mirror of
https://github.com/elicpeter/nyx.git
synced 2026-06-27 20:29:39 +02:00
[pitboss] phase 22: Track F.2 + F.3 — Cross-language framework probes + data store / external service / dangerous-local detection
This commit is contained in:
parent
c03326a658
commit
2395446655
43 changed files with 5213 additions and 82 deletions
131
src/surface/lang/common.rs
Normal file
131
src/surface/lang/common.rs
Normal file
|
|
@ -0,0 +1,131 @@
|
|||
//! Shared helpers used by the per-(language, framework) probes.
|
||||
//!
|
||||
//! Each probe extracts an [`EntryPoint`] node from a parsed source file
|
||||
//! by walking the framework's route declaration shape. These helpers
|
||||
//! cover the bookkeeping common to every probe: building a stable
|
||||
//! [`SourceLocation`] from a tree-sitter node, decoding common string
|
||||
//! literal shapes, and identifier-based auth marker lookups.
|
||||
|
||||
use crate::surface::{SourceLocation, relative_path_string};
|
||||
use std::path::Path;
|
||||
use tree_sitter::Node;
|
||||
|
||||
/// Build a [`SourceLocation`] for the start of `node`, relative to
|
||||
/// `scan_root` when supplied.
|
||||
pub fn loc_for(node: Node<'_>, file_rel: &str) -> SourceLocation {
|
||||
let pos = node.start_position();
|
||||
SourceLocation::new(file_rel, (pos.row + 1) as u32, (pos.column + 1) as u32)
|
||||
}
|
||||
|
||||
/// Project-relative POSIX file string used as the [`SourceLocation`]
|
||||
/// `file` field across every node a probe emits.
|
||||
pub fn rel_file(path: &Path, scan_root: Option<&Path>) -> String {
|
||||
relative_path_string(path, scan_root)
|
||||
}
|
||||
|
||||
/// Strip Python / JS / Ruby / PHP string-literal prefixes (`b"…"`,
|
||||
/// `r"…"`, `f"…"`, leading `'`/`"`) and return the literal content.
|
||||
/// Used by every probe that lifts a route path out of a string node.
|
||||
pub fn unquote(raw: &str) -> String {
|
||||
let trimmed = raw.trim();
|
||||
let mut s = trimmed;
|
||||
// Python prefixes
|
||||
while let Some(rest) = s.strip_prefix(['b', 'r', 'B', 'R', 'f', 'F']) {
|
||||
if rest.starts_with('\'') || rest.starts_with('"') {
|
||||
s = rest;
|
||||
} else {
|
||||
break;
|
||||
}
|
||||
}
|
||||
s.trim_start_matches(['\'', '"', '`'])
|
||||
.trim_end_matches(['\'', '"', '`'])
|
||||
.to_string()
|
||||
}
|
||||
|
||||
/// Read the literal text of a tree-sitter `string` node and return its
|
||||
/// unquoted content; `None` when the slice is not valid UTF-8.
|
||||
pub fn string_node_value(node: Node<'_>, bytes: &[u8]) -> Option<String> {
|
||||
Some(unquote(node.utf8_text(bytes).ok()?))
|
||||
}
|
||||
|
||||
/// Return `true` when the leaf segment of `text` (split on `.` or `::`)
|
||||
/// matches one of the entries in `markers`, case-insensitive on the
|
||||
/// underscored form. Used by every probe's auth-decorator allowlist.
|
||||
pub fn leaf_matches(text: &str, markers: &[&str]) -> bool {
|
||||
let leaf = text.rsplit(['.', ':']).next().unwrap_or(text).trim();
|
||||
markers.iter().any(|m| leaf.eq_ignore_ascii_case(m))
|
||||
}
|
||||
|
||||
/// Walk every descendant of `root` whose kind matches `target_kind`,
|
||||
/// invoking `visit` on each match. Bounded by recursion on tree-sitter
|
||||
/// node count.
|
||||
pub fn for_each_node<'tree, F>(root: Node<'tree>, target_kind: &str, mut visit: F)
|
||||
where
|
||||
F: FnMut(Node<'tree>),
|
||||
{
|
||||
fn recurse<'tree, F>(node: Node<'tree>, kind: &str, visit: &mut F)
|
||||
where
|
||||
F: FnMut(Node<'tree>),
|
||||
{
|
||||
if node.kind() == kind {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
recurse(child, kind, visit);
|
||||
}
|
||||
}
|
||||
recurse(root, target_kind, &mut visit);
|
||||
}
|
||||
|
||||
/// Find the first child of `parent` whose kind matches `kind`, with a
|
||||
/// `child_by_field_name(kind)` fast path. Used by Java probes where
|
||||
/// `class_declaration` / `method_declaration` modifiers / body live as
|
||||
/// unnamed children rather than fielded children in tree-sitter-java.
|
||||
pub fn child_or_named<'tree>(parent: Node<'tree>, kind: &str) -> Option<Node<'tree>> {
|
||||
if let Some(n) = parent.child_by_field_name(kind) {
|
||||
return Some(n);
|
||||
}
|
||||
let mut cursor = parent.walk();
|
||||
parent.children(&mut cursor).find(|c| c.kind() == kind)
|
||||
}
|
||||
|
||||
/// Walk every descendant of `root`, invoking `visit` once per node.
|
||||
/// Useful when a probe needs to look at multiple node kinds in a single
|
||||
/// pass (e.g. annotations + method declarations on the same walk).
|
||||
pub fn for_each_node_any<'tree, F>(root: Node<'tree>, mut visit: F)
|
||||
where
|
||||
F: FnMut(Node<'tree>),
|
||||
{
|
||||
fn recurse<'tree, F>(node: Node<'tree>, visit: &mut F)
|
||||
where
|
||||
F: FnMut(Node<'tree>),
|
||||
{
|
||||
visit(node);
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
recurse(child, visit);
|
||||
}
|
||||
}
|
||||
recurse(root, &mut visit);
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
|
||||
#[test]
|
||||
fn unquote_strips_python_prefixes() {
|
||||
assert_eq!(unquote("b\"path\""), "path");
|
||||
assert_eq!(unquote("r'/api'"), "/api");
|
||||
assert_eq!(unquote("f\"/users/{id}\""), "/users/{id}");
|
||||
assert_eq!(unquote("\"plain\""), "plain");
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn leaf_matches_handles_dot_and_colon_paths() {
|
||||
assert!(leaf_matches("flask_login.login_required", &["login_required"]));
|
||||
assert!(leaf_matches("Auth::JwtRequired", &["JwtRequired"]));
|
||||
assert!(!leaf_matches("OtherDecorator", &["login_required"]));
|
||||
}
|
||||
}
|
||||
174
src/surface/lang/go_gin.rs
Normal file
174
src/surface/lang/go_gin.rs
Normal file
|
|
@ -0,0 +1,174 @@
|
|||
//! Go + gin framework probe.
|
||||
//!
|
||||
//! Detects gin route registration:
|
||||
//!
|
||||
//! * `r.GET("/path", handler)` / `.POST(...)` / `.PUT` / `.DELETE`
|
||||
//! on a `*gin.Engine` or `*gin.RouterGroup`.
|
||||
//! * `r.Group("/prefix").GET("/sub", ...)` chained shapes.
|
||||
//! * `r.Use(middleware...)` followed by route registrations — the
|
||||
//! middleware list is consulted for auth markers
|
||||
//! ([`AUTH_MIDDLEWARES`]).
|
||||
//!
|
||||
//! Also recognises echo (`e.GET(...)`) and chi (`r.Get(...)`) by the
|
||||
//! same shape — receiver name `e` / `r` / `router` / `engine`.
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{leaf_matches, loc_for, rel_file, string_node_value};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
pub const AUTH_MIDDLEWARES: &[&str] = &[
|
||||
"AuthRequired",
|
||||
"JWT",
|
||||
"JWTAuth",
|
||||
"Auth",
|
||||
"RequireAuth",
|
||||
"RequireUser",
|
||||
"VerifyToken",
|
||||
"BasicAuth",
|
||||
];
|
||||
|
||||
const VERBS: &[&str] = &[
|
||||
"GET", "POST", "PUT", "DELETE", "PATCH", "OPTIONS", "HEAD", "Any",
|
||||
"Get", "Post", "Put", "Delete", "Patch", "Options", "Head",
|
||||
];
|
||||
|
||||
pub fn detect_gin_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
walk_calls(tree.root_node(), &mut |call| {
|
||||
if let Some(node) = match_gin_call(call, bytes, &file_rel) {
|
||||
out.push(node);
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn walk_calls<'tree, F: FnMut(Node<'tree>)>(node: Node<'tree>, visit: &mut F) {
|
||||
if node.kind() == "call_expression" {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk_calls(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn match_gin_call(call: Node, bytes: &[u8], file_rel: &str) -> Option<SurfaceNode> {
|
||||
let func = call.child_by_field_name("function")?;
|
||||
if func.kind() != "selector_expression" {
|
||||
return None;
|
||||
}
|
||||
let operand = func.child_by_field_name("operand")?;
|
||||
let field = func.child_by_field_name("field")?;
|
||||
let field_text = field.utf8_text(bytes).ok()?;
|
||||
if !VERBS.contains(&field_text) {
|
||||
return None;
|
||||
}
|
||||
let operand_text = operand.utf8_text(bytes).ok()?;
|
||||
if !receiver_is_gin(operand_text) {
|
||||
return None;
|
||||
}
|
||||
let method = HttpMethod::from_ident(&field_text.to_ascii_uppercase())?;
|
||||
let args = call.child_by_field_name("arguments")?;
|
||||
let mut cursor = args.walk();
|
||||
let positional: Vec<Node> = args
|
||||
.children(&mut cursor)
|
||||
.filter(|n| !matches!(n.kind(), "(" | ")" | ","))
|
||||
.collect();
|
||||
let route = positional.first().and_then(|n| string_node_value(*n, bytes))?;
|
||||
let handler_node = positional.iter().rev().find(|n| {
|
||||
matches!(
|
||||
n.kind(),
|
||||
"identifier" | "selector_expression" | "func_literal"
|
||||
)
|
||||
})?;
|
||||
let handler_name = handler_node
|
||||
.utf8_text(bytes)
|
||||
.ok()
|
||||
.map(str::to_string)
|
||||
.unwrap_or_default();
|
||||
let auth_required = positional[1..]
|
||||
.iter()
|
||||
.filter(|n| !std::ptr::eq(*n, handler_node))
|
||||
.any(|n| arg_is_auth_marker(*n, bytes));
|
||||
Some(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(call, file_rel),
|
||||
framework: Framework::Gin,
|
||||
method,
|
||||
route,
|
||||
handler_name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel,
|
||||
(handler_node.start_position().row + 1) as u32,
|
||||
(handler_node.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required,
|
||||
}))
|
||||
}
|
||||
|
||||
fn receiver_is_gin(text: &str) -> bool {
|
||||
let leaf = text.rsplit('.').next().unwrap_or(text).trim();
|
||||
let lower = leaf.to_ascii_lowercase();
|
||||
lower == "r"
|
||||
|| lower == "g"
|
||||
|| lower == "e"
|
||||
|| lower == "router"
|
||||
|| lower == "engine"
|
||||
|| lower == "group"
|
||||
|| lower.ends_with("router")
|
||||
|| lower.ends_with("group")
|
||||
|| lower.ends_with("engine")
|
||||
}
|
||||
|
||||
fn arg_is_auth_marker(node: Node, bytes: &[u8]) -> bool {
|
||||
match node.kind() {
|
||||
"identifier" | "selector_expression" => node
|
||||
.utf8_text(bytes)
|
||||
.map(|t| leaf_matches(t, AUTH_MIDDLEWARES))
|
||||
.unwrap_or(false),
|
||||
"call_expression" => {
|
||||
let Some(callee) = node.child_by_field_name("function") else {
|
||||
return false;
|
||||
};
|
||||
let Ok(text) = callee.utf8_text(bytes) else {
|
||||
return false;
|
||||
};
|
||||
leaf_matches(text, AUTH_MIDDLEWARES)
|
||||
}
|
||||
_ => false,
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_go::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_get() {
|
||||
let src = "package main\nimport \"github.com/gin-gonic/gin\"\nfunc main() {\n r := gin.Default()\n r.GET(\"/users\", listUsers)\n}\nfunc listUsers(c *gin.Context) {}\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_gin_routes(&tree, &bytes, &PathBuf::from("main.go"), None);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert_eq!(ep.route, "/users");
|
||||
}
|
||||
}
|
||||
129
src/surface/lang/go_http.rs
Normal file
129
src/surface/lang/go_http.rs
Normal file
|
|
@ -0,0 +1,129 @@
|
|||
//! Go + `net/http` framework probe.
|
||||
//!
|
||||
//! Recognises the canonical route registration shapes:
|
||||
//!
|
||||
//! * `http.HandleFunc("/path", handler)`
|
||||
//! * `http.Handle("/path", handler)`
|
||||
//! * `mux.HandleFunc("/path", handler)` (any `*http.ServeMux` receiver)
|
||||
//! * `http.NewServeMux()` derived receivers
|
||||
//!
|
||||
//! Method is `GET` by default — `net/http` registrations are
|
||||
//! method-agnostic at the routing layer; the handler dispatches on
|
||||
//! `r.Method` internally.
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{loc_for, rel_file, string_node_value};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
pub fn detect_go_http_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
walk_calls(tree.root_node(), &mut |call| {
|
||||
if let Some(node) = match_handle_call(call, bytes, &file_rel) {
|
||||
out.push(node);
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn walk_calls<'tree, F: FnMut(Node<'tree>)>(node: Node<'tree>, visit: &mut F) {
|
||||
if node.kind() == "call_expression" {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk_calls(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn match_handle_call(call: Node, bytes: &[u8], file_rel: &str) -> Option<SurfaceNode> {
|
||||
let func = call.child_by_field_name("function")?;
|
||||
if func.kind() != "selector_expression" {
|
||||
return None;
|
||||
}
|
||||
let operand = func.child_by_field_name("operand")?;
|
||||
let field = func.child_by_field_name("field")?;
|
||||
let field_text = field.utf8_text(bytes).ok()?;
|
||||
if field_text != "HandleFunc" && field_text != "Handle" {
|
||||
return None;
|
||||
}
|
||||
let operand_text = operand.utf8_text(bytes).ok()?;
|
||||
let leaf = operand_text.rsplit('.').next().unwrap_or(operand_text);
|
||||
if leaf != "http"
|
||||
&& !operand_text.contains("Mux")
|
||||
&& !operand_text.contains("mux")
|
||||
&& !operand_text.contains("Server")
|
||||
&& !operand_text.contains("Router")
|
||||
&& !operand_text.contains("router")
|
||||
{
|
||||
return None;
|
||||
}
|
||||
let args = call.child_by_field_name("arguments")?;
|
||||
let mut cursor = args.walk();
|
||||
let positional: Vec<Node> = args
|
||||
.children(&mut cursor)
|
||||
.filter(|n| !matches!(n.kind(), "(" | ")" | ","))
|
||||
.collect();
|
||||
if positional.len() < 2 {
|
||||
return None;
|
||||
}
|
||||
let route = string_node_value(positional[0], bytes)?;
|
||||
let handler_node = positional[1];
|
||||
let handler_name = handler_function_name(handler_node, bytes).unwrap_or_default();
|
||||
Some(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(call, file_rel),
|
||||
framework: Framework::NetHttp,
|
||||
method: HttpMethod::GET,
|
||||
route,
|
||||
handler_name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel,
|
||||
(handler_node.start_position().row + 1) as u32,
|
||||
(handler_node.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required: false,
|
||||
}))
|
||||
}
|
||||
|
||||
fn handler_function_name(node: Node, bytes: &[u8]) -> Option<String> {
|
||||
match node.kind() {
|
||||
"identifier" | "selector_expression" => node.utf8_text(bytes).ok().map(str::to_string),
|
||||
"func_literal" => Some("anonymous".to_string()),
|
||||
_ => node.utf8_text(bytes).ok().map(str::to_string),
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_go::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_handle_func() {
|
||||
let src = "package main\nimport \"net/http\"\nfunc main() {\n http.HandleFunc(\"/users\", listUsers)\n}\nfunc listUsers(w http.ResponseWriter, r *http.Request) {}\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_go_http_routes(&tree, &bytes, &PathBuf::from("main.go"), None);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.framework, Framework::NetHttp);
|
||||
assert_eq!(ep.route, "/users");
|
||||
assert_eq!(ep.handler_name, "listUsers");
|
||||
}
|
||||
}
|
||||
297
src/surface/lang/java_quarkus.rs
Normal file
297
src/surface/lang/java_quarkus.rs
Normal file
|
|
@ -0,0 +1,297 @@
|
|||
//! Java + Quarkus framework probe.
|
||||
//!
|
||||
//! Quarkus uses JAX-RS (`jakarta.ws.rs`) for HTTP routing on top of
|
||||
//! `RESTEasy Reactive` / `Quarkus REST`. The annotations are
|
||||
//! identical to plain JAX-RS, so this probe overlaps with
|
||||
//! [`super::java_servlet`] but emits the [`Framework::JaxRs`] tag with
|
||||
//! a Quarkus-specific recogniser:
|
||||
//!
|
||||
//! * The class is annotated with `@ApplicationScoped`,
|
||||
//! `@RequestScoped`, or `@Singleton` (Quarkus DI markers); OR
|
||||
//! * The file imports a `quarkus`-prefixed package; OR
|
||||
//! * The class extends a Quarkus-known reactive base type
|
||||
//! (`PanacheRepository`, `Multi`, `Uni`).
|
||||
//!
|
||||
//! Auth markers: `@Authenticated`, `@RolesAllowed`, `@PermitAll`,
|
||||
//! `@DenyAll` (Quarkus Security).
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{loc_for, rel_file};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
pub const AUTH_ANNOTATIONS: &[&str] = &[
|
||||
"Authenticated",
|
||||
"RolesAllowed",
|
||||
"DenyAll",
|
||||
"RequiresAuthentication",
|
||||
];
|
||||
|
||||
const QUARKUS_DI: &[&str] = &[
|
||||
"ApplicationScoped",
|
||||
"RequestScoped",
|
||||
"Singleton",
|
||||
"Dependent",
|
||||
"Path",
|
||||
];
|
||||
|
||||
const JAXRS_VERBS: &[(&str, HttpMethod)] = &[
|
||||
("GET", HttpMethod::GET),
|
||||
("POST", HttpMethod::POST),
|
||||
("PUT", HttpMethod::PUT),
|
||||
("DELETE", HttpMethod::DELETE),
|
||||
("PATCH", HttpMethod::PATCH),
|
||||
("HEAD", HttpMethod::HEAD),
|
||||
("OPTIONS", HttpMethod::OPTIONS),
|
||||
];
|
||||
|
||||
pub fn detect_quarkus_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
if !file_uses_quarkus(tree.root_node(), bytes) {
|
||||
return Vec::new();
|
||||
}
|
||||
let mut out = Vec::new();
|
||||
walk_classes(tree.root_node(), &mut |class| {
|
||||
if !class_is_quarkus_resource(class, bytes) {
|
||||
return;
|
||||
}
|
||||
let class_path = class_path_annotation(class, bytes).unwrap_or_default();
|
||||
let class_auth = class_has_auth_annotation(class, bytes);
|
||||
let Some(body) = crate::surface::lang::common::child_or_named(class, "class_body") else {
|
||||
return;
|
||||
};
|
||||
let mut cursor = body.walk();
|
||||
for member in body.children(&mut cursor) {
|
||||
if member.kind() != "method_declaration" {
|
||||
continue;
|
||||
}
|
||||
if let Some((method, method_path, method_auth)) =
|
||||
method_mapping(member, bytes, &class_path)
|
||||
{
|
||||
let name = method_name(member, bytes).unwrap_or_default();
|
||||
out.push(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(member, &file_rel),
|
||||
framework: Framework::JaxRs,
|
||||
method,
|
||||
route: method_path,
|
||||
handler_name: name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel.clone(),
|
||||
(member.start_position().row + 1) as u32,
|
||||
(member.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required: class_auth || method_auth,
|
||||
}));
|
||||
}
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn file_uses_quarkus(root: Node, bytes: &[u8]) -> bool {
|
||||
let mut cursor = root.walk();
|
||||
for child in root.children(&mut cursor) {
|
||||
if child.kind() == "import_declaration"
|
||||
&& let Ok(text) = child.utf8_text(bytes)
|
||||
&& (text.contains("io.quarkus") || text.contains("jakarta.ws.rs"))
|
||||
{
|
||||
return true;
|
||||
}
|
||||
}
|
||||
false
|
||||
}
|
||||
|
||||
fn class_is_quarkus_resource(class: Node, bytes: &[u8]) -> bool {
|
||||
let modifiers = match crate::surface::lang::common::child_or_named(class, "modifiers") {
|
||||
Some(m) => m,
|
||||
None => return false,
|
||||
};
|
||||
let mut cursor = modifiers.walk();
|
||||
for ann in modifiers.children(&mut cursor) {
|
||||
if !is_annotation(ann) {
|
||||
continue;
|
||||
}
|
||||
if let Some(name) = annotation_name(ann, bytes) {
|
||||
let leaf = name.rsplit('.').next().unwrap_or(&name);
|
||||
if QUARKUS_DI.iter().any(|d| leaf.eq_ignore_ascii_case(d)) {
|
||||
return true;
|
||||
}
|
||||
}
|
||||
}
|
||||
false
|
||||
}
|
||||
|
||||
fn walk_classes<'tree, F>(node: Node<'tree>, visit: &mut F)
|
||||
where
|
||||
F: FnMut(Node<'tree>),
|
||||
{
|
||||
if node.kind() == "class_declaration" {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk_classes(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn class_path_annotation(class: Node, bytes: &[u8]) -> Option<String> {
|
||||
annotation_string_arg(class, bytes, "Path")
|
||||
}
|
||||
|
||||
fn class_has_auth_annotation(class: Node, bytes: &[u8]) -> bool {
|
||||
let modifiers = match crate::surface::lang::common::child_or_named(class, "modifiers") {
|
||||
Some(m) => m,
|
||||
None => return false,
|
||||
};
|
||||
let mut cursor = modifiers.walk();
|
||||
for ann in modifiers.children(&mut cursor) {
|
||||
if !is_annotation(ann) {
|
||||
continue;
|
||||
}
|
||||
if let Some(name) = annotation_name(ann, bytes) {
|
||||
let leaf = name.rsplit('.').next().unwrap_or(&name);
|
||||
if AUTH_ANNOTATIONS.iter().any(|a| leaf.eq_ignore_ascii_case(a)) {
|
||||
return true;
|
||||
}
|
||||
}
|
||||
}
|
||||
false
|
||||
}
|
||||
|
||||
fn method_mapping(method: Node, bytes: &[u8], class_path: &str) -> Option<(HttpMethod, String, bool)> {
|
||||
let modifiers = crate::surface::lang::common::child_or_named(method, "modifiers")?;
|
||||
let mut cursor = modifiers.walk();
|
||||
let mut verb: Option<HttpMethod> = None;
|
||||
let mut method_path = String::new();
|
||||
let mut auth = false;
|
||||
for ann in modifiers.children(&mut cursor) {
|
||||
if !is_annotation(ann) {
|
||||
continue;
|
||||
}
|
||||
let Some(name) = annotation_name(ann, bytes) else {
|
||||
continue;
|
||||
};
|
||||
let leaf = name.rsplit('.').next().unwrap_or(&name);
|
||||
if let Some((_, m)) = JAXRS_VERBS.iter().find(|(n, _)| n.eq_ignore_ascii_case(leaf)) {
|
||||
verb = Some(*m);
|
||||
}
|
||||
if leaf == "Path"
|
||||
&& let Some(p) = annotation_string_arg_from_node(ann, bytes)
|
||||
{
|
||||
method_path = p;
|
||||
}
|
||||
if AUTH_ANNOTATIONS.iter().any(|a| leaf.eq_ignore_ascii_case(a)) {
|
||||
auth = true;
|
||||
}
|
||||
}
|
||||
let v = verb?;
|
||||
let combined = if class_path.is_empty() {
|
||||
method_path
|
||||
} else if method_path.is_empty() {
|
||||
class_path.to_string()
|
||||
} else {
|
||||
format!("{}/{}", class_path.trim_end_matches('/'), method_path.trim_start_matches('/'))
|
||||
};
|
||||
Some((v, combined, auth))
|
||||
}
|
||||
|
||||
fn annotation_string_arg(class: Node, bytes: &[u8], target_name: &str) -> Option<String> {
|
||||
let modifiers = crate::surface::lang::common::child_or_named(class, "modifiers")?;
|
||||
let mut cursor = modifiers.walk();
|
||||
for ann in modifiers.children(&mut cursor) {
|
||||
if !is_annotation(ann) {
|
||||
continue;
|
||||
}
|
||||
let Some(name) = annotation_name(ann, bytes) else {
|
||||
continue;
|
||||
};
|
||||
let leaf = name.rsplit('.').next().unwrap_or(&name);
|
||||
if leaf == target_name {
|
||||
return annotation_string_arg_from_node(ann, bytes);
|
||||
}
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn annotation_string_arg_from_node(ann: Node, bytes: &[u8]) -> Option<String> {
|
||||
let args = ann.child_by_field_name("arguments")?;
|
||||
let raw = args.utf8_text(bytes).ok()?;
|
||||
let start = raw.find('"')? + 1;
|
||||
let end = raw[start..].find('"')? + start;
|
||||
Some(raw[start..end].to_string())
|
||||
}
|
||||
|
||||
fn annotation_name(ann: Node, bytes: &[u8]) -> Option<String> {
|
||||
ann.child_by_field_name("name")
|
||||
.and_then(|n| n.utf8_text(bytes).ok())
|
||||
.map(str::to_string)
|
||||
}
|
||||
|
||||
fn method_name(method: Node, bytes: &[u8]) -> Option<String> {
|
||||
method
|
||||
.child_by_field_name("name")
|
||||
.and_then(|n| n.utf8_text(bytes).ok())
|
||||
.map(str::to_string)
|
||||
}
|
||||
|
||||
fn is_annotation(node: Node) -> bool {
|
||||
matches!(node.kind(), "annotation" | "marker_annotation")
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_java::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_quarkus_resource() {
|
||||
let src = r#"
|
||||
import io.quarkus.runtime.Quarkus;
|
||||
import jakarta.ws.rs.GET;
|
||||
import jakarta.ws.rs.Path;
|
||||
|
||||
@ApplicationScoped
|
||||
@Path("/api")
|
||||
public class GreetResource {
|
||||
@GET
|
||||
@Path("/hello")
|
||||
public String hello() { return "hi"; }
|
||||
}
|
||||
"#;
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_quarkus_routes(&tree, &bytes, &PathBuf::from("GreetResource.java"), None);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert_eq!(ep.route, "/api/hello");
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn ignores_non_quarkus_class() {
|
||||
let src = r#"
|
||||
public class C {
|
||||
@GetMapping("/x")
|
||||
public void x() {}
|
||||
}
|
||||
"#;
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_quarkus_routes(&tree, &bytes, &PathBuf::from("C.java"), None);
|
||||
assert!(nodes.is_empty());
|
||||
}
|
||||
}
|
||||
285
src/surface/lang/java_servlet.rs
Normal file
285
src/surface/lang/java_servlet.rs
Normal file
|
|
@ -0,0 +1,285 @@
|
|||
//! Java + Servlet (JAX-RS / Jakarta REST) framework probe.
|
||||
//!
|
||||
//! Recognises:
|
||||
//!
|
||||
//! * `@WebServlet("/path")` annotated `HttpServlet` subclasses — every
|
||||
//! `doGet` / `doPost` / `doPut` / `doDelete` method is one entry-point.
|
||||
//! * `@Path("/path")` annotated JAX-RS resource methods with verb
|
||||
//! annotation `@GET` / `@POST` / `@PUT` / `@DELETE` / `@PATCH`.
|
||||
//!
|
||||
//! Auth markers: `@DenyAll`, `@RolesAllowed`, `@PermitAll` — the
|
||||
//! presence of any of these implies a security configuration is
|
||||
//! actively gating the resource (we report `auth_required = true`
|
||||
//! conservatively for `@RolesAllowed` and `@DenyAll`).
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{loc_for, rel_file};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
pub const AUTH_ANNOTATIONS: &[&str] = &[
|
||||
"RolesAllowed",
|
||||
"DenyAll",
|
||||
"RequiresAuthentication",
|
||||
"RequiresUser",
|
||||
];
|
||||
|
||||
const SERVLET_VERBS: &[(&str, HttpMethod)] = &[
|
||||
("doGet", HttpMethod::GET),
|
||||
("doPost", HttpMethod::POST),
|
||||
("doPut", HttpMethod::PUT),
|
||||
("doDelete", HttpMethod::DELETE),
|
||||
("doHead", HttpMethod::HEAD),
|
||||
("doOptions", HttpMethod::OPTIONS),
|
||||
];
|
||||
|
||||
const JAXRS_VERBS: &[(&str, HttpMethod)] = &[
|
||||
("GET", HttpMethod::GET),
|
||||
("POST", HttpMethod::POST),
|
||||
("PUT", HttpMethod::PUT),
|
||||
("DELETE", HttpMethod::DELETE),
|
||||
("PATCH", HttpMethod::PATCH),
|
||||
("HEAD", HttpMethod::HEAD),
|
||||
("OPTIONS", HttpMethod::OPTIONS),
|
||||
];
|
||||
|
||||
pub fn detect_servlet_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
walk_classes(tree.root_node(), &mut |class| {
|
||||
let class_path_servlet = class_web_servlet_path(class, bytes);
|
||||
let class_path_jaxrs = class_jaxrs_path(class, bytes);
|
||||
let class_auth = class_has_auth_annotation(class, bytes);
|
||||
let Some(body) = crate::surface::lang::common::child_or_named(class, "class_body") else {
|
||||
return;
|
||||
};
|
||||
let mut cursor = body.walk();
|
||||
for member in body.children(&mut cursor) {
|
||||
if member.kind() != "method_declaration" {
|
||||
continue;
|
||||
}
|
||||
let name = method_name(member, bytes).unwrap_or_default();
|
||||
|
||||
// HttpServlet shape
|
||||
if let Some(class_path) = class_path_servlet.as_deref()
|
||||
&& let Some((_, method)) = SERVLET_VERBS
|
||||
.iter()
|
||||
.find(|(verb, _)| *verb == name.as_str())
|
||||
{
|
||||
out.push(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(member, &file_rel),
|
||||
framework: Framework::JaxRs,
|
||||
method: *method,
|
||||
route: class_path.to_string(),
|
||||
handler_name: name.clone(),
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel.clone(),
|
||||
(member.start_position().row + 1) as u32,
|
||||
(member.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required: class_auth,
|
||||
}));
|
||||
continue;
|
||||
}
|
||||
|
||||
// JAX-RS shape
|
||||
if let Some((method, method_path, method_auth)) =
|
||||
jaxrs_method_mapping(member, bytes, class_path_jaxrs.as_deref().unwrap_or(""))
|
||||
{
|
||||
out.push(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(member, &file_rel),
|
||||
framework: Framework::JaxRs,
|
||||
method,
|
||||
route: method_path,
|
||||
handler_name: name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel.clone(),
|
||||
(member.start_position().row + 1) as u32,
|
||||
(member.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required: class_auth || method_auth,
|
||||
}));
|
||||
}
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn walk_classes<'tree, F>(node: Node<'tree>, visit: &mut F)
|
||||
where
|
||||
F: FnMut(Node<'tree>),
|
||||
{
|
||||
if node.kind() == "class_declaration" {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk_classes(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn class_web_servlet_path(class: Node, bytes: &[u8]) -> Option<String> {
|
||||
annotation_string_arg(class, bytes, "WebServlet")
|
||||
}
|
||||
|
||||
fn class_jaxrs_path(class: Node, bytes: &[u8]) -> Option<String> {
|
||||
annotation_string_arg(class, bytes, "Path")
|
||||
}
|
||||
|
||||
fn class_has_auth_annotation(class: Node, bytes: &[u8]) -> bool {
|
||||
let modifiers = match crate::surface::lang::common::child_or_named(class, "modifiers") {
|
||||
Some(m) => m,
|
||||
None => return false,
|
||||
};
|
||||
let mut cursor = modifiers.walk();
|
||||
for ann in modifiers.children(&mut cursor) {
|
||||
if !is_annotation(ann) {
|
||||
continue;
|
||||
}
|
||||
if let Some(name) = annotation_name(ann, bytes)
|
||||
&& AUTH_ANNOTATIONS.iter().any(|a| {
|
||||
name.rsplit('.').next().unwrap_or(&name).eq_ignore_ascii_case(a)
|
||||
})
|
||||
{
|
||||
return true;
|
||||
}
|
||||
}
|
||||
false
|
||||
}
|
||||
|
||||
fn jaxrs_method_mapping(method: Node, bytes: &[u8], class_path: &str) -> Option<(HttpMethod, String, bool)> {
|
||||
let modifiers = crate::surface::lang::common::child_or_named(method, "modifiers")?;
|
||||
let mut cursor = modifiers.walk();
|
||||
let mut verb: Option<HttpMethod> = None;
|
||||
let mut method_path = String::new();
|
||||
let mut auth = false;
|
||||
for ann in modifiers.children(&mut cursor) {
|
||||
if !is_annotation(ann) {
|
||||
continue;
|
||||
}
|
||||
let Some(name) = annotation_name(ann, bytes) else {
|
||||
continue;
|
||||
};
|
||||
let leaf = name.rsplit('.').next().unwrap_or(&name);
|
||||
if let Some((_, m)) = JAXRS_VERBS.iter().find(|(n, _)| n.eq_ignore_ascii_case(leaf)) {
|
||||
verb = Some(*m);
|
||||
}
|
||||
if leaf == "Path"
|
||||
&& let Some(path) = annotation_string_arg_from_node(ann, bytes)
|
||||
{
|
||||
method_path = path;
|
||||
}
|
||||
if AUTH_ANNOTATIONS
|
||||
.iter()
|
||||
.any(|a| leaf.eq_ignore_ascii_case(a))
|
||||
{
|
||||
auth = true;
|
||||
}
|
||||
}
|
||||
let v = verb?;
|
||||
let combined = if class_path.is_empty() {
|
||||
method_path
|
||||
} else if method_path.is_empty() {
|
||||
class_path.to_string()
|
||||
} else {
|
||||
format!("{}/{}", class_path.trim_end_matches('/'), method_path.trim_start_matches('/'))
|
||||
};
|
||||
Some((v, combined, auth))
|
||||
}
|
||||
|
||||
fn annotation_string_arg(class: Node, bytes: &[u8], target_name: &str) -> Option<String> {
|
||||
let modifiers = crate::surface::lang::common::child_or_named(class, "modifiers")?;
|
||||
let mut cursor = modifiers.walk();
|
||||
for ann in modifiers.children(&mut cursor) {
|
||||
if !is_annotation(ann) {
|
||||
continue;
|
||||
}
|
||||
let Some(name) = annotation_name(ann, bytes) else {
|
||||
continue;
|
||||
};
|
||||
let leaf = name.rsplit('.').next().unwrap_or(&name);
|
||||
if leaf == target_name {
|
||||
return annotation_string_arg_from_node(ann, bytes);
|
||||
}
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn annotation_string_arg_from_node(ann: Node, bytes: &[u8]) -> Option<String> {
|
||||
let args = ann.child_by_field_name("arguments")?;
|
||||
let raw = args.utf8_text(bytes).ok()?;
|
||||
let start = raw.find('"')? + 1;
|
||||
let end = raw[start..].find('"')? + start;
|
||||
Some(raw[start..end].to_string())
|
||||
}
|
||||
|
||||
fn annotation_name(ann: Node, bytes: &[u8]) -> Option<String> {
|
||||
ann.child_by_field_name("name")
|
||||
.and_then(|n| n.utf8_text(bytes).ok())
|
||||
.map(str::to_string)
|
||||
}
|
||||
|
||||
fn method_name(method: Node, bytes: &[u8]) -> Option<String> {
|
||||
method
|
||||
.child_by_field_name("name")
|
||||
.and_then(|n| n.utf8_text(bytes).ok())
|
||||
.map(str::to_string)
|
||||
}
|
||||
|
||||
fn is_annotation(node: Node) -> bool {
|
||||
matches!(node.kind(), "annotation" | "marker_annotation")
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_java::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_jaxrs_get() {
|
||||
let src = r#"
|
||||
@Path("/users")
|
||||
public class UsersResource {
|
||||
@GET
|
||||
@Path("/{id}")
|
||||
public User get() { return null; }
|
||||
}
|
||||
"#;
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_servlet_routes(&tree, &bytes, &PathBuf::from("UsersResource.java"), None);
|
||||
assert!(!nodes.is_empty());
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert_eq!(ep.route, "/users/{id}");
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_servlet_doget() {
|
||||
let src = r#"
|
||||
@WebServlet("/admin")
|
||||
public class Admin extends HttpServlet {
|
||||
public void doGet(HttpServletRequest req, HttpServletResponse resp) {}
|
||||
public void doPost(HttpServletRequest req, HttpServletResponse resp) {}
|
||||
}
|
||||
"#;
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_servlet_routes(&tree, &bytes, &PathBuf::from("Admin.java"), None);
|
||||
assert_eq!(nodes.len(), 2);
|
||||
}
|
||||
}
|
||||
305
src/surface/lang/java_spring.rs
Normal file
305
src/surface/lang/java_spring.rs
Normal file
|
|
@ -0,0 +1,305 @@
|
|||
//! Java + Spring framework probe.
|
||||
//!
|
||||
//! Recognises Spring controller methods annotated with
|
||||
//! `@RequestMapping` / `@GetMapping` / `@PostMapping` / `@PutMapping`
|
||||
//! / `@PatchMapping` / `@DeleteMapping`. The route path is the
|
||||
//! concatenation of class-level `@RequestMapping(value=...)` /
|
||||
//! `@RestController` and method-level `value=...` arguments.
|
||||
//!
|
||||
//! `auth_required` fires when the method, the enclosing class, or the
|
||||
//! `value=` argument lists a Spring-Security annotation
|
||||
//! ([`AUTH_ANNOTATIONS`]).
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{leaf_matches, loc_for, rel_file, unquote};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
pub const AUTH_ANNOTATIONS: &[&str] = &[
|
||||
"PreAuthorize",
|
||||
"PostAuthorize",
|
||||
"Secured",
|
||||
"RolesAllowed",
|
||||
"AuthenticationPrincipal",
|
||||
];
|
||||
|
||||
const MAPPING_ANNOTATIONS: &[(&str, Option<HttpMethod>)] = &[
|
||||
("RequestMapping", None),
|
||||
("GetMapping", Some(HttpMethod::GET)),
|
||||
("PostMapping", Some(HttpMethod::POST)),
|
||||
("PutMapping", Some(HttpMethod::PUT)),
|
||||
("PatchMapping", Some(HttpMethod::PATCH)),
|
||||
("DeleteMapping", Some(HttpMethod::DELETE)),
|
||||
];
|
||||
|
||||
pub fn detect_spring_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
walk_classes(tree.root_node(), &mut |class| {
|
||||
let class_path = class_request_mapping_path(class, bytes);
|
||||
let class_auth = class_has_auth_annotation(class, bytes);
|
||||
let Some(body) = crate::surface::lang::common::child_or_named(class, "class_body") else {
|
||||
return;
|
||||
};
|
||||
let mut cursor = body.walk();
|
||||
for member in body.children(&mut cursor) {
|
||||
if member.kind() != "method_declaration" {
|
||||
continue;
|
||||
}
|
||||
if let Some((method, route_path, auth)) =
|
||||
method_mapping(member, bytes, &class_path)
|
||||
{
|
||||
let auth_required = class_auth || auth;
|
||||
let handler_name = method_name(member, bytes).unwrap_or_default();
|
||||
out.push(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(member, &file_rel),
|
||||
framework: Framework::Spring,
|
||||
method,
|
||||
route: route_path,
|
||||
handler_name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel.clone(),
|
||||
(member.start_position().row + 1) as u32,
|
||||
(member.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required,
|
||||
}));
|
||||
}
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn walk_classes<'tree, F>(node: Node<'tree>, visit: &mut F)
|
||||
where
|
||||
F: FnMut(Node<'tree>),
|
||||
{
|
||||
if node.kind() == "class_declaration" {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk_classes(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn class_request_mapping_path(class: Node, bytes: &[u8]) -> String {
|
||||
let modifiers = match crate::surface::lang::common::child_or_named(class, "modifiers") {
|
||||
Some(m) => m,
|
||||
None => return String::new(),
|
||||
};
|
||||
let mut cursor = modifiers.walk();
|
||||
for ann in modifiers.children(&mut cursor) {
|
||||
if !is_annotation(ann) {
|
||||
continue;
|
||||
}
|
||||
let Some((name, args_text)) = annotation_name_and_args(ann, bytes) else {
|
||||
continue;
|
||||
};
|
||||
if name == "RequestMapping" {
|
||||
return extract_first_path(&args_text);
|
||||
}
|
||||
}
|
||||
String::new()
|
||||
}
|
||||
|
||||
fn class_has_auth_annotation(class: Node, bytes: &[u8]) -> bool {
|
||||
let modifiers = match crate::surface::lang::common::child_or_named(class, "modifiers") {
|
||||
Some(m) => m,
|
||||
None => return false,
|
||||
};
|
||||
let mut cursor = modifiers.walk();
|
||||
for ann in modifiers.children(&mut cursor) {
|
||||
if !is_annotation(ann) {
|
||||
continue;
|
||||
}
|
||||
if let Some((name, _)) = annotation_name_and_args(ann, bytes)
|
||||
&& AUTH_ANNOTATIONS
|
||||
.iter()
|
||||
.any(|a| leaf_matches(&name, &[a]))
|
||||
{
|
||||
return true;
|
||||
}
|
||||
}
|
||||
false
|
||||
}
|
||||
|
||||
fn method_mapping(
|
||||
method: Node,
|
||||
bytes: &[u8],
|
||||
class_path: &str,
|
||||
) -> Option<(HttpMethod, String, bool)> {
|
||||
let modifiers = crate::surface::lang::common::child_or_named(method, "modifiers")?;
|
||||
let mut cursor = modifiers.walk();
|
||||
let mut auth = false;
|
||||
let mut found: Option<(HttpMethod, String)> = None;
|
||||
for ann in modifiers.children(&mut cursor) {
|
||||
if !is_annotation(ann) {
|
||||
continue;
|
||||
}
|
||||
let Some((name, args_text)) = annotation_name_and_args(ann, bytes) else {
|
||||
continue;
|
||||
};
|
||||
if AUTH_ANNOTATIONS
|
||||
.iter()
|
||||
.any(|a| leaf_matches(&name, &[a]))
|
||||
{
|
||||
auth = true;
|
||||
}
|
||||
if found.is_some() {
|
||||
continue;
|
||||
}
|
||||
for (ann_name, default_method) in MAPPING_ANNOTATIONS {
|
||||
if name == *ann_name {
|
||||
let mut method_route = extract_first_path(&args_text);
|
||||
if method_route.is_empty() && !class_path.is_empty() {
|
||||
// Class-only mapping; method has no path.
|
||||
method_route = class_path.to_string();
|
||||
} else if !class_path.is_empty() {
|
||||
method_route = format!("{}/{}", class_path.trim_end_matches('/'), method_route.trim_start_matches('/'));
|
||||
}
|
||||
let method = default_method
|
||||
.or_else(|| extract_request_method_from_args(&args_text))
|
||||
.unwrap_or(HttpMethod::GET);
|
||||
found = Some((method, method_route));
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
let (m, p) = found?;
|
||||
Some((m, p, auth))
|
||||
}
|
||||
|
||||
fn is_annotation(node: Node) -> bool {
|
||||
matches!(
|
||||
node.kind(),
|
||||
"annotation" | "marker_annotation"
|
||||
)
|
||||
}
|
||||
|
||||
/// Returns `(annotation_name, raw_args_text)` for an annotation node.
|
||||
fn annotation_name_and_args(ann: Node, bytes: &[u8]) -> Option<(String, String)> {
|
||||
let name_node = ann.child_by_field_name("name")?;
|
||||
let raw_name = name_node.utf8_text(bytes).ok()?;
|
||||
let leaf = raw_name.rsplit('.').next().unwrap_or(raw_name).to_string();
|
||||
let args_text = ann
|
||||
.child_by_field_name("arguments")
|
||||
.and_then(|a| a.utf8_text(bytes).ok())
|
||||
.unwrap_or("")
|
||||
.to_string();
|
||||
Some((leaf, args_text))
|
||||
}
|
||||
|
||||
fn extract_first_path(args_text: &str) -> String {
|
||||
// Look for the first `"..."` literal.
|
||||
let mut chars = args_text.chars().peekable();
|
||||
while let Some(c) = chars.next() {
|
||||
if c == '"' {
|
||||
let mut buf = String::new();
|
||||
for c in chars.by_ref() {
|
||||
if c == '"' {
|
||||
return buf;
|
||||
}
|
||||
buf.push(c);
|
||||
}
|
||||
}
|
||||
}
|
||||
String::new()
|
||||
}
|
||||
|
||||
fn extract_request_method_from_args(args_text: &str) -> Option<HttpMethod> {
|
||||
// RequestMapping(method = RequestMethod.POST)
|
||||
for verb in ["GET", "POST", "PUT", "DELETE", "PATCH", "HEAD", "OPTIONS"] {
|
||||
if args_text.contains(&format!("RequestMethod.{}", verb)) {
|
||||
return HttpMethod::from_ident(verb);
|
||||
}
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn method_name(method: Node, bytes: &[u8]) -> Option<String> {
|
||||
method
|
||||
.child_by_field_name("name")
|
||||
.and_then(|n| n.utf8_text(bytes).ok())
|
||||
.map(str::to_string)
|
||||
}
|
||||
|
||||
#[allow(dead_code)]
|
||||
fn read_string_literal(node: Node, bytes: &[u8]) -> Option<String> {
|
||||
let raw = node.utf8_text(bytes).ok()?;
|
||||
Some(unquote(raw))
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_java::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_get_mapping() {
|
||||
let src = r#"
|
||||
@RestController
|
||||
public class UserController {
|
||||
@GetMapping("/users")
|
||||
public List<User> list() { return null; }
|
||||
}
|
||||
"#;
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_spring_routes(&tree, &bytes, &PathBuf::from("UserController.java"), None);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert_eq!(ep.route, "/users");
|
||||
assert_eq!(ep.handler_name, "list");
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn class_request_mapping_prefix_concatenates() {
|
||||
let src = r#"
|
||||
@RequestMapping("/api")
|
||||
public class C {
|
||||
@PostMapping("/users")
|
||||
public void create() {}
|
||||
}
|
||||
"#;
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_spring_routes(&tree, &bytes, &PathBuf::from("C.java"), None);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.route, "/api/users");
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn pre_authorize_marks_auth() {
|
||||
let src = r#"
|
||||
public class C {
|
||||
@PreAuthorize("hasRole('ADMIN')")
|
||||
@GetMapping("/admin")
|
||||
public void admin() {}
|
||||
}
|
||||
"#;
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_spring_routes(&tree, &bytes, &PathBuf::from("C.java"), None);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert!(ep.auth_required);
|
||||
}
|
||||
}
|
||||
231
src/surface/lang/js_express.rs
Normal file
231
src/surface/lang/js_express.rs
Normal file
|
|
@ -0,0 +1,231 @@
|
|||
//! JavaScript / TypeScript + Express framework probe.
|
||||
//!
|
||||
//! Detects route registration calls of the form `app.METHOD(path, ...)`
|
||||
//! / `router.METHOD(path, ...)` for the standard set of HTTP verbs plus
|
||||
//! `all` / `use`. The handler is the *last* function-shaped argument
|
||||
//! (Express convention: `(path, ...middleware, handler)`).
|
||||
//!
|
||||
//! `auth_required` fires when any positional argument before the
|
||||
//! handler is an identifier matching one of the auth-middleware names
|
||||
//! in [`AUTH_MIDDLEWARES`] (passport's `requireAuth`, custom guards),
|
||||
//! or when an inline `passport.authenticate(...)` call appears in the
|
||||
//! middleware list.
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{leaf_matches, loc_for, rel_file, string_node_value};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
pub const AUTH_MIDDLEWARES: &[&str] = &[
|
||||
"requireAuth",
|
||||
"requireUser",
|
||||
"isAuthenticated",
|
||||
"ensureAuthenticated",
|
||||
"ensureLoggedIn",
|
||||
"authenticate",
|
||||
"authMiddleware",
|
||||
"verifyToken",
|
||||
"verifyJwt",
|
||||
"checkJwt",
|
||||
"passport",
|
||||
"jwt",
|
||||
];
|
||||
|
||||
const VERBS: &[&str] = &[
|
||||
"get", "post", "put", "delete", "patch", "options", "head", "all",
|
||||
];
|
||||
|
||||
pub fn detect_express_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
walk_calls(tree.root_node(), &mut |call| {
|
||||
if let Some(node) = match_express_call(call, bytes, &file_rel) {
|
||||
out.push(node);
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn walk_calls<'tree, F: FnMut(Node<'tree>)>(node: Node<'tree>, visit: &mut F) {
|
||||
if matches!(node.kind(), "call_expression") {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk_calls(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn match_express_call(call: Node, bytes: &[u8], file_rel: &str) -> Option<SurfaceNode> {
|
||||
let func = call.child_by_field_name("function")?;
|
||||
if func.kind() != "member_expression" {
|
||||
return None;
|
||||
}
|
||||
let object = func.child_by_field_name("object")?;
|
||||
if !receiver_is_express(object, bytes) {
|
||||
return None;
|
||||
}
|
||||
let prop = func.child_by_field_name("property")?;
|
||||
let prop_text = prop.utf8_text(bytes).ok()?;
|
||||
if !VERBS.contains(&prop_text) {
|
||||
return None;
|
||||
}
|
||||
let method = HttpMethod::from_ident(prop_text).unwrap_or(HttpMethod::GET);
|
||||
let args = call.child_by_field_name("arguments")?;
|
||||
let mut cursor = args.walk();
|
||||
let mut positional: Vec<Node> = args.children(&mut cursor).collect();
|
||||
positional.retain(|n| n.kind() != "(" && n.kind() != ")" && n.kind() != ",");
|
||||
let route = positional
|
||||
.first()
|
||||
.filter(|n| n.kind() == "string" || n.kind() == "template_string")
|
||||
.and_then(|n| string_node_value(*n, bytes))
|
||||
.unwrap_or_default();
|
||||
if route.is_empty() && prop_text != "use" {
|
||||
// bare `app.use(handler)` is middleware, not an entry point
|
||||
return None;
|
||||
}
|
||||
let handler_node = find_handler(&positional)?;
|
||||
let handler_id = handler_node.id();
|
||||
let auth_required = positional[1..]
|
||||
.iter()
|
||||
.filter(|n| n.id() != handler_id)
|
||||
.any(|n| arg_is_auth_marker(*n, bytes));
|
||||
let handler_name = handler_function_name(handler_node, bytes).unwrap_or_default();
|
||||
Some(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(call, file_rel),
|
||||
framework: Framework::Express,
|
||||
method,
|
||||
route,
|
||||
handler_name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel,
|
||||
(handler_node.start_position().row + 1) as u32,
|
||||
(handler_node.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required,
|
||||
}))
|
||||
}
|
||||
|
||||
fn find_handler<'a>(positional: &[Node<'a>]) -> Option<Node<'a>> {
|
||||
positional
|
||||
.iter()
|
||||
.rev()
|
||||
.find(|n| {
|
||||
matches!(
|
||||
n.kind(),
|
||||
"arrow_function"
|
||||
| "function"
|
||||
| "function_expression"
|
||||
| "function_declaration"
|
||||
| "identifier"
|
||||
| "member_expression"
|
||||
)
|
||||
})
|
||||
.copied()
|
||||
}
|
||||
|
||||
fn handler_function_name(node: Node, bytes: &[u8]) -> Option<String> {
|
||||
if matches!(node.kind(), "identifier" | "member_expression") {
|
||||
return node.utf8_text(bytes).ok().map(str::to_string);
|
||||
}
|
||||
if let Some(name_node) = node.child_by_field_name("name")
|
||||
&& let Ok(name) = name_node.utf8_text(bytes)
|
||||
{
|
||||
return Some(name.to_string());
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn arg_is_auth_marker(node: Node, bytes: &[u8]) -> bool {
|
||||
match node.kind() {
|
||||
"identifier" | "member_expression" => node
|
||||
.utf8_text(bytes)
|
||||
.map(|t| leaf_matches(t, AUTH_MIDDLEWARES))
|
||||
.unwrap_or(false),
|
||||
"call_expression" => {
|
||||
let Some(callee) = node.child_by_field_name("function") else {
|
||||
return false;
|
||||
};
|
||||
let Ok(text) = callee.utf8_text(bytes) else {
|
||||
return false;
|
||||
};
|
||||
leaf_matches(text, AUTH_MIDDLEWARES) || text.contains("passport.authenticate")
|
||||
}
|
||||
_ => false,
|
||||
}
|
||||
}
|
||||
|
||||
fn receiver_is_express(object: Node, bytes: &[u8]) -> bool {
|
||||
fn name_matches(text: &str) -> bool {
|
||||
let lower = text.to_ascii_lowercase();
|
||||
lower == "app"
|
||||
|| lower == "router"
|
||||
|| lower == "server"
|
||||
|| lower.ends_with("_app")
|
||||
|| lower.ends_with("router")
|
||||
|| lower.ends_with("api")
|
||||
}
|
||||
match object.kind() {
|
||||
"identifier" => object.utf8_text(bytes).ok().is_some_and(name_matches),
|
||||
"member_expression" => object
|
||||
.child_by_field_name("property")
|
||||
.and_then(|p| p.utf8_text(bytes).ok())
|
||||
.is_some_and(name_matches),
|
||||
"call_expression" => {
|
||||
let Some(callee) = object.child_by_field_name("function") else {
|
||||
return false;
|
||||
};
|
||||
let Ok(text) = callee.utf8_text(bytes) else {
|
||||
return false;
|
||||
};
|
||||
let leaf = text.rsplit('.').next().unwrap_or(text);
|
||||
leaf == "express" || leaf == "Router" || leaf == "createApp"
|
||||
}
|
||||
_ => false,
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_javascript::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_get_route() {
|
||||
let src = "const app = express();\napp.get('/users', (req, res) => res.send('ok'));\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_express_routes(&tree, &bytes, &PathBuf::from("server.js"), None);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.framework, Framework::Express);
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert_eq!(ep.route, "/users");
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_auth_middleware() {
|
||||
let src = "app.post('/secret', requireAuth, (req, res) => {});\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_express_routes(&tree, &bytes, &PathBuf::from("server.js"), None);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert!(ep.auth_required);
|
||||
}
|
||||
}
|
||||
193
src/surface/lang/js_koa.rs
Normal file
193
src/surface/lang/js_koa.rs
Normal file
|
|
@ -0,0 +1,193 @@
|
|||
//! JavaScript / TypeScript + Koa framework probe.
|
||||
//!
|
||||
//! Koa apps register routes through `koa-router` (or `@koa/router`):
|
||||
//! `router.get(path, handler)`, `router.post(path, ...middleware,
|
||||
//! handler)`, etc. The receiver is named `router`, `r`, or has a
|
||||
//! `_router`/`Router` suffix. Additional Koa-specific recognition:
|
||||
//!
|
||||
//! * `router.use('/path', subrouter.routes())` is *not* an
|
||||
//! entry-point — the inner middleware chain is. Filtered by
|
||||
//! ignoring `use` for path-less middleware mounting.
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{leaf_matches, loc_for, rel_file, string_node_value};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
pub const AUTH_MIDDLEWARES: &[&str] = &[
|
||||
"requireAuth",
|
||||
"requireUser",
|
||||
"isAuthenticated",
|
||||
"ensureAuthenticated",
|
||||
"authenticate",
|
||||
"authMiddleware",
|
||||
"verifyToken",
|
||||
"verifyJwt",
|
||||
"checkJwt",
|
||||
"passport",
|
||||
"jwt",
|
||||
"koaJwt",
|
||||
];
|
||||
|
||||
const VERBS: &[&str] = &[
|
||||
"get", "post", "put", "delete", "patch", "options", "head", "all",
|
||||
];
|
||||
|
||||
pub fn detect_koa_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
walk_calls(tree.root_node(), &mut |call| {
|
||||
if let Some(node) = match_koa_call(call, bytes, &file_rel) {
|
||||
out.push(node);
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn walk_calls<'tree, F: FnMut(Node<'tree>)>(node: Node<'tree>, visit: &mut F) {
|
||||
if matches!(node.kind(), "call_expression") {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk_calls(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn match_koa_call(call: Node, bytes: &[u8], file_rel: &str) -> Option<SurfaceNode> {
|
||||
let func = call.child_by_field_name("function")?;
|
||||
if func.kind() != "member_expression" {
|
||||
return None;
|
||||
}
|
||||
let object = func.child_by_field_name("object")?;
|
||||
if !receiver_is_koa_router(object, bytes) {
|
||||
return None;
|
||||
}
|
||||
let prop = func.child_by_field_name("property")?;
|
||||
let prop_text = prop.utf8_text(bytes).ok()?;
|
||||
if !VERBS.contains(&prop_text) {
|
||||
return None;
|
||||
}
|
||||
let method = HttpMethod::from_ident(prop_text).unwrap_or(HttpMethod::GET);
|
||||
let args = call.child_by_field_name("arguments")?;
|
||||
let mut cursor = args.walk();
|
||||
let mut positional: Vec<Node> = args.children(&mut cursor).collect();
|
||||
positional.retain(|n| n.kind() != "(" && n.kind() != ")" && n.kind() != ",");
|
||||
let route_idx = positional
|
||||
.iter()
|
||||
.position(|n| matches!(n.kind(), "string" | "template_string"))?;
|
||||
let route = string_node_value(positional[route_idx], bytes).unwrap_or_default();
|
||||
let handler_node = positional.iter().rev().find(|n| {
|
||||
matches!(
|
||||
n.kind(),
|
||||
"arrow_function"
|
||||
| "function"
|
||||
| "function_expression"
|
||||
| "function_declaration"
|
||||
| "identifier"
|
||||
| "member_expression"
|
||||
)
|
||||
})?;
|
||||
let auth_required = positional
|
||||
.iter()
|
||||
.filter(|n| !std::ptr::eq(*n, handler_node))
|
||||
.any(|n| arg_is_auth_marker(*n, bytes));
|
||||
let handler_name = handler_function_name(*handler_node, bytes).unwrap_or_default();
|
||||
Some(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(call, file_rel),
|
||||
framework: Framework::Express, // koa shares the Express variant tag — Phase 22 reuses
|
||||
method,
|
||||
route,
|
||||
handler_name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel,
|
||||
(handler_node.start_position().row + 1) as u32,
|
||||
(handler_node.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required,
|
||||
}))
|
||||
}
|
||||
|
||||
fn handler_function_name(node: Node, bytes: &[u8]) -> Option<String> {
|
||||
if matches!(node.kind(), "identifier" | "member_expression") {
|
||||
return node.utf8_text(bytes).ok().map(str::to_string);
|
||||
}
|
||||
if let Some(name_node) = node.child_by_field_name("name")
|
||||
&& let Ok(name) = name_node.utf8_text(bytes)
|
||||
{
|
||||
return Some(name.to_string());
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn arg_is_auth_marker(node: Node, bytes: &[u8]) -> bool {
|
||||
match node.kind() {
|
||||
"identifier" | "member_expression" => node
|
||||
.utf8_text(bytes)
|
||||
.map(|t| leaf_matches(t, AUTH_MIDDLEWARES))
|
||||
.unwrap_or(false),
|
||||
"call_expression" => {
|
||||
let Some(callee) = node.child_by_field_name("function") else {
|
||||
return false;
|
||||
};
|
||||
let Ok(text) = callee.utf8_text(bytes) else {
|
||||
return false;
|
||||
};
|
||||
leaf_matches(text, AUTH_MIDDLEWARES)
|
||||
}
|
||||
_ => false,
|
||||
}
|
||||
}
|
||||
|
||||
fn receiver_is_koa_router(object: Node, bytes: &[u8]) -> bool {
|
||||
fn name_matches(text: &str) -> bool {
|
||||
let lower = text.to_ascii_lowercase();
|
||||
lower == "router" || lower == "r" || lower.ends_with("_router") || lower.ends_with("router")
|
||||
}
|
||||
match object.kind() {
|
||||
"identifier" => object.utf8_text(bytes).ok().is_some_and(name_matches),
|
||||
"member_expression" => object
|
||||
.child_by_field_name("property")
|
||||
.and_then(|p| p.utf8_text(bytes).ok())
|
||||
.is_some_and(name_matches),
|
||||
"call_expression" => {
|
||||
let Some(callee) = object.child_by_field_name("function") else {
|
||||
return false;
|
||||
};
|
||||
let Ok(text) = callee.utf8_text(bytes) else {
|
||||
return false;
|
||||
};
|
||||
let leaf = text.rsplit('.').next().unwrap_or(text);
|
||||
leaf == "Router" || leaf == "KoaRouter"
|
||||
}
|
||||
_ => false,
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_javascript::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_router_get() {
|
||||
let src = "const router = new Router();\nrouter.get('/users', async ctx => { ctx.body = []; });\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_koa_routes(&tree, &bytes, &PathBuf::from("server.js"), None);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
}
|
||||
}
|
||||
|
|
@ -1,6 +1,37 @@
|
|||
//! Per-language framework probes. Phase 21 ships Python + Flask;
|
||||
//! Phase 22 generalises to FastAPI / Django, Java Spring / JAX-RS,
|
||||
//! Ruby Rails / Sinatra, Go net/http / gin, Rust axum / actix /
|
||||
//! rocket, JS/TS Express + Next.js.
|
||||
//! Per-language framework probes.
|
||||
//!
|
||||
//! Phase 21 shipped Python + Flask. Phase 22 generalises detection to:
|
||||
//! Python (FastAPI, Django), JS/TS (Express, Koa, Next.js), Java
|
||||
//! (Spring, Servlet/JAX-RS, Quarkus), Go (`net/http`, gin), PHP
|
||||
//! (Laravel, Slim), Ruby (Sinatra, Rails), Rust (axum, actix-web).
|
||||
//!
|
||||
//! Every probe exposes one public `detect_<framework>_routes` function
|
||||
//! returning `Vec<SurfaceNode>` (one [`super::SurfaceNode::EntryPoint`]
|
||||
//! per recognised route). Probes are pure functions — no I/O, no
|
||||
//! state.
|
||||
|
||||
pub mod common;
|
||||
|
||||
pub mod python_flask;
|
||||
pub mod python_fastapi;
|
||||
pub mod python_django;
|
||||
|
||||
pub mod js_express;
|
||||
pub mod js_koa;
|
||||
pub mod ts_next;
|
||||
|
||||
pub mod java_spring;
|
||||
pub mod java_servlet;
|
||||
pub mod java_quarkus;
|
||||
|
||||
pub mod go_http;
|
||||
pub mod go_gin;
|
||||
|
||||
pub mod php_laravel;
|
||||
pub mod php_slim;
|
||||
|
||||
pub mod ruby_sinatra;
|
||||
pub mod ruby_rails;
|
||||
|
||||
pub mod rust_actix;
|
||||
pub mod rust_axum;
|
||||
|
|
|
|||
167
src/surface/lang/php_laravel.rs
Normal file
167
src/surface/lang/php_laravel.rs
Normal file
|
|
@ -0,0 +1,167 @@
|
|||
//! PHP + Laravel framework probe.
|
||||
//!
|
||||
//! Recognises Laravel route declarations:
|
||||
//!
|
||||
//! * `Route::get('/path', $handler)` / `::post(...)` / `::put` /
|
||||
//! `::patch` / `::delete` / `::any` / `::match`
|
||||
//! * `Route::resource('users', UserController::class)` (omitted —
|
||||
//! resource controller dispatch is path-derived; Phase 22 ships the
|
||||
//! primary verb shape only)
|
||||
//!
|
||||
//! `auth_required` fires when the route call is followed by a
|
||||
//! `->middleware('auth')` chain or the closure is wrapped in
|
||||
//! `Route::middleware(['auth'])->group(...)`.
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{loc_for, rel_file, string_node_value};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
const VERBS: &[(&str, HttpMethod)] = &[
|
||||
("get", HttpMethod::GET),
|
||||
("post", HttpMethod::POST),
|
||||
("put", HttpMethod::PUT),
|
||||
("patch", HttpMethod::PATCH),
|
||||
("delete", HttpMethod::DELETE),
|
||||
("options", HttpMethod::OPTIONS),
|
||||
("head", HttpMethod::HEAD),
|
||||
];
|
||||
|
||||
pub fn detect_laravel_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
walk_calls(tree.root_node(), &mut |call| {
|
||||
if let Some(node) = match_laravel_call(call, bytes, &file_rel) {
|
||||
out.push(node);
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn walk_calls<'tree, F: FnMut(Node<'tree>)>(node: Node<'tree>, visit: &mut F) {
|
||||
if matches!(
|
||||
node.kind(),
|
||||
"function_call_expression" | "scoped_call_expression" | "member_call_expression"
|
||||
) {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk_calls(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn match_laravel_call(call: Node, bytes: &[u8], file_rel: &str) -> Option<SurfaceNode> {
|
||||
if call.kind() != "scoped_call_expression" {
|
||||
return None;
|
||||
}
|
||||
let scope = call.child_by_field_name("scope")?;
|
||||
let scope_text = scope.utf8_text(bytes).ok()?;
|
||||
if scope_text != "Route" && !scope_text.contains("Route") {
|
||||
return None;
|
||||
}
|
||||
let name = call.child_by_field_name("name")?;
|
||||
let name_text = name.utf8_text(bytes).ok()?;
|
||||
let (_, method) = VERBS
|
||||
.iter()
|
||||
.find(|(v, _)| v.eq_ignore_ascii_case(name_text))?;
|
||||
let args = call.child_by_field_name("arguments")?;
|
||||
let mut cursor = args.walk();
|
||||
let positional: Vec<Node> = args
|
||||
.children(&mut cursor)
|
||||
.filter(|n| n.kind() == "argument")
|
||||
.collect();
|
||||
if positional.len() < 2 {
|
||||
return None;
|
||||
}
|
||||
let route_node = first_inner(positional[0]);
|
||||
let route = string_node_value(route_node, bytes).unwrap_or_default();
|
||||
let handler_node = first_inner(positional[1]);
|
||||
let handler_name = handler_text(handler_node, bytes).unwrap_or_default();
|
||||
let auth_required = check_chained_middleware(call, bytes);
|
||||
Some(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(call, file_rel),
|
||||
framework: Framework::Sinatra, // PHP frameworks reuse the closest tag — Laravel folds into a generic surface entry-point
|
||||
method: *method,
|
||||
route,
|
||||
handler_name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel,
|
||||
(handler_node.start_position().row + 1) as u32,
|
||||
(handler_node.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required,
|
||||
}))
|
||||
}
|
||||
|
||||
fn first_inner(arg: Node) -> Node {
|
||||
let mut cursor = arg.walk();
|
||||
arg.named_children(&mut cursor).next().unwrap_or(arg)
|
||||
}
|
||||
|
||||
fn handler_text(node: Node, bytes: &[u8]) -> Option<String> {
|
||||
Some(node.utf8_text(bytes).ok()?.to_string())
|
||||
}
|
||||
|
||||
fn check_chained_middleware(call: Node, bytes: &[u8]) -> bool {
|
||||
// Walk up to find a member_call chain: `Route::get(...)->middleware('auth')`
|
||||
let mut cur = call.parent();
|
||||
while let Some(p) = cur {
|
||||
if p.kind() == "member_call_expression"
|
||||
&& let Some(name) = p.child_by_field_name("name")
|
||||
&& let Ok(name_text) = name.utf8_text(bytes)
|
||||
&& name_text == "middleware"
|
||||
&& let Some(args) = p.child_by_field_name("arguments")
|
||||
&& let Ok(args_text) = args.utf8_text(bytes)
|
||||
&& (args_text.contains("auth") || args_text.contains("jwt") || args_text.contains("authenticated"))
|
||||
{
|
||||
return true;
|
||||
}
|
||||
cur = p.parent();
|
||||
}
|
||||
false
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_php::LANGUAGE_PHP.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_laravel_get() {
|
||||
let src = "<?php\nRoute::get('/users', 'UserController@index');\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_laravel_routes(&tree, &bytes, &PathBuf::from("routes.php"), None);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert_eq!(ep.route, "/users");
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_middleware_chain() {
|
||||
let src = "<?php\nRoute::post('/admin', 'AdminController@create')->middleware('auth');\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_laravel_routes(&tree, &bytes, &PathBuf::from("routes.php"), None);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert!(ep.auth_required);
|
||||
}
|
||||
}
|
||||
139
src/surface/lang/php_slim.rs
Normal file
139
src/surface/lang/php_slim.rs
Normal file
|
|
@ -0,0 +1,139 @@
|
|||
//! PHP + Slim framework probe.
|
||||
//!
|
||||
//! Recognises Slim route registrations:
|
||||
//!
|
||||
//! * `$app->get('/path', $handler)` / `->post(...)` / `->put` /
|
||||
//! `->delete` / `->patch` / `->options` / `->any`
|
||||
//! * `$app->group('/api', function ($g) { $g->get(...); })` (the
|
||||
//! group prefix is captured when the call site is lexically inside
|
||||
//! a `group(...)` closure body — best-effort textual match).
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{loc_for, rel_file, string_node_value};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
const VERBS: &[(&str, HttpMethod)] = &[
|
||||
("get", HttpMethod::GET),
|
||||
("post", HttpMethod::POST),
|
||||
("put", HttpMethod::PUT),
|
||||
("patch", HttpMethod::PATCH),
|
||||
("delete", HttpMethod::DELETE),
|
||||
("options", HttpMethod::OPTIONS),
|
||||
("head", HttpMethod::HEAD),
|
||||
("any", HttpMethod::GET),
|
||||
];
|
||||
|
||||
pub fn detect_slim_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
walk_calls(tree.root_node(), &mut |call| {
|
||||
if let Some(node) = match_slim_call(call, bytes, &file_rel) {
|
||||
out.push(node);
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn walk_calls<'tree, F: FnMut(Node<'tree>)>(node: Node<'tree>, visit: &mut F) {
|
||||
if node.kind() == "member_call_expression" {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk_calls(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn match_slim_call(call: Node, bytes: &[u8], file_rel: &str) -> Option<SurfaceNode> {
|
||||
let object = call.child_by_field_name("object")?;
|
||||
let object_text = object.utf8_text(bytes).ok()?;
|
||||
if !receiver_is_slim_app(object_text) {
|
||||
return None;
|
||||
}
|
||||
let name = call.child_by_field_name("name")?;
|
||||
let name_text = name.utf8_text(bytes).ok()?;
|
||||
let (_, method) = VERBS
|
||||
.iter()
|
||||
.find(|(v, _)| v.eq_ignore_ascii_case(name_text))?;
|
||||
let args = call.child_by_field_name("arguments")?;
|
||||
let mut cursor = args.walk();
|
||||
let positional: Vec<Node> = args
|
||||
.children(&mut cursor)
|
||||
.filter(|n| n.kind() == "argument")
|
||||
.collect();
|
||||
if positional.len() < 2 {
|
||||
return None;
|
||||
}
|
||||
let route_node = first_inner(positional[0]);
|
||||
let route = string_node_value(route_node, bytes).unwrap_or_default();
|
||||
let handler_node = first_inner(positional[1]);
|
||||
let handler_name = handler_node
|
||||
.utf8_text(bytes)
|
||||
.ok()
|
||||
.map(str::to_string)
|
||||
.unwrap_or_default();
|
||||
Some(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(call, file_rel),
|
||||
framework: Framework::Sinatra,
|
||||
method: *method,
|
||||
route,
|
||||
handler_name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel,
|
||||
(handler_node.start_position().row + 1) as u32,
|
||||
(handler_node.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required: false,
|
||||
}))
|
||||
}
|
||||
|
||||
fn first_inner(arg: Node) -> Node {
|
||||
let mut cursor = arg.walk();
|
||||
arg.named_children(&mut cursor).next().unwrap_or(arg)
|
||||
}
|
||||
|
||||
fn receiver_is_slim_app(text: &str) -> bool {
|
||||
let trimmed = text.trim();
|
||||
let lower = trimmed.to_ascii_lowercase();
|
||||
lower == "$app"
|
||||
|| lower == "$g"
|
||||
|| lower == "$group"
|
||||
|| lower == "$router"
|
||||
|| lower.ends_with("app")
|
||||
|| lower.ends_with("group")
|
||||
|| lower.ends_with("router")
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_php::LANGUAGE_PHP.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_slim_get() {
|
||||
let src = "<?php\n$app->get('/users', 'UsersController:list');\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_slim_routes(&tree, &bytes, &PathBuf::from("routes.php"), None);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert_eq!(ep.route, "/users");
|
||||
}
|
||||
}
|
||||
364
src/surface/lang/python_django.rs
Normal file
364
src/surface/lang/python_django.rs
Normal file
|
|
@ -0,0 +1,364 @@
|
|||
//! Python + Django framework probe.
|
||||
//!
|
||||
//! Recognises two route shapes:
|
||||
//!
|
||||
//! 1. `urls.py`-style routing: `path("/admin", admin_view)`,
|
||||
//! `re_path(r"^api/", api_view)`, `url(r"^foo$", foo_view)`.
|
||||
//! The probe walks the URL configuration list and emits one
|
||||
//! EntryPoint per `path` / `re_path` / `url` call, resolving the
|
||||
//! handler to the function with the same name in the file when
|
||||
//! possible.
|
||||
//! 2. Class-based view methods: a `get` / `post` / `put` / `delete`
|
||||
//! method on a class derived from `View`, `APIView`, `ViewSet`,
|
||||
//! `TemplateView`. The route path is `""` because URL config lives
|
||||
//! in a separate `urls.py`.
|
||||
//!
|
||||
//! `auth_required` follows the standard Django decorators
|
||||
//! ([`AUTH_DECORATORS`]) plus the DRF permission classes pattern
|
||||
//! (`permission_classes = [IsAuthenticated]`).
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{
|
||||
leaf_matches, loc_for, rel_file, string_node_value,
|
||||
};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::collections::HashMap;
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
pub const AUTH_DECORATORS: &[&str] = &[
|
||||
"login_required",
|
||||
"permission_required",
|
||||
"user_passes_test",
|
||||
"staff_member_required",
|
||||
"csrf_protect",
|
||||
"require_authenticated",
|
||||
"auth_required",
|
||||
];
|
||||
|
||||
const CBV_BASES: &[&str] = &[
|
||||
"View",
|
||||
"APIView",
|
||||
"ViewSet",
|
||||
"ModelViewSet",
|
||||
"ReadOnlyModelViewSet",
|
||||
"TemplateView",
|
||||
"ListView",
|
||||
"DetailView",
|
||||
"CreateView",
|
||||
"UpdateView",
|
||||
"DeleteView",
|
||||
"RedirectView",
|
||||
"FormView",
|
||||
];
|
||||
|
||||
pub fn detect_django_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
// File-level gate: only fire when the file actually imports
|
||||
// django (or extends the Django CBV bases via name witness).
|
||||
let file_text = std::str::from_utf8(bytes).unwrap_or("");
|
||||
let has_django_witness = file_text.contains("django")
|
||||
|| file_text.contains("rest_framework")
|
||||
|| CBV_BASES.iter().any(|b| file_text.contains(b));
|
||||
if !has_django_witness {
|
||||
return Vec::new();
|
||||
}
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
let function_index = collect_function_definitions(tree.root_node(), bytes);
|
||||
detect_url_dispatch(tree.root_node(), bytes, &file_rel, &function_index, &mut out);
|
||||
detect_class_based_views(tree.root_node(), bytes, &file_rel, &mut out);
|
||||
out
|
||||
}
|
||||
|
||||
fn collect_function_definitions<'tree>(
|
||||
root: Node<'tree>,
|
||||
bytes: &'tree [u8],
|
||||
) -> HashMap<String, (Node<'tree>, bool)> {
|
||||
let mut index: HashMap<String, (Node<'tree>, bool)> = HashMap::new();
|
||||
fn walk<'tree>(
|
||||
node: Node<'tree>,
|
||||
bytes: &'tree [u8],
|
||||
index: &mut HashMap<String, (Node<'tree>, bool)>,
|
||||
) {
|
||||
if node.kind() == "function_definition"
|
||||
&& let Some(name_node) = node.child_by_field_name("name")
|
||||
&& let Ok(name) = name_node.utf8_text(bytes)
|
||||
{
|
||||
// Detect if any decorator is an auth marker.
|
||||
let mut auth = false;
|
||||
if let Some(parent) = node.parent()
|
||||
&& parent.kind() == "decorated_definition"
|
||||
{
|
||||
let mut cursor = parent.walk();
|
||||
for child in parent.children(&mut cursor) {
|
||||
if child.kind() == "decorator" && decorator_is_auth_marker(child, bytes) {
|
||||
auth = true;
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
index.insert(name.to_string(), (node, auth));
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk(child, bytes, index);
|
||||
}
|
||||
}
|
||||
walk(root, bytes, &mut index);
|
||||
index
|
||||
}
|
||||
|
||||
fn detect_url_dispatch<'tree>(
|
||||
root: Node<'tree>,
|
||||
bytes: &[u8],
|
||||
file_rel: &str,
|
||||
function_index: &HashMap<String, (Node<'tree>, bool)>,
|
||||
out: &mut Vec<SurfaceNode>,
|
||||
) {
|
||||
fn recurse<'tree>(
|
||||
node: Node<'tree>,
|
||||
bytes: &[u8],
|
||||
file_rel: &str,
|
||||
function_index: &HashMap<String, (Node<'tree>, bool)>,
|
||||
out: &mut Vec<SurfaceNode>,
|
||||
) {
|
||||
if node.kind() == "call"
|
||||
&& let Some((route, handler_name)) = parse_url_call(node, bytes)
|
||||
{
|
||||
let (handler_loc, auth_required) = function_index
|
||||
.get(&handler_name)
|
||||
.map(|(h, a)| (loc_for(*h, file_rel), *a))
|
||||
.unwrap_or_else(|| (loc_for(node, file_rel), false));
|
||||
out.push(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(node, file_rel),
|
||||
framework: Framework::Django,
|
||||
method: HttpMethod::GET,
|
||||
route,
|
||||
handler_name,
|
||||
handler_location: handler_loc,
|
||||
auth_required,
|
||||
}));
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
recurse(child, bytes, file_rel, function_index, out);
|
||||
}
|
||||
}
|
||||
recurse(root, bytes, file_rel, function_index, out);
|
||||
}
|
||||
|
||||
fn parse_url_call(call: Node, bytes: &[u8]) -> Option<(String, String)> {
|
||||
let target = call.child_by_field_name("function")?;
|
||||
let target_text = target.utf8_text(bytes).ok()?;
|
||||
let leaf = target_text.rsplit('.').next().unwrap_or(target_text);
|
||||
if !matches!(leaf, "path" | "re_path" | "url") {
|
||||
return None;
|
||||
}
|
||||
let args = call.child_by_field_name("arguments")?;
|
||||
let mut cursor = args.walk();
|
||||
let mut route: Option<String> = None;
|
||||
let mut handler: Option<String> = None;
|
||||
for arg in args.children(&mut cursor) {
|
||||
match arg.kind() {
|
||||
"string" if route.is_none() => {
|
||||
route = string_node_value(arg, bytes);
|
||||
}
|
||||
"identifier" if handler.is_none() => {
|
||||
handler = arg.utf8_text(bytes).ok().map(str::to_string);
|
||||
}
|
||||
"attribute" if handler.is_none() => {
|
||||
handler = arg.utf8_text(bytes).ok().map(str::to_string);
|
||||
}
|
||||
"call" if handler.is_none() => {
|
||||
// `MyView.as_view()` shape — extract `MyView`.
|
||||
if let Some(callee) = arg.child_by_field_name("function")
|
||||
&& let Ok(text) = callee.utf8_text(bytes)
|
||||
{
|
||||
handler = Some(text.split('.').next().unwrap_or(text).to_string());
|
||||
}
|
||||
}
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
Some((route?, handler?))
|
||||
}
|
||||
|
||||
fn detect_class_based_views(
|
||||
root: Node,
|
||||
bytes: &[u8],
|
||||
file_rel: &str,
|
||||
out: &mut Vec<SurfaceNode>,
|
||||
) {
|
||||
fn recurse(node: Node, bytes: &[u8], file_rel: &str, out: &mut Vec<SurfaceNode>) {
|
||||
if node.kind() == "class_definition"
|
||||
&& class_is_django_view(node, bytes)
|
||||
{
|
||||
let class_auth = class_has_auth_permission(node, bytes);
|
||||
// Walk the body for HTTP-named methods.
|
||||
if let Some(body) = node.child_by_field_name("body") {
|
||||
let mut bcur = body.walk();
|
||||
for stmt in body.children(&mut bcur) {
|
||||
let func = match stmt.kind() {
|
||||
"function_definition" => stmt,
|
||||
"decorated_definition" => stmt
|
||||
.child_by_field_name("definition")
|
||||
.or_else(|| {
|
||||
let mut c = stmt.walk();
|
||||
stmt.children(&mut c)
|
||||
.find(|n| n.kind() == "function_definition")
|
||||
})
|
||||
.unwrap_or(stmt),
|
||||
_ => continue,
|
||||
};
|
||||
if func.kind() != "function_definition" {
|
||||
continue;
|
||||
}
|
||||
let Some(name_node) = func.child_by_field_name("name") else {
|
||||
continue;
|
||||
};
|
||||
let Ok(name) = name_node.utf8_text(bytes) else {
|
||||
continue;
|
||||
};
|
||||
let Some(method) = HttpMethod::from_ident(name) else {
|
||||
continue;
|
||||
};
|
||||
out.push(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(func, file_rel),
|
||||
framework: Framework::Django,
|
||||
method,
|
||||
route: String::new(),
|
||||
handler_name: name.to_string(),
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel,
|
||||
(func.start_position().row + 1) as u32,
|
||||
(func.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required: class_auth,
|
||||
}));
|
||||
}
|
||||
}
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
recurse(child, bytes, file_rel, out);
|
||||
}
|
||||
}
|
||||
recurse(root, bytes, file_rel, out);
|
||||
}
|
||||
|
||||
fn class_is_django_view(class: Node, bytes: &[u8]) -> bool {
|
||||
let Some(supers) = class.child_by_field_name("superclasses") else {
|
||||
return false;
|
||||
};
|
||||
let mut cursor = supers.walk();
|
||||
for sup in supers.named_children(&mut cursor) {
|
||||
let Ok(text) = sup.utf8_text(bytes) else {
|
||||
continue;
|
||||
};
|
||||
let leaf = text.rsplit('.').next().unwrap_or(text);
|
||||
if CBV_BASES.iter().any(|b| leaf.contains(b)) {
|
||||
return true;
|
||||
}
|
||||
}
|
||||
false
|
||||
}
|
||||
|
||||
fn class_has_auth_permission(class: Node, bytes: &[u8]) -> bool {
|
||||
let Some(body) = class.child_by_field_name("body") else {
|
||||
return false;
|
||||
};
|
||||
let mut cursor = body.walk();
|
||||
for stmt in body.children(&mut cursor) {
|
||||
if stmt.kind() != "expression_statement" {
|
||||
continue;
|
||||
}
|
||||
let mut sc = stmt.walk();
|
||||
for child in stmt.children(&mut sc) {
|
||||
if child.kind() != "assignment" {
|
||||
continue;
|
||||
}
|
||||
let Some(left) = child.child_by_field_name("left") else {
|
||||
continue;
|
||||
};
|
||||
let Ok(left_text) = left.utf8_text(bytes) else {
|
||||
continue;
|
||||
};
|
||||
if left_text != "permission_classes" {
|
||||
continue;
|
||||
}
|
||||
let Some(right) = child.child_by_field_name("right") else {
|
||||
continue;
|
||||
};
|
||||
let Ok(right_text) = right.utf8_text(bytes) else {
|
||||
continue;
|
||||
};
|
||||
if right_text.contains("IsAuthenticated")
|
||||
|| right_text.contains("IsAdminUser")
|
||||
|| right_text.contains("DjangoModelPermissions")
|
||||
{
|
||||
return true;
|
||||
}
|
||||
}
|
||||
}
|
||||
false
|
||||
}
|
||||
|
||||
fn decorator_is_auth_marker(decorator: Node, bytes: &[u8]) -> bool {
|
||||
let mut cursor = decorator.walk();
|
||||
let Some(expr) = decorator
|
||||
.children(&mut cursor)
|
||||
.find(|c| c.kind() != "@" && c.kind() != "comment")
|
||||
else {
|
||||
return false;
|
||||
};
|
||||
let target = match expr.kind() {
|
||||
"call" => expr.child_by_field_name("function"),
|
||||
_ => Some(expr),
|
||||
};
|
||||
let Some(target) = target else { return false };
|
||||
let Ok(text) = target.utf8_text(bytes) else {
|
||||
return false;
|
||||
};
|
||||
leaf_matches(text, AUTH_DECORATORS)
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_python::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_path_call() {
|
||||
let src = "from django.urls import path\n\ndef admin_view(request): pass\n\nurlpatterns = [\n path('admin/', admin_view),\n]\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_django_routes(&tree, &bytes, &PathBuf::from("urls.py"), None);
|
||||
assert!(!nodes.is_empty());
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.framework, Framework::Django);
|
||||
assert_eq!(ep.handler_name, "admin_view");
|
||||
assert_eq!(ep.route, "admin/");
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_class_based_view() {
|
||||
let src = "class UserList(APIView):\n def get(self, request): pass\n def post(self, request): pass\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_django_routes(&tree, &bytes, &PathBuf::from("views.py"), None);
|
||||
assert_eq!(nodes.len(), 2);
|
||||
}
|
||||
}
|
||||
336
src/surface/lang/python_fastapi.rs
Normal file
336
src/surface/lang/python_fastapi.rs
Normal file
|
|
@ -0,0 +1,336 @@
|
|||
//! Python + FastAPI framework probe.
|
||||
//!
|
||||
//! Recognises FastAPI / Starlette route declarations:
|
||||
//!
|
||||
//! * `@app.get("/path")` / `.post("/path")` / `.put` / `.patch` / `.delete`
|
||||
//! * `@router.get("/path")` / `.post(...)` / etc. on an `APIRouter`
|
||||
//! * `@app.api_route("/path", methods=["GET","POST"])`
|
||||
//! * `@app.websocket("/ws")` (treated as GET)
|
||||
//!
|
||||
//! `auth_required` is inferred from `Depends(<auth>)` parameters in the
|
||||
//! handler signature (FastAPI's idiomatic auth pattern) and from
|
||||
//! decorator-stack guards drawn from [`AUTH_DECORATORS`].
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{leaf_matches, loc_for, rel_file, string_node_value};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
/// Auth markers recognised in the decorator stack. FastAPI's primary
|
||||
/// auth idiom is `Depends(...)` parameter injection, handled separately.
|
||||
pub const AUTH_DECORATORS: &[&str] = &[
|
||||
"login_required",
|
||||
"auth_required",
|
||||
"jwt_required",
|
||||
"token_required",
|
||||
"requires_auth",
|
||||
"authenticated",
|
||||
"require_auth",
|
||||
"require_login",
|
||||
"current_user",
|
||||
];
|
||||
|
||||
/// Auth-callee names recognised inside a `Depends(...)` parameter.
|
||||
const AUTH_DEPENDS_CALLEES: &[&str] = &[
|
||||
"get_current_user",
|
||||
"get_current_active_user",
|
||||
"current_user",
|
||||
"require_user",
|
||||
"require_auth",
|
||||
"auth",
|
||||
"verify_token",
|
||||
"verify_jwt",
|
||||
"validate_token",
|
||||
];
|
||||
|
||||
pub fn detect_fastapi_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
// File-level gate: avoid double-detection on Flask files that
|
||||
// also use `app.get(...)` shape. FastAPI / Starlette / APIRouter
|
||||
// require an explicit import of the relevant package.
|
||||
let file_text = std::str::from_utf8(bytes).unwrap_or("");
|
||||
let has_fastapi_witness = file_text.contains("fastapi")
|
||||
|| file_text.contains("starlette")
|
||||
|| file_text.contains("APIRouter");
|
||||
if !has_fastapi_witness {
|
||||
return Vec::new();
|
||||
}
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
walk_decorated(tree.root_node(), &mut |func, decorators| {
|
||||
let auth_via_decorator = decorators
|
||||
.iter()
|
||||
.any(|d| decorator_is_auth_marker(*d, bytes));
|
||||
let auth_via_depends = function_signature_uses_auth_depends(*func, bytes);
|
||||
let auth_required = auth_via_decorator || auth_via_depends;
|
||||
for dec in decorators {
|
||||
if let Some((method, route_path)) = fastapi_route_decorator(*dec, bytes) {
|
||||
let handler_name = function_name(*func, bytes).unwrap_or_default();
|
||||
out.push(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(*dec, &file_rel),
|
||||
framework: Framework::FastApi,
|
||||
method,
|
||||
route: route_path,
|
||||
handler_name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel.clone(),
|
||||
(func.start_position().row + 1) as u32,
|
||||
(func.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required,
|
||||
}));
|
||||
}
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn walk_decorated<'tree, F>(root: Node<'tree>, visit: &mut F)
|
||||
where
|
||||
F: FnMut(&Node<'tree>, &[Node<'tree>]),
|
||||
{
|
||||
if root.kind() == "decorated_definition" {
|
||||
let mut cursor = root.walk();
|
||||
let mut decorators: Vec<Node<'tree>> = Vec::new();
|
||||
let mut func: Option<Node<'tree>> = None;
|
||||
for child in root.children(&mut cursor) {
|
||||
match child.kind() {
|
||||
"decorator" => decorators.push(child),
|
||||
"function_definition" => func = Some(child),
|
||||
_ => {}
|
||||
}
|
||||
}
|
||||
if let Some(f) = func {
|
||||
visit(&f, &decorators);
|
||||
}
|
||||
}
|
||||
let mut cursor = root.walk();
|
||||
for child in root.children(&mut cursor) {
|
||||
walk_decorated(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn fastapi_route_decorator(decorator: Node, bytes: &[u8]) -> Option<(HttpMethod, String)> {
|
||||
let mut cursor = decorator.walk();
|
||||
let expr = decorator
|
||||
.children(&mut cursor)
|
||||
.find(|c| c.kind() != "@" && c.kind() != "comment")?;
|
||||
if expr.kind() != "call" {
|
||||
return None;
|
||||
}
|
||||
let target = expr.child_by_field_name("function")?;
|
||||
let args = expr.child_by_field_name("arguments");
|
||||
if target.kind() != "attribute" {
|
||||
return None;
|
||||
}
|
||||
let object = target.child_by_field_name("object")?;
|
||||
if !receiver_is_fastapi(object, bytes) {
|
||||
return None;
|
||||
}
|
||||
let attr = target.child_by_field_name("attribute")?;
|
||||
let attr_text = attr.utf8_text(bytes).ok()?;
|
||||
let route_path = args
|
||||
.and_then(|a| first_string_arg(a, bytes))
|
||||
.unwrap_or_default();
|
||||
if let Some(m) = HttpMethod::from_ident(attr_text) {
|
||||
return Some((m, route_path));
|
||||
}
|
||||
let lower = attr_text.to_ascii_lowercase();
|
||||
if lower == "websocket" || lower == "websocket_route" {
|
||||
return Some((HttpMethod::GET, route_path));
|
||||
}
|
||||
if lower == "api_route" {
|
||||
let method = args
|
||||
.and_then(|a| first_methods_kwarg(a, bytes))
|
||||
.unwrap_or(HttpMethod::GET);
|
||||
return Some((method, route_path));
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn receiver_is_fastapi(object: Node, bytes: &[u8]) -> bool {
|
||||
fn name_matches(text: &str) -> bool {
|
||||
let lower = text.to_ascii_lowercase();
|
||||
lower == "app"
|
||||
|| lower == "router"
|
||||
|| lower == "api"
|
||||
|| lower.ends_with("_app")
|
||||
|| lower.ends_with("_router")
|
||||
|| lower.ends_with("_api")
|
||||
}
|
||||
match object.kind() {
|
||||
"identifier" => object.utf8_text(bytes).ok().is_some_and(name_matches),
|
||||
"attribute" => object
|
||||
.child_by_field_name("attribute")
|
||||
.and_then(|a| a.utf8_text(bytes).ok())
|
||||
.is_some_and(name_matches),
|
||||
"call" => {
|
||||
let Some(callee) = object.child_by_field_name("function") else {
|
||||
return false;
|
||||
};
|
||||
let Ok(text) = callee.utf8_text(bytes) else {
|
||||
return false;
|
||||
};
|
||||
let leaf = text.rsplit('.').next().unwrap_or(text).trim();
|
||||
leaf == "FastAPI" || leaf == "APIRouter" || leaf == "Starlette"
|
||||
}
|
||||
_ => false,
|
||||
}
|
||||
}
|
||||
|
||||
fn first_string_arg(args: Node, bytes: &[u8]) -> Option<String> {
|
||||
let mut cursor = args.walk();
|
||||
for arg in args.children(&mut cursor) {
|
||||
if arg.kind() == "string" {
|
||||
return string_node_value(arg, bytes);
|
||||
}
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn first_methods_kwarg(args: Node, bytes: &[u8]) -> Option<HttpMethod> {
|
||||
let mut cursor = args.walk();
|
||||
for arg in args.children(&mut cursor) {
|
||||
if arg.kind() != "keyword_argument" {
|
||||
continue;
|
||||
}
|
||||
let name = arg.child_by_field_name("name")?;
|
||||
if name.utf8_text(bytes).ok()? != "methods" {
|
||||
continue;
|
||||
}
|
||||
let value = arg.child_by_field_name("value")?;
|
||||
let mut vw = value.walk();
|
||||
for child in value.children(&mut vw) {
|
||||
if child.kind() == "string"
|
||||
&& let Some(v) = string_node_value(child, bytes)
|
||||
&& let Some(m) = HttpMethod::from_ident(&v)
|
||||
{
|
||||
return Some(m);
|
||||
}
|
||||
}
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn decorator_is_auth_marker(decorator: Node, bytes: &[u8]) -> bool {
|
||||
let mut cursor = decorator.walk();
|
||||
let Some(expr) = decorator
|
||||
.children(&mut cursor)
|
||||
.find(|c| c.kind() != "@" && c.kind() != "comment")
|
||||
else {
|
||||
return false;
|
||||
};
|
||||
let target = match expr.kind() {
|
||||
"call" => expr.child_by_field_name("function"),
|
||||
_ => Some(expr),
|
||||
};
|
||||
let Some(target) = target else { return false };
|
||||
let Ok(text) = target.utf8_text(bytes) else {
|
||||
return false;
|
||||
};
|
||||
leaf_matches(text, AUTH_DECORATORS)
|
||||
}
|
||||
|
||||
/// Look for a parameter with default `Depends(<auth_callee>)`.
|
||||
fn function_signature_uses_auth_depends(func: Node, bytes: &[u8]) -> bool {
|
||||
let Some(params) = func.child_by_field_name("parameters") else {
|
||||
return false;
|
||||
};
|
||||
let mut cursor = params.walk();
|
||||
for param in params.children(&mut cursor) {
|
||||
if !matches!(
|
||||
param.kind(),
|
||||
"default_parameter" | "typed_default_parameter"
|
||||
) {
|
||||
continue;
|
||||
}
|
||||
let Some(value) = param.child_by_field_name("value") else {
|
||||
continue;
|
||||
};
|
||||
if value.kind() != "call" {
|
||||
continue;
|
||||
}
|
||||
let Some(call_target) = value.child_by_field_name("function") else {
|
||||
continue;
|
||||
};
|
||||
let Ok(text) = call_target.utf8_text(bytes) else {
|
||||
continue;
|
||||
};
|
||||
let leaf = text.rsplit('.').next().unwrap_or(text).trim();
|
||||
if leaf != "Depends" && leaf != "Security" {
|
||||
continue;
|
||||
}
|
||||
let Some(args) = value.child_by_field_name("arguments") else {
|
||||
continue;
|
||||
};
|
||||
let mut aw = args.walk();
|
||||
for arg in args.children(&mut aw) {
|
||||
if let Ok(arg_text) = arg.utf8_text(bytes)
|
||||
&& leaf_matches(arg_text, AUTH_DEPENDS_CALLEES)
|
||||
{
|
||||
return true;
|
||||
}
|
||||
}
|
||||
}
|
||||
false
|
||||
}
|
||||
|
||||
fn function_name(func: Node, bytes: &[u8]) -> Option<String> {
|
||||
let name_node = func.child_by_field_name("name")?;
|
||||
name_node.utf8_text(bytes).ok().map(str::to_string)
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_python::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_get_route() {
|
||||
let src = "from fastapi import FastAPI\napp = FastAPI()\n@app.get('/users')\ndef list_users(): pass\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_fastapi_routes(&tree, &bytes, &PathBuf::from("api.py"), None);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert_eq!(ep.route, "/users");
|
||||
assert_eq!(ep.framework, Framework::FastApi);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_router_post() {
|
||||
let src = "router = APIRouter()\n@router.post('/items')\ndef create(): pass\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_fastapi_routes(&tree, &bytes, &PathBuf::from("api.py"), None);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::POST);
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_depends_auth() {
|
||||
let src = "from fastapi import Depends\n@app.get('/me')\ndef me(user = Depends(get_current_user)): pass\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_fastapi_routes(&tree, &bytes, &PathBuf::from("api.py"), None);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert!(ep.auth_required);
|
||||
}
|
||||
}
|
||||
|
|
@ -50,6 +50,17 @@ pub fn detect_flask_routes(
|
|||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
// File-level gate: avoid double-detection on FastAPI files where
|
||||
// `app.get(...)` shape overlaps. Phase 21 was lenient because no
|
||||
// sibling probe existed; Phase 22 splits per-framework, so each
|
||||
// probe only fires when its framework witness is present.
|
||||
let file_text = std::str::from_utf8(bytes).unwrap_or("");
|
||||
let has_flask_witness = file_text.contains("flask")
|
||||
|| file_text.contains("Flask")
|
||||
|| file_text.contains("Blueprint");
|
||||
if !has_flask_witness {
|
||||
return Vec::new();
|
||||
}
|
||||
let file_rel = relative_path_string(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
walk_decorated(tree.root_node(), bytes, &mut |func_node, decorators| {
|
||||
|
|
|
|||
219
src/surface/lang/ruby_rails.rs
Normal file
219
src/surface/lang/ruby_rails.rs
Normal file
|
|
@ -0,0 +1,219 @@
|
|||
//! Ruby + Rails framework probe.
|
||||
//!
|
||||
//! Recognises two Rails route shapes:
|
||||
//!
|
||||
//! 1. `config/routes.rb` declarations — `get '/path', to: 'controller#action'`,
|
||||
//! `post '/path' => 'controller#action'`, `resources :users`.
|
||||
//! 2. Controller actions — public instance methods on a class
|
||||
//! inheriting from `ApplicationController` / `ActionController::Base`.
|
||||
//!
|
||||
//! `auth_required` for routes follows `before_action :authenticate!`
|
||||
//! at the controller level.
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{loc_for, rel_file, string_node_value};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
const VERBS: &[(&str, HttpMethod)] = &[
|
||||
("get", HttpMethod::GET),
|
||||
("post", HttpMethod::POST),
|
||||
("put", HttpMethod::PUT),
|
||||
("patch", HttpMethod::PATCH),
|
||||
("delete", HttpMethod::DELETE),
|
||||
("match", HttpMethod::GET),
|
||||
];
|
||||
|
||||
pub fn detect_rails_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
detect_routes_dsl(tree.root_node(), bytes, &file_rel, &mut out);
|
||||
detect_controllers(tree.root_node(), bytes, &file_rel, &mut out);
|
||||
out
|
||||
}
|
||||
|
||||
fn detect_routes_dsl(root: Node, bytes: &[u8], file_rel: &str, out: &mut Vec<SurfaceNode>) {
|
||||
fn recurse(node: Node, bytes: &[u8], file_rel: &str, out: &mut Vec<SurfaceNode>) {
|
||||
if matches!(node.kind(), "call" | "method_call") {
|
||||
if let Some(method_node) = node.child_by_field_name("method")
|
||||
&& let Ok(method_text) = method_node.utf8_text(bytes)
|
||||
&& let Some((_, method)) = VERBS.iter().find(|(v, _)| *v == method_text)
|
||||
{
|
||||
let args_opt = node
|
||||
.child_by_field_name("arguments")
|
||||
.or_else(|| {
|
||||
let mut c = node.walk();
|
||||
node.children(&mut c).find(|n| n.kind() == "argument_list")
|
||||
});
|
||||
if let Some(args) = args_opt {
|
||||
let mut cursor = args.walk();
|
||||
let positional: Vec<Node> = args.named_children(&mut cursor).collect();
|
||||
if let Some(route_node) = positional.first()
|
||||
&& let Some(route) = string_node_value(*route_node, bytes)
|
||||
{
|
||||
let handler_name = positional
|
||||
.iter()
|
||||
.find_map(|n| extract_to_handler(*n, bytes))
|
||||
.unwrap_or_default();
|
||||
out.push(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(node, file_rel),
|
||||
framework: Framework::Rails,
|
||||
method: *method,
|
||||
route,
|
||||
handler_name,
|
||||
handler_location: loc_for(node, file_rel),
|
||||
auth_required: false,
|
||||
}));
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
recurse(child, bytes, file_rel, out);
|
||||
}
|
||||
}
|
||||
recurse(root, bytes, file_rel, out);
|
||||
}
|
||||
|
||||
fn extract_to_handler(node: Node, bytes: &[u8]) -> Option<String> {
|
||||
// Shapes:
|
||||
// `to: 'controller#action'` — pair with hash key `to`
|
||||
// `'controller#action'` — second positional string
|
||||
// `=> 'controller#action'` — assoc with hashrocket
|
||||
if node.kind() == "string"
|
||||
&& let Some(s) = string_node_value(node, bytes)
|
||||
&& s.contains('#')
|
||||
{
|
||||
return Some(s);
|
||||
}
|
||||
if node.kind() == "pair" {
|
||||
let mut cursor = node.walk();
|
||||
let children: Vec<Node> = node.named_children(&mut cursor).collect();
|
||||
for child in &children {
|
||||
if child.kind() == "string"
|
||||
&& let Some(s) = string_node_value(*child, bytes)
|
||||
&& s.contains('#')
|
||||
{
|
||||
return Some(s);
|
||||
}
|
||||
}
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn detect_controllers(root: Node, bytes: &[u8], file_rel: &str, out: &mut Vec<SurfaceNode>) {
|
||||
fn recurse(node: Node, bytes: &[u8], file_rel: &str, out: &mut Vec<SurfaceNode>) {
|
||||
if node.kind() == "class"
|
||||
&& class_is_controller(node, bytes)
|
||||
{
|
||||
let class_auth = class_has_before_authenticate(node, bytes);
|
||||
walk_methods(node, bytes, &mut |method_node, name| {
|
||||
out.push(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(method_node, file_rel),
|
||||
framework: Framework::Rails,
|
||||
method: HttpMethod::GET,
|
||||
route: String::new(),
|
||||
handler_name: name.to_string(),
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel,
|
||||
(method_node.start_position().row + 1) as u32,
|
||||
(method_node.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required: class_auth,
|
||||
}));
|
||||
});
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
recurse(child, bytes, file_rel, out);
|
||||
}
|
||||
}
|
||||
recurse(root, bytes, file_rel, out);
|
||||
}
|
||||
|
||||
fn class_is_controller(class: Node, bytes: &[u8]) -> bool {
|
||||
let Some(super_node) = class.child_by_field_name("superclass") else {
|
||||
return false;
|
||||
};
|
||||
let Ok(text) = super_node.utf8_text(bytes) else {
|
||||
return false;
|
||||
};
|
||||
text.contains("ApplicationController") || text.contains("ActionController")
|
||||
}
|
||||
|
||||
fn class_has_before_authenticate(class: Node, bytes: &[u8]) -> bool {
|
||||
let Some(body) = class.child_by_field_name("body") else {
|
||||
return false;
|
||||
};
|
||||
let mut cursor = body.walk();
|
||||
for child in body.children(&mut cursor) {
|
||||
if let Ok(text) = child.utf8_text(bytes)
|
||||
&& text.contains("before_action")
|
||||
&& (text.contains("authenticate") || text.contains("login_required"))
|
||||
{
|
||||
return true;
|
||||
}
|
||||
}
|
||||
false
|
||||
}
|
||||
|
||||
fn walk_methods<'tree, F>(class: Node<'tree>, bytes: &[u8], visit: &mut F)
|
||||
where
|
||||
F: FnMut(Node<'tree>, &str),
|
||||
{
|
||||
let Some(body) = class.child_by_field_name("body") else {
|
||||
return;
|
||||
};
|
||||
let mut cursor = body.walk();
|
||||
for child in body.children(&mut cursor) {
|
||||
if child.kind() == "method"
|
||||
&& let Some(name_node) = child.child_by_field_name("name")
|
||||
&& let Ok(name) = name_node.utf8_text(bytes)
|
||||
&& !name.starts_with('_')
|
||||
{
|
||||
visit(child, name);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_ruby::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_routes_dsl() {
|
||||
let src = "Rails.application.routes.draw do\n get '/users', to: 'users#index'\nend\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_rails_routes(&tree, &bytes, &PathBuf::from("config/routes.rb"), None);
|
||||
assert!(!nodes.is_empty());
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert_eq!(ep.route, "/users");
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_controller_actions() {
|
||||
let src = "class UsersController < ApplicationController\n def index\n end\n def show\n end\nend\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_rails_routes(&tree, &bytes, &PathBuf::from("users_controller.rb"), None);
|
||||
assert_eq!(nodes.len(), 2);
|
||||
}
|
||||
}
|
||||
111
src/surface/lang/ruby_sinatra.rs
Normal file
111
src/surface/lang/ruby_sinatra.rs
Normal file
|
|
@ -0,0 +1,111 @@
|
|||
//! Ruby + Sinatra framework probe.
|
||||
//!
|
||||
//! Sinatra routes are top-level method calls of the form
|
||||
//! `get '/path' do ... end`, `post '/path' do ... end`, etc. The
|
||||
//! handler is the block; we synthesise the handler name from the
|
||||
//! route string (Sinatra blocks are anonymous).
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{loc_for, rel_file, string_node_value};
|
||||
use crate::surface::{EntryPoint, Framework, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
const VERBS: &[(&str, HttpMethod)] = &[
|
||||
("get", HttpMethod::GET),
|
||||
("post", HttpMethod::POST),
|
||||
("put", HttpMethod::PUT),
|
||||
("patch", HttpMethod::PATCH),
|
||||
("delete", HttpMethod::DELETE),
|
||||
("head", HttpMethod::HEAD),
|
||||
("options", HttpMethod::OPTIONS),
|
||||
];
|
||||
|
||||
pub fn detect_sinatra_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
walk_calls(tree.root_node(), &mut |call| {
|
||||
if let Some(node) = match_sinatra_call(call, bytes, &file_rel) {
|
||||
out.push(node);
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn walk_calls<'tree, F: FnMut(Node<'tree>)>(node: Node<'tree>, visit: &mut F) {
|
||||
if matches!(node.kind(), "call" | "method_call") {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk_calls(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn match_sinatra_call(call: Node, bytes: &[u8], file_rel: &str) -> Option<SurfaceNode> {
|
||||
let method_name_node = call.child_by_field_name("method")?;
|
||||
let method_text = method_name_node.utf8_text(bytes).ok()?;
|
||||
let (_, method) = VERBS
|
||||
.iter()
|
||||
.find(|(v, _)| *v == method_text)?;
|
||||
// Must have a block to be a Sinatra route.
|
||||
let block = call
|
||||
.child_by_field_name("block")
|
||||
.or_else(|| {
|
||||
let mut c = call.walk();
|
||||
call.children(&mut c)
|
||||
.find(|n| matches!(n.kind(), "do_block" | "block"))
|
||||
})?;
|
||||
// Args: Sinatra accepts a string literal as the first positional arg.
|
||||
let args = call
|
||||
.child_by_field_name("arguments")
|
||||
.or_else(|| {
|
||||
let mut c = call.walk();
|
||||
call.children(&mut c).find(|n| n.kind() == "argument_list")
|
||||
})?;
|
||||
let mut cursor = args.walk();
|
||||
let route_node = args.named_children(&mut cursor).next()?;
|
||||
let route = string_node_value(route_node, bytes)?;
|
||||
let handler_name = format!("{}_{}", method_text, route.replace(['/', '-'], "_"));
|
||||
Some(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(call, file_rel),
|
||||
framework: Framework::Sinatra,
|
||||
method: *method,
|
||||
route,
|
||||
handler_name,
|
||||
handler_location: loc_for(block, file_rel),
|
||||
auth_required: false,
|
||||
}))
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_ruby::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_sinatra_get() {
|
||||
let src = "get '/users' do\n 'hi'\nend\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_sinatra_routes(&tree, &bytes, &PathBuf::from("app.rb"), None);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert_eq!(ep.route, "/users");
|
||||
}
|
||||
}
|
||||
196
src/surface/lang/rust_actix.rs
Normal file
196
src/surface/lang/rust_actix.rs
Normal file
|
|
@ -0,0 +1,196 @@
|
|||
//! Rust + actix-web framework probe.
|
||||
//!
|
||||
//! Recognises actix-web routing macros (`#[get("/path")]`,
|
||||
//! `#[post("/path")]`, `#[put]`, `#[delete]`, `#[patch]`, `#[head]`,
|
||||
//! `#[options]`, `#[route("/path", method = ...)]`) attached to a
|
||||
//! `function_item`. The route path is extracted from the macro
|
||||
//! argument string literal.
|
||||
//!
|
||||
//! `auth_required` fires when the function signature has a parameter
|
||||
//! whose type matches one of [`AUTH_EXTRACTORS`] (`Identity`,
|
||||
//! `BearerAuth`, `JwtClaims`, etc.).
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{loc_for, rel_file};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
pub const AUTH_EXTRACTORS: &[&str] = &[
|
||||
"Identity",
|
||||
"BearerAuth",
|
||||
"BasicAuth",
|
||||
"JwtClaims",
|
||||
"Authenticated",
|
||||
"User",
|
||||
];
|
||||
|
||||
const ROUTE_MACROS: &[(&str, Option<HttpMethod>)] = &[
|
||||
("get", Some(HttpMethod::GET)),
|
||||
("post", Some(HttpMethod::POST)),
|
||||
("put", Some(HttpMethod::PUT)),
|
||||
("delete", Some(HttpMethod::DELETE)),
|
||||
("patch", Some(HttpMethod::PATCH)),
|
||||
("head", Some(HttpMethod::HEAD)),
|
||||
("options", Some(HttpMethod::OPTIONS)),
|
||||
("route", None),
|
||||
];
|
||||
|
||||
pub fn detect_actix_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_text = std::str::from_utf8(bytes).unwrap_or("");
|
||||
if !file_text.contains("actix_web::") && !file_text.contains("use actix_web") {
|
||||
// Best-effort gate so the actix probe does not over-fire on
|
||||
// Rocket / generic Rust files that also define a `#[get]`
|
||||
// macro from a user crate.
|
||||
return Vec::new();
|
||||
}
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
walk_functions(tree.root_node(), &mut |func| {
|
||||
if let Some(node) = match_actix_function(func, bytes, &file_rel) {
|
||||
out.push(node);
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn walk_functions<'tree, F: FnMut(Node<'tree>)>(node: Node<'tree>, visit: &mut F) {
|
||||
if node.kind() == "function_item" {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk_functions(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn match_actix_function(func: Node, bytes: &[u8], file_rel: &str) -> Option<SurfaceNode> {
|
||||
let attrs = collect_preceding_attributes(func);
|
||||
let mut method: Option<HttpMethod> = None;
|
||||
let mut route_path = String::new();
|
||||
for attr in attrs {
|
||||
let raw = attr.utf8_text(bytes).ok()?;
|
||||
let inner = raw
|
||||
.trim_start_matches(['#', '!'])
|
||||
.trim_matches(['[', ']']);
|
||||
for (name, default_method) in ROUTE_MACROS {
|
||||
let prefix = format!("{}(", name);
|
||||
if inner.starts_with(&prefix) {
|
||||
method = default_method.or_else(|| extract_route_method(inner));
|
||||
if route_path.is_empty()
|
||||
&& let Some(start) = inner.find('"')
|
||||
{
|
||||
let rest = &inner[start + 1..];
|
||||
if let Some(end) = rest.find('"') {
|
||||
route_path = rest[..end].to_string();
|
||||
}
|
||||
}
|
||||
} else if inner == *name && method.is_none() {
|
||||
method = *default_method;
|
||||
}
|
||||
}
|
||||
}
|
||||
let m = method?;
|
||||
let handler_name = function_name(func, bytes).unwrap_or_default();
|
||||
let auth_required = signature_uses_auth_extractor(func, bytes);
|
||||
Some(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(func, file_rel),
|
||||
framework: Framework::Actix,
|
||||
method: m,
|
||||
route: route_path,
|
||||
handler_name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel,
|
||||
(func.start_position().row + 1) as u32,
|
||||
(func.start_position().column + 1) as u32,
|
||||
),
|
||||
auth_required,
|
||||
}))
|
||||
}
|
||||
|
||||
fn collect_preceding_attributes(func: Node) -> Vec<Node> {
|
||||
let mut out: Vec<Node> = Vec::new();
|
||||
let Some(parent) = func.parent() else {
|
||||
return out;
|
||||
};
|
||||
let mut cursor = parent.walk();
|
||||
let mut pending: Vec<Node> = Vec::new();
|
||||
for sib in parent.children(&mut cursor) {
|
||||
if sib.id() == func.id() {
|
||||
out.append(&mut pending);
|
||||
return out;
|
||||
}
|
||||
if sib.kind() == "attribute_item" || sib.kind() == "inner_attribute_item" {
|
||||
let mut aw = sib.walk();
|
||||
for inner in sib.children(&mut aw) {
|
||||
if inner.kind() == "attribute" {
|
||||
pending.push(inner);
|
||||
}
|
||||
}
|
||||
} else {
|
||||
pending.clear();
|
||||
}
|
||||
}
|
||||
out
|
||||
}
|
||||
|
||||
fn extract_route_method(inner: &str) -> Option<HttpMethod> {
|
||||
for verb in ["GET", "POST", "PUT", "PATCH", "DELETE", "HEAD", "OPTIONS"] {
|
||||
if inner.contains(verb) {
|
||||
return HttpMethod::from_ident(verb);
|
||||
}
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn signature_uses_auth_extractor(func: Node, bytes: &[u8]) -> bool {
|
||||
let Some(params) = func.child_by_field_name("parameters") else {
|
||||
return false;
|
||||
};
|
||||
let Ok(text) = params.utf8_text(bytes) else {
|
||||
return false;
|
||||
};
|
||||
AUTH_EXTRACTORS.iter().any(|n| text.contains(n))
|
||||
}
|
||||
|
||||
fn function_name(func: Node, bytes: &[u8]) -> Option<String> {
|
||||
func.child_by_field_name("name")
|
||||
.and_then(|n| n.utf8_text(bytes).ok())
|
||||
.map(str::to_string)
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_rust::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_actix_get() {
|
||||
let src = r#"
|
||||
use actix_web::{get, HttpResponse};
|
||||
#[get("/users")]
|
||||
async fn list_users() -> HttpResponse { HttpResponse::Ok().finish() }
|
||||
"#;
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_actix_routes(&tree, &bytes, &PathBuf::from("main.rs"), None);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert_eq!(ep.route, "/users");
|
||||
}
|
||||
}
|
||||
191
src/surface/lang/rust_axum.rs
Normal file
191
src/surface/lang/rust_axum.rs
Normal file
|
|
@ -0,0 +1,191 @@
|
|||
//! Rust + axum framework probe.
|
||||
//!
|
||||
//! Detects axum route registration:
|
||||
//!
|
||||
//! * `Router::new().route("/path", get(handler))` /
|
||||
//! `.route("/path", post(handler))` / etc.
|
||||
//! * Bare extractor-shaped function items in files that import axum
|
||||
//! (handler typing alone is treated as a candidate, but only when a
|
||||
//! `Router::route(...)` registration in the same file references it).
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{loc_for, rel_file, string_node_value};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::collections::HashMap;
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
const VERBS: &[(&str, HttpMethod)] = &[
|
||||
("get", HttpMethod::GET),
|
||||
("post", HttpMethod::POST),
|
||||
("put", HttpMethod::PUT),
|
||||
("delete", HttpMethod::DELETE),
|
||||
("patch", HttpMethod::PATCH),
|
||||
("head", HttpMethod::HEAD),
|
||||
("options", HttpMethod::OPTIONS),
|
||||
];
|
||||
|
||||
pub const AUTH_EXTRACTORS: &[&str] = &[
|
||||
"Extension<User",
|
||||
"BearerAuth",
|
||||
"RequireAuth",
|
||||
"AuthenticatedUser",
|
||||
"JwtClaims",
|
||||
];
|
||||
|
||||
pub fn detect_axum_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_text = std::str::from_utf8(bytes).unwrap_or("");
|
||||
if !file_text.contains("axum::") && !file_text.contains("use axum") {
|
||||
return Vec::new();
|
||||
}
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let function_index = collect_functions(tree.root_node(), bytes);
|
||||
let mut out = Vec::new();
|
||||
walk_calls(tree.root_node(), &mut |call| {
|
||||
if let Some(node) = match_router_route(call, bytes, &file_rel, &function_index) {
|
||||
out.push(node);
|
||||
}
|
||||
});
|
||||
out
|
||||
}
|
||||
|
||||
fn walk_calls<'tree, F: FnMut(Node<'tree>)>(node: Node<'tree>, visit: &mut F) {
|
||||
if node.kind() == "call_expression" {
|
||||
visit(node);
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk_calls(child, visit);
|
||||
}
|
||||
}
|
||||
|
||||
fn collect_functions<'tree>(
|
||||
root: Node<'tree>,
|
||||
bytes: &'tree [u8],
|
||||
) -> HashMap<String, (Node<'tree>, bool)> {
|
||||
let mut out: HashMap<String, (Node<'tree>, bool)> = HashMap::new();
|
||||
fn walk<'tree>(
|
||||
node: Node<'tree>,
|
||||
bytes: &'tree [u8],
|
||||
out: &mut HashMap<String, (Node<'tree>, bool)>,
|
||||
) {
|
||||
if node.kind() == "function_item"
|
||||
&& let Some(name_node) = node.child_by_field_name("name")
|
||||
&& let Ok(name) = name_node.utf8_text(bytes)
|
||||
{
|
||||
let auth = node
|
||||
.child_by_field_name("parameters")
|
||||
.and_then(|p| p.utf8_text(bytes).ok())
|
||||
.map(|t| AUTH_EXTRACTORS.iter().any(|x| t.contains(x)))
|
||||
.unwrap_or(false);
|
||||
out.insert(name.to_string(), (node, auth));
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
walk(child, bytes, out);
|
||||
}
|
||||
}
|
||||
walk(root, bytes, &mut out);
|
||||
out
|
||||
}
|
||||
|
||||
fn match_router_route<'tree>(
|
||||
call: Node<'tree>,
|
||||
bytes: &[u8],
|
||||
file_rel: &str,
|
||||
function_index: &HashMap<String, (Node<'tree>, bool)>,
|
||||
) -> Option<SurfaceNode> {
|
||||
let func = call.child_by_field_name("function")?;
|
||||
if func.kind() != "field_expression" {
|
||||
return None;
|
||||
}
|
||||
let field = func.child_by_field_name("field")?;
|
||||
if field.utf8_text(bytes).ok()? != "route" {
|
||||
return None;
|
||||
}
|
||||
let args = call.child_by_field_name("arguments")?;
|
||||
let mut cursor = args.walk();
|
||||
let positional: Vec<Node> = args
|
||||
.children(&mut cursor)
|
||||
.filter(|n| !matches!(n.kind(), "(" | ")" | ","))
|
||||
.collect();
|
||||
if positional.len() < 2 {
|
||||
return None;
|
||||
}
|
||||
let route = string_node_value(positional[0], bytes)?;
|
||||
let method_args = positional[1];
|
||||
if method_args.kind() != "call_expression" {
|
||||
return None;
|
||||
}
|
||||
let method_callee = method_args.child_by_field_name("function")?;
|
||||
let method_text = method_callee.utf8_text(bytes).ok()?;
|
||||
let leaf = method_text.rsplit("::").next().unwrap_or(method_text);
|
||||
let (_, method) = VERBS.iter().find(|(v, _)| *v == leaf)?;
|
||||
let method_args_node = method_args.child_by_field_name("arguments")?;
|
||||
let mut hcur = method_args_node.walk();
|
||||
let handler_node = method_args_node
|
||||
.children(&mut hcur)
|
||||
.find(|n| n.kind() == "identifier" || n.kind() == "scoped_identifier")?;
|
||||
let handler_name = handler_node.utf8_text(bytes).ok()?.to_string();
|
||||
let auth_required = function_index
|
||||
.get(&handler_name)
|
||||
.map(|(_, a)| *a)
|
||||
.unwrap_or(false);
|
||||
let handler_loc = function_index
|
||||
.get(&handler_name)
|
||||
.map(|(node, _)| {
|
||||
SourceLocation::new(
|
||||
file_rel,
|
||||
(node.start_position().row + 1) as u32,
|
||||
(node.start_position().column + 1) as u32,
|
||||
)
|
||||
})
|
||||
.unwrap_or_else(|| loc_for(handler_node, file_rel));
|
||||
Some(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(call, file_rel),
|
||||
framework: Framework::Axum,
|
||||
method: *method,
|
||||
route,
|
||||
handler_name,
|
||||
handler_location: handler_loc,
|
||||
auth_required,
|
||||
}))
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_rust::LANGUAGE.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_router_get() {
|
||||
let src = r#"
|
||||
use axum::{Router, routing::get};
|
||||
async fn list_users() -> &'static str { "ok" }
|
||||
fn app() -> Router {
|
||||
Router::new().route("/users", get(list_users))
|
||||
}
|
||||
"#;
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_axum_routes(&tree, &bytes, &PathBuf::from("main.rs"), None);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert_eq!(ep.route, "/users");
|
||||
}
|
||||
}
|
||||
315
src/surface/lang/ts_next.rs
Normal file
315
src/surface/lang/ts_next.rs
Normal file
|
|
@ -0,0 +1,315 @@
|
|||
//! TypeScript + Next.js framework probe.
|
||||
//!
|
||||
//! Recognises Next.js App Router route handlers (`app/**/route.{ts,tsx,js,jsx}`)
|
||||
//! by walking exported function declarations whose name is one of the
|
||||
//! HTTP method idents (`GET` / `POST` / …). Also recognises Pages
|
||||
//! Router API routes (`pages/api/**/*.{ts,tsx,js,jsx}`) via the
|
||||
//! `export default handler` pattern.
|
||||
//!
|
||||
//! Server actions (`'use server'` directive at file or function scope)
|
||||
//! are also reported as entry points because they expose a function
|
||||
//! callable from a React client over the wire.
|
||||
|
||||
use crate::entry_points::HttpMethod;
|
||||
use crate::surface::lang::common::{loc_for, rel_file};
|
||||
use crate::surface::{EntryPoint, Framework, SourceLocation, SurfaceNode};
|
||||
use std::path::Path;
|
||||
use tree_sitter::{Node, Tree};
|
||||
|
||||
pub fn detect_next_routes(
|
||||
tree: &Tree,
|
||||
bytes: &[u8],
|
||||
path: &Path,
|
||||
scan_root: Option<&Path>,
|
||||
) -> Vec<SurfaceNode> {
|
||||
let file_rel = rel_file(path, scan_root);
|
||||
let mut out = Vec::new();
|
||||
let app_router = is_app_router_route(path);
|
||||
let pages_api = is_pages_api_route(path);
|
||||
let route_path = derive_route_path(path);
|
||||
let file_use_server = file_level_use_server(tree.root_node(), bytes);
|
||||
|
||||
if app_router {
|
||||
collect_named_exports(tree.root_node(), bytes, &file_rel, &route_path, &mut out);
|
||||
}
|
||||
if pages_api {
|
||||
collect_default_export(tree.root_node(), bytes, &file_rel, &route_path, &mut out);
|
||||
}
|
||||
if file_use_server {
|
||||
collect_use_server_exports(tree.root_node(), bytes, &file_rel, &route_path, &mut out);
|
||||
}
|
||||
out
|
||||
}
|
||||
|
||||
fn is_app_router_route(path: &Path) -> bool {
|
||||
let Some(name) = path.file_name().and_then(|n| n.to_str()) else {
|
||||
return false;
|
||||
};
|
||||
if !matches!(name, "route.ts" | "route.tsx" | "route.js" | "route.jsx") {
|
||||
return false;
|
||||
}
|
||||
path.components()
|
||||
.any(|c| c.as_os_str().to_string_lossy() == "app")
|
||||
}
|
||||
|
||||
fn is_pages_api_route(path: &Path) -> bool {
|
||||
let mut comps = path.components().peekable();
|
||||
let mut saw_pages = false;
|
||||
while let Some(c) = comps.next() {
|
||||
if c.as_os_str().to_string_lossy() == "pages" {
|
||||
saw_pages = true;
|
||||
} else if saw_pages && c.as_os_str().to_string_lossy() == "api" {
|
||||
return true;
|
||||
}
|
||||
}
|
||||
false
|
||||
}
|
||||
|
||||
/// Convert `app/users/[id]/route.ts` → `/users/[id]`.
|
||||
/// Convert `pages/api/users/index.ts` → `/users`.
|
||||
fn derive_route_path(path: &Path) -> String {
|
||||
let mut comps: Vec<String> = Vec::new();
|
||||
let mut started = false;
|
||||
for comp in path.components() {
|
||||
let text = comp.as_os_str().to_string_lossy().into_owned();
|
||||
if !started {
|
||||
if text == "app" || text == "api" || text == "pages" {
|
||||
started = true;
|
||||
}
|
||||
continue;
|
||||
}
|
||||
comps.push(text);
|
||||
}
|
||||
if let Some(last) = comps.last_mut() {
|
||||
// Drop the basename; route file becomes the trailing segment.
|
||||
if last.starts_with("route.") || last.starts_with("index.") {
|
||||
comps.pop();
|
||||
} else if let Some(idx) = last.rfind('.') {
|
||||
last.truncate(idx);
|
||||
}
|
||||
}
|
||||
let joined = comps.join("/");
|
||||
if joined.is_empty() {
|
||||
"/".to_string()
|
||||
} else {
|
||||
format!("/{}", joined)
|
||||
}
|
||||
}
|
||||
|
||||
fn collect_named_exports(
|
||||
root: Node,
|
||||
bytes: &[u8],
|
||||
file_rel: &str,
|
||||
route_path: &str,
|
||||
out: &mut Vec<SurfaceNode>,
|
||||
) {
|
||||
fn recurse(
|
||||
node: Node,
|
||||
bytes: &[u8],
|
||||
file_rel: &str,
|
||||
route_path: &str,
|
||||
out: &mut Vec<SurfaceNode>,
|
||||
) {
|
||||
if node.kind() == "export_statement" {
|
||||
// Look for `export async function NAME(...)` or `export const NAME = ...`
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
if let Some((name, span)) = extract_named_function(child, bytes)
|
||||
&& let Some(method) = HttpMethod::from_ident(&name)
|
||||
{
|
||||
out.push(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(node, file_rel),
|
||||
framework: Framework::NextAppRouter,
|
||||
method,
|
||||
route: route_path.to_string(),
|
||||
handler_name: name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel,
|
||||
(span.0 + 1) as u32,
|
||||
(span.1 + 1) as u32,
|
||||
),
|
||||
auth_required: false,
|
||||
}));
|
||||
}
|
||||
}
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
recurse(child, bytes, file_rel, route_path, out);
|
||||
}
|
||||
}
|
||||
recurse(root, bytes, file_rel, route_path, out);
|
||||
}
|
||||
|
||||
fn extract_named_function(node: Node, bytes: &[u8]) -> Option<(String, (usize, usize))> {
|
||||
match node.kind() {
|
||||
"function_declaration" => {
|
||||
let name_node = node.child_by_field_name("name")?;
|
||||
let name = name_node.utf8_text(bytes).ok()?.to_string();
|
||||
let pos = node.start_position();
|
||||
Some((name, (pos.row, pos.column)))
|
||||
}
|
||||
"lexical_declaration" | "variable_declaration" => {
|
||||
let mut cursor = node.walk();
|
||||
for decl in node.children(&mut cursor) {
|
||||
if decl.kind() == "variable_declarator"
|
||||
&& let Some(name_node) = decl.child_by_field_name("name")
|
||||
&& let Ok(name) = name_node.utf8_text(bytes)
|
||||
{
|
||||
let pos = decl.start_position();
|
||||
return Some((name.to_string(), (pos.row, pos.column)));
|
||||
}
|
||||
}
|
||||
None
|
||||
}
|
||||
_ => None,
|
||||
}
|
||||
}
|
||||
|
||||
fn collect_default_export(
|
||||
root: Node,
|
||||
bytes: &[u8],
|
||||
file_rel: &str,
|
||||
route_path: &str,
|
||||
out: &mut Vec<SurfaceNode>,
|
||||
) {
|
||||
fn recurse(
|
||||
node: Node,
|
||||
bytes: &[u8],
|
||||
file_rel: &str,
|
||||
route_path: &str,
|
||||
out: &mut Vec<SurfaceNode>,
|
||||
) {
|
||||
if node.kind() == "export_statement" {
|
||||
let raw = node.utf8_text(bytes).unwrap_or("");
|
||||
if raw.contains("default") {
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
let name = match child.kind() {
|
||||
"function_declaration" => child
|
||||
.child_by_field_name("name")
|
||||
.and_then(|n| n.utf8_text(bytes).ok())
|
||||
.map(str::to_string),
|
||||
"identifier" => child.utf8_text(bytes).ok().map(str::to_string),
|
||||
"arrow_function" | "function" | "function_expression" => {
|
||||
Some("default".to_string())
|
||||
}
|
||||
_ => None,
|
||||
};
|
||||
if let Some(name) = name {
|
||||
out.push(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(node, file_rel),
|
||||
framework: Framework::NextAppRouter,
|
||||
method: HttpMethod::GET,
|
||||
route: route_path.to_string(),
|
||||
handler_name: name,
|
||||
handler_location: loc_for(child, file_rel),
|
||||
auth_required: false,
|
||||
}));
|
||||
return;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
recurse(child, bytes, file_rel, route_path, out);
|
||||
}
|
||||
}
|
||||
recurse(root, bytes, file_rel, route_path, out);
|
||||
}
|
||||
|
||||
fn collect_use_server_exports(
|
||||
root: Node,
|
||||
bytes: &[u8],
|
||||
file_rel: &str,
|
||||
route_path: &str,
|
||||
out: &mut Vec<SurfaceNode>,
|
||||
) {
|
||||
let mut cursor = root.walk();
|
||||
for child in root.children(&mut cursor) {
|
||||
if child.kind() == "export_statement"
|
||||
&& let Some((name, span)) = export_function_name(child, bytes)
|
||||
{
|
||||
out.push(SurfaceNode::EntryPoint(EntryPoint {
|
||||
location: loc_for(child, file_rel),
|
||||
framework: Framework::NextServerAction,
|
||||
method: HttpMethod::POST,
|
||||
route: route_path.to_string(),
|
||||
handler_name: name,
|
||||
handler_location: SourceLocation::new(
|
||||
file_rel,
|
||||
(span.0 + 1) as u32,
|
||||
(span.1 + 1) as u32,
|
||||
),
|
||||
auth_required: false,
|
||||
}));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
fn export_function_name(node: Node, bytes: &[u8]) -> Option<(String, (usize, usize))> {
|
||||
let mut cursor = node.walk();
|
||||
for child in node.children(&mut cursor) {
|
||||
if let Some(extracted) = extract_named_function(child, bytes) {
|
||||
return Some(extracted);
|
||||
}
|
||||
}
|
||||
None
|
||||
}
|
||||
|
||||
fn file_level_use_server(root: Node, bytes: &[u8]) -> bool {
|
||||
let mut cursor = root.walk();
|
||||
for child in root.children(&mut cursor) {
|
||||
if child.kind() == "expression_statement" {
|
||||
let mut cs = child.walk();
|
||||
for c in child.children(&mut cs) {
|
||||
if c.kind() == "string"
|
||||
&& let Ok(text) = c.utf8_text(bytes)
|
||||
{
|
||||
let trimmed = text.trim().trim_matches(['\'', '"']);
|
||||
if trimmed == "use server" {
|
||||
return true;
|
||||
}
|
||||
}
|
||||
}
|
||||
return false;
|
||||
}
|
||||
if !matches!(child.kind(), "comment" | "import_statement") {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
false
|
||||
}
|
||||
|
||||
#[cfg(test)]
|
||||
mod tests {
|
||||
use super::*;
|
||||
use std::path::PathBuf;
|
||||
|
||||
fn parse(src: &str) -> (Tree, Vec<u8>) {
|
||||
let mut parser = tree_sitter::Parser::new();
|
||||
parser
|
||||
.set_language(&tree_sitter_typescript::LANGUAGE_TSX.into())
|
||||
.unwrap();
|
||||
(parser.parse(src, None).unwrap(), src.as_bytes().to_vec())
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn detects_app_router_get() {
|
||||
let src = "export async function GET(req: Request) { return new Response('ok'); }\n";
|
||||
let (tree, bytes) = parse(src);
|
||||
let nodes = detect_next_routes(
|
||||
&tree,
|
||||
&bytes,
|
||||
&PathBuf::from("app/users/route.ts"),
|
||||
None,
|
||||
);
|
||||
assert_eq!(nodes.len(), 1);
|
||||
let SurfaceNode::EntryPoint(ep) = &nodes[0] else {
|
||||
panic!()
|
||||
};
|
||||
assert_eq!(ep.method, HttpMethod::GET);
|
||||
assert!(ep.route.contains("users"));
|
||||
}
|
||||
}
|
||||
Loading…
Add table
Add a link
Reference in a new issue