refactor: change Tsconfig::parse to accept owned string; add replace_bom_with_whitespace (#859)

Author: Boshen

Cargo.lock

@@ -79,7 +79,7 @@ name = "resolver"
 [dependencies]
 cfg-if = "1"
 indexmap = { version = "2", features = ["serde"] }
-json-strip-comments = "3"
+json-strip-comments = "3.1"
 once_cell = "1" # Use `std::sync::OnceLock::get_or_try_init` when it is stable.
 papaya = "0.2"
 parking_lot = "0.12"

Cargo.toml

@@ -227,12 +227,12 @@ impl<Fs: FileSystem> Cache<Fs> {
             os_string.push(".json");
             Cow::Owned(PathBuf::from(os_string))
         };
-        let mut tsconfig_string = self
+        let tsconfig_string = self
             .fs
             .read_to_string(&tsconfig_path)
             .map_err(|_| ResolveError::TsconfigNotFound(path.to_path_buf()))?;
         let mut tsconfig =
-            TsConfig::parse(root, &tsconfig_path, &mut tsconfig_string).map_err(|error| {
+            TsConfig::parse(root, &tsconfig_path, tsconfig_string).map_err(|error| {
                 ResolveError::from_serde_json_error(tsconfig_path.to_path_buf(), &error)
             })?;
         callback(&mut tsconfig)?;

src/cache/cache_impl.rs

@@ -1926,3 +1926,13 @@ fn resolve_file_protocol(specifier: &str) -> Result<Cow<'_, str>, ResolveError>
         Ok(Cow::Borrowed(specifier))
     }
 }
+
+/// Strip BOM in place by replacing with spaces (no reallocation)
+/// UTF-8 BOM is 3 bytes: 0xEF, 0xBB, 0xBF
+pub(crate) fn replace_bom_with_whitespace(s: &mut [u8]) {
+    if s.starts_with(b"\xEF\xBB\xBF") {
+        s[0] = b' ';
+        s[1] = b' ';
+        s[2] = b' ';
+    }
+}

src/lib.rs

@@ -9,8 +9,9 @@ use std::{
 
 use serde_json::Value;
 
+use crate::{FileSystem, JSONError, ResolveError, path::PathUtil, replace_bom_with_whitespace};
+
 use super::{ImportsExportsKind, PackageType, SideEffects};
-use crate::{FileSystem, JSONError, ResolveError, path::PathUtil};
 
 /// Serde implementation for the deserialized `package.json`.
 ///
@@ -226,20 +227,15 @@ impl PackageJson {
         realpath: PathBuf,
         json: Vec<u8>,
     ) -> Result<Self, JSONError> {
-        // Strip BOM - UTF-8 BOM is 3 bytes: 0xEF, 0xBB, 0xBF
-        let json_bytes = if json.starts_with(b"\xEF\xBB\xBF") { &json[3..] } else { &json[..] };
-
-        // Check if empty after BOM stripping
-        super::check_if_empty(json_bytes, &path)?;
-
-        // Parse JSON directly from bytes
-        let value = serde_json::from_slice::<Value>(json_bytes).map_err(|error| JSONError {
+        let mut json = json;
+        replace_bom_with_whitespace(&mut json);
+        super::check_if_empty(&json, &path)?;
+        let value = serde_json::from_slice::<Value>(&json).map_err(|error| JSONError {
             path: path.clone(),
             message: error.to_string(),
             line: error.line(),
             column: error.column(),
         })?;
-
         Ok(Self { path, realpath, value })
     }
 

src/package_json/serde.rs

@@ -11,7 +11,7 @@ use self_cell::MutBorrow;
 use simd_json::{BorrowedValue, prelude::*};
 
 use super::{ImportsExportsKind, PackageType, SideEffects};
-use crate::{FileSystem, JSONError, ResolveError, path::PathUtil};
+use crate::{FileSystem, JSONError, ResolveError, path::PathUtil, replace_bom_with_whitespace};
 
 // Use simd_json's Object type which handles the hasher correctly based on features
 type BorrowedObject<'a> = simd_json::value::borrowed::Object<'a>;
@@ -260,19 +260,14 @@ impl PackageJson {
         realpath: PathBuf,
         json: Vec<u8>,
     ) -> Result<Self, JSONError> {
-        // Strip BOM in place by replacing with spaces (no reallocation)
-        let mut json_bytes = json;
-        if json_bytes.starts_with(b"\xEF\xBB\xBF") {
-            json_bytes[0] = b' ';
-            json_bytes[1] = b' ';
-            json_bytes[2] = b' ';
-        }
+        let mut json = json;
+        replace_bom_with_whitespace(&mut json);
 
         // Check if empty after BOM stripping
-        super::check_if_empty(&json_bytes, &path)?;
+        super::check_if_empty(&json, &path)?;
 
         // Create the self-cell with the JSON bytes and parsed BorrowedValue
-        let cell = PackageJsonCell::try_new(MutBorrow::new(json_bytes), |bytes| {
+        let cell = PackageJsonCell::try_new(MutBorrow::new(json), |bytes| {
             // Use MutBorrow to safely get mutable access for simd_json parsing
             simd_json::to_borrowed_value(bytes.borrow_mut())
         })

src/package_json/simd.rs

@@ -162,7 +162,7 @@ fn test_extend_tsconfig_no_override_existing() {
     let parent_path = Path::new("/parent/tsconfig.json");
     let child_path = Path::new("/child/tsconfig.json");
 
-    let mut parent_config = serde_json::json!({
+    let parent_config = serde_json::json!({
         "compilerOptions": {
             "baseUrl": "./src",
             "jsx": "react-jsx",
@@ -171,15 +171,15 @@ fn test_extend_tsconfig_no_override_existing() {
     })
     .to_string();
 
-    let mut child_config = serde_json::json!({
+    let child_config = serde_json::json!({
         "compilerOptions": {
             "jsx": "preserve"  // This should NOT be overridden
         }
     })
     .to_string();
 
-    let parent_tsconfig = TsConfig::parse(true, parent_path, &mut parent_config).unwrap().build();
-    let mut child_tsconfig = TsConfig::parse(true, child_path, &mut child_config).unwrap();
+    let parent_tsconfig = TsConfig::parse(true, parent_path, parent_config).unwrap().build();
+    let mut child_tsconfig = TsConfig::parse(true, child_path, child_config).unwrap();
 
     // Perform the extension
     child_tsconfig.extend_tsconfig(&parent_tsconfig);

src/tests/tsconfig_extends.rs

@@ -175,7 +175,7 @@ fn empty() {
 #[test]
 fn test_paths() {
     let path = Path::new("/foo/tsconfig.json");
-    let mut tsconfig_json = serde_json::json!({
+    let tsconfig_json = serde_json::json!({
         "compilerOptions": {
             "paths": {
                 "jquery": ["node_modules/jquery/dist/jquery"],
@@ -188,7 +188,7 @@ fn test_paths() {
         }
     })
     .to_string();
-    let tsconfig = TsConfig::parse(true, path, &mut tsconfig_json).unwrap().build();
+    let tsconfig = TsConfig::parse(true, path, tsconfig_json).unwrap().build();
 
     let data = [
         ("jquery", vec!["/foo/node_modules/jquery/dist/jquery"]),
@@ -212,13 +212,13 @@ fn test_paths() {
 #[test]
 fn test_base_url() {
     let path = Path::new("/foo/tsconfig.json");
-    let mut tsconfig_json = serde_json::json!({
+    let tsconfig_json = serde_json::json!({
         "compilerOptions": {
             "baseUrl": "./src"
         }
     })
     .to_string();
-    let tsconfig = TsConfig::parse(true, path, &mut tsconfig_json).unwrap().build();
+    let tsconfig = TsConfig::parse(true, path, tsconfig_json).unwrap().build();
 
     let data = [
         ("foo", vec!["/foo/src/foo"]),
@@ -237,7 +237,7 @@ fn test_base_url() {
 #[test]
 fn test_paths_and_base_url() {
     let path = Path::new("/foo/tsconfig.json");
-    let mut tsconfig_json = serde_json::json!({
+    let tsconfig_json = serde_json::json!({
         "compilerOptions": {
             "baseUrl": "./src",
             "paths": {
@@ -249,7 +249,7 @@ fn test_paths_and_base_url() {
         }
     })
     .to_string();
-    let tsconfig = TsConfig::parse(true, path, &mut tsconfig_json).unwrap().build();
+    let tsconfig = TsConfig::parse(true, path, tsconfig_json).unwrap().build();
 
     let data = [
         ("test", vec!["/foo/src/generated/test", "/foo/src/test"]),

src/tests/tsconfig_paths.rs

@@ -9,7 +9,7 @@ use indexmap::IndexMap;
 use rustc_hash::FxHasher;
 use serde::Deserialize;
 
-use crate::{TsconfigReferences, path::PathUtil};
+use crate::{TsconfigReferences, path::PathUtil, replace_bom_with_whitespace};
 
 const TEMPLATE_VARIABLE: &str = "${configDir}";
 
@@ -23,7 +23,7 @@ pub struct ProjectReference {
     pub path: PathBuf,
 }
 
-#[derive(Debug, Deserialize)]
+#[derive(Debug, Default, Deserialize)]
 #[serde(rename_all = "camelCase")]
 pub struct TsConfig {
     /// Whether this is the caller tsconfig.
@@ -66,11 +66,15 @@ impl TsConfig {
     /// # Errors
     ///
     /// * Any error that can be returned by `serde_json::from_str()`.
-    pub fn parse(root: bool, path: &Path, json: &mut str) -> Result<Self, serde_json::Error> {
-        let json = trim_start_matches_mut(json, '\u{feff}'); // strip bom
-        _ = json_strip_comments::strip(json);
-        let mut tsconfig: Self =
-            serde_json::from_str(if json.trim().is_empty() { "{}" } else { json })?;
+    pub fn parse(root: bool, path: &Path, json: String) -> Result<Self, serde_json::Error> {
+        let mut json = json.into_bytes();
+        replace_bom_with_whitespace(&mut json);
+        _ = json_strip_comments::strip_slice(&mut json);
+        let mut tsconfig: Self = if json.iter().all(u8::is_ascii_whitespace) {
+            Self::default()
+        } else {
+            serde_json::from_slice(&json)?
+        };
         tsconfig.root = root;
         tsconfig.path = path.to_path_buf();
         Ok(tsconfig)
@@ -488,12 +492,3 @@ pub enum ExtendsField {
     Single(String),
     Multiple(Vec<String>),
 }
-
-fn trim_start_matches_mut(s: &mut str, pat: char) -> &mut str {
-    if s.starts_with(pat) {
-        // trim the prefix
-        &mut s[pat.len_utf8()..]
-    } else {
-        s
-    }
-}