perf: using Arc<str> instead of String (#84)

Author: underfin

src/builder.rs

@@ -5,6 +5,7 @@ use std::env;
 use std::fs;
 use std::io::Read;
 use std::path::{Path, PathBuf};
+use std::sync::Arc;
 
 use debugid::DebugId;
 use rustc_hash::FxHashMap;
@@ -19,14 +20,14 @@ use crate::types::{RawToken, SourceMap, Token};
 /// objects is generally not very comfortable.  As a general aid this
 /// type can help.
 pub struct SourceMapBuilder {
-    file: Option<String>,
-    name_map: FxHashMap<String, u32>,
-    names: Vec<String>,
+    file: Option<Arc<str>>,
+    name_map: FxHashMap<Arc<str>, u32>,
+    names: Vec<Arc<str>>,
     tokens: Vec<RawToken>,
-    source_map: FxHashMap<String, u32>,
-    source_root: Option<String>,
-    sources: Vec<String>,
-    source_contents: Vec<Option<String>>,
+    source_map: FxHashMap<Arc<str>, u32>,
+    source_root: Option<Arc<str>>,
+    sources: Vec<Arc<str>>,
+    source_contents: Vec<Option<Arc<str>>>,
     sources_mapping: Vec<u32>,
     debug_id: Option<DebugId>,
 }
@@ -52,7 +53,7 @@ impl SourceMapBuilder {
     /// Creates a new source map builder and sets the file.
     pub fn new(file: Option<&str>) -> SourceMapBuilder {
         SourceMapBuilder {
-            file: file.map(str::to_owned),
+            file: file.map(Into::into),
             name_map: FxHashMap::default(),
             names: vec![],
             tokens: vec![],
@@ -71,7 +72,7 @@ impl SourceMapBuilder {
     }
 
     /// Sets the file for the sourcemap (optional)
-    pub fn set_file<T: Into<String>>(&mut self, value: Option<T>) {
+    pub fn set_file<T: Into<Arc<str>>>(&mut self, value: Option<T>) {
         self.file = value.map(Into::into);
     }
 
@@ -81,7 +82,7 @@ impl SourceMapBuilder {
     }
 
     /// Sets a new value for the source_root.
-    pub fn set_source_root<T: Into<String>>(&mut self, value: Option<T>) {
+    pub fn set_source_root<T: Into<Arc<str>>>(&mut self, value: Option<T>) {
         self.source_root = value.map(Into::into);
     }
 
@@ -108,7 +109,7 @@ impl SourceMapBuilder {
     /// Changes the source name for an already set source.
     pub fn set_source(&mut self, src_id: u32, src: &str) {
         assert!(src_id != !0, "Cannot set sources for tombstone source id");
-        self.sources[src_id as usize] = src.to_string();
+        self.sources[src_id as usize] = src.into();
     }
 
     /// Looks up a source name for an ID.
@@ -122,7 +123,7 @@ impl SourceMapBuilder {
         if self.sources.len() > self.source_contents.len() {
             self.source_contents.resize(self.sources.len(), None);
         }
-        self.source_contents[src_id as usize] = contents.map(str::to_owned);
+        self.source_contents[src_id as usize] = contents.map(Into::into);
     }
 
     /// Returns the current source contents for a source.
@@ -272,7 +273,7 @@ impl SourceMapBuilder {
                     prefix.push('/');
                 }
                 if source.starts_with(&prefix) {
-                    *source = source[prefix.len()..].to_string();
+                    *source = source[prefix.len()..].into();
                     break;
                 }
             }

src/decoder.rs

@@ -187,26 +187,34 @@ pub fn decode_regular(rsm: RawSourceMap) -> Result<SourceMap> {
         }
     }
 
-    let sources = sources.into_iter().map(Option::unwrap_or_default).collect();
+    let sources = sources
+        .into_iter()
+        .map(Option::unwrap_or_default)
+        .map(Into::into)
+        .collect();
 
     // apparently we can encounter some non string types in real world
     // sourcemaps :(
     let names = names
         .into_iter()
         .map(|val| match val {
-            Value::String(s) => s,
-            Value::Number(num) => num.to_string(),
+            Value::String(s) => s.into(),
+            Value::Number(num) => num.to_string().into(),
             _ => "".into(),
         })
-        .collect::<Vec<String>>();
+        .collect::<Vec<_>>();
 
     // file sometimes is not a string for unexplicable reasons
     let file = rsm.file.map(|val| match val {
-        Value::String(s) => s,
+        Value::String(s) => s.into(),
         _ => "<invalid>".into(),
     });
 
-    let mut sm = SourceMap::new(file, tokens, names, sources, rsm.sources_content);
+    let source_content = rsm
+        .sources_content
+        .map(|x| x.into_iter().map(|v| v.map(Into::into)).collect::<Vec<_>>());
+
+    let mut sm = SourceMap::new(file, tokens, names, sources, source_content);
     sm.set_source_root(rsm.source_root);
     sm.set_debug_id(rsm.debug_id);
 

src/ram_bundle.rs

@@ -70,9 +70,9 @@ impl<'a> RamBundleModule<'a> {
     /// Returns a source view of the data.
     ///
     /// This operation fails if the source code is not valid UTF-8.
-    pub fn source_view(&self) -> Result<SourceView<'a>> {
+    pub fn source_view(&self) -> Result<SourceView> {
         match std::str::from_utf8(self.data) {
-            Ok(s) => Ok(SourceView::new(s)),
+            Ok(s) => Ok(SourceView::new(s.into())),
             Err(e) => Err(Error::Utf8(e)),
         }
     }
@@ -361,7 +361,7 @@ impl<'a> SplitRamBundleModuleIter<'a> {
     fn split_module(
         &self,
         module: RamBundleModule<'a>,
-    ) -> Result<Option<(String, SourceView<'a>, SourceMap)>> {
+    ) -> Result<Option<(String, SourceView, SourceMap)>> {
         let module_offset = self
             .offsets
             .get(module.id())
@@ -377,7 +377,7 @@ impl<'a> SplitRamBundleModuleIter<'a> {
             return Err(Error::InvalidRamBundleEntry);
         }
 
-        let source: SourceView<'a> = module.source_view()?;
+        let source: SourceView = module.source_view()?;
         let line_count = source.line_count() as u32;
         let ending_line = starting_line + line_count;
         let last_line_len = source
@@ -416,7 +416,7 @@ impl<'a> SplitRamBundleModuleIter<'a> {
 }
 
 impl<'a> Iterator for SplitRamBundleModuleIter<'a> {
-    type Item = Result<(String, SourceView<'a>, SourceMap)>;
+    type Item = Result<(String, SourceView, SourceMap)>;
 
     fn next(&mut self) -> Option<Self::Item> {
         while let Some(module_result) = self.ram_bundle_iter.next() {

src/sourceview.rs

@@ -1,9 +1,9 @@
-use std::borrow::Cow;
 use std::fmt;
 use std::slice;
 use std::str;
 use std::sync::atomic::AtomicUsize;
 use std::sync::atomic::Ordering;
+use std::sync::Arc;
 use std::sync::Mutex;
 
 use if_chain::if_chain;
@@ -14,19 +14,13 @@ use crate::js_identifiers::{get_javascript_token, is_valid_javascript_identifier
 use crate::types::{idx_from_token, sourcemap_from_token, Token};
 
 /// An iterator that iterates over tokens in reverse.
-pub struct RevTokenIter<'view, 'viewbase, 'map>
-where
-    'viewbase: 'view,
-{
-    sv: &'view SourceView<'viewbase>,
+pub struct RevTokenIter<'view, 'map> {
+    sv: &'view SourceView,
     token: Option<Token<'map>>,
     source_line: Option<(&'view str, usize, usize, usize)>,
 }
 
-impl<'view, 'viewbase, 'map> Iterator for RevTokenIter<'view, 'viewbase, 'map>
-where
-    'viewbase: 'view,
-{
+impl<'view, 'map> Iterator for RevTokenIter<'view, 'map> {
     type Item = (Token<'map>, Option<&'view str>);
 
     fn next(&mut self) -> Option<(Token<'map>, Option<&'view str>)> {
@@ -118,7 +112,7 @@ where
 }
 
 pub struct Lines<'a> {
-    sv: &'a SourceView<'a>,
+    sv: &'a SourceView,
     idx: u32,
 }
 
@@ -139,14 +133,14 @@ impl<'a> Iterator for Lines<'a> {
 ///
 /// This type is used to implement fairly efficient source mapping
 /// operations.
-pub struct SourceView<'a> {
-    source: Cow<'a, str>,
+pub struct SourceView {
+    source: Arc<str>,
     processed_until: AtomicUsize,
     lines: Mutex<Vec<&'static str>>,
 }
 
-impl<'a> Clone for SourceView<'a> {
-    fn clone(&self) -> SourceView<'a> {
+impl Clone for SourceView {
+    fn clone(&self) -> SourceView {
         SourceView {
             source: self.source.clone(),
             processed_until: AtomicUsize::new(0),
@@ -155,28 +149,28 @@ impl<'a> Clone for SourceView<'a> {
     }
 }
 
-impl<'a> fmt::Debug for SourceView<'a> {
+impl fmt::Debug for SourceView {
     fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
         f.debug_struct("SourceView")
             .field("source", &self.source())
             .finish()
     }
 }
 
-impl<'a> SourceView<'a> {
+impl SourceView {
     /// Creates an optimized view of a given source.
-    pub fn new(source: &'a str) -> SourceView<'a> {
+    pub fn new(source: Arc<str>) -> SourceView {
         SourceView {
-            source: Cow::Borrowed(source),
+            source,
             processed_until: AtomicUsize::new(0),
             lines: Mutex::new(vec![]),
         }
     }
 
     /// Creates an optimized view from a given source string
-    pub fn from_string(source: String) -> SourceView<'static> {
+    pub fn from_string(source: String) -> SourceView {
         SourceView {
-            source: Cow::Owned(source),
+            source: source.into(),
             processed_until: AtomicUsize::new(0),
             lines: Mutex::new(vec![]),
         }
@@ -264,7 +258,7 @@ impl<'a> SourceView<'a> {
     }
 
     /// Returns an iterator over all lines.
-    pub fn lines(&'a self) -> Lines<'a> {
+    pub fn lines(&self) -> Lines {
         Lines { sv: self, idx: 0 }
     }
 
@@ -273,10 +267,7 @@ impl<'a> SourceView<'a> {
         &self.source
     }
 
-    fn rev_token_iter<'this, 'map>(
-        &'this self,
-        token: Token<'map>,
-    ) -> RevTokenIter<'this, 'a, 'map> {
+    fn rev_token_iter<'this, 'map>(&'this self, token: Token<'map>) -> RevTokenIter<'this, 'map> {
         RevTokenIter {
             sv: self,
             token: Some(token),
@@ -332,7 +323,7 @@ impl<'a> SourceView<'a> {
 #[test]
 #[allow(clippy::cognitive_complexity)]
 fn test_minified_source_view() {
-    let view = SourceView::new("a\nb\nc");
+    let view = SourceView::new("a\nb\nc".into());
     assert_eq!(view.get_line(0), Some("a"));
     assert_eq!(view.get_line(0), Some("a"));
     assert_eq!(view.get_line(2), Some("c"));
@@ -341,7 +332,7 @@ fn test_minified_source_view() {
 
     assert_eq!(view.line_count(), 3);
 
-    let view = SourceView::new("a\r\nb\r\nc");
+    let view = SourceView::new("a\r\nb\r\nc".into());
     assert_eq!(view.get_line(0), Some("a"));
     assert_eq!(view.get_line(0), Some("a"));
     assert_eq!(view.get_line(2), Some("c"));
@@ -350,7 +341,7 @@ fn test_minified_source_view() {
 
     assert_eq!(view.line_count(), 3);
 
-    let view = SourceView::new("abc👌def\nblah");
+    let view = SourceView::new("abc👌def\nblah".into());
     assert_eq!(view.get_line_slice(0, 0, 3), Some("abc"));
     assert_eq!(view.get_line_slice(0, 3, 1), Some("👌"));
     assert_eq!(view.get_line_slice(0, 3, 2), Some("👌"));
@@ -362,7 +353,7 @@ fn test_minified_source_view() {
     assert_eq!(view.get_line_slice(1, 0, 5), None);
     assert_eq!(view.get_line_slice(1, 0, 12), None);
 
-    let view = SourceView::new("a\nb\nc\n");
+    let view = SourceView::new("a\nb\nc\n".into());
     assert_eq!(view.get_line(0), Some("a"));
     assert_eq!(view.get_line(1), Some("b"));
     assert_eq!(view.get_line(2), Some("c"));
@@ -371,6 +362,6 @@ fn test_minified_source_view() {
 
     fn is_send<T: Send>() {}
     fn is_sync<T: Sync>() {}
-    is_send::<SourceView<'static>>();
-    is_sync::<SourceView<'static>>();
+    is_send::<SourceView>();
+    is_sync::<SourceView>();
 }