nodejieba-plus 3.5.13 → 3.5.17

package/test/load_user_dict_test.js

@@ -76,15 +76,59 @@ describe("nodejieba.loadUserDict", function() {
   });
 
   it("nodejieba.loadUserDict should filter empty strings", function() {
-    // 测试空字符串被过滤，不会导致断言失败
     var dictLines = [
       "有效词1",
-      "",  // 空字符串
+      "",
       "有效词2",
-      "",  // 空字符串
-      "   "  // 只有空格的字符串（也会被保留，因为不是完全空）
+      "",
+      "   ",
+      "\t",
+      "\n",
+      "有效词3"
     ];
     var loadResult = nodejieba.loadUserDict(dictLines);
     loadResult.should.eql(true);
   });
+
+  it("nodejieba.loadUserDict with space-containing keywords", function() {
+    var dictLines = [
+      "深度 学习",
+      "机器 学习 200 n",
+      "人工 智能 300 nz"
+    ];
+    var loadResult = nodejieba.loadUserDict(dictLines);
+    loadResult.should.eql(true);
+  });
+
+  it("nodejieba.loadUserDict should throw error for non-string array elements", function() {
+    var dictLines = [
+      "有效词",
+      123,
+      "另一个有效词"
+    ];
+    
+    (function() {
+      nodejieba.loadUserDict(dictLines);
+    }).should.throw();
+  });
+
+  it("nodejieba.loadUserDict should return false for null", function() {
+    var loadResult = nodejieba.loadUserDict(null);
+    loadResult.should.eql(false);
+  });
+
+  it("nodejieba.loadUserDict should return false for undefined", function() {
+    var loadResult = nodejieba.loadUserDict(undefined);
+    loadResult.should.eql(false);
+  });
+
+  it("nodejieba.loadUserDict should throw TypeError for invalid type", function() {
+    (function() {
+      nodejieba.loadUserDict(123);
+    }).should.throw(TypeError);
+    
+    (function() {
+      nodejieba.loadUserDict({});
+    }).should.throw(TypeError);
+  });
 });

package/test_1_3x_weight.js

@@ -0,0 +1,86 @@
+var nodejieba = require("./index.js");
+var fs = require("fs");
+
+console.log("=== 测试 1.3 倍 IDF 权重 ===\n");
+
+const content = "疯狂动物城 疯狂动物城 疯狂动物城 这是一个二次开发的项目，整合了原版的动画及Open Claw打包制作了MAC安装包，它可以出Open Claw现在你的系统桌面的任何地方，也会随互动有特定的动作，还蛮有意思的项目地址：https://github.com/justaLoli/VPet-Mac云盘：https://pan.quark.cn/s/62596470429a功能：✅开始、关闭、正常效果的动画播放✅拖动效果✅「互动」菜单里的互动，即睡觉、学习、工作等（带计时器，但没有经验、金钱加成）✅自动事件（发呆、待机、睡觉等）✅桌宠自动移动✅摸头预览";
+
+console.log("【测试 1: 未加载用户词典】");
+nodejieba.load();
+var result1 = nodejieba.extract(content, 10);
+console.log("关键词及权重:");
+result1.forEach((item, i) => {
+  console.log(`  ${i + 1}. ${item.word}: ${item.weight.toFixed(2)}`);
+});
+
+console.log("\n【测试 2: 加载用户词典（自动 1.3 倍权重）】");
+nodejieba.loadUserDict("Open Claw 10 n");
+var result2 = nodejieba.extract(content, 10);
+console.log("关键词及权重:");
+result2.forEach((item, i) => {
+  console.log(`  ${i + 1}. ${item.word}: ${item.weight.toFixed(2)}`);
+});
+
+console.log("\n【验证 1.3 倍权重】");
+var openClawWeight1 = result1.find(r => r.word === "Open" || r.word === "Claw");
+var openClawWeight2 = result2.find(r => r.word === "Open Claw");
+if (openClawWeight1 && openClawWeight2) {
+  console.log(`未加载时权重: ${openClawWeight1.weight.toFixed(2)}`);
+  console.log(`加载后权重: ${openClawWeight2.weight.toFixed(2)}`);
+  console.log(`权重倍数: ${(openClawWeight2.weight / openClawWeight1.weight).toFixed(2)} ✅`);
+}
+
+console.log("\n=== 测试 IDF 词典支持空格关键词 ===\n");
+
+console.log("【测试 3: 创建包含空格关键词的 IDF 词典】");
+var testIdfPath = "./test_idf_with_spaces.txt";
+var testIdfContent = "Open Claw 30.0\nMachine Learning 25.0\nDeep Learning 28.0\n互动 12.0\n";
+fs.writeFileSync(testIdfPath, testIdfContent, "utf8");
+console.log("测试 IDF 词典内容:");
+console.log(testIdfContent);
+
+console.log("\n【测试 4: 加载包含空格的 IDF 词典 + 用户词典】");
+delete require.cache[require.resolve('./index.js')];
+var nodejieba2 = require("./index.js");
+
+nodejieba2.load({
+  idfDict: testIdfPath
+});
+
+nodejieba2.loadUserDict([
+  "Open Claw",
+  "Machine Learning",
+  "Deep Learning"
+]);
+
+console.log("\n【测试 5: 验证空格关键词识别】");
+var testContent = "Open Claw和Machine Learning都是Deep Learning的基础";
+var keywords = nodejieba2.extract(testContent, 5);
+console.log("测试文本:", testContent);
+console.log("关键词提取结果:");
+keywords.forEach((item, i) => {
+  console.log(`  ${i + 1}. ${item.word}: ${item.weight.toFixed(2)}`);
+});
+
+console.log("\n【验证结果】");
+var hasOpenClaw = keywords.some(r => r.word === "Open Claw");
+var hasMachineLearning = keywords.some(r => r.word === "Machine Learning");
+var hasDeepLearning = keywords.some(r => r.word === "Deep Learning");
+
+console.log(`识别到 'Open Claw': ${hasOpenClaw ? '✅' : '❌'}`);
+console.log(`识别到 'Machine Learning': ${hasMachineLearning ? '✅' : '❌'}`);
+console.log(`识别到 'Deep Learning': ${hasDeepLearning ? '✅' : '❌'}`);
+
+if (hasOpenClaw) {
+  var openClaw = keywords.find(r => r.word === "Open Claw");
+  console.log(`\n'Open Claw' 权重: ${openClaw.weight.toFixed(2)}`);
+  console.log(`IDF 词典中设置的权重: 30.0`);
+  console.log(`用户词典自动提升倍数: 1.3`);
+  console.log(`理论权重: ${(30.0 * 1.3).toFixed(2)} (30.0 × 1.3)`);
+}
+
+console.log("\n【清理测试文件】");
+fs.unlinkSync(testIdfPath);
+console.log("测试文件已删除");
+
+console.log("\n=== 测试完成 ===");

package/test_assertion_fix.js

@@ -0,0 +1,60 @@
+var nodejieba = require("./index.js");
+
+nodejieba.load();
+
+console.log("测试1: 加载包含空白字符的词典条目");
+try {
+  var result = nodejieba.loadUserDict([
+    "有效词1",
+    "",
+    "   ",
+    "\t",
+    "\n",
+    "有效词2",
+    " 测试词 ",
+    "  空格词  "
+  ]);
+  console.log("✅ 加载成功:", result);
+} catch (e) {
+  console.log("❌ 加载失败:", e.message);
+}
+
+console.log("\n测试2: 使用包含空白字符的词典进行分词");
+try {
+  var result = nodejieba.cut("有效词1和有效词2以及测试词");
+  console.log("✅ 分词成功:", result);
+  if (result.includes("有效词1") && result.includes("有效词2") && result.includes("测试词")) {
+    console.log("✅ 词典条目正确识别");
+  }
+} catch (e) {
+  console.log("❌ 分词失败:", e.message);
+}
+
+console.log("\n测试3: 加载大量空白字符的词典");
+try {
+  var largeDict = [];
+  for (var i = 0; i < 100; i++) {
+    largeDict.push("词" + i);
+    largeDict.push("");
+    largeDict.push("   ");
+    largeDict.push("\t\n");
+  }
+  var result = nodejieba.loadUserDict(largeDict);
+  console.log("✅ 大量词典加载成功:", result);
+} catch (e) {
+  console.log("❌ 大量词典加载失败:", e.message);
+}
+
+console.log("\n测试4: Buffer 包含空白行");
+try {
+  var bufferContent = "词A\n\n   \n\t\n词B\n  词C  \n";
+  var result = nodejieba.loadUserDict(Buffer.from(bufferContent));
+  console.log("✅ Buffer 加载成功:", result);
+  
+  var cutResult = nodejieba.cut("词A和词B以及词C");
+  console.log("✅ Buffer 词典分词成功:", cutResult);
+} catch (e) {
+  console.log("❌ Buffer 加载失败:", e.message);
+}
+
+console.log("\n✅ 所有测试完成，断言错误已修复！");

package/test_idf_feature.js

@@ -0,0 +1,43 @@
+var nodejieba = require("./index.js");
+
+console.log("=== 测试 IDF 权重提升功能 ===\n");
+
+const content = "疯狂动物城 疯狂动物城 疯狂动物城 这是一个二次开发的项目，整合了原版的动画及Open Claw打包制作了MAC安装包，它可以出Open Claw现在你的系统桌面的任何地方，也会随互动有特定的动作，还蛮有意思的项目地址：https://github.com/justaLoli/VPet-Mac云盘：https://pan.quark.cn/s/62596470429a功能：✅开始、关闭、正常效果的动画播放✅拖动效果✅「互动」菜单里的互动，即睡觉、学习、工作等（带计时器，但没有经验、金钱加成）✅自动事件（发呆、待机、睡觉等）✅桌宠自动移动✅摸头预览";
+
+console.log("【测试 1: 默认提取（未加载用户词典）】");
+nodejieba.load();
+var result1 = nodejieba.extract(content, 20);
+console.log("关键词及权重:");
+result1.forEach((item, i) => {
+  console.log(`  ${i + 1}. ${item.word}: ${item.weight.toFixed(2)}`);
+});
+
+console.log("\n【测试 2: 加载用户词典（自动提升 IDF 权重）】");
+nodejieba.loadUserDict("Open Claw 10 n");
+var result2 = nodejieba.extract(content, 20);
+console.log("关键词及权重:");
+result2.forEach((item, i) => {
+  console.log(`  ${i + 1}. ${item.word}: ${item.weight.toFixed(2)}`);
+});
+
+console.log("\n【测试 3: 手动设置 IDF 权重】");
+nodejieba.setIdf("Open Claw", 30.0);
+var result3 = nodejieba.extract(content, 20);
+console.log("关键词及权重:");
+result3.forEach((item, i) => {
+  console.log(`  ${i + 1}. ${item.word}: ${item.weight.toFixed(2)}`);
+});
+
+console.log("\n【测试 4: 使用倍数提升 IDF 权重】");
+nodejieba.setIdf("Open Claw", null, 3.0);  // 3倍权重
+var result4 = nodejieba.extract(content, 20);
+console.log("关键词及权重:");
+result4.forEach((item, i) => {
+  console.log(`  ${i + 1}. ${item.word}: ${item.weight.toFixed(2)}`);
+});
+
+console.log("\n【对比结果】");
+console.log("未加载词典时 'Open Claw' 排名:", result1.findIndex(r => r.word === "Open Claw") + 1 || "未出现");
+console.log("加载词典后 'Open Claw' 排名:", result2.findIndex(r => r.word === "Open Claw") + 1);
+console.log("手动设置IDF后 'Open Claw' 排名:", result3.findIndex(r => r.word === "Open Claw") + 1);
+console.log("3倍权重后 'Open Claw' 排名:", result4.findIndex(r => r.word === "Open Claw") + 1);

package/test_open_claw.js

@@ -0,0 +1,65 @@
+var nodejieba = require("./index.js");
+
+const content = "疯狂动物城 疯狂动物城 疯狂动物城 这是一个二次开发的项目，整合了原版的动画及Open Claw打包制作了MAC安装包，它可以出Open Claw现在你的系统桌面的任何地方，也会随互动有特定的动作，还蛮有意思的项目地址：https://github.com/justaLoli/VPet-Mac云盘：https://pan.quark.cn/s/62596470429a功能：✅开始、关闭、正常效果的动画播放✅拖动效果✅「互动」菜单里的互动，即睡觉、学习、工作等（带计时器，但没有经验、金钱加成）✅自动事件（发呆、待机、睡觉等）✅桌宠自动移动✅摸头预览";
+
+console.log("=== 测试 1: 默认加载词典 ===");
+nodejieba.load();
+
+console.log("\n测试 1.1: 分词结果（未加载用户词典）");
+var cutResult1 = nodejieba.cut(content);
+console.log("分词结果:", cutResult1);
+console.log("包含 'Open Claw':", cutResult1.includes("Open Claw") || cutResult1.includes("Open") || cutResult1.includes("Claw"));
+
+console.log("\n测试 1.2: 关键词提取（未加载用户词典）");
+var extractResult1 = nodejieba.extract(content, 20);
+console.log("关键词提取结果:", extractResult1.map(r => r.word));
+
+console.log("\n=== 测试 2: 加载用户词典后 ===");
+console.log("加载用户词典: 'Open Claw 10 n'");
+var loadResult = nodejieba.loadUserDict("Open Claw 10 n");
+console.log("加载结果:", loadResult);
+
+console.log("\n测试 2.1: 分词结果（已加载用户词典）");
+var cutResult2 = nodejieba.cut(content);
+console.log("分词结果:", cutResult2);
+console.log("包含 'Open Claw':", cutResult2.includes("Open Claw"));
+
+console.log("\n测试 2.2: 关键词提取（已加载用户词典）");
+var extractResult2 = nodejieba.extract(content, 20);
+console.log("关键词提取结果:", extractResult2.map(r => r.word));
+console.log("包含 'Open Claw':", extractResult2.some(r => r.word === "Open Claw"));
+
+console.log("\n=== 测试 3: 使用 insertWord ===");
+nodejieba.insertWord("Open Claw", "n");
+console.log("插入词: 'Open Claw'");
+
+console.log("\n测试 3.1: 分词结果（使用 insertWord）");
+var cutResult3 = nodejieba.cut(content);
+console.log("分词结果:", cutResult3);
+console.log("包含 'Open Claw':", cutResult3.includes("Open Claw"));
+
+console.log("\n测试 3.2: 关键词提取（使用 insertWord）");
+var extractResult3 = nodejieba.extract(content, 20);
+console.log("关键词提取结果:", extractResult3.map(r => r.word));
+console.log("包含 'Open Claw':", extractResult3.some(r => r.word === "Open Claw"));
+
+console.log("\n=== 测试 4: 检查文本中的 Open Claw ===");
+var testText1 = "Open Claw是一个好游戏";
+var testText2 = "我喜欢Open Claw";
+var testText3 = "OpenClaw很好玩";
+
+console.log("\n测试文本1:", testText1);
+console.log("分词:", nodejieba.cut(testText1));
+console.log("关键词:", nodejieba.extract(testText1, 5).map(r => r.word));
+
+console.log("\n测试文本2:", testText2);
+console.log("分词:", nodejieba.cut(testText2));
+console.log("关键词:", nodejieba.extract(testText2, 5).map(r => r.word));
+
+console.log("\n测试文本3:", testText3);
+console.log("分词:", nodejieba.cut(testText3));
+console.log("关键词:", nodejieba.extract(testText3, 5).map(r => r.word));
+
+console.log("\n=== 测试 5: 检查 IDF 词典 ===");
+console.log("说明: 关键词提取需要 IDF 词典支持");
+console.log("用户词典只影响分词，不影响关键词提取的权重计算");

package/test_simple.js

@@ -0,0 +1,17 @@
+var nodejieba = require("./index.js");
+
+console.log("=== 开始测试 ===\n");
+
+try {
+  console.log("调用 load()...");
+  nodejieba.load();
+  console.log("load() 完成");
+  
+  console.log("\n测试分词...");
+  var result = nodejieba.cut("测试");
+  console.log("分词结果:", result);
+} catch (e) {
+  console.error("错误:", e);
+}
+
+console.log("\n=== 测试结束 ===");

package/test_space_keyword.js

@@ -0,0 +1,66 @@
+var nodejieba = require("./index.js");
+
+console.log("=== 测试包含空格的关键词匹配功能 ===\n");
+
+// 加载词典
+nodejieba.load();
+
+// 测试1: 加载包含空格的关键词（带词频和词性）
+console.log("测试1: 加载 'Open Claw 2 n'");
+nodejieba.loadUserDict(["Open Claw 2 n"]);
+
+var testCases1 = [
+  "Open Claw",
+  "OpenClaw", 
+  "Openclaw",
+  "OPENCLAW",
+  "open claw",
+  "OPEN CLAW"
+];
+
+console.log("测试各种大小写变体:");
+testCases1.forEach(function(testText) {
+  var result = nodejieba.cut(testText);
+  console.log("  '" + testText + "' ->", result);
+});
+
+console.log("\n");
+
+// 测试2: 加载包含空格的关键词（只有关键词）
+console.log("测试2: 加载 'Game Master' (只有关键词)");
+nodejieba.loadUserDict("Game Master");
+
+var testCases2 = [
+  "Game Master",
+  "GameMaster",
+  "gamemaster",
+  "GAMEMASTER",
+  "GAME MASTER"
+];
+
+console.log("测试各种大小写变体:");
+testCases2.forEach(function(testText) {
+  var result = nodejieba.cut(testText);
+  console.log("  '" + testText + "' ->", result);
+});
+
+console.log("\n");
+
+// 测试3: 在句子中匹配
+console.log("测试3: 在句子中匹配包含空格的关键词");
+var sentence1 = "I like Open Claw game very much";
+var result1 = nodejieba.cut(sentence1);
+console.log("  句子: '" + sentence1 + "'");
+console.log("  分词结果:", result1);
+
+var sentence2 = "Open Claw和Game Master都是好游戏";
+var result2 = nodejieba.cut(sentence2);
+console.log("  句子: '" + sentence2 + "'");
+console.log("  分词结果:", result2);
+
+var sentence3 = "OPENCLAW和gamemaster都是好游戏";
+var result3 = nodejieba.cut(sentence3);
+console.log("  句子: '" + sentence3 + "'");
+console.log("  分词结果:", result3);
+
+console.log("\n=== 测试完成 ===");

package/types/index.d.ts

@@ -28,4 +28,5 @@ declare module "nodejieba" {
   export function insertWord(word: string, tag?: string): boolean;
   export function cutSmall(sentence: string, small: number): string[];
   export function loadUserDict(dict: string | string[] | Set<string> | Buffer): boolean;
+  export function setIdf(word: string, idf?: number, multiplier?: number): boolean;
 }