perf: avoid file text string clone in wasm (#639)

Author: dsherret

Cargo.lock

@@ -31,7 +31,7 @@ harness = false
 
 [dependencies]
 anyhow = "1.0.64"
-deno_ast = { version = "0.38.0", features = ["view"] }
+deno_ast = { version = "0.39.0", features = ["view"] }
 dprint-core = { version = "0.66.2", features = ["formatting"] }
 dprint-core-macros = "0.1.0"
 percent-encoding = "2.3.1"

Cargo.toml

@@ -1,4 +1,5 @@
 use std::path::Path;
+use std::sync::Arc;
 
 use anyhow::Result;
 use deno_ast::ParsedSource;
@@ -33,21 +34,24 @@ use super::swc::parse_swc_ast;
 ///
 /// // now format many files (it is recommended to parallelize this)
 /// let files_to_format = vec![(PathBuf::from("path/to/file.ts"), "const  t  =  5 ;")];
-/// for (file_path, file_text) in files_to_format.iter() {
-///     let result = format_text(file_path, file_text, &config);
+/// for (file_path, file_text) in files_to_format {
+///     let result = format_text(&file_path, file_text.into(), &config);
 ///     // save result here...
 /// }
 /// ```
-pub fn format_text(file_path: &Path, file_text: &str, config: &Configuration) -> Result<Option<String>> {
-  if super::utils::file_text_has_ignore_comment(file_text, &config.ignore_file_comment_text) {
+pub fn format_text(file_path: &Path, file_text: String, config: &Configuration) -> Result<Option<String>> {
+  if super::utils::file_text_has_ignore_comment(&file_text, &config.ignore_file_comment_text) {
     Ok(None)
   } else {
+    let had_bom = file_text.starts_with("\u{FEFF}");
+    let file_text = if had_bom { file_text[3..].to_string() } else { file_text };
+    let file_text: Arc<str> = file_text.into();
     let parsed_source = parse_swc_ast(file_path, file_text)?;
     match inner_format(&parsed_source, config)? {
       Some(new_text) => Ok(Some(new_text)),
       None => {
-        if let Some(stripped) = file_text.strip_prefix("\u{FEFF}") {
-          Ok(Some(stripped.to_string()))
+        if had_bom {
+          Ok(Some(parsed_source.text().to_string()))
         } else {
           Ok(None)
         }
@@ -58,7 +62,7 @@ pub fn format_text(file_path: &Path, file_text: &str, config: &Configuration) ->
 
 /// Formats an already parsed source. This is useful as a performance optimization.
 pub fn format_parsed_source(source: &ParsedSource, config: &Configuration) -> Result<Option<String>> {
-  if super::utils::file_text_has_ignore_comment(source.text_info().text_str(), &config.ignore_file_comment_text) {
+  if super::utils::file_text_has_ignore_comment(source.text(), &config.ignore_file_comment_text) {
     Ok(None)
   } else {
     ensure_no_specific_syntax_errors(source)?;
@@ -74,9 +78,9 @@ fn inner_format(parsed_source: &ParsedSource, config: &Configuration) -> Result<
       // println!("{}", print_items.get_as_text());
       print_items
     },
-    config_to_print_options(parsed_source.text_info().text_str(), config),
+    config_to_print_options(parsed_source.text(), config),
   );
-  if result == parsed_source.text_info().text_str() {
+  if result == parsed_source.text().as_ref() {
     Ok(None)
   } else {
     Ok(Some(result))
@@ -107,7 +111,7 @@ mod test {
   fn strips_bom() {
     for input_text in ["\u{FEFF}const t = 5;\n", "\u{FEFF}const t =   5;"] {
       let config = crate::configuration::ConfigurationBuilder::new().build();
-      let result = format_text(&std::path::PathBuf::from("test.ts"), input_text, &config).unwrap().unwrap();
+      let result = format_text(&std::path::PathBuf::from("test.ts"), input_text.into(), &config).unwrap().unwrap();
       assert_eq!(result, "const t = 5;\n");
     }
   }

src/format_text.rs

@@ -5,11 +5,10 @@ use deno_ast::swc::parser::error::SyntaxError;
 use deno_ast::swc::parser::Syntax;
 use deno_ast::ModuleSpecifier;
 use deno_ast::ParsedSource;
-use deno_ast::SourceTextInfo;
 use std::path::Path;
+use std::sync::Arc;
 
-pub fn parse_swc_ast(file_path: &Path, file_text: &str) -> Result<ParsedSource> {
-  let file_text = SourceTextInfo::from_string(file_text.to_string());
+pub fn parse_swc_ast(file_path: &Path, file_text: Arc<str>) -> Result<ParsedSource> {
   match parse_inner(file_path, file_text.clone()) {
     Ok(result) => Ok(result),
     Err(err) => {
@@ -28,13 +27,13 @@ pub fn parse_swc_ast(file_path: &Path, file_text: &str) -> Result<ParsedSource>
   }
 }
 
-fn parse_inner(file_path: &Path, text_info: SourceTextInfo) -> Result<ParsedSource> {
-  let parsed_source = parse_inner_no_diagnostic_check(file_path, text_info)?;
+fn parse_inner(file_path: &Path, text: Arc<str>) -> Result<ParsedSource> {
+  let parsed_source = parse_inner_no_diagnostic_check(file_path, text)?;
   ensure_no_specific_syntax_errors(&parsed_source)?;
   Ok(parsed_source)
 }
 
-fn parse_inner_no_diagnostic_check(file_path: &Path, text_info: SourceTextInfo) -> Result<ParsedSource> {
+fn parse_inner_no_diagnostic_check(file_path: &Path, text: Arc<str>) -> Result<ParsedSource> {
   let media_type = deno_ast::MediaType::from_path(file_path);
   let mut syntax = deno_ast::get_syntax(media_type);
   if let Syntax::Es(es) = &mut syntax {
@@ -47,7 +46,7 @@ fn parse_inner_no_diagnostic_check(file_path: &Path, text_info: SourceTextInfo)
     maybe_syntax: Some(syntax),
     media_type,
     scope_analysis: false,
-    text_info,
+    text,
   })
   .map_err(|diagnostic| anyhow!("{:#}", &diagnostic))
 }
@@ -247,7 +246,7 @@ mod tests {
 
   fn run_fatal_diagnostic_test(file_path: &str, text: &str, expected: &str) {
     let file_path = PathBuf::from(file_path);
-    assert_eq!(parse_swc_ast(&file_path, text).err().unwrap().to_string(), expected);
+    assert_eq!(parse_swc_ast(&file_path, text.into()).err().unwrap().to_string(), expected);
   }
 
   #[test]
@@ -343,11 +342,11 @@ Merge conflict marker encountered. at file:///test.ts:6:1
 
   fn run_non_fatal_diagnostic_test(file_path: &str, text: &str, expected: &str) {
     let file_path = PathBuf::from(file_path);
-    assert_eq!(format!("{}", parse_swc_ast(&file_path, text).err().unwrap()), expected);
+    assert_eq!(format!("{}", parse_swc_ast(&file_path, text.into()).err().unwrap()), expected);
 
     // this error should also be surfaced in `format_parsed_source` if someone provides
     // a source file that had a non-fatal diagnostic
-    let parsed_source = parse_inner_no_diagnostic_check(&file_path, SourceTextInfo::from_string(text.to_string())).unwrap();
+    let parsed_source = parse_inner_no_diagnostic_check(&file_path, text.into()).unwrap();
     let config = ConfigurationBuilder::new().build();
     assert_eq!(crate::format_parsed_source(&parsed_source, &config).err().unwrap().to_string(), expected);
   }

src/swc.rs

@@ -58,7 +58,7 @@ impl SyncPluginHandler<Configuration> for TypeScriptPluginHandler {
     _format_with_host: impl FnMut(&Path, Vec<u8>, &ConfigKeyMap) -> FormatResult,
   ) -> FormatResult {
     let file_text = String::from_utf8(file_bytes)?;
-    super::format_text(file_path, &file_text, config).map(|maybe_text| maybe_text.map(|t| t.into_bytes()))
+    super::format_text(file_path, file_text, config).map(|maybe_text| maybe_text.map(|t| t.into_bytes()))
   }
 }
 

src/wasm_plugin.rs

@@ -28,7 +28,7 @@ fn main() {
         let config_result = resolve_config(spec_config, &global_config);
         ensure_no_diagnostics(&config_result.diagnostics);
 
-        format_text(file_name, file_text, &config_result.config)
+        format_text(file_name, file_text.into(), &config_result.config)
       })
     },
     Arc::new(move |_file_name, _file_text, _spec_config| {