fix : 修正文本提取工具,优化工作经验解析

This commit is contained in:
2025-12-29 10:41:42 +08:00
parent e0c5ad5cc9
commit f78a0b84d8

View File

@@ -46,62 +46,90 @@ export const useAudioSpeak = (config = {}) => {
// 文本提取工具函数
function extractSpeechText(markdown) {
if (!markdown || markdown.indexOf('job-json') === -1) {
return markdown;
}
const jobRegex = /``` job-json\s*({[\s\S]*?})\s*```/g;
const jobs = [];
let match;
let lastJobEndIndex = 0;
let firstJobStartIndex = -1;
if (!markdown || markdown.indexOf('job-json') === -1) {
return markdown;
}
const jobRegex = /``` job-json\s*({[\s\S]*?})\s*```/g;
const jobs = [];
let match;
let lastJobEndIndex = 0;
let firstJobStartIndex = -1;
while ((match = jobRegex.exec(markdown)) !== null) {
const jobStr = match[1];
try {
const job = JSON.parse(jobStr);
jobs.push(job);
if (firstJobStartIndex === -1) {
firstJobStartIndex = match.index;
}
lastJobEndIndex = jobRegex.lastIndex;
} catch (e) {
console.warn('JSON 解析失败', e);
while ((match = jobRegex.exec(markdown)) !== null) {
const jobStr = match[1];
try {
const job = JSON.parse(jobStr);
jobs.push(job);
if (firstJobStartIndex === -1) {
firstJobStartIndex = match.index;
}
lastJobEndIndex = jobRegex.lastIndex;
} catch (e) {
console.warn('JSON 解析失败', e);
}
}
const guideText = firstJobStartIndex > 0 ?
markdown.slice(0, firstJobStartIndex).trim() : '';
const endingText = lastJobEndIndex < markdown.length ?
markdown.slice(lastJobEndIndex).trim() : '';
const jobTexts = jobs.map((job, index) => {
// 处理薪资格式
let salaryText = job.salary;
if (salaryText) {
// 匹配 "XXXXX-XXXXX元/月" 格式
const rangeMatch = salaryText.match(/(\d+)-(\d+)元\/月/);
if (rangeMatch) {
const minSalary = parseInt(rangeMatch[1], 10);
const maxSalary = parseInt(rangeMatch[2], 10);
// 转换为千位单位
const minK = Math.round(minSalary / 1000);
const maxK = Math.round(maxSalary / 1000);
salaryText = `${minK}千到${maxK}千每月`;
}
// 如果不是 "XXXXX-XXXXX元/月" 格式,保持原样
}
const guideText = firstJobStartIndex > 0 ?
markdown.slice(0, firstJobStartIndex).trim() : '';
const endingText = lastJobEndIndex < markdown.length ?
markdown.slice(lastJobEndIndex).trim() : '';
const jobTexts = jobs.map((job, index) => {
// 处理薪资格式
let salaryText = job.salary;
if (salaryText) {
// 匹配 "XXXXX-XXXXX元/月" 格式
const rangeMatch = salaryText.match(/(\d+)-(\d+)元\/月/);
if (rangeMatch) {
const minSalary = parseInt(rangeMatch[1], 10);
const maxSalary = parseInt(rangeMatch[2], 10);
// 转换为千位单位
const minK = Math.round(minSalary / 1000);
const maxK = Math.round(maxSalary / 1000);
salaryText = `${minK}千到${maxK}千每月`;
}
// 如果不是 "XXXXX-XXXXX元/月" 格式,保持原样
// 处理经验要求格式
let experienceText = job.experience;
if (experienceText) {
// 匹配 "X-X年" 格式,如 "1-3年"、"5-10年"
const expRangeMatch = experienceText.match(/(\d+)-(\d+)年/);
if (expRangeMatch) {
const minExp = parseInt(expRangeMatch[1], 10);
const maxExp = parseInt(expRangeMatch[2], 10);
experienceText = `${minExp}${maxExp}`;
}
return `${index + 1} 个岗位,岗位名称是:${job.jobTitle},公司是:${job.companyName},薪资:${salaryText},地点:${job.location},学历要求:${job.education},经验要求:${job.experience}`;
});
// 还可以处理其他常见格式
else if (experienceText.includes('年以下')) {
// 如 "3年以下"
experienceText = experienceText.replace('年以下', '年以下');
} else if (experienceText.includes('年以上')) {
// 如 "5年以上"
experienceText = experienceText.replace('年以上', '年以上');
} else if (experienceText === '不限' || experienceText === '无经验要求') {
experienceText = '经验不限';
}
// 其他格式保持原样
}
const finalTextParts = [];
if (guideText) finalTextParts.push(guideText);
finalTextParts.push(...jobTexts);
if (endingText) finalTextParts.push(endingText);
// 同样可以处理学历要求(如果需要的话)
let educationText = job.education;
// 这里可以添加学历格式的转换逻辑
return finalTextParts.join('\n');
return `${index + 1} 个岗位,岗位名称是:${job.jobTitle},公司是:${job.companyName},薪资:${salaryText},地点:${job.location},学历要求:${educationText},经验要求:${experienceText}`;
});
const finalTextParts = [];
if (guideText) finalTextParts.push(guideText);
finalTextParts.push(...jobTexts);
if (endingText) finalTextParts.push(endingText);
return finalTextParts.join('\n');
}
/**