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

fix: revert

Browse Source
This commit is contained in:
karishmas6
2024-09-03 03:45:06 +05:30
parent f491d1bf3d
commit 8b20b8223b
1 changed files with 34 additions and 65 deletions
Download Patch File Download Diff File Expand all files Collapse all files

99
server/src/workflow-management/selector.ts
View File

@@ -721,23 +721,6 @@ export const getSelectors = async (page: Page, coordinates: Coordinates) => {
return null; return null;
}; };
function generateNonUniqueSelector(element: HTMLElement): string {
let selector = element.tagName.toLowerCase();
if (element.className) {
const classes = element.className.split(/\s+/).filter((cls: string) => Boolean(cls));
if (classes.length > 0) {
const validClasses = classes.filter((cls: string) => !cls.startsWith('!') && !cls.includes(':'));
if (validClasses.length > 0) {
selector += '.' + validClasses.map(cls => CSS.escape(cls)).join('.');
}
}
}
return selector;
}
interface SelectorResult { interface SelectorResult {
generalSelector: string; generalSelector: string;
@@ -755,23 +738,23 @@ export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates
try { try {
const selectors = await page.evaluate(({ x, y }: { x: number, y: number }) => { const selectors = await page.evaluate(({ x, y }: { x: number, y: number }) => {
// function getNonUniqueSelector(element: HTMLElement): string { function getNonUniqueSelector(element: HTMLElement): string {
// let selector = element.tagName.toLowerCase(); let selector = element.tagName.toLowerCase();
// // Avoid using IDs to maintain non-uniqueness // Avoid using IDs to maintain non-uniqueness
// if (element.className) { if (element.className) {
// const classes = element.className.split(/\s+/).filter((cls: string) => Boolean(cls)); const classes = element.className.split(/\s+/).filter((cls: string) => Boolean(cls));
// if (classes.length > 0) { if (classes.length > 0) {
// // Exclude utility classes and escape special characters // Exclude utility classes and escape special characters
// 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('.');
// } }
// } }
// } }
// return selector; return selector;
// } }
function getSelectorPath(element: HTMLElement | null): string { function getSelectorPath(element: HTMLElement | null): string {
const path: string[] = []; const path: string[] = [];
@@ -779,7 +762,7 @@ export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates
const maxDepth = 2; // Limiting the depth of the selector path const maxDepth = 2; // Limiting the depth of the selector path
while (element && element !== document.body && depth < maxDepth) { while (element && element !== document.body && depth < maxDepth) {
const selector = generateNonUniqueSelector(element); const selector = getNonUniqueSelector(element);
path.unshift(selector); path.unshift(selector);
element = element.parentElement; element = element.parentElement;
@@ -806,40 +789,27 @@ export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates
} }
}; };
export const getChildSelectors = async (page: Page, parentSelector: string, maxDepth: number = 2): Promise<string[]> => { export const getChildSelectors = async (page: Page, parentSelector: string): Promise<string[]> => {
try { try {
const childSelectors = await page.evaluate(({ parentSelector, maxDepth }: { parentSelector: string, maxDepth: number }) => { const childSelectors = await page.evaluate((parentSelector: string) => {
function getSelectors(element: HTMLElement, currentDepth: number): string[] { const parentElement = document.querySelector(parentSelector);
if (currentDepth > maxDepth) return [];
const selectors: string[] = [];
const childElements = Array.from(element.children);
for (const child of childElements) {
let selector = child.tagName.toLowerCase();
if (child.className) {
const classes = child.className.split(/\s+/).filter((cls: string) => Boolean(cls));
if (classes.length > 0) {
const validClasses = classes.filter((cls: string) => !cls.startsWith('!') && !cls.includes(':'));
if (validClasses.length > 0) {
selector += '.' + validClasses.map(cls => CSS.escape(cls)).join('.');
}
}
}
selectors.push(selector);
// Recursively get selectors for deeper levels
selectors.push(...getSelectors(child as HTMLElement, currentDepth + 1));
}
return selectors;
}
const parentElement = document.querySelector(parentSelector) as HTMLElement;
if (!parentElement) return []; if (!parentElement) return [];
return getSelectors(parentElement, 0); const childElements = Array.from(parentElement.children);
}, { parentSelector, maxDepth }); return childElements.map(child => {
let selector = child.tagName.toLowerCase();
if (child.className) {
const classes = child.className.split(/\s+/).filter((cls: string) => Boolean(cls));
if (classes.length > 0) {
const validClasses = classes.filter((cls: string) => !cls.startsWith('!') && !cls.includes(':'));
if (validClasses.length > 0) {
selector += '.' + validClasses.map(cls => CSS.escape(cls)).join('.');
}
}
}
return selector;
});
}, parentSelector);
return childSelectors || []; return childSelectors || [];
} catch (error) { } catch (error) {
@@ -848,7 +818,6 @@ export const getChildSelectors = async (page: Page, parentSelector: string, maxD
} }
}; };
/** /**
* 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.