perf(es/preset_env): Use phf for corejs3 entry

Author: quininer

Cargo.lock

@@ -26,6 +26,9 @@ serde      = { workspace = true, features = ["derive"], optional = true }
 serde_json = { workspace = true }
 st-map     = { workspace = true }
 
+precomputed-map = { workspace = true }
+xxhash-rust     = { workspace = true, features = [ "xxh3" ] }
+
 preset_env_base = { version = "3.0.2", path = "../preset_env_base" }
 rustc-hash = { workspace = true }
 string_enum = { version = "1.0.1", path = "../string_enum" }

crates/swc_ecma_preset_env/Cargo.toml

@@ -81,6 +81,16 @@ fn bench_cases(c: &mut Criterion) {
 
         run(b, SOURCE, Default::default())
     });
+
+    c.bench_function("es/preset-env/entry/import", |b| {
+        const SOURCE: &str = r#"
+        import {} from "aaa.js";
+        "#;
+
+        let mut config = Config::default();
+        config.mode = Some(swc_ecma_preset_env::Mode::Entry);
+        run(b, SOURCE, config)
+    });    
 }
 
 criterion_group!(benches, bench_cases);

crates/swc_ecma_preset_env/benches/polyfills.rs

@@ -1,35 +1,47 @@
 use std::sync::Arc;
 
 use indexmap::IndexSet;
-use once_cell::sync::Lazy;
 use preset_env_base::{
     version::{should_enable, Version},
     Versions,
 };
-use rustc_hash::{FxBuildHasher, FxHashMap};
+use rustc_hash::FxBuildHasher;
 use swc_atoms::atom;
 use swc_common::DUMMY_SP;
 use swc_ecma_ast::*;
 use swc_ecma_visit::VisitMut;
+use std::ops::Range;
+use crate::util::SwcXxh3;
 
 use super::{compat::DATA as CORE_JS_COMPAT_DATA, data::MODULES_BY_VERSION};
 
-static ENTRIES: Lazy<FxHashMap<String, Vec<&'static str>>> = Lazy::new(|| {
-    serde_json::from_str::<FxHashMap<String, Vec<String>>>(include_str!(
-        "../../data/core-js-compat/entries.json"
-    ))
-    .expect("failed to parse entries.json from core js 3")
-    .into_iter()
-    .map(|(k, v)| {
-        (
-            k,
-            v.into_iter()
-                .map(|s: String| &*Box::leak(s.into_boxed_str()))
-                .collect::<Vec<_>>(),
-        )
-    })
-    .collect()
-});
+include!("../generated/corejs3_entries.rs");
+
+pub struct FeatureSet(Range<u32>);
+        
+pub fn entries_get(name: &str) -> Option<FeatureSet> {{
+    let index = ENTRY_INDEX.get(name)?;
+    ENTRY_VALUES_LIST
+        .get(index)
+        .cloned()
+        .map(FeatureSet)
+}}
+
+impl FeatureSet {
+    pub fn iter(&self) -> impl ExactSizeIterator<Item = &'static str> {
+        use precomputed_map::store::AccessSeq;
+        
+        self.0
+            .clone()
+            .map(|idx| ENTRY_VALUES_STRING_ID.index(idx as usize))
+            .map(|id| {
+                let offset = id & ((1 << 24) - 1);
+                let len = id >> 24;
+
+                &ENTRY_VALUES_STRING_STORE[(offset as usize)..][..(len as usize)]
+            })
+    }
+}
 
 #[derive(Debug)]
 pub struct Entry {
@@ -68,9 +80,9 @@ impl Entry {
             return true;
         }
 
-        if let Some(features) = ENTRIES.get(src) {
+        if let Some(features) = entries_get(src) {
             self.imports.extend(features.iter().filter(|f| {
-                let feature = CORE_JS_COMPAT_DATA.get(&***f);
+                let feature = CORE_JS_COMPAT_DATA.get(*f);
 
                 if !*is_any_target {
                     if let Some(feature) = feature {
@@ -80,7 +92,7 @@ impl Entry {
                     }
                 }
 
-                if let Some(version) = MODULES_BY_VERSION.get(**f) {
+                if let Some(version) = MODULES_BY_VERSION.get(f) {
                     return version <= corejs_version;
                 }
 

crates/swc_ecma_preset_env/src/corejs3/entry.rs

@@ -0,0 +1 @@
+es.symboles.symbol.descriptiones.symbol.async-iteratores.symbol.has-instancees.symbol.is-concat-spreadablees.symbol.iteratores.symbol.matches.symbol.match-alles.symbol.replacees.symbol.searches.symbol.specieses.symbol.splites.symbol.to-primitivees.symbol.to-string-tages.symbol.unscopableses.error.causees.error.to-stringes.aggregate-errores.aggregate-error.causees.array.ates.array.concates.array.copy-withines.array.everyes.array.filles.array.filteres.array.findes.array.find-indexes.array.find-lastes.array.find-last-indexes.array.flates.array.flat-mapes.array.for-eaches.array.fromes.array.includeses.array.index-ofes.array.is-arrayes.array.iteratores.array.joines.array.last-index-ofes.array.mapes.array.ofes.array.pushes.array.reducees.array.reduce-rightes.array.reversees.array.slicees.array.somees.array.sortes.array.specieses.array.splicees.array.to-reversedes.array.to-sortedes.array.to-splicedes.array.unscopables.flates.array.unscopables.flat-mapes.array.unshiftes.array.withes.array-buffer.constructores.array-buffer.is-viewes.array-buffer.slicees.data-viewes.array-buffer.detachedes.array-buffer.transferes.array-buffer.transfer-to-fixed-lengthes.date.get-yeares.date.nowes.date.set-yeares.date.to-gmt-stringes.date.to-iso-stringes.date.to-jsones.date.to-primitivees.date.to-stringes.escapees.function.bindes.function.has-instancees.function.namees.global-thises.json.stringifyes.json.to-string-tages.mapes.map.group-byes.math.acoshes.math.asinhes.math.atanhes.math.cbrtes.math.clz32es.math.coshes.math.expm1es.math.froundes.math.hypotes.math.imules.math.log10es.math.log1pes.math.log2es.math.signes.math.sinhes.math.tanhes.math.to-string-tages.math.trunces.number.constructores.number.epsilones.number.is-finitees.number.is-integeres.number.is-nanes.number.is-safe-integeres.number.max-safe-integeres.number.min-safe-integeres.number.parse-floates.number.parse-intes.number.to-exponentiales.number.to-fixedes.number.to-precisiones.object.assignes.object.createes.object.define-getteres.object.define-propertieses.object.define-propertyes.object.define-setteres.object.entrieses.object.freezees.object.from-entrieses.object.get-own-property-descriptores.object.get-own-property-descriptorses.object.get-own-property-nameses.object.get-prototype-ofes.object.group-byes.object.has-ownes.object.ises.object.is-extensiblees.object.is-frozenes.object.is-sealedes.object.keyses.object.lookup-getteres.object.lookup-setteres.object.prevent-extensionses.object.protoes.object.seales.object.set-prototype-ofes.object.to-stringes.object.valueses.parse-floates.parse-intes.promisees.promise.all-settledes.promise.anyes.promise.finallyes.promise.with-resolverses.reflect.applyes.reflect.constructes.reflect.define-propertyes.reflect.delete-propertyes.reflect.getes.reflect.get-own-property-descriptores.reflect.get-prototype-ofes.reflect.hases.reflect.is-extensiblees.reflect.own-keyses.reflect.prevent-extensionses.reflect.setes.reflect.set-prototype-ofes.reflect.to-string-tages.regexp.constructores.regexp.dot-alles.regexp.execes.regexp.flagses.regexp.stickyes.regexp.testes.regexp.to-stringes.setes.set.difference.v2es.set.intersection.v2es.set.is-disjoint-from.v2es.set.is-subset-of.v2es.set.is-superset-of.v2es.set.symmetric-difference.v2es.set.union.v2es.string.at-alternativees.string.code-point-ates.string.ends-withes.string.from-code-pointes.string.includeses.string.is-well-formedes.string.iteratores.string.matches.string.match-alles.string.pad-endes.string.pad-startes.string.rawes.string.repeates.string.replacees.string.replace-alles.string.searches.string.splites.string.starts-withes.string.substres.string.to-well-formedes.string.trimes.string.trim-endes.string.trim-startes.string.anchores.string.biges.string.blinkes.string.boldes.string.fixedes.string.fontcolores.string.fontsizees.string.italicses.string.linkes.string.smalles.string.strikees.string.subes.string.supes.typed-array.float32-arrayes.typed-array.float64-arrayes.typed-array.int8-arrayes.typed-array.int16-arrayes.typed-array.int32-arrayes.typed-array.uint8-arrayes.typed-array.uint8-clamped-arrayes.typed-array.uint16-arrayes.typed-array.uint32-arrayes.typed-array.ates.typed-array.copy-withines.typed-array.everyes.typed-array.filles.typed-array.filteres.typed-array.findes.typed-array.find-indexes.typed-array.find-lastes.typed-array.find-last-indexes.typed-array.for-eaches.typed-array.fromes.typed-array.includeses.typed-array.index-ofes.typed-array.iteratores.typed-array.joines.typed-array.last-index-ofes.typed-array.mapes.typed-array.ofes.typed-array.reducees.typed-array.reduce-rightes.typed-array.reversees.typed-array.setes.typed-array.slicees.typed-array.somees.typed-array.sortes.typed-array.subarrayes.typed-array.to-locale-stringes.typed-array.to-reversedes.typed-array.to-sortedes.typed-array.to-stringes.typed-array.withes.unescapees.weak-mapes.weak-setesnext.aggregate-erroresnext.suppressed-error.constructoresnext.array.from-asyncesnext.array.atesnext.array.filter-outesnext.array.filter-rejectesnext.array.find-lastesnext.array.find-last-indexesnext.array.groupesnext.array.group-byesnext.array.group-by-to-mapesnext.array.group-to-mapesnext.array.is-template-objectesnext.array.last-indexesnext.array.last-itemesnext.array.to-reversedesnext.array.to-sortedesnext.array.to-splicedesnext.array.unique-byesnext.array.withesnext.array-buffer.detachedesnext.array-buffer.transferesnext.array-buffer.transfer-to-fixed-lengthesnext.async-disposable-stack.constructoresnext.async-iterator.constructoresnext.async-iterator.as-indexed-pairsesnext.async-iterator.async-disposeesnext.async-iterator.dropesnext.async-iterator.everyesnext.async-iterator.filteresnext.async-iterator.findesnext.async-iterator.flat-mapesnext.async-iterator.for-eachesnext.async-iterator.fromesnext.async-iterator.indexedesnext.async-iterator.mapesnext.async-iterator.reduceesnext.async-iterator.someesnext.async-iterator.takeesnext.async-iterator.to-arrayesnext.bigint.rangeesnext.composite-keyesnext.composite-symbolesnext.data-view.get-float16esnext.data-view.get-uint8-clampedesnext.data-view.set-float16esnext.data-view.set-uint8-clampedesnext.disposable-stack.constructoresnext.function.demethodizeesnext.function.is-callableesnext.function.is-constructoresnext.function.metadataesnext.function.un-thisesnext.global-thisesnext.iterator.constructoresnext.iterator.as-indexed-pairsesnext.iterator.disposeesnext.iterator.dropesnext.iterator.everyesnext.iterator.filteresnext.iterator.findesnext.iterator.flat-mapesnext.iterator.for-eachesnext.iterator.fromesnext.iterator.indexedesnext.iterator.mapesnext.iterator.rangeesnext.iterator.reduceesnext.iterator.someesnext.iterator.takeesnext.iterator.to-arrayesnext.iterator.to-asyncesnext.json.is-raw-jsonesnext.json.parseesnext.json.raw-jsonesnext.map.delete-allesnext.map.emplaceesnext.map.everyesnext.map.filteresnext.map.findesnext.map.find-keyesnext.map.fromesnext.map.group-byesnext.map.includesesnext.map.key-byesnext.map.key-ofesnext.map.map-keysesnext.map.map-valuesesnext.map.mergeesnext.map.ofesnext.map.reduceesnext.map.someesnext.map.updateesnext.map.update-or-insertesnext.map.upsertesnext.math.clampesnext.math.deg-per-radesnext.math.degreesesnext.math.fscaleesnext.math.f16roundesnext.math.iaddhesnext.math.imulhesnext.math.isubhesnext.math.rad-per-degesnext.math.radiansesnext.math.scaleesnext.math.seeded-prngesnext.math.signbitesnext.math.sum-preciseesnext.math.umulhesnext.number.from-stringesnext.number.rangeesnext.object.has-ownesnext.object.iterate-entriesesnext.object.iterate-keysesnext.object.iterate-valuesesnext.object.group-byesnext.observableesnext.promise.all-settledesnext.promise.anyesnext.promise.tryesnext.promise.with-resolversesnext.reflect.define-metadataesnext.reflect.delete-metadataesnext.reflect.get-metadataesnext.reflect.get-metadata-keysesnext.reflect.get-own-metadataesnext.reflect.get-own-metadata-keysesnext.reflect.has-metadataesnext.reflect.has-own-metadataesnext.reflect.metadataesnext.regexp.escapeesnext.set.add-allesnext.set.delete-allesnext.set.difference.v2esnext.set.differenceesnext.set.everyesnext.set.filteresnext.set.findesnext.set.fromesnext.set.intersection.v2esnext.set.intersectionesnext.set.is-disjoint-from.v2esnext.set.is-disjoint-fromesnext.set.is-subset-of.v2esnext.set.is-subset-ofesnext.set.is-superset-of.v2esnext.set.is-superset-ofesnext.set.joinesnext.set.mapesnext.set.ofesnext.set.reduceesnext.set.someesnext.set.symmetric-difference.v2esnext.set.symmetric-differenceesnext.set.union.v2esnext.set.unionesnext.string.atesnext.string.cookedesnext.string.code-pointsesnext.string.dedentesnext.string.is-well-formedesnext.string.match-allesnext.string.replace-allesnext.string.to-well-formedesnext.symbol.async-disposeesnext.symbol.custom-matcheresnext.symbol.disposeesnext.symbol.is-registered-symbolesnext.symbol.is-registeredesnext.symbol.is-well-known-symbolesnext.symbol.is-well-knownesnext.symbol.matcheresnext.symbol.metadataesnext.symbol.metadata-keyesnext.symbol.observableesnext.symbol.pattern-matchesnext.symbol.replace-allesnext.typed-array.from-asyncesnext.typed-array.atesnext.typed-array.filter-outesnext.typed-array.filter-rejectesnext.typed-array.find-lastesnext.typed-array.find-last-indexesnext.typed-array.group-byesnext.typed-array.to-reversedesnext.typed-array.to-sortedesnext.typed-array.to-splicedesnext.typed-array.unique-byesnext.typed-array.withesnext.uint8-array.from-base64esnext.uint8-array.from-hexesnext.uint8-array.set-from-base64esnext.uint8-array.set-from-hexesnext.uint8-array.to-base64esnext.uint8-array.to-hexesnext.weak-map.delete-allesnext.weak-map.fromesnext.weak-map.ofesnext.weak-map.emplaceesnext.weak-map.upsertesnext.weak-set.add-allesnext.weak-set.delete-allesnext.weak-set.fromesnext.weak-set.ofweb.atobweb.btoaweb.dom-collections.for-eachweb.dom-collections.iteratorweb.dom-exception.constructorweb.dom-exception.stackweb.dom-exception.to-string-tagweb.immediateweb.queue-microtaskweb.selfweb.structured-cloneweb.timersweb.urlweb.url.can-parseweb.url.parseweb.url.to-jsonweb.url-search-paramsweb.url-search-params.deleteweb.url-search-params.hasweb.url-search-params.sizees.aggregate-error.constructores.data-view.constructores.map.constructores.object.get-own-property-symbolses.promise.alles.promise.catches.promise.constructores.promise.racees.promise.rejectes.promise.resolvees.set.constructores.string.trim-leftes.string.trim-rightes.symbol.constructores.symbol.fores.symbol.key-fores.weak-map.constructores.weak-set.constructoresnext.observable.constructoresnext.observable.fromesnext.observable.ofesnext.string.at-alternativeweb.clear-immediateweb.set-immediateweb.set-intervalweb.set-timeoutweb.url-search-params.constructorweb.url.constructor

crates/swc_ecma_preset_env/src/generated/corejs3_entries.rs

@@ -687,3 +687,17 @@ impl<T> DataMapExt<T> for DataMap<T> {
         self
     }
 }
+
+pub struct PooledStr(u32);
+
+pub struct SwcXxh3;
+
+impl precomputed_map::phf::HashOne for SwcXxh3 {
+    fn hash_one<T: std::hash::Hash>(k: u64, v: T) -> u64 {
+        use std::hash::Hasher;
+        
+        let mut hasher = xxhash_rust::xxh3::Xxh3::with_seed(k);
+        v.hash(&mut hasher);
+        hasher.finish()
+    }
+}

crates/swc_ecma_preset_env/src/generated/corejs3_entries.str

@@ -1,6 +1,7 @@
 use std::fs;
 use std::io::Write;
 use std::path::Path;
+use std::collections::HashMap;
 use clap::Args;
 use anyhow::Context;
 
@@ -12,29 +13,34 @@ impl CodegenCmd {
         let dir = Path::new(env!("CARGO_MANIFEST_DIR"));
         let dir = dir.parent().unwrap();
 
-        es_minifier_js_environment_props(dir)
+        es_preset_env_corejs3_entry(dir)
     }
 }
 
-fn es_minifier_js_environment_props(dir: &Path) -> anyhow::Result<()> {
-    let crate_dir = dir.join("crates/swc_ecma_minifier/");
+fn es_preset_env_corejs3_entry(dir: &Path) -> anyhow::Result<()> {
+    use std::collections::BTreeMap;
 
-    let domprops = fs::read_to_string(crate_dir.join("src/lists/domprops.json"))?;
-    let jsprops = fs::read_to_string(crate_dir.join("src/lists/jsprops.json"))?;
+    let crate_dir = dir.join("crates/swc_ecma_preset_env/");
+    
+    let entry_data = fs::read_to_string(crate_dir.join("data/core-js-compat/entries.json"))?;
+    let entry_data: BTreeMap<&str, Vec<&str>> = serde_json::from_str(&entry_data)
+        .context("failed to parse entries.json from core js 3")?;
+    let (keys, values): (Vec<_>, Vec<_>) = entry_data.into_iter().unzip();
 
-    let props = {
-        let mut domprops: Vec<&str> = serde_json::from_str(&domprops)
-            .context("failed to parse domprops.json for property mangler")?;
-        let mut jsprops: Vec<&str> = serde_json::from_str(&jsprops)
-            .context("Failed to parse jsprops.json for property mangler")?;
-        domprops.append(&mut jsprops);
-        domprops.sort_unstable();
-        domprops.dedup();
-        domprops
-    };
+    let mut strpool = StrPool::default();
+    let mut values_strid = Vec::new();
+    let mut values_index = Vec::new();
+    for list in values {
+        let start: u32 = values_strid.len().try_into().unwrap();
+        for s in list {
+            values_strid.push(strpool.insert(s));
+        }
+        let end: u32 = values_strid.len().try_into().unwrap();
+        values_index.push(start..end);
+    }
 
-    let mapout = precomputed_map::builder::MapBuilder::new(&props)
-        .set_ord(&|&x, &y| x.cmp(&y))
+    let mapout = precomputed_map::builder::MapBuilder::new(&keys)
+        .set_seed(16416001479773392852)
         .set_hash(&|seed, &v| {
             use std::hash::{ Hash, Hasher };
 
@@ -43,33 +49,59 @@ fn es_minifier_js_environment_props(dir: &Path) -> anyhow::Result<()> {
             hasher.finish()
         })
         .set_next_seed(|seed, c| {
-            dbg!(c);
             xxhash_rust::xxh3::xxh3_64_with_seed(&c.to_le_bytes(), seed)
         })
         .build()?;
+    println!("seed: {:?}", mapout.seed());
+    
     let mut builder = precomputed_map::builder::CodeBuilder::new(
-        "props".into(),
+        "corejs3_entries".into(),
         "SwcXxh3".into(),
         crate_dir.join("src/generated")
     );
 
     let _ = fs::create_dir(crate_dir.join("src/generated"));
 
-    let k = builder.create_str_seq("PROPS_KEYS".into(), mapout.reorder(&props))?;
-    mapout.create_map("PROPS".into(), k, &mut builder)?;
+    let k = builder.create_str_seq("ENTRY_KEYS".into(), mapout.reorder(&keys))?;
+    builder.create_u32_seq("ENTRY_VALUES_STRING_ID".into(), values_strid.iter().copied())?;
+    mapout.create_map("ENTRY_INDEX".into(), k, &mut builder)?;
 
-    let mut codeout = fs::File::create(crate_dir.join("src/generated/props.rs"))?;
+    let mut codeout = fs::File::create(crate_dir.join("src/generated/corejs3_entries.rs"))?;
     builder.write_to(&mut codeout)?;
 
+    fs::write(crate_dir.join("src/generated/corejs3_entries.strpool"), strpool.pool.as_bytes())?;
+
     writeln!(codeout,
-        r#"
-use crate::util::SwcXxh3;
-        
-pub fn is_exist(name: &str) -> bool {{
-    PROPS.get(name).is_some()
-}}
-"#
+        "static ENTRY_VALUES_STRING_STORE: &str = include_str!(\"corejs3_entries.strpool\");
+        static ENTRY_VALUES_LIST: &[Range<u32>] = &["
     )?;
+    for range in mapout.reorder(&values_index) {
+        writeln!(codeout, "{}..{},", range.start, range.end)?;
+    }
+    writeln!(codeout, "];")?;
 
     Ok(())
 }
+
+#[derive(Default)]
+struct StrPool<'s> {
+    pool: String,
+    map: HashMap<&'s str, u32>,
+}
+
+impl<'s> StrPool<'s> {
+    pub fn insert(&mut self, s: &'s str) -> u32 {
+        *self.map.entry(s).or_insert_with(|| {
+            let offset = self.pool.len();
+            self.pool.push_str(&s);
+            let len: u8 = (self.pool.len() - offset).try_into().unwrap();
+            let offset: u32 = offset.try_into().unwrap();
+
+            if offset > (1 << 24) {
+                panic!("string too large");
+            }
+
+            offset | (u32::from(len) << 24)
+        })
+    }
+}