use std::time::Instant;
use tree_sitter::{Node as TsNode, Parser, Tree};
use crate::extraction::complexity::{count_complexity, ComplexityMetrics, FSHARP_COMPLEXITY};
use crate::extraction::ts_state::ExtractionState;
use crate::types::{
generate_node_id, Edge, EdgeKind, ExtractionResult, Node, NodeKind, UnresolvedRef, Visibility,
};
pub struct FSharpExtractor;
impl FSharpExtractor {
pub fn extract_fsharp(file_path: &str, source: &str) -> ExtractionResult {
let start = Instant::now();
let mut state = ExtractionState::new(file_path, source);
let tree = match Self::parse_source(source) {
Ok(t) => t,
Err(msg) => {
state.errors.push(msg);
return state.build_result(start);
}
};
let file_node = Node {
id: generate_node_id(file_path, &NodeKind::File, file_path, 0),
kind: NodeKind::File,
name: file_path.to_string(),
qualified_name: file_path.to_string(),
file_path: file_path.to_string(),
start_line: 0,
attrs_start_line: 0,
end_line: source.lines().count().saturating_sub(1) as u32,
start_column: 0,
end_column: 0,
signature: None,
docstring: None,
visibility: Visibility::Pub,
is_async: false,
branches: 0,
loops: 0,
returns: 0,
max_nesting: 0,
unsafe_blocks: 0,
unchecked_calls: 0,
assertions: 0,
cognitive_complexity: 0,
distinct_operators: 0,
distinct_operands: 0,
total_operators: 0,
total_operands: 0,
updated_at: state.timestamp,
parent_id: None,
};
let file_node_id = file_node.id.clone();
state.nodes.push(file_node);
state.node_stack.push((file_path.to_string(), file_node_id));
let root = tree.root_node();
Self::visit_children(&mut state, root);
state.node_stack.pop();
state.build_result(start)
}
fn parse_source(source: &str) -> Result<Tree, String> {
let mut parser = Parser::new();
let language = crate::extraction::ts_provider::language("fsharp");
parser
.set_language(&language)
.map_err(|e| format!("failed to load F# grammar: {e}"))?;
parser
.parse(source, None)
.ok_or_else(|| "tree-sitter parse returned None".to_string())
}
fn visit_children(state: &mut ExtractionState, node: TsNode<'_>) {
let mut cursor = node.walk();
if cursor.goto_first_child() {
loop {
Self::visit_node(state, cursor.node());
if !cursor.goto_next_sibling() {
break;
}
}
}
}
fn visit_node(state: &mut ExtractionState, node: TsNode<'_>) {
match node.kind() {
"function_or_value_defn" => Self::visit_function_or_value(state, node),
"type_definition" => Self::visit_type_definition(state, node),
"module_defn" | "named_module" => Self::visit_module(state, node),
"namespace" => Self::visit_namespace(state, node),
"open_decl" => Self::visit_open(state, node),
_ => Self::visit_children(state, node),
}
}
fn visit_function_or_value(state: &mut ExtractionState, node: TsNode<'_>) {
let name = Self::extract_binding_name(state, node);
let Some(name) = name else { return };
let is_fn = Self::has_params(node);
let kind = if is_fn {
NodeKind::Function
} else {
NodeKind::Const
};
let sig = Self::first_line(state, node);
let docstring = Self::extract_docstring(state, node);
let start_line = node.start_position().row as u32;
let qualified_name = format!("{}::{}", state.qualified_prefix(), name);
let id = generate_node_id(&state.file_path, &kind, &name, start_line);
let metrics = if is_fn && node.child_count() > 0 {
count_complexity(node, &FSHARP_COMPLEXITY, &state.source)
} else {
ComplexityMetrics::default()
};
let graph_node = Node {
id: id.clone(),
kind,
name: name.clone(),
qualified_name,
file_path: state.file_path.clone(),
start_line,
attrs_start_line: start_line,
end_line: node.end_position().row as u32,
start_column: node.start_position().column as u32,
end_column: node.end_position().column as u32,
signature: sig,
docstring,
visibility: Visibility::Pub,
is_async: false,
branches: metrics.branches,
loops: metrics.loops,
returns: metrics.returns,
max_nesting: metrics.max_nesting,
unsafe_blocks: 0,
unchecked_calls: metrics.unchecked_calls,
assertions: metrics.assertions,
cognitive_complexity: metrics.cognitive_complexity,
distinct_operators: metrics.distinct_operators,
distinct_operands: metrics.distinct_operands,
total_operators: metrics.total_operators,
total_operands: metrics.total_operands,
updated_at: state.timestamp,
parent_id: None,
};
state.nodes.push(graph_node);
if let Some(parent_id) = state.parent_node_id() {
state.edges.push(Edge {
source: parent_id.to_string(),
target: id.clone(),
kind: EdgeKind::Contains,
line: Some(start_line),
});
}
Self::extract_calls(state, node, &id);
}
fn visit_type_definition(state: &mut ExtractionState, node: TsNode<'_>) {
let name = Self::find_child_text(state, node, "type_name")
.or_else(|| Self::find_child_text(state, node, "identifier"));
let Some(name) = name else { return };
let sig = Self::first_line(state, node);
let start_line = node.start_position().row as u32;
let qualified_name = format!("{}::{}", state.qualified_prefix(), name);
let id = generate_node_id(&state.file_path, &NodeKind::Class, &name, start_line);
let graph_node = Node {
id: id.clone(),
kind: NodeKind::Class,
name,
qualified_name,
file_path: state.file_path.clone(),
start_line,
attrs_start_line: start_line,
end_line: node.end_position().row as u32,
start_column: node.start_position().column as u32,
end_column: node.end_position().column as u32,
signature: sig,
docstring: None,
visibility: Visibility::Pub,
is_async: false,
branches: 0,
loops: 0,
returns: 0,
max_nesting: 0,
unsafe_blocks: 0,
unchecked_calls: 0,
assertions: 0,
cognitive_complexity: 0,
distinct_operators: 0,
distinct_operands: 0,
total_operators: 0,
total_operands: 0,
updated_at: state.timestamp,
parent_id: None,
};
state.nodes.push(graph_node);
if let Some(parent_id) = state.parent_node_id() {
state.edges.push(Edge {
source: parent_id.to_string(),
target: id,
kind: EdgeKind::Contains,
line: Some(start_line),
});
}
}
fn visit_module(state: &mut ExtractionState, node: TsNode<'_>) {
let name = Self::find_child_text(state, node, "long_identifier")
.or_else(|| Self::find_child_text(state, node, "identifier"));
let Some(name) = name else { return };
let start_line = node.start_position().row as u32;
let qualified_name = format!("{}::{}", state.qualified_prefix(), name);
let id = generate_node_id(&state.file_path, &NodeKind::Module, &name, start_line);
let graph_node = Node {
id: id.clone(),
kind: NodeKind::Module,
name: name.clone(),
qualified_name,
file_path: state.file_path.clone(),
start_line,
attrs_start_line: start_line,
end_line: node.end_position().row as u32,
start_column: node.start_position().column as u32,
end_column: node.end_position().column as u32,
signature: None,
docstring: None,
visibility: Visibility::Pub,
is_async: false,
branches: 0,
loops: 0,
returns: 0,
max_nesting: 0,
unsafe_blocks: 0,
unchecked_calls: 0,
assertions: 0,
cognitive_complexity: 0,
distinct_operators: 0,
distinct_operands: 0,
total_operators: 0,
total_operands: 0,
updated_at: state.timestamp,
parent_id: None,
};
state.nodes.push(graph_node);
if let Some(parent_id) = state.parent_node_id() {
state.edges.push(Edge {
source: parent_id.to_string(),
target: id.clone(),
kind: EdgeKind::Contains,
line: Some(start_line),
});
}
state.node_stack.push((name, id));
Self::visit_children(state, node);
state.node_stack.pop();
}
fn visit_namespace(state: &mut ExtractionState, node: TsNode<'_>) {
let name = Self::find_child_text(state, node, "long_identifier")
.or_else(|| Self::find_child_text(state, node, "identifier"))
.unwrap_or_else(|| "?".to_string());
let start_line = node.start_position().row as u32;
let qualified_name = format!("{}::{}", state.file_path, name);
let id = generate_node_id(&state.file_path, &NodeKind::Module, &name, start_line);
let graph_node = Node {
id: id.clone(),
kind: NodeKind::Module,
name: name.clone(),
qualified_name,
file_path: state.file_path.clone(),
start_line,
attrs_start_line: start_line,
end_line: node.end_position().row as u32,
start_column: 0,
end_column: 0,
signature: None,
docstring: None,
visibility: Visibility::Pub,
is_async: false,
branches: 0,
loops: 0,
returns: 0,
max_nesting: 0,
unsafe_blocks: 0,
unchecked_calls: 0,
assertions: 0,
cognitive_complexity: 0,
distinct_operators: 0,
distinct_operands: 0,
total_operators: 0,
total_operands: 0,
updated_at: state.timestamp,
parent_id: None,
};
state.nodes.push(graph_node);
if let Some(parent_id) = state.parent_node_id() {
state.edges.push(Edge {
source: parent_id.to_string(),
target: id.clone(),
kind: EdgeKind::Contains,
line: Some(start_line),
});
}
state.node_stack.push((name, id));
Self::visit_children(state, node);
state.node_stack.pop();
}
fn visit_open(state: &mut ExtractionState, node: TsNode<'_>) {
let text = state.node_text(node);
let name = text.trim_start_matches("open").trim().to_string();
let start_line = node.start_position().row as u32;
let id = generate_node_id(&state.file_path, &NodeKind::Use, &name, start_line);
let graph_node = Node {
id: id.clone(),
kind: NodeKind::Use,
name,
qualified_name: format!("{}::open", state.file_path),
file_path: state.file_path.clone(),
start_line,
attrs_start_line: start_line,
end_line: node.end_position().row as u32,
start_column: node.start_position().column as u32,
end_column: node.end_position().column as u32,
signature: Some(text.trim().to_string()),
docstring: None,
visibility: Visibility::Private,
is_async: false,
branches: 0,
loops: 0,
returns: 0,
max_nesting: 0,
unsafe_blocks: 0,
unchecked_calls: 0,
assertions: 0,
cognitive_complexity: 0,
distinct_operators: 0,
distinct_operands: 0,
total_operators: 0,
total_operands: 0,
updated_at: state.timestamp,
parent_id: None,
};
state.nodes.push(graph_node);
if let Some(parent_id) = state.parent_node_id() {
state.edges.push(Edge {
source: parent_id.to_string(),
target: id,
kind: EdgeKind::Contains,
line: Some(start_line),
});
}
}
fn extract_binding_name(state: &ExtractionState, node: TsNode<'_>) -> Option<String> {
let mut cursor = node.walk();
if cursor.goto_first_child() {
loop {
let child = cursor.node();
match child.kind() {
"function_declaration_left" | "value_declaration_left" => {
return Self::find_child_text(state, child, "identifier")
.or_else(|| Self::find_child_text(state, child, "operator_name"));
}
_ => {}
}
if !cursor.goto_next_sibling() {
break;
}
}
}
None
}
fn has_params(node: TsNode<'_>) -> bool {
let mut cursor = node.walk();
if cursor.goto_first_child() {
loop {
let child = cursor.node();
if child.kind() == "function_declaration_left" {
return child.named_child_count() > 1;
}
if !cursor.goto_next_sibling() {
break;
}
}
}
false
}
fn find_child_text(state: &ExtractionState, node: TsNode<'_>, kind: &str) -> Option<String> {
let mut cursor = node.walk();
if cursor.goto_first_child() {
loop {
let child = cursor.node();
if child.kind() == kind {
return Some(state.node_text(child));
}
if !cursor.goto_next_sibling() {
break;
}
}
}
None
}
fn extract_calls(state: &mut ExtractionState, node: TsNode<'_>, fn_id: &str) {
let mut cursor = node.walk();
if cursor.goto_first_child() {
loop {
let child = cursor.node();
match child.kind() {
"application_expression" => {
if let Some(callee) = child.child(0) {
let name = state.node_text(callee);
state.unresolved_refs.push(UnresolvedRef {
from_node_id: fn_id.to_string(),
reference_name: name,
reference_kind: EdgeKind::Calls,
line: child.start_position().row as u32,
column: child.start_position().column as u32,
file_path: state.file_path.clone(),
});
}
Self::extract_calls(state, child, fn_id);
}
"function_or_value_defn" | "type_definition" => {}
_ => Self::extract_calls(state, child, fn_id),
}
if !cursor.goto_next_sibling() {
break;
}
}
}
}
fn extract_docstring(state: &ExtractionState, node: TsNode<'_>) -> Option<String> {
let mut comments = Vec::new();
let mut prev = node.prev_named_sibling();
while let Some(p) = prev {
if p.kind() == "xml_doc" || p.kind() == "line_comment" {
let text = state.node_text(p);
if text.starts_with("///") {
comments.push(text.trim_start_matches("///").trim().to_string());
prev = p.prev_named_sibling();
} else {
break;
}
} else {
break;
}
}
if comments.is_empty() {
return None;
}
comments.reverse();
Some(comments.join("\n"))
}
fn first_line(state: &ExtractionState, node: TsNode<'_>) -> Option<String> {
let text = state.node_text(node);
text.lines().next().map(|l| l.trim().to_string())
}
}
impl crate::extraction::LanguageExtractor for FSharpExtractor {
fn extensions(&self) -> &[&str] {
&["fs", "fsi", "fsx"]
}
fn language_name(&self) -> &'static str {
"F#"
}
fn extract(&self, file_path: &str, source: &str) -> ExtractionResult {
Self::extract_fsharp(file_path, source)
}
}