feat(jieba): support weight value in extract

Author: Brooooooklyn

packages/jieba/Cargo.toml

@@ -8,7 +8,7 @@ edition = "2018"
 crate-type = ["cdylib"]
 
 [dependencies]
-jieba-rs = { version = "0.5", features = ["default-dict", "tfidf", "textrank"] }
+jieba-rs = { version = "0.6.0-alpha.1", features = ["default-dict", "tfidf", "textrank"] }
 napi = { version = "0.4"}
 napi-derive = { version = "0.4" }
 once_cell = "1.4"

packages/jieba/__tests__/jieba.spec.ts

@@ -18,7 +18,13 @@ test('extract should be equal to nodejieba', (t) => {
     '今天纽约的天气真好啊，京华大酒店的张尧经理吃了一只北京烤鸭。后天纽约的天气不好，昨天纽约的天气也不好，北京烤鸭真好吃'
   const topn = 3
   t.deepEqual(
-    extract(sentence, topn),
-    nodejieba.extract(sentence, topn).map((t) => t.word),
+    extract(sentence, topn).map((t) => ({
+      keyword: t.keyword,
+      weight: typeof t.weight,
+    })),
+    nodejieba.extract(sentence, topn).map((t) => ({
+      keyword: t.word,
+      weight: typeof t.weight,
+    })),
   )
 })

packages/jieba/index.d.ts

@@ -8,6 +8,11 @@ export interface TagResult {
   tag: string
 }
 
+export interface Keyword {
+  keyword: string
+  weight: number
+}
+
 export function tag(sentence: string | Buffer, hmm?: boolean): TagResult[]
 
-export function extract(sentence: string | Buffer, topn: number, allowedPos?: string[]): string[]
+export function extract(sentence: string | Buffer, topn: number, allowedPos?: string[]): Keyword[]

packages/jieba/index.js

@@ -26,7 +26,6 @@ module.exports = {
 
   extract: function extract(sentence, topn, allowedPos = []) {
     const input = Buffer.isBuffer(sentence) ? sentence : Buffer.from(sentence)
-    const output = native.extract(input, topn, allowedPos.join(','))
-    return output.split(',')
+    return native.extract(input, topn, allowedPos.join(','))
   },
 }

packages/jieba/src/lib.rs

@@ -5,8 +5,8 @@ extern crate napi_derive;
 
 use jieba_rs::{Jieba, KeywordExtract, TFIDF};
 use napi::{
-  CallContext, Env, Error, JsBoolean, JsBuffer, JsNumber, JsString, JsUndefined, Module, Result,
-  Status,
+  CallContext, Env, Error, JsBoolean, JsBuffer, JsNumber, JsObject, JsString, JsUndefined, Module,
+  Result, Status,
 };
 use once_cell::sync::OnceCell;
 use std::convert::TryInto;
@@ -133,7 +133,7 @@ fn tag(ctx: CallContext) -> Result<JsString> {
 }
 
 #[js_function(3)]
-fn extract(ctx: CallContext) -> Result<JsString> {
+fn extract(ctx: CallContext) -> Result<JsObject> {
   let sentence = ctx.get::<JsBuffer>(0)?;
   let topn = ctx.get::<JsNumber>(1)?;
   let allowed_pos = ctx
@@ -159,8 +159,16 @@ fn extract(ctx: CallContext) -> Result<JsString> {
     topn,
     allowed_pos,
   );
+  let mut js_tags = ctx.env.create_array_with_length(tags.len())?;
 
-  ctx.env.create_string(tags.join(",").as_str())
+  for (index, t) in tags.iter().enumerate() {
+    let mut tag_value = ctx.env.create_object()?;
+    tag_value.set_named_property("keyword", ctx.env.create_string(t.keyword.as_str())?)?;
+    tag_value.set_named_property("weight", ctx.env.create_double(t.weight)?)?;
+    js_tags.set_index(index, tag_value)?;
+  }
+
+  Ok(js_tags)
 }
 
 #[js_function]