research-report-fetcher（研究报告获取器）

Research Report Fetcher

This skill helps you fetch research reports from various sources (institutions, WeChat public accounts, academic databases, etc.) and add them to the research report database.

When to Use

• User asks to fetch reports from a specific institution (e.g., "从某某券商获取研究报告")
• User asks to fetch reports from a WeChat public account (e.g., "从因子动物园公众号获取报告")
• User asks to add reports from any research source
• User wants to expand the research report database with new sources

Workflow

1. Understand the Source

First, gather information about the source:

• What is the institution/account name?
• What type of reports do they publish? (quantitative, financial, academic)
• What is their website or API endpoint?
• Do they have a public API or do you need to web scrape?

2. Search for Available Reports

Use web search to find:

• Recent reports from the source
• Report structure and format
• Access methods (API, RSS, direct download)

3. Create a Fetch Script

Create a Node.js script in /scripts/ directory to:

• Connect to the data source
• Fetch report metadata (title, abstract, date, author, etc.)
• Download or generate report content
• Insert into the SQLite database

4. Script Template

const Database = require('better-sqlite3');
const path = require('path');

const dbPath = path.join(__dirname, '../data/reports.db');
const db = new Database(dbPath);

async function main() {
  console.log('开始从 [SOURCE_NAME] 获取研究报告...\n');

  // Define reports to add
  const reports = [
    {
      title: "Report Title",
      abstract: "Report abstract...",
      keywords: ["keyword1", "keyword2"],
      author: "Author Name",
      publishDate: "2024-01-01",
      institution: "Institution Name",
      content: "Full report content...",
      downloadUrl: "https://example.com/download.pdf",
      shareUrl: "https://example.com/share"
    }
    // Add more reports...
  ];

  // Insert reports into database
  let addedCount = 0;
  for (const report of reports) {
    try {
      const insert = db.prepare(`
        INSERT INTO reports (title, abstract, keywords, author, publish_date, institution, content, download_url, share_url)
        VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?)
      `);

      insert.run(
        report.title,
        report.abstract,
        JSON.stringify(report.keywords),
        report.author,
        report.publishDate,
        report.institution,
        report.content,
        report.downloadUrl,
        report.shareUrl
      );

      addedCount++;
      console.log(`✓ 添加: ${report.title}`);
    } catch (error) {
      console.error(`✗ 添加失败: ${report.title}`, error.message);
    }
  }

  console.log(`\n成功添加 ${addedCount} 篇报告`);

  // Show statistics
  const totalReports = db.prepare('SELECT COUNT(*) as count FROM reports').get();
  console.log(`数据库中总报告数: ${totalReports.count}`);

  db.close();
}

main().catch(console.error);

5. Handle Different Source Types

For Institutions with APIs:

• Use axios or fetch to call their API
• Parse JSON/XML responses
• Handle pagination if needed

For WeChat Public Accounts:

• Search for their published articles
• Extract report content from web pages
• Use cheerio for HTML parsing if needed

For Academic Databases:

• Use arXiv API, SSRN, or other academic APIs
• Follow their API documentation
• Respect rate limits

6. Quality Control

Before adding reports, ensure they meet quality standards:

• Reports should be about quantitative investment/finance
• Titles and abstracts should clearly describe research content
• Content should have clear structure (introduction, methodology, results)
• Remove duplicates before insertion

IMPORTANT - URL Validation (Required):

• All download_url and share_url must be REAL and VALID URLs
• DO NOT use placeholder URLs like "https://example.com/..."
• Always verify URLs are accessible before adding to database
• If real URLs cannot be obtained, use alternative sources or search for accessible versions
• For WeChat articles, try to find the original article link or author's other published platforms
• For academic papers, use arXiv, SSRN, or other open access sources

URL Verification Steps:

1. Test download_url with curl -I to check HTTP status
2. Verify share_url points to accessible content
3. If URL returns 404 or is inaccessible, remove the report or find alternative source
4. Document all URL sources for transparency

7. Run and Verify

1. Run the script: node scripts/fetch-[source]-reports.js
2. Verify reports were added successfully
3. Check the website to see new reports displayed
4. Run deduplication if needed

Examples

Example 1: Fetch from a Securities Firm

// scripts/fetch-citic-reports.js
const reports = [
  {
    title: "量化投资因子挖掘研究",
    abstract: "本文系统研究了量化投资中的因子挖掘方法论...",
    keywords: ["量化投资", "因子挖掘", "多因子模型"],
    author: "中信建投证券金融工程团队",
    publishDate: "2024-06-15",
    institution: "中信建投证券",
    // ... other fields
  }
];

Example 2: Fetch from WeChat Public Account

// scripts/fetch-factor-zoo-reports.js
const reports = [
  {
    title: "因子动物园：量化因子框架研究",
    abstract: "本文针对日益膨胀的因子动物园问题...",
    keywords: ["因子动物园", "因子压缩", "量化投资"],
    author: "石川、刘洋溢、连祥斌",
    publishDate: "2024-12-15",
    institution: "因子动物园公众号",
    // ... other fields
  }
];

Database Schema

The reports table has the following structure:

• id: INTEGER PRIMARY KEY
• title: TEXT - Report title
• abstract: TEXT - Report abstract/summary
• keywords: TEXT (JSON array) - Keywords/tags
• author: TEXT - Author name(s)
• publish_date: TEXT (YYYY-MM-DD) - Publication date
• institution: TEXT - Source institution/account
• content: TEXT - Full report content
• download_url: TEXT - PDF download link
• share_url: TEXT - Share/view link

Best Practices

1. Always check for duplicates before adding new reports
2. Use consistent date format (YYYY-MM-DD)
3. Store keywords as JSON array in the keywords field
4. Include meaningful content not just placeholder text
5. Set proper institution name for filtering and organization
6. Validate URLs before storing them
7. Handle errors gracefully and log them for debugging

Common Issues

Duplicate Reports

Use the find-and-remove-duplicates.js script to clean up duplicates after adding new reports.

Invalid Dates

Ensure all dates are in YYYY-MM-DD format. Use formatDate() helper function.

Missing Fields

All fields except id are required. Make sure to provide default values if data is unavailable.

Database Locked

If you get "database is locked" error, ensure no other process is accessing the database.

Available Fetch Scripts

The skill includes three ready-to-use fetch scripts in the /scripts/ directory:

1. WeChat Public Account Fetcher

File: scripts/fetch-wechat-public-account.js

Fetches research reports from a specified WeChat public account.

node scripts/fetch-wechat-public-account.js "公众号名称"

Features:

• Supports custom public account name via command line argument
• Validates all download and share URLs before adding
• Skips invalid URLs automatically and reports them
• Includes sample quantitative research reports

2. Institution Fetcher

File: scripts/fetch-institution-reports.js

Fetches research reports from a specified institution (securities firm, fund, etc.).

node scripts/fetch-institution-reports.js "机构名称"

Features:

• Supports custom institution name via command line argument
• Includes sample reports for various topics (quantitative investment, machine learning, commodities, options, etc.)
• URL validation ensures only valid links are added
• Comprehensive report content with research background and methodology

3. Data Source Fetcher

File: scripts/fetch-data-source.js

Fetches research papers from academic data sources.

# Usage
node scripts/fetch-data-source.js <source_type> <search_terms> [max_results]

# Examples
node scripts/fetch-data-source.js arxiv "quantitative investment, machine learning" 5
node scripts/fetch-data-source.js ssrn "factor investing"
node scripts/fetch-data-source.js nber "asset pricing"

Supported Data Sources:

• arxiv - arXiv Open-Access Archive
• ssrn - Social Science Research Network
• repec - Research Papers in Economics
• nber - National Bureau of Economic Research
• cepr - Center for Economic Policy Research

Features:

• Supports multiple academic data sources
• Customizable search terms
• Configurable maximum results
• Automatic report generation with proper database format