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

Merge branch 'develop' into iframe

Browse Source
This commit is contained in:
Rohit
2025-01-04 15:51:51 +05:30
committed by GitHub
parent 7b08471ba1 170e4f19a0
commit 5e775fa83f
56 changed files with 3858 additions and 657 deletions
Download Patch File Download Diff File Expand all files Collapse all files

170
server/src/workflow-management/selector.ts
View File

@@ -223,24 +223,44 @@ export const getElementInformation = async (
if (originalEl) {
let element = originalEl;
while (element.parentElement) {
const parentRect = element.parentElement.getBoundingClientRect();
const childRect = element.getBoundingClientRect();
if (element.tagName === 'TD' || element.tagName === 'TH') {
const tableParent = element.closest('table');
if (tableParent) {
element = tableParent;
}
}
const fullyContained =
parentRect.left <= childRect.left &&
parentRect.right >= childRect.right &&
parentRect.top <= childRect.top &&
parentRect.bottom >= childRect.bottom;
if (element.tagName !== 'TABLE') {
while (element.parentElement) {
if (element.tagName.toLowerCase() === 'body' ||
element.tagName.toLowerCase() === 'html') {
break;
}
const significantOverlap =
(childRect.width * childRect.height) /
(parentRect.width * parentRect.height) > 0.5;
const parentRect = element.parentElement.getBoundingClientRect();
const childRect = element.getBoundingClientRect();
if (fullyContained && significantOverlap) {
element = element.parentElement;
} else {
break;
const fullyContained =
parentRect.left <= childRect.left &&
parentRect.right >= childRect.right &&
parentRect.top <= childRect.top &&
parentRect.bottom >= childRect.bottom;
const significantOverlap =
(childRect.width * childRect.height) /
(parentRect.width * parentRect.height) > 0.5;
if (fullyContained && significantOverlap) {
const nextParent = element.parentElement;
if (nextParent.tagName.toLowerCase() !== 'body' &&
nextParent.tagName.toLowerCase() !== 'html') {
element = nextParent;
} else {
break;
}
} else {
break;
}
}
}
@@ -512,24 +532,44 @@ export const getRect = async (page: Page, coordinates: Coordinates, listSelector
if (originalEl) {
let element = originalEl;
while (element.parentElement) {
const parentRect = element.parentElement.getBoundingClientRect();
const childRect = element.getBoundingClientRect();
if (element.tagName === 'TD' || element.tagName === 'TH') {
const tableParent = element.closest('table');
if (tableParent) {
element = tableParent;
}
}
const fullyContained =
parentRect.left <= childRect.left &&
parentRect.right >= childRect.right &&
parentRect.top <= childRect.top &&
parentRect.bottom >= childRect.bottom;
if (element.tagName !== 'TABLE') {
while (element.parentElement) {
if (element.tagName.toLowerCase() === 'body' ||
element.tagName.toLowerCase() === 'html') {
break;
}
const significantOverlap =
(childRect.width * childRect.height) /
(parentRect.width * parentRect.height) > 0.5;
const parentRect = element.parentElement.getBoundingClientRect();
const childRect = element.getBoundingClientRect();
if (fullyContained && significantOverlap) {
element = element.parentElement;
} else {
break;
const fullyContained =
parentRect.left <= childRect.left &&
parentRect.right >= childRect.right &&
parentRect.top <= childRect.top &&
parentRect.bottom >= childRect.bottom;
const significantOverlap =
(childRect.width * childRect.height) /
(parentRect.width * parentRect.height) > 0.5;
if (fullyContained && significantOverlap) {
const nextParent = element.parentElement;
if (nextParent.tagName.toLowerCase() !== 'body' &&
nextParent.tagName.toLowerCase() !== 'html') {
element = nextParent;
} else {
break;
}
} else {
break;
}
}
}
@@ -1386,6 +1426,13 @@ export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates
function getNonUniqueSelector(element: HTMLElement): string {
let selector = element.tagName.toLowerCase();
if (selector === 'td' && element.parentElement) {
// Find position among td siblings
const siblings = Array.from(element.parentElement.children);
const position = siblings.indexOf(element) + 1;
return `${selector}:nth-child(${position})`;
}
if (element.className) {
const classes = element.className.split(/\s+/).filter((cls: string) => Boolean(cls));
if (classes.length > 0) {
@@ -1474,25 +1521,45 @@ export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates
let element = originalEl;
if (element.tagName === 'TD' || element.tagName === 'TH') {
const tableParent = element.closest('table');
if (tableParent) {
element = tableParent;
}
}
// if (listSelector === '') {
while (element.parentElement) {
const parentRect = element.parentElement.getBoundingClientRect();
const childRect = element.getBoundingClientRect();
if (element.tagName !== 'TABLE') {
while (element.parentElement) {
if (element.tagName.toLowerCase() === 'body' ||
element.tagName.toLowerCase() === 'html') {
break;
}
const fullyContained =
parentRect.left <= childRect.left &&
parentRect.right >= childRect.right &&
parentRect.top <= childRect.top &&
parentRect.bottom >= childRect.bottom;
const parentRect = element.parentElement.getBoundingClientRect();
const childRect = element.getBoundingClientRect();
const significantOverlap =
(childRect.width * childRect.height) /
(parentRect.width * parentRect.height) > 0.5;
const fullyContained =
parentRect.left <= childRect.left &&
parentRect.right >= childRect.right &&
parentRect.top <= childRect.top &&
parentRect.bottom >= childRect.bottom;
if (fullyContained && significantOverlap) {
element = element.parentElement;
} else {
break;
const significantOverlap =
(childRect.width * childRect.height) /
(parentRect.width * parentRect.height) > 0.5;
if (fullyContained && significantOverlap) {
const nextParent = element.parentElement;
if (nextParent.tagName.toLowerCase() !== 'body' &&
nextParent.tagName.toLowerCase() !== 'html') {
element = nextParent;
} else {
break;
}
} else {
break;
}
}
}
// }
@@ -1504,7 +1571,6 @@ export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates
}, coordinates);
return selectors || { generalSelector: '' };
} else {
console.log(`NON UNIQUE: MODE 2`)
const selectors = await page.evaluate(({ x, y }: { x: number, y: number }) => {
const getDeepestElementFromPoint = (x: number, y: number): HTMLElement | null => {
// First, get the element at the specified coordinates in the main document
@@ -1555,6 +1621,12 @@ export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates
function getNonUniqueSelector(element: HTMLElement): string {
let selector = element.tagName.toLowerCase();
if (selector === 'td' && element.parentElement) {
const siblings = Array.from(element.parentElement.children);
const position = siblings.indexOf(element) + 1;
return `${selector}:nth-child(${position})`;
}
if (element.className) {
const classes = element.className.split(/\s+/).filter((cls: string) => Boolean(cls));
if (classes.length > 0) {
@@ -1664,6 +1736,12 @@ export const getChildSelectors = async (page: Page, parentSelector: string): Pro
function getNonUniqueSelector(element: HTMLElement): string {
let selector = element.tagName.toLowerCase();
if (selector === 'td' && element.parentElement) {
const siblings = Array.from(element.parentElement.children);
const position = siblings.indexOf(element) + 1;
return `${selector}:nth-child(${position})`;
}
const className = typeof element.className === 'string' ? element.className : '';
if (className) {
const classes = className.split(/\s+/).filter((cls: string) => Boolean(cls));