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

feat: produce full path from child element up to top-level parent

Browse Source
This commit is contained in:
karishmas6
2024-09-03 09:06:27 +05:30
parent f09cb3fc73
commit 3b398d1027
1 changed files with 7 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
server/src/workflow-management/selector.ts
View File

@@ -786,13 +786,12 @@ export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates
}; };
export const getChildSelectors = async (page: Page, parentSelector: string, maxDepth: number = 5): 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 getNonUniqueSelector(element: HTMLElement): string { function getNonUniqueSelector(element: HTMLElement): string {
let selector = element.tagName.toLowerCase(); let selector = element.tagName.toLowerCase();
// Ensure that className is a string before splitting
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));
@@ -809,27 +808,23 @@ export const getChildSelectors = async (page: Page, parentSelector: string, maxD
function getSelectorPath(element: HTMLElement | null): string { function getSelectorPath(element: HTMLElement | null): string {
const path: string[] = []; const path: string[] = [];
let depth = 0;
while (element && element !== document.body && depth < maxDepth) { while (element && element !== document.body) {
const selector = getNonUniqueSelector(element); const selector = getNonUniqueSelector(element);
path.unshift(selector); path.unshift(selector);
element = element.parentElement; element = element.parentElement;
depth++;
} }
return path.join(' > '); return path.join(' > ');
} }
function getAllDescendantSelectors(element: HTMLElement, currentDepth: number = 0): string[] { function getAllDescendantSelectors(element: HTMLElement): string[] {
if (currentDepth >= maxDepth) return [];
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));
selectors = selectors.concat(getAllDescendantSelectors(child, currentDepth + 1)); selectors = selectors.concat(getAllDescendantSelectors(child));
} }
return selectors; return selectors;
@@ -839,7 +834,7 @@ export const getChildSelectors = async (page: Page, parentSelector: string, maxD
if (!parentElement) return []; if (!parentElement) return [];
return getAllDescendantSelectors(parentElement); return getAllDescendantSelectors(parentElement);
}, { parentSelector, maxDepth }); }, parentSelector);
return childSelectors || []; return childSelectors || [];
} catch (error) { } catch (error) {
@@ -850,6 +845,7 @@ 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.