perf(atoms): Improve static atom set of JsWord

Cargo.toml

@@ -41,3 +41,6 @@ opt-level = 3
 
 [profile.test.package.pretty_assertions]
 opt-level = 3
+
+[patch.crates-io]
+string_cache = { git = 'https://github.com/kdy1/string-cache', branch = 'drop' }

crates/dbg-swc/src/extra.rs

@@ -0,0 +1,67 @@
+use std::{path::PathBuf, sync::Arc};
+
+use anyhow::{anyhow, Result};
+use clap::{Args, Subcommand};
+use swc_atoms::JsWord;
+use swc_common::SourceMap;
+use swc_ecma_ast::{EsVersion, Ident};
+use swc_ecma_parser::{parse_file_as_program, TsConfig};
+use swc_ecma_visit::{Visit, VisitWith};
+
+#[derive(Debug, Subcommand)]
+pub enum ExtraCommand {
+    PrintAtoms(PrintAtomsComamnd),
+}
+
+impl ExtraCommand {
+    pub fn run(self, cm: Arc<SourceMap>) -> Result<()> {
+        match self {
+            ExtraCommand::PrintAtoms(cmd) => cmd.run(cm),
+        }
+    }
+}
+
+#[derive(Debug, Args)]
+pub struct PrintAtomsComamnd {
+    pub path: PathBuf,
+}
+
+impl PrintAtomsComamnd {
+    pub fn run(self, cm: Arc<SourceMap>) -> Result<()> {
+        let fm = cm.load_file(&self.path)?;
+
+        let p = parse_file_as_program(
+            &fm,
+            swc_ecma_parser::Syntax::Typescript(TsConfig {
+                dts: true,
+                ..Default::default()
+            }),
+            EsVersion::latest(),
+            None,
+            &mut vec![],
+        )
+        .map_err(|err| anyhow!("failed to parse: {:?}", err))?;
+
+        let mut v = IdentCollector::default();
+        p.visit_with(&mut v);
+        v.ident.sort();
+        v.ident.dedup();
+
+        for i in v.ident {
+            println!("{}", i);
+        }
+
+        Ok(())
+    }
+}
+
+#[derive(Default)]
+struct IdentCollector {
+    ident: Vec<JsWord>,
+}
+
+impl Visit for IdentCollector {
+    fn visit_ident(&mut self, ident: &Ident) {
+        self.ident.push(ident.sym.clone());
+    }
+}

crates/dbg-swc/src/main.rs

@@ -13,13 +13,15 @@ use tracing_subscriber::EnvFilter;
 
 use self::{
     bundle::BundleCommand,
+    extra::ExtraCommand,
     minify::MinifyCommand,
     test::TestCommand,
     util::{minifier::get_esbuild_output, print_js},
 };
 use crate::util::minifier::{get_minified, get_terser_output};
 
 mod bundle;
+mod extra;
 mod minify;
 mod test;
 mod util;
@@ -42,6 +44,9 @@ enum Cmd {
     Minify(MinifyCommand),
     #[clap(subcommand)]
     Test(TestCommand),
+
+    #[clap(subcommand)]
+    X(ExtraCommand),
 }
 
 fn init() -> Result<()> {
@@ -150,6 +155,7 @@ fn main() -> Result<()> {
                     Cmd::Bundle(_) => todo!(),
                     Cmd::Minify(cmd) => cmd.run(cm),
                     Cmd::Test(cmd) => cmd.run(cm),
+                    Cmd::X(cmd) => cmd.run(cm),
                 })
             })
         },

crates/swc_atoms/build.rs

@@ -1,6 +1,9 @@
 use std::{env, path::Path};
 
 fn main() {
+    print!("cargo:rerun-if-changed=build.rs");
+    print!("cargo:rerun-if-changed=words.txt");
+
     let strs = include_str!("words.txt")
         .lines()
         .map(|l| l.trim())

crates/swc_atoms/scripts/add-all.sh

@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+set -eu
+
+cargo build -p dbg-swc
+BINARY="$(cargo metadata --format-version 1 | jq -r '.target_directory')/debug/dbg-swc"
+echo "Binary:$BINARY"
+# TypeScript libraries has lots of commonly used identifiers.
+ls ../../node_modules/typescript/lib/lib.*.d.ts | xargs -L 1 -I {} $BINARY x print-atoms {} | sed -r '/^.{,3}$/d' >> words.txt || true
+
+./scripts/sort.sh