element" {
            if let Some(var_node) = child.child_by_field_name("name") {
                let name = node_text(&var_node, source);
                if !name.is_empty() {
                    symbols.push(Symbol {
                        name: name.to_string(),
                        kind: SymbolKind::Const,
                        span: Span::from_node(node),
                        signature: node_text(node, source)
                            .trim_end_matches(';')
                            .trim()
                            .to_string(),
                        doc_comment: extract_doc_comment(node, source),
                        parent: parent_name.map(|s| s.to_string()),
                        children: Vec::new(),
                    });
                }
            }
        }
    }
}

fn find_child_of_kind<'a>(
    node: &'a tree_sitter::Node,
    kind: &str,
) -> Option<tree_sitter::Node<'a>> {
    let mut cursor = node.walk();
    for child in node.children(&mut cursor) {
        if child.kind() == kind {
            return Some(child);
        }
    }
    None
}

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

    fn parse_php(source: &str) -> (Vec<Symbol>, Vec<Import>) {
        let mut parser = tree_sitter::Parser::new();
        parser
            .set_language(&tree_sitter_php::LANGUAGE_PHP.into())
            .unwrap();
        let tree = parser.parse(source, None).unwrap();
        extract(&tree, source.as_bytes())
    }

    #[test]
    fn test_class_with_methods() {
        let source = r#"<?php
namespace App\Models;

use App\Contracts\Serializable;

interface Loggable {
    public function log(): void;
}

trait Timestamped {
    public function createdAt(): string {
        return $this->created;
    }
}

class User extends Model implements Serializable {
    const MAX_NAME_LENGTH = 255;

    public function getName(): string {
        return $this->name;
    }

    public function setName(string $name): void {
        $this->name = $name;
    }
}
"#;
        let (symbols, imports) = parse_php(source);

        // use statement
        assert!(
            imports.len() >= 1,
            "expected at least 1 import, got {}",
            imports.len()
        );
        assert!(
            imports.iter().any(|i| i.path.contains("Serializable")),
            "missing Serializable import in {:?}",
            imports
        );

        // Namespace
        let ns = symbols.iter().find(|s| s.kind == SymbolKind::Module);
        assert!(
            ns.is_some(),
            "missing namespace in {:?}",
            symbols
                .iter()
                .map(|s| (&s.name, &s.kind))
                .collect::<Vec<_>>()
        );
        let ns = ns.unwrap();
        assert!(
            ns.name.contains("Models"),
            "namespace name should contain Models: {}",
            ns.name
        );

        let ns_children = &ns.children;
        let names: Vec<&str> = ns_children.iter().map(|s| s.name.as_str()).collect();

        // Interface
        let iface = ns_children.iter().find(|s| s.name == "Loggable");
        assert!(iface.is_some(), "missing Loggable in {:?}", names);
        assert_eq!(iface.unwrap().kind, SymbolKind::Interface);

        // Trait
        let tr = ns_children.iter().find(|s| s.name == "Timestamped");
        assert!(tr.is_some(), "missing Timestamped in {:?}", names);
        assert_eq!(tr.unwrap().kind, SymbolKind::Trait);

        // Class
        let cls = ns_children.iter().find(|s| s.name == "User");
        assert!(cls.is_some(), "missing User in {:?}", names);
        let cls = cls.unwrap();
        assert_eq!(cls.kind, SymbolKind::Class);

        // Class methods
        let method_names: Vec<&str> = cls.children.iter().map(|s| s.name.as_str()).collect();
        assert!(
            method_names.contains(&"getName"),
            "missing getName in {:?}",
            method_names
        );
        assert!(
            method_names.contains(&"setName"),
            "missing setName in {:?}",
            method_names
        );
    }

    #[test]
    fn test_standalone_function() {
        let source = r#"<?php
function add(int $a, int $b): int {
    return $a + $b;
}
"#;
        let (symbols, _imports) = parse_php(source);
        assert_eq!(symbols.len(), 1);
        assert_eq!(symbols[0].name, "add");
        assert_eq!(symbols[0].kind, SymbolKind::Function);
    }

    #[test]
    fn test_enum_declaration() {
        let source = r#"<?php
enum Status {
    case Active;
    case Inactive;
}
"#;
        let (symbols, _imports) = parse_php(source);
        let enm = symbols.iter().find(|s| s.name == "Status");
        assert!(enm.is_some(), "missing Status enum");
        assert_eq!(enm.unwrap().kind, SymbolKind::Enum);
    }
}