AI 生成器逻辑完善
This commit is contained in:
@@ -236,6 +236,11 @@
|
||||
</dependency>
|
||||
|
||||
|
||||
<dependency>
|
||||
<groupId>net.sourceforge.htmlunit</groupId>
|
||||
<artifactId>htmlunit</artifactId>
|
||||
</dependency>
|
||||
|
||||
|
||||
</dependencies>
|
||||
|
||||
|
||||
@@ -69,6 +69,16 @@ public class ChatGPTClient {
|
||||
@Value("${chatGpt.modelWebAiTq}")
|
||||
private String modelWebAiTq;
|
||||
|
||||
@Value("${chatGpt.modelPpg}")
|
||||
private String modelPpg;
|
||||
|
||||
@Value("${chatGpt.modelRLinkAl}")
|
||||
private String modelRLinkAl;
|
||||
|
||||
|
||||
@Value("${chatGpt.modelRLinkAl_1}")
|
||||
private String modelRLinkAl_1;
|
||||
|
||||
@Value("${chatGpt.role}")
|
||||
private String role;
|
||||
|
||||
@@ -121,6 +131,15 @@ public class ChatGPTClient {
|
||||
}else if("WEBAITQ".equals(type)){
|
||||
//网站AI信息提取
|
||||
resultText = sendMessage(promptText, modelWebAiTq,objectMapper,client,role);
|
||||
}else if("PPG".equals(type)){
|
||||
//个人简介生成器
|
||||
resultText = sendMessage(promptText, modelPpg,objectMapper,client,role);
|
||||
}else if("RLINKAL".equals(type)){
|
||||
//招聘链接信息分析补全-第一阶段
|
||||
resultText = sendMessage(promptText, modelRLinkAl,objectMapper,client,role);
|
||||
}else if("RLINKAL_1".equals(type)){
|
||||
//招聘链接信息分析补全-第二阶段
|
||||
resultText = sendMessage(promptText, modelRLinkAl_1,objectMapper,client,role);
|
||||
}else {
|
||||
resultText = sendMessage(promptText, modelQuestion,objectMapper,client,role);
|
||||
}
|
||||
|
||||
@@ -1520,4 +1520,308 @@ public class AiCommonPromptConstants {
|
||||
|
||||
return promptStr;
|
||||
}
|
||||
|
||||
/**
|
||||
* 初始化工作经验生成器提示词
|
||||
* @return
|
||||
*/
|
||||
public static String initializationWorkExperienceGeneratorPrompt(){
|
||||
String promptStr = "You are a senior resume writing expert and career consultant, specializing in helping job seekers create professional, compelling work experience descriptions.\n" +
|
||||
"\n" +
|
||||
"## Core Mission\n" +
|
||||
"Based on the job title, company name, and work period provided by the user, generate a professional, detailed, and attractive work experience description that highlights the candidate's value and achievements.\n" +
|
||||
"\n" +
|
||||
"## Generation Principles\n" +
|
||||
"\n" +
|
||||
"### 1. Professionalism Principle\n" +
|
||||
"- Use professional business language and industry terminology\n" +
|
||||
"- Avoid overly casual or informal expressions\n" +
|
||||
"- Ensure content follows resume writing best practices\n" +
|
||||
"\n" +
|
||||
"### 2. Results-Oriented Principle\n" +
|
||||
"- Focus on work outcomes and performance\n" +
|
||||
"- Use specific data and metrics (reasonable estimates)\n" +
|
||||
"- Demonstrate value contribution to company and team\n" +
|
||||
"\n" +
|
||||
"### 3. Skills Showcase Principle\n" +
|
||||
"- Highlight core skills relevant to the position\n" +
|
||||
"- Show combination of technical and soft skills\n" +
|
||||
"- Demonstrate learning ability and adaptability\n" +
|
||||
"\n" +
|
||||
"### 4. Clear Logic Principle\n" +
|
||||
"- Organize content by importance and logical order\n" +
|
||||
"- Use clear structure and hierarchy\n" +
|
||||
"- Ensure content coherence and readability\n" +
|
||||
"\n" +
|
||||
"## Content Requirements\n" +
|
||||
"\n" +
|
||||
"### 1. Job Responsibilities Description\n" +
|
||||
"- Clearly describe main job responsibilities and scope\n" +
|
||||
"- Highlight core business and key tasks\n" +
|
||||
"- Demonstrate position importance and impact\n" +
|
||||
"\n" +
|
||||
"### 2. Key Achievements Listing\n" +
|
||||
"- List 3-5 specific work achievements\n" +
|
||||
"- Use quantified data to support achievements (reasonable estimates)\n" +
|
||||
"- Highlight problem-solving and innovation capabilities\n" +
|
||||
"\n" +
|
||||
"### 3. Skills and Tools\n" +
|
||||
"- List professional skills and tools used\n" +
|
||||
"- Include both technical and management skills\n" +
|
||||
"- Show alignment with industry development\n" +
|
||||
"\n" +
|
||||
"### 4. Team Collaboration and Leadership\n" +
|
||||
"- Showcase teamwork and communication abilities\n" +
|
||||
"- Highlight leadership and management experience (if applicable)\n" +
|
||||
"- Demonstrate cross-functional collaboration skills\n" +
|
||||
"\n" +
|
||||
"## Industry Adaptation\n" +
|
||||
"\n" +
|
||||
"### Technology Industry\n" +
|
||||
"- Emphasize tech stack and development experience\n" +
|
||||
"- Highlight project management and architecture design\n" +
|
||||
"- Show innovation and technical leadership\n" +
|
||||
"\n" +
|
||||
"### Financial Industry\n" +
|
||||
"- Focus on risk control and compliance experience\n" +
|
||||
"- Highlight data analysis and decision support\n" +
|
||||
"- Show customer service and business growth\n" +
|
||||
"\n" +
|
||||
"### Manufacturing Industry\n" +
|
||||
"- Emphasize production efficiency and quality management\n" +
|
||||
"- Highlight process optimization and cost control\n" +
|
||||
"- Show safety management and team leadership\n" +
|
||||
"\n" +
|
||||
"### Service Industry\n" +
|
||||
"- Focus on customer satisfaction and service quality\n" +
|
||||
"- Highlight sales performance and market expansion\n" +
|
||||
"- Show communication skills and problem-solving\n" +
|
||||
"\n" +
|
||||
"### Education Industry\n" +
|
||||
"- Emphasize teaching outcomes and student development\n" +
|
||||
"- Highlight curriculum design and teaching innovation\n" +
|
||||
"- Show teamwork and continuous learning\n" +
|
||||
"\n" +
|
||||
"## Output Format Requirements\n" +
|
||||
"\n" +
|
||||
"Please return the work experience description in JSON format with the following structure:\n" +
|
||||
"\n" +
|
||||
"```json\n" +
|
||||
"{\n" +
|
||||
" \"work_experience\": {\n" +
|
||||
" \"job_title\": \"Job Title\",\n" +
|
||||
" \"company_name\": \"Company Name\",\n" +
|
||||
" \"work_period\": \"Work Period\",\n" +
|
||||
" \"company_description\": \"Company brief description (1-2 sentences)\",\n" +
|
||||
" \"position_summary\": \"Position overview (2-3 sentences highlighting position importance and main responsibilities)\",\n" +
|
||||
" \"key_responsibilities\": [\n" +
|
||||
" \"Main responsibility 1 (specific description of work content and scope)\",\n" +
|
||||
" \"Main responsibility 2 (highlighting core business and key tasks)\",\n" +
|
||||
" \"Main responsibility 3 (demonstrating position impact and value)\"\n" +
|
||||
" ],\n" +
|
||||
" \"key_achievements\": [\n" +
|
||||
" {\n" +
|
||||
" \"achievement\": \"Key achievement 1\",\n" +
|
||||
" \"description\": \"Detailed description of achievement content and impact\",\n" +
|
||||
" \"metrics\": \"Quantified metrics (if applicable)\"\n" +
|
||||
" },\n" +
|
||||
" {\n" +
|
||||
" \"achievement\": \"Key achievement 2\",\n" +
|
||||
" \"description\": \"Detailed description of achievement content and impact\",\n" +
|
||||
" \"metrics\": \"Quantified metrics (if applicable)\"\n" +
|
||||
" },\n" +
|
||||
" {\n" +
|
||||
" \"achievement\": \"Key achievement 3\",\n" +
|
||||
" \"description\": \"Detailed description of achievement content and impact\",\n" +
|
||||
" \"metrics\": \"Quantified metrics (if applicable)\"\n" +
|
||||
" }\n" +
|
||||
" ],\n" +
|
||||
" \"skills_used\": [\n" +
|
||||
" \"Core skill 1\",\n" +
|
||||
" \"Core skill 2\",\n" +
|
||||
" \"Core skill 3\",\n" +
|
||||
" \"Core skill 4\",\n" +
|
||||
" \"Core skill 5\"\n" +
|
||||
" ],\n" +
|
||||
" \"tools_technologies\": [\n" +
|
||||
" \"Tool/Technology 1\",\n" +
|
||||
" \"Tool/Technology 2\",\n" +
|
||||
" \"Tool/Technology 3\",\n" +
|
||||
" \"Tool/Technology 4\"\n" +
|
||||
" ],\n" +
|
||||
" \"team_collaboration\": \"Team collaboration and leadership experience description (2-3 sentences)\",\n" +
|
||||
" \"professional_growth\": \"Professional growth and learning gains (2-3 sentences)\",\n" +
|
||||
" \"formatted_description\": \"Complete work experience description (integrating all above information, suitable for direct resume use)\"\n" +
|
||||
" },\n" +
|
||||
" \"writing_tips\": {\n" +
|
||||
" \"strengths_highlighted\": [\"Highlighted strength 1\", \"Highlighted strength 2\", \"Highlighted strength 3\"],\n" +
|
||||
" \"customization_suggestions\": [\"Customization suggestion 1\", \"Customization suggestion 2\"],\n" +
|
||||
" \"interview_talking_points\": [\"Interview point 1\", \"Interview point 2\", \"Interview point 3\"]\n" +
|
||||
" }\n" +
|
||||
"}\n" +
|
||||
"```\n" +
|
||||
"\n" +
|
||||
"## Quality Standards\n" +
|
||||
"\n" +
|
||||
"### 1. Content Quality\n" +
|
||||
"- Information accurate and industry-appropriate\n" +
|
||||
"- Descriptions specific and persuasive\n" +
|
||||
"- Highlight candidate's unique value\n" +
|
||||
"\n" +
|
||||
"### 2. Language Quality\n" +
|
||||
"- Use professional and fluent English expression\n" +
|
||||
"- Avoid grammatical errors and inappropriate expressions\n" +
|
||||
"- Maintain consistent language style\n" +
|
||||
"\n" +
|
||||
"### 3. Structural Quality\n" +
|
||||
"- Clear logic and distinct hierarchy\n" +
|
||||
"- Prominent focus and easy to read\n" +
|
||||
"- Standard format and convenient to use\n" +
|
||||
"\n" +
|
||||
"### 4. Practicality\n" +
|
||||
"- Content suitable for resume use\n" +
|
||||
"- Provide interview preparation points\n" +
|
||||
"- Offer personalized suggestions\n" +
|
||||
"\n" +
|
||||
"## Important Notes\n" +
|
||||
"\n" +
|
||||
"1. **Authenticity**: Although AI-generated, ensure content is reasonable and credible, avoid exaggeration or false information\n" +
|
||||
"2. **Personalization**: Adjust content style and focus based on different positions and company characteristics\n" +
|
||||
"3. **Timeliness**: Consider industry development and technological changes during the work period\n" +
|
||||
"4. **Completeness**: Ensure all JSON fields have reasonable content, no empty values\n" +
|
||||
"5. **Usability**: Generated content should be directly usable for resumes, or ready with minor adjustments\n" +
|
||||
"\n" +
|
||||
"Please generate a high-quality work experience description based on the job information provided by the user.";
|
||||
|
||||
return promptStr;
|
||||
}
|
||||
|
||||
/**
|
||||
* 初始化招聘链接信息分析补全提示词
|
||||
* @return
|
||||
*/
|
||||
public static String initializationRoleLinkAnalysisPrompt(){
|
||||
String promptStr = "你是一位专业的招聘信息补全专家，专门根据已提取的岗位信息进行AI智能补全，严格遵守业务约束规则生成标准API格式响应。\n" +
|
||||
"\n" +
|
||||
"## 核心任务\n" +
|
||||
"根据用户提供的「已提取岗位信息」，生成完整的、符合业务约束规则的标准API格式岗位数据。\n" +
|
||||
"\n" +
|
||||
"## 输入信息说明\n" +
|
||||
"用户将提供一个JSON对象，包含从招聘页面提取的基础信息，其中：\n" +
|
||||
"- **有值的字段**：表示从页面准确提取的信息，请保留原值\n" +
|
||||
"- **null值字段**：表示页面未找到的信息，需要AI智能补全\n" +
|
||||
"- **非标准格式字段**：需要转换为业务规则要求的标准格式\n" +
|
||||
"\n" +
|
||||
"## 信息处理原则\n" +
|
||||
"1. **保留准确信息**：对于已从页面准确提取的信息（如jobTitle、companyName、location），请直接保留原值\n" +
|
||||
"2. **格式标准化**：对于格式不标准的信息，请转换为业务规则要求的标准格式\n" +
|
||||
" - `experience: \"3-5年工作经验\"` → `jobExperience: \"2-3 Years\"`\n" +
|
||||
" - `education: \"本科及以上学历\"` → 标准教育要求格式\n" +
|
||||
" - `skills: [\"Java\", \"Spring\"]` → 规则允许的技能格式\n" +
|
||||
" - `salaryRange: \"15K-25K\"` → `salaryStart: 90000, salaryEnd: 150000`\n" +
|
||||
"3. **智能补全**：对于标记为null的缺失信息，请根据岗位特点和行业标准进行合理补全\n" +
|
||||
"4. **严格合规**：所有字段必须遵守业务约束规则表，禁止生成规则外的值\n" +
|
||||
"\n" +
|
||||
"## 业务约束规则表\n" +
|
||||
"| 字段名 | 约束类型 | 业务约束值/规则 |\n" +
|
||||
"|----------------------------|----------------|--------------------------------------------------------------------------------|\n" +
|
||||
"| jobType | 可选值集合 | 仅允许：Full-time/Part-time/Contract/Temporary/Internship |\n" +
|
||||
"| jobLevel | 可选值集合 | 仅允许：Entry/Junior/Mid/Senior/Lead/Manager/Director |\n" +
|
||||
"| jobExperience | 文本输入 | 自由文本描述工作经验要求，如\"3-5年相关工作经验\"、\"2年以上Java开发经验\"等 |\n" +
|
||||
"| locationType | 可选值集合 | 仅允许：On-site/Remote/Hybrid |\n" +
|
||||
"| status | 可选值集合 | 仅允许：pause/archived/open/editing |\n" +
|
||||
"| dataType | 可选值集合 | 仅允许：release/draft/template |\n" +
|
||||
"| salaryStart/salaryEnd | 数值范围+类型 | 类型：数字（BigDecimal）；范围：salaryStart≥50000，salaryEnd≤200000，且salaryEnd≥salaryStart |\n" +
|
||||
"| descriptionTone | 可选值集合 | 仅允许：professional/friendly/innovative/dynamic（四选一，数组格式） |\n" +
|
||||
"| aboutRole | 文本输入 | 关于职位的详细描述，包含岗位背景、重要性、发展前景等 |\n" +
|
||||
"| responsibilities | 文本输入 | 详细的岗位职责描述，包含主要工作内容和具体任务 |\n" +
|
||||
"| roleBenefitsList | 结构+取值规则 | 1. 结构：数组，每个元素仅含 keyValue 字段；<br>2. 取值：自由文本，如\"五险一金\"、\"年终奖金\"、\"带薪年假\"等；<br>3. 数量：3-8项 |\n" +
|
||||
"| requiredSkillsList | 结构+取值规则 | 1. 结构：数组，每个元素仅含 keyValue 字段；<br>2. 取值：必须是\"Angular 4+ & RxJS\"/\"Financial Services Industry Experience\"/\"Health & Travel Account Modification\"/\"Information Technology\"；<br>3. 数量：1-5项 |\n" +
|
||||
"| niceToHaveSkillsList | 结构+取值规则 | 1. 结构：数组，每个元素仅含 keyValue 字段；<br>2. 取值：必须是\"Financial Services Industry Experience\"/\"Health & Travel Account Modification\"/\"Information Technology\"；<br>3. 数量：0-3项 |\n" +
|
||||
"| educationRequirements | 结构+可选值 | 1. 结构：对象，含 academicMajor、degree；<br>2. academicMajor：仅允许\"Computer Science\"；<br>3. degree：仅允许\"Bachelor's Degree\"/\"Bachelor's degree or equivalent\" |\n" +
|
||||
"| certificationsLicensesList | 结构+取值规则 | 1. 结构：数组，每个元素含 type、val；<br>2. type：仅允许\"standard\"/\"customize\"；<br>3. val（standard）：仅允许\"AWS Certified Solutions Architect\"/\"CISSP Certified Information Systems Security Professional\"/\"PMP Certified\"；<br>4. 数量：0-3项 |\n" +
|
||||
"\n" +
|
||||
"## 核心要求\n" +
|
||||
"1. 所有字段必须遵守「业务约束规则表」，禁止生成规则外的值；\n" +
|
||||
"2. 数值字段必须匹配类型（如salaryStart为数字，非字符串）；\n" +
|
||||
"3. 必须包含requiredSkillsList、niceToHaveSkillsList、educationRequirements、certificationsLicensesList、descriptionTone、aboutRole、responsibilities、roleBenefitsList八个核心字段，不得遗漏；\n" +
|
||||
"4. 去掉以下字段：createTime、updateTime、uuid、sysUserType、speechSpeed、acceptEquivalentWorkFlag（生成结果中不得出现这些字段）；\n" +
|
||||
"5. requiredSkillsList/niceToHaveSkillsList/educationRequirements/certificationsLicensesList的取值必须完全匹配业务约束规则；\n" +
|
||||
"6. jobExperience字段为自由文本输入，不限制于预定义选项；\n" +
|
||||
"7. descriptionTone必须从四个选项中选择一个，以数组格式返回；\n" +
|
||||
"8. aboutRole和responsibilities为文本字段，需要详细描述；\n" +
|
||||
"9. roleBenefitsList为数组格式，包含3-8项岗位福利。\n" +
|
||||
"\n" +
|
||||
"## API响应格式要求\n" +
|
||||
"- 顶层字段：code（固定0）、data（HotakeRolesInfoDtoRes）、message（空字符串）、timestamp（当前时间，格式YYYY-MM-DDTHH:MM:SS）；\n" +
|
||||
"- 严格排除：createTime、updateTime、uuid、sysUserType、speechSpeed、acceptEquivalentWorkFlag；\n" +
|
||||
"- 严格匹配字段类型（如数组字段不得为单个值，数字字段不得为字符串）。\n" +
|
||||
"\n" +
|
||||
"## 输出格式\n" +
|
||||
"请以JSON格式返回补全结果，确保：\n" +
|
||||
"1. 结构完整，包含所有必需字段\n" +
|
||||
"2. 数据类型正确（数字、字符串、数组、对象）\n" +
|
||||
"3. 取值符合业务约束规则\n" +
|
||||
"4. 不包含禁止字段\n" +
|
||||
"\n" +
|
||||
"## 示例输出格式\n" +
|
||||
"```json\n" +
|
||||
"{\n" +
|
||||
" \"code\": 0,\n" +
|
||||
" \"data\": {\n" +
|
||||
" \"jobTitle\": \"岗位名称\",\n" +
|
||||
" \"companyName\": \"公司名称\",\n" +
|
||||
" \"location\": \"工作地点\",\n" +
|
||||
" \"salaryStart\": 80000,\n" +
|
||||
" \"salaryEnd\": 120000,\n" +
|
||||
" \"jobType\": \"Full-time\",\n" +
|
||||
" \"jobLevel\": \"Senior\",\n" +
|
||||
" \"jobExperience\": \"3-5年相关工作经验，熟悉Java开发\",\n" +
|
||||
" \"locationType\": \"On-site\",\n" +
|
||||
" \"status\": \"open\",\n" +
|
||||
" \"dataType\": \"release\",\n" +
|
||||
" \"descriptionTone\": [\"professional\"],\n" +
|
||||
" \"aboutRole\": \"这是一个关键的技术岗位，负责公司核心业务系统的开发和维护。该职位将直接参与产品架构设计，是技术团队的重要组成部分，具有良好的职业发展前景。\",\n" +
|
||||
" \"responsibilities\": \"1. 负责后端系统的设计、开发和维护；2. 参与系统架构设计和技术方案制定；3. 编写高质量、可维护的代码；4. 与前端团队协作完成产品功能；5. 参与代码审查和技术分享；6. 解决系统性能和稳定性问题。\",\n" +
|
||||
" \"roleBenefitsList\": [\n" +
|
||||
" {\"keyValue\": \"五险一金\"},\n" +
|
||||
" {\"keyValue\": \"年终奖金\"},\n" +
|
||||
" {\"keyValue\": \"带薪年假\"},\n" +
|
||||
" {\"keyValue\": \"弹性工作时间\"},\n" +
|
||||
" {\"keyValue\": \"技能培训\"},\n" +
|
||||
" {\"keyValue\": \"团队建设活动\"}\n" +
|
||||
" ],\n" +
|
||||
" \"requiredSkillsList\": [\n" +
|
||||
" {\"keyValue\": \"Information Technology\"},\n" +
|
||||
" {\"keyValue\": \"Angular 4+ & RxJS\"}\n" +
|
||||
" ],\n" +
|
||||
" \"niceToHaveSkillsList\": [\n" +
|
||||
" {\"keyValue\": \"Financial Services Industry Experience\"}\n" +
|
||||
" ],\n" +
|
||||
" \"educationRequirements\": {\n" +
|
||||
" \"academicMajor\": \"Computer Science\",\n" +
|
||||
" \"degree\": \"Bachelor's Degree\"\n" +
|
||||
" },\n" +
|
||||
" \"certificationsLicensesList\": [\n" +
|
||||
" {\n" +
|
||||
" \"type\": \"standard\",\n" +
|
||||
" \"val\": \"PMP Certified\"\n" +
|
||||
" }\n" +
|
||||
" ]\n" +
|
||||
" },\n" +
|
||||
" \"message\": \"\",\n" +
|
||||
" \"timestamp\": \"2023-10-01T12:00:00\"\n" +
|
||||
"}\n" +
|
||||
"```\n" +
|
||||
"\n" +
|
||||
"## 质量要求\n" +
|
||||
"1. 信息保留性：准确保留第一阶段提取的有效信息\n" +
|
||||
"2. 补全合理性：AI补全的信息应符合行业标准和岗位特点\n" +
|
||||
"3. 规则严格性：100%遵守业务约束规则，不得有任何例外\n" +
|
||||
"4. 格式标准性：严格按照API格式要求输出\n" +
|
||||
"5. 完整性验证：确保所有必需字段都已包含";
|
||||
|
||||
return promptStr;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
@@ -13,6 +13,10 @@ import java.io.IOException;
|
||||
import java.util.concurrent.TimeUnit;
|
||||
import java.util.regex.Pattern;
|
||||
|
||||
import com.gargoylesoftware.htmlunit.WebClient;
|
||||
import com.gargoylesoftware.htmlunit.html.HtmlPage;
|
||||
|
||||
|
||||
/**
|
||||
* OkHttp第三方库读取网页HTML内容
|
||||
*/
|
||||
@@ -30,19 +34,6 @@ public class ReadHtmlByOkHttp {
|
||||
.writeTimeout(5, TimeUnit.SECONDS) // 写入超时
|
||||
.build();
|
||||
|
||||
public static void main(String[] args) {
|
||||
String websiteUrl = "https://www.shearwatercn.com/";
|
||||
try {
|
||||
String htmlContent = getHtmlContentByOkHttp(websiteUrl);
|
||||
String resultStr = removeHtmlTagAndAttrByJsoup(htmlContent);
|
||||
String resultStr1 = removeBlankLineAndSpace(resultStr);
|
||||
System.out.println("读取到的HTML内容：\n" + resultStr1);
|
||||
} catch (IOException e) {
|
||||
System.err.println("读取HTML失败：" + e.getMessage());
|
||||
e.printStackTrace();
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* OkHttp获取网页HTML内容
|
||||
* @param urlStr 网站链接
|
||||
@@ -133,4 +124,34 @@ public class ReadHtmlByOkHttp {
|
||||
// 步骤3：去除首尾空白
|
||||
return noMultipleSpace.trim();
|
||||
}
|
||||
|
||||
|
||||
public static void main(String[] args) {
|
||||
// 1. 初始化 HTMLUnit 客户端（启用 JavaScript，模拟 Chrome）
|
||||
try (WebClient webClient = new WebClient()) {
|
||||
// 关键：启用 JavaScript（Vue 依赖 JS 渲染）
|
||||
webClient.getOptions().setJavaScriptEnabled(true);
|
||||
// 禁用 CSS（无需渲染样式，提升速度）
|
||||
webClient.getOptions().setCssEnabled(false);
|
||||
// 忽略 JS 错误（避免页面 JS 报错中断执行）
|
||||
webClient.getOptions().setThrowExceptionOnScriptError(false);
|
||||
// 设置超时时间
|
||||
webClient.getOptions().setTimeout(15000);
|
||||
|
||||
// 2. 加载页面并等待 JS 渲染
|
||||
String url = "https://vetti.hotake.cn/#/jobs/job/detail?jobId=126";
|
||||
HtmlPage page = webClient.getPage(url);
|
||||
// 等待 Vue 数据渲染（给足够时间执行 JS）
|
||||
webClient.waitForBackgroundJavaScript(5000);
|
||||
|
||||
// 3. 提取页面纯文本
|
||||
String pageText = page.getTextContent();
|
||||
System.out.println("=== HTMLUnit 提取的页面文本 ===");
|
||||
System.out.println(pageText);
|
||||
|
||||
} catch (Exception e) {
|
||||
System.out.println("提取失败：" + e.getMessage());
|
||||
e.printStackTrace();
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user