Merge pull request #3 from Brooooooklyn/jieba

feat(jieba): impl jieba-rs binding

Author: Brooooooklyn

.github/workflows/ci.yaml

@@ -90,7 +90,9 @@ jobs:
         if: matrix.os == 'ubuntu-latest'
 
       - name: Run build
-        run: npx lerna run build --stream -- --platform
+        run: |
+          cargo build --release
+          npx lerna run build --stream -- --platform
 
       - name: Upload artifact
         uses: actions/upload-artifact@v2
@@ -196,6 +198,7 @@ jobs:
         shell: bash
       - name: Lerna publish
         run: |
+          find ./packages/ -type d -maxdepth 1 -exec cp LICENSE {} \;
           echo "//registry.npmjs.org/:_authToken=$NPM_TOKEN" >> ~/.npmrc
           npx lerna publish from-package --no-verify-access --yes
         env:

Cargo.toml

@@ -1,4 +1,5 @@
 [workspace]
 members = [
-  "./packages/crc32"
+  "./packages/crc32",
+  "./packages/jieba"
 ]

packages/crc32/LICENSE renamed to LICENSE

@@ -16,6 +16,7 @@ Make rust crates binding to NodeJS use [napi-rs](https://github.com/Brooooooklyn
 
 # Packages
 
-| Package                                        | Status                                                              | Description                                 |
-| ---------------------------------------------- | ------------------------------------------------------------------- | ------------------------------------------- |
-| [`@node-rs/crc32`](./packages/crc32/README.md) | ![](https://github.com/Brooooooklyn/node-rs/workflows/CI/badge.svg) | Fastest `CRC32` implementation using `SIMD` |
+| Package                                        | Status                                                              | Description                                                |
+| ---------------------------------------------- | ------------------------------------------------------------------- | ---------------------------------------------------------- |
+| [`@node-rs/crc32`](./packages/crc32/README.md) | ![](https://github.com/Brooooooklyn/node-rs/workflows/CI/badge.svg) | Fastest `CRC32` implementation using `SIMD`                |
+| [`@node-rs/jieba`](./packages/jieba/README,md) | ![](https://github.com/Brooooooklyn/node-rs/workflows/CI/badge.svg) | [`jieba-rs`](https://github.com/messense/jieba-rs) binding |

README.md

@@ -8,7 +8,7 @@
   "private": true,
   "workspaces": ["packages/*"],
   "scripts": {
-    "bench": "lerna run bench --stream --no-prefix",
+    "bench": "lerna run bench --concurrency 1 --stream --no-prefix",
     "build:ts": "tsc -b tsconfig.project.json -verbose",
     "lint": "eslint . -c ./.eslintrc.yml 'packages/**/*.{ts,js}'",
     "test": "ava",

package.json

@@ -12,5 +12,8 @@ napi-rs = { version = "0.2" }
 napi-rs-derive = { version = "0.1" }
 crc32fast = "1.2"
 
+[target.'cfg(unix)'.dependencies]
+jemallocator = { version = "0.3", features = ["disable_initial_exec_tls"] }
+
 [build-dependencies]
 napi-build = { version = "0.1" }

packages/crc32/Cargo.toml

@@ -11,11 +11,11 @@ The 4 tested implementations are:
 - **js_crc32c** Javascript implemented CRC-32C
 - **js_crc32** Javascript implemented CRC-32 from [buffer-crc32](https://github.com/brianloveswords/buffer-crc32)
 
-# Performance
+## Performance
 
 ```bash
-sse4_crc32c_node_rs for inputs 1024B x 5,108,123 ops/sec ±1.86% (89 runs sampled)
-sse4_crc32c_node_rs for inputs 16931844B, avg 2066B x 271 ops/sec ±1.15% (85 runs sampled)
+@node-rs/crc32 for inputs 1024B x 5,108,123 ops/sec ±1.86% (89 runs sampled)
+@node-rs/crc32 for inputs 16931844B, avg 2066B x 271 ops/sec ±1.15% (85 runs sampled)
 sse4_crc32c_hw for inputs 1024B x 3,543,443 ops/sec ±1.39% (93 runs sampled)
 sse4_crc32c_hw for inputs 16931844B, avg 2066B x 209 ops/sec ±0.78% (76 runs sampled)
 sse4_crc32c_sw for inputs 1024B x 1,460,284 ops/sec ±2.35% (90 runs sampled)
@@ -27,7 +27,7 @@ js_crc32 for inputs 16931844B, avg 2066B x 22.12 ops/sec ±5.20% (40 runs sample
 +---------------------+-------------------+----------------------+
 |                     │ 1024B             │ 16931844B, avg 2066B |
 +---------------------+-------------------+----------------------+
-| sse4_crc32c_node_rs │ 5,108,123 ops/sec │ 271 ops/sec          |
+| @node-rs/crc32      │ 5,108,123 ops/sec │ 271 ops/sec          |
 +---------------------+-------------------+----------------------+
 | sse4_crc32c_hw      │ 3,543,443 ops/sec │ 209 ops/sec          |
 +---------------------+-------------------+----------------------+
@@ -39,13 +39,13 @@ js_crc32 for inputs 16931844B, avg 2066B x 22.12 ops/sec ±5.20% (40 runs sample
 +---------------------+-------------------+----------------------+
 ```
 
-# Support matrix
+## Support matrix
 
 |                   | node 10 | node12 | node13 | node14 |
 | ----------------- | ------- | ------ | ------ | ------ |
-| Windows 64 latest | ✅      | ✅     | ✅     | ✅     |
-| macOS latest      | ✅      | ✅     | ✅     | ✅     |
-| Linux             | ✅      | ✅     | ✅     | ✅     |
+| Windows 64 latest | ✓       | ✓      | ✓      | ✓      |
+| macOS latest      | ✓       | ✓      | ✓      | ✓      |
+| Linux             | ✓       | ✓      | ✓      | ✓      |
 
 ## API
 

packages/crc32/README.md

@@ -23,13 +23,13 @@ egestas tempus leo. Duis condimentum turpis duis.`)
 const initialCrc32 = crc32Node(TEST_BUFFER)
 const initialCrc32c = Sse4Crc32.calculate(TEST_BUFFER)
 
-console.assert(crc32(TEST_BUFFER), initialCrc32)
-console.assert(crc32c(TEST_BUFFER), initialCrc32c)
+console.assert(crc32(TEST_BUFFER) === initialCrc32)
+console.assert(crc32c(TEST_BUFFER) === initialCrc32c)
 
 const suite = new Suite('crc32c without initial crc')
 
 suite
-  .add('SIMD + NAPI', () => {
+  .add('@node/rs crc32c', () => {
     crc32c(TEST_BUFFER)
   })
   .add('sse4_crc32', () => {
@@ -46,7 +46,7 @@ suite
 const suite2 = new Suite('crc32c with initial crc')
 
 suite2
-  .add('SIMD + NAPI', () => {
+  .add('@node/rs crc32c', () => {
     crc32c(TEST_BUFFER, initialCrc32c)
   })
   .add('sse4_crc32', () => {
@@ -63,7 +63,7 @@ suite2
 const suite3 = new Suite('crc32 without initial crc')
 
 suite3
-  .add('SIMD + NAPI', () => {
+  .add('@node/rs crc32', () => {
     crc32(TEST_BUFFER)
   })
   .add('Node crc', () => {
@@ -80,7 +80,7 @@ suite3
 const suite4 = new Suite('crc32 with initial crc')
 
 suite4
-  .add('SIMD + NAPI', () => {
+  .add('@node/rs crc32', () => {
     crc32(TEST_BUFFER, initialCrc32)
   })
   .add('Node crc32', () => {

packages/crc32/benchmark/crc32.js

@@ -19,8 +19,8 @@
   },
   "scripts": {
     "bench": "cross-env NODE_ENV=production node benchmark/crc32.js",
-    "build": "cargo build --release && napi --release ./crc32",
-    "build:debug": "cargo build && napi ./index"
+    "build": "napi --release ./crc32",
+    "build:debug": "napi ./index"
   },
   "bugs": {
     "url": "https://github.com/Brooooooklyn/node-rs/issues"

packages/crc32/package.json

@@ -8,6 +8,10 @@ use crc32fast::Hasher;
 use napi::{Buffer, CallContext, Env, Number, Object, Result, Value};
 use std::convert::TryInto;
 
+#[cfg(unix)]
+#[global_allocator]
+static ALLOC: jemallocator::Jemalloc = jemallocator::Jemalloc;
+
 mod bytes;
 mod crc32;
 mod crc32_table;
@@ -24,7 +28,7 @@ fn init<'env>(
 }
 
 #[js_function(2)]
-fn crc32c<'a>(ctx: CallContext<'a>) -> Result<Value<'a, Number>> {
+fn crc32c(ctx: CallContext) -> Result<Value<Number>> {
   let input_data = ctx.get::<Buffer>(0)?;
   let init_state = ctx.get::<Number>(1);
   let result = if init_state.is_ok() {
@@ -36,7 +40,7 @@ fn crc32c<'a>(ctx: CallContext<'a>) -> Result<Value<'a, Number>> {
 }
 
 #[js_function(2)]
-fn crc32<'a>(ctx: CallContext<'a>) -> Result<Value<'a, Number>> {
+fn crc32(ctx: CallContext) -> Result<Value<Number>> {
   let input_data = ctx.get::<Buffer>(0)?;
   let init_state = ctx.get::<Number>(1);
   let mut hasher = if init_state.is_ok() {