vodorod/parcer
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: traverse up the parent selector

Browse Source
This commit is contained in:
amhsirak
2024-11-20 01:07:42 +05:30
parent 58847afe24
commit 316531ea57
2 changed files with 15 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
server/src/workflow-management/classes/Generator.ts
View File

@@ -559,6 +559,8 @@ export class WorkflowGenerator {
if (this.listSelector !== '') { if (this.listSelector !== '') {
const childSelectors = await getChildSelectors(page, this.listSelector || ''); const childSelectors = await getChildSelectors(page, this.listSelector || '');
this.socket.emit('highlighter', { rect, selector: displaySelector, elementInfo, childSelectors }) this.socket.emit('highlighter', { rect, selector: displaySelector, elementInfo, childSelectors })
console.log(`Child Selectors: ${childSelectors}`)
console.log(`List sekector ${this.listSelector}`)
} else { } else {
this.socket.emit('highlighter', { rect, selector: displaySelector, elementInfo }); this.socket.emit('highlighter', { rect, selector: displaySelector, elementInfo });
} }

20
server/src/workflow-management/selector.ts
View File

@@ -790,20 +790,21 @@ export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates
} }
}; };
export const getChildSelectors = async (page: Page, parentSelector: string): Promise<string[]> => { export const getChildSelectors = async (page: Page, parentSelector: string): Promise<string[]> => {
try { try {
const childSelectors = await page.evaluate((parentSelector: string) => { const childSelectors = await page.evaluate((parentSelector: string) => {
// Function to generate non-unique selector for an element
function getNonUniqueSelector(element: HTMLElement): string { function getNonUniqueSelector(element: HTMLElement): string {
let selector = element.tagName.toLowerCase(); let selector = element.tagName.toLowerCase();
// Add class names if available
const className = typeof element.className === 'string' ? element.className : ''; const className = typeof element.className === 'string' ? element.className : '';
if (className) { if (className) {
const classes = className.split(/\s+/).filter((cls: string) => Boolean(cls)); const classes = className.split(/\s+/).filter((cls: string) => Boolean(cls));
if (classes.length > 0) { if (classes.length > 0) {
const validClasses = classes.filter((cls: string) => !cls.startsWith('!') && !cls.includes(':')); const validClasses = classes.filter((cls: string) => !cls.startsWith('!') && !cls.includes(':'));
if (validClasses.length > 0) { if (validClasses.length > 0) {
selector += '.' + validClasses.map(cls => CSS.escape(cls)).join('.'); selector += '.' + validClasses.map(cls => CSS.escape(cls)).join('.');
} }
} }
} }
@@ -811,30 +812,34 @@ export const getChildSelectors = async (page: Page, parentSelector: string): Pro
return selector; return selector;
} }
function getSelectorPath(element: HTMLElement | null): string { // Function to get the full CSS selector path from the current element to the parentSelector
function getSelectorPath(element: HTMLElement | null, parentElement: HTMLElement): string {
if (!element || !element.parentElement) return ''; if (!element || !element.parentElement) return '';
const parentSelector = getNonUniqueSelector(element.parentElement); const parentSelector = getNonUniqueSelector(parentElement);
const elementSelector = getNonUniqueSelector(element); const elementSelector = getNonUniqueSelector(element);
return `${parentSelector} > ${elementSelector}`; return `${parentSelector} > ${elementSelector}`;
} }
function getAllDescendantSelectors(element: HTMLElement, stopAtParent: HTMLElement | null): string[] { // Function to recursively gather all descendant selectors
function getAllDescendantSelectors(element: HTMLElement, parentElement: HTMLElement): string[] {
let selectors: string[] = []; let selectors: string[] = [];
const children = Array.from(element.children) as HTMLElement[]; const children = Array.from(element.children) as HTMLElement[];
for (const child of children) { for (const child of children) {
selectors.push(getSelectorPath(child)); selectors.push(getSelectorPath(child, parentElement));
selectors = selectors.concat(getAllDescendantSelectors(child, stopAtParent)); selectors = selectors.concat(getAllDescendantSelectors(child, parentElement));
} }
return selectors; return selectors;
} }
// Get the parent element based on the selector
const parentElement = document.querySelector(parentSelector) as HTMLElement; const parentElement = document.querySelector(parentSelector) as HTMLElement;
if (!parentElement) return []; if (!parentElement) return [];
// Gather all descendant selectors starting from the parent element
return getAllDescendantSelectors(parentElement, parentElement); return getAllDescendantSelectors(parentElement, parentElement);
}, parentSelector); }, parentSelector);
@@ -845,6 +850,7 @@ export const getChildSelectors = async (page: Page, parentSelector: string): Pro
} }
}; };
/** /**
* Returns the first pair from the given workflow that contains the given selector * Returns the first pair from the given workflow that contains the given selector
* inside the where condition, and it is the only selector there. * inside the where condition, and it is the only selector there.