fin.cx/opendata
3
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(HandelsRegister): Refined HandelsRegister functionality for better error handling and response capture.

Browse Source
This commit is contained in:
Philipp Kunz
2025-01-03 02:19:07 +01:00
parent 1c0e04cb0d
commit e4a8d371f7
3 changed files with 97 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
changelog.md
View File

@@ -1,5 +1,12 @@
# Changelog # Changelog
## 2025-01-03 - 1.3.1 - fix(HandelsRegister)
Refined HandelsRegister functionality for better error handling and response capture.
- Improved parsing logic in parseGermanRegistration function.
- Enhanced navigateToPage and clickFindButton methods with error messages for clarity.
- Implemented a new responseListener to handle and log HTTP responses correctly.
## 2025-01-03 - 1.3.0 - feat(core) ## 2025-01-03 - 1.3.0 - feat(core)
Enhanced data handling capabilities and improved company search functionalities. Enhanced data handling capabilities and improved company search functionalities.

2
ts/00_commitinfo_data.ts
View File

@@ -3,6 +3,6 @@
*/ */
export const commitinfo = { export const commitinfo = {
name: '@fin.cx/opendata', name: '@fin.cx/opendata',
version: '1.3.0', version: '1.3.1',
description: 'A TypeScript library for accessing, managing, and updating open business data, focused on German companies and integrating with MongoDB.' description: 'A TypeScript library for accessing, managing, and updating open business data, focused on German companies and integrating with MongoDB.'
} }

115
ts/classes.handelsregister.ts
View File

@@ -114,7 +114,8 @@ export class HandelsRegister {
private async parseGermanRegistration( private async parseGermanRegistration(
input: string input: string
): Promise<BusinessRecord['data']['germanParsedRegistration']> { ): Promise<BusinessRecord['data']['germanParsedRegistration']> {
const regex = /District court (\p{L}[\p{L}\s-]*?(?:\s*\([\p{L}\s-]+\))?)\s+(HRA|HRB|GnR|VR|PR|GsR)\s+(\d+)/u; const regex =
/District court (\p{L}[\p{L}\s-]*?(?:\s*\([\p{L}\s-]+\))?)\s+(HRA|HRB|GnR|VR|PR|GsR)\s+(\d+)/u;
const match = input.match(regex); const match = input.match(regex);
if (match) { if (match) {
@@ -193,39 +194,101 @@ export class HandelsRegister {
await page.waitForSelector('#form\\:schlagwoerter', { timeout: 5000 }); await page.waitForSelector('#form\\:schlagwoerter', { timeout: 5000 });
// 1) Type of Register: // 1) Type of Register:
// Open the dropdown to reveal options // Open the dropdown to reveal options
await page.waitForSelector('#form\\:registerArt_label'); await page.waitForSelector('#form\\:registerArt_label');
await page.click('#form\\:registerArt_label'); // Open the dropdown await page.click('#form\\:registerArt_label'); // Open the dropdown
// Wait for the options and select the one matching companyArg.type // Wait for the options and select the one matching companyArg.type
await page.waitForSelector('#form\\:registerArt_items'); // Ensure dropdown options are loaded await page.waitForSelector('#form\\:registerArt_items'); // Ensure dropdown options are loaded
await page.evaluate((type) => { await page.evaluate((type) => {
const options = Array.from(document.querySelectorAll('#form\\:registerArt_items li')); const options = Array.from(document.querySelectorAll('#form\\:registerArt_items li'));
const targetOption = options.find((option) => option.textContent?.trim() === type); // Match type dynamically const targetOption = options.find((option) => option.textContent?.trim() === type); // Match type dynamically
(targetOption as any)?.click(); (targetOption as any)?.click();
}, companyArg.type); // Pass companyArg.type to the browser context }, companyArg.type); // Pass companyArg.type to the browser context
// 2) Register number: // 2) Register number:
// Fill in the register number // Fill in the register number
await page.waitForSelector('#form\\:registerNummer'); await page.waitForSelector('#form\\:registerNummer');
await page.type('#form\\:registerNummer', companyArg.number); await page.type('#form\\:registerNummer', companyArg.number);
// 3) Register court: // 3) Register court:
// Open the dropdown for the register court // Open the dropdown for the register court
await page.waitForSelector('#form\\:registergericht_label'); await page.waitForSelector('#form\\:registergericht_label');
await page.click('#form\\:registergericht_label'); // Open the dropdown await page.click('#form\\:registergericht_label'); // Open the dropdown
// Wait for the options and select the one matching companyArg.court // Wait for the options and select the one matching companyArg.court
await page.waitForSelector('#form\\:registergericht_items'); // Ensure dropdown options are loaded await page.waitForSelector('#form\\:registergericht_items'); // Ensure dropdown options are loaded
await page.evaluate((court) => { await page.evaluate((court) => {
const options = Array.from(document.querySelectorAll('#form\\:registergericht_items li')); const options = Array.from(document.querySelectorAll('#form\\:registergericht_items li'));
const targetOption = options.find((option) => option.textContent?.trim() === court); // Match court dynamically const targetOption = options.find((option) => option.textContent?.trim() === court); // Match court dynamically
(targetOption as any)?.click(); (targetOption as any)?.click();
}, companyArg.court); // Pass companyArg.court to the browser context }, companyArg.court); // Pass companyArg.court to the browser context
await this.clickFindButton(page); await this.clickFindButton(page);
const businessRecords = await this.waitForResults(page); const businessRecords = await this.waitForResults(page);
console.log(businessRecords); console.log(businessRecords);
// Define the response listener
const responseListener = async (
response: plugins.smartbrowser.smartpuppeteer.puppeteer.HTTPResponse
) => {
// Ignore preflight (OPTIONS) requests
if (response.request().method() === 'OPTIONS') {
console.log(`Ignoring preflight request: ${response.url()}`);
return;
}
// Check for downloads (Content-Disposition header)
const contentDisposition = response.headers()['content-disposition'];
if (contentDisposition && contentDisposition.includes('attachment')) {
console.log(`Download detected: ${response.url()}`);
try {
const buffer = await response.buffer();
console.log(`Downloaded file size: ${buffer.length} bytes`);
} catch (error) {
console.error('Error downloading file:', error);
}
}
};
page.on('response', responseListener);
// Click the element
await page.evaluate(() => {
// Locate the table body
const tableBody = document.querySelector(
'#ergebnissForm\\:selectedSuchErgebnisFormTable_data'
);
if (!tableBody) {
throw new Error('Table body not found');
}
// Locate the first row
const firstRow = tableBody.querySelector('tr:nth-child(1)');
if (!firstRow) {
throw new Error('First row not found');
}
// Locate the last cell in the first row
const lastCell = firstRow.querySelector('td:last-child');
if (!lastCell) {
throw new Error('Last cell not found in the first row');
}
// Locate the last <a> element in the last cell
const lastLink = lastCell.querySelector('a:last-of-type');
if (!lastLink) {
throw new Error('Last link not found in the last cell');
}
// Simulate a click on the last <a> element
(lastLink as HTMLElement).click();
});
// Optional: Wait for some response or navigation triggered by the click
await page.waitForTimeout(10000);
page.off('response', responseListener);
} }
} }