Merge pull request #275 from getmaxun/iframe

feat: support `iframe` extraction
This commit is contained in:
Rohit
2025-01-05 00:34:48 +05:30
committed by GitHub
6 changed files with 1416 additions and 629 deletions

View File

@@ -189,111 +189,145 @@ function scrapableHeuristics(maxCountPerPage = 50, minArea = 20000, scrolls = 3,
* @returns {Array.<Object.<string, string>>}
*/
window.scrapeSchema = function(lists) {
// Utility functions remain the same
function omap(object, f, kf = (x) => x) {
return Object.fromEntries(
Object.entries(object)
.map(([k, v]) => [kf(k), f(v)]),
);
return Object.fromEntries(
Object.entries(object)
.map(([k, v]) => [kf(k), f(v)]),
);
}
function ofilter(object, f) {
return Object.fromEntries(
Object.entries(object)
.filter(([k, v]) => f(k, v)),
);
}
function findAllElements(config) {
if (!config.shadow || !config.selector.includes('>>')) {
return Array.from(document.querySelectorAll(config.selector));
}
// For shadow DOM, we'll get all possible combinations
const parts = config.selector.split('>>').map(s => s.trim());
let currentElements = [document];
for (let i = 0; i < parts.length; i++) {
const part = parts[i];
const nextElements = [];
for (const element of currentElements) {
let targets;
if (i === 0) {
// First selector is queried from document
targets = Array.from(element.querySelectorAll(part))
.filter(el => {
// Only include elements that either:
// 1. Have an open shadow root
// 2. Don't need shadow root (last part of selector)
if (i === parts.length - 1) return true;
const shadowRoot = el.shadowRoot;
return shadowRoot && shadowRoot.mode === 'open';
});
} else {
// For subsequent selectors, only use elements with open shadow roots
const shadowRoot = element.shadowRoot;
if (!shadowRoot || shadowRoot.mode !== 'open') continue;
targets = Array.from(shadowRoot.querySelectorAll(part));
}
nextElements.push(...targets);
}
if (nextElements.length === 0) return [];
currentElements = nextElements;
}
return currentElements;
}
function getElementValue(element, attribute) {
if (!element) return null;
switch (attribute) {
case 'href': {
const relativeHref = element.getAttribute('href');
return relativeHref ? new URL(relativeHref, window.location.origin).href : null;
}
case 'src': {
const relativeSrc = element.getAttribute('src');
return relativeSrc ? new URL(relativeSrc, window.location.origin).href : null;
}
case 'innerText':
return element.innerText?.trim();
case 'textContent':
return element.textContent?.trim();
default:
return element.getAttribute(attribute) || element.innerText?.trim();
}
return Object.fromEntries(
Object.entries(object)
.filter(([k, v]) => f(k, v)),
);
}
// Get the seed key based on the maximum number of elements found
function findAllElements(config) {
// Regular DOM query if no special delimiters
if (!config.selector.includes('>>') && !config.selector.includes(':>>')) {
return Array.from(document.querySelectorAll(config.selector));
}
// Split by both types of delimiters
const parts = config.selector.split(/(?:>>|:>>)/).map(s => s.trim());
const delimiters = config.selector.match(/(?:>>|:>>)/g) || [];
let currentElements = [document];
for (let i = 0; i < parts.length; i++) {
const part = parts[i];
const nextElements = [];
const isLast = i === parts.length - 1;
const delimiter = delimiters[i] || '';
const isIframeTraversal = delimiter === ':>>';
for (const element of currentElements) {
try {
let targets;
if (i === 0) {
// First selector is queried from main document
targets = Array.from(element.querySelectorAll(part))
.filter(el => {
if (isLast) return true;
// For iframe traversal, only include iframes
if (isIframeTraversal) return el.tagName === 'IFRAME';
// For shadow DOM traversal, only include elements with shadow root
return el.shadowRoot && el.shadowRoot.mode === 'open';
});
} else {
if (isIframeTraversal) {
// Handle iframe traversal
const iframeDocument = element.contentDocument || element.contentWindow?.document;
if (!iframeDocument) continue;
targets = Array.from(iframeDocument.querySelectorAll(part));
if (!isLast) {
targets = targets.filter(el => el.tagName === 'IFRAME');
}
} else {
// Handle shadow DOM traversal
const shadowRoot = element.shadowRoot;
if (!shadowRoot || shadowRoot.mode !== 'open') continue;
targets = Array.from(shadowRoot.querySelectorAll(part));
if (!isLast) {
targets = targets.filter(el => el.shadowRoot && el.shadowRoot.mode === 'open');
}
}
}
nextElements.push(...targets);
} catch (error) {
console.warn('Cannot access content:', error);
continue;
}
}
if (nextElements.length === 0) return [];
currentElements = nextElements;
}
return currentElements;
}
// Modified to handle iframe context for URL resolution
function getElementValue(element, attribute) {
if (!element) return null;
// Get the base URL for resolving relative URLs
const baseURL = element.ownerDocument?.location?.href || window.location.origin;
switch (attribute) {
case 'href': {
const relativeHref = element.getAttribute('href');
return relativeHref ? new URL(relativeHref, baseURL).href : null;
}
case 'src': {
const relativeSrc = element.getAttribute('src');
return relativeSrc ? new URL(relativeSrc, baseURL).href : null;
}
case 'innerText':
return element.innerText?.trim();
case 'textContent':
return element.textContent?.trim();
default:
return element.getAttribute(attribute) || element.innerText?.trim();
}
}
// Rest of the functions remain largely the same
function getSeedKey(listObj) {
const maxLength = Math.max(...Object.values(
omap(listObj, (x) => findAllElements(x).length)
));
return Object.keys(
ofilter(listObj, (_, v) => findAllElements(v).length === maxLength)
)[0];
const maxLength = Math.max(...Object.values(
omap(listObj, (x) => findAllElements(x).length)
));
return Object.keys(
ofilter(listObj, (_, v) => findAllElements(v).length === maxLength)
)[0];
}
// Find minimal bounding elements
function getMBEs(elements) {
return elements.map((element) => {
let candidate = element;
const isUniqueChild = (e) => elements
.filter((elem) => e.parentNode?.contains(elem))
.length === 1;
while (candidate && isUniqueChild(candidate)) {
candidate = candidate.parentNode;
}
return candidate;
let candidate = element;
const isUniqueChild = (e) => elements
.filter((elem) => {
// Handle both iframe and shadow DOM boundaries
const sameContext = elem.getRootNode() === e.getRootNode() &&
elem.ownerDocument === e.ownerDocument;
return sameContext && e.parentNode?.contains(elem);
})
.length === 1;
while (candidate && isUniqueChild(candidate)) {
candidate = candidate.parentNode;
}
return candidate;
});
}
// First try the MBE approach
const seedName = getSeedKey(lists);
const seedElements = findAllElements(lists[seedName]);
const MBEs = getMBEs(seedElements);
@@ -347,164 +381,210 @@ function scrapableHeuristics(maxCountPerPage = 50, minArea = 20000, scrolls = 3,
* @returns {Array.<Array.<Object>>} Array of arrays of scraped items, one sub-array per list
*/
window.scrapeList = async function ({ listSelector, fields, limit = 10 }) {
// Shadow DOM query functions remain unchanged
const queryShadowDOM = (rootElement, selector) => {
if (!selector.includes('>>')) {
// Enhanced query function to handle both iframe and shadow DOM
const queryElement = (rootElement, selector) => {
if (!selector.includes('>>') && !selector.includes(':>>')) {
return rootElement.querySelector(selector);
}
}
const parts = selector.split('>>').map(part => part.trim());
let currentElement = rootElement;
const parts = selector.split(/(?:>>|:>>)/).map(part => part.trim());
let currentElement = rootElement;
for (let i = 0; i < parts.length; i++) {
if (!currentElement) return null;
for (let i = 0; i < parts.length; i++) {
if (!currentElement) return null;
if (!currentElement.querySelector && !currentElement.shadowRoot) {
currentElement = document.querySelector(parts[i]);
continue;
}
// Handle iframe traversal
if (currentElement.tagName === 'IFRAME') {
try {
const iframeDoc = currentElement.contentDocument || currentElement.contentWindow.document;
currentElement = iframeDoc.querySelector(parts[i]);
continue;
} catch (e) {
console.warn('Cannot access iframe content:', e);
return null;
}
}
let nextElement = currentElement.querySelector(parts[i]);
// Try regular DOM first
let nextElement = currentElement.querySelector(parts[i]);
if (!nextElement && currentElement.shadowRoot) {
nextElement = currentElement.shadowRoot.querySelector(parts[i]);
}
// Try shadow DOM if not found
if (!nextElement && currentElement.shadowRoot) {
nextElement = currentElement.shadowRoot.querySelector(parts[i]);
}
if (!nextElement) {
const allChildren = Array.from(currentElement.children || []);
for (const child of allChildren) {
if (child.shadowRoot) {
nextElement = child.shadowRoot.querySelector(parts[i]);
if (nextElement) break;
}
}
}
// Check children's shadow roots if still not found
if (!nextElement) {
const children = Array.from(currentElement.children || []);
for (const child of children) {
if (child.shadowRoot) {
nextElement = child.shadowRoot.querySelector(parts[i]);
if (nextElement) break;
}
}
}
currentElement = nextElement;
}
currentElement = nextElement;
}
return currentElement;
return currentElement;
};
const queryShadowDOMAll = (rootElement, selector) => {
if (!selector.includes('>>')) {
// Enhanced query all function for both contexts
const queryElementAll = (rootElement, selector) => {
if (!selector.includes('>>') && !selector.includes(':>>')) {
return rootElement.querySelectorAll(selector);
}
}
const parts = selector.split('>>').map(part => part.trim());
let currentElements = [rootElement];
for (const part of parts) {
const nextElements = [];
for (const element of currentElements) {
if (element.querySelectorAll) {
nextElements.push(...element.querySelectorAll(part));
}
if (element.shadowRoot) {
nextElements.push(...element.shadowRoot.querySelectorAll(part));
}
const children = Array.from(element.children || []);
for (const child of children) {
if (child.shadowRoot) {
nextElements.push(...child.shadowRoot.querySelectorAll(part));
}
}
}
currentElements = nextElements;
}
return currentElements;
const parts = selector.split(/(?:>>|:>>)/).map(part => part.trim());
let currentElements = [rootElement];
for (const part of parts) {
const nextElements = [];
for (const element of currentElements) {
// Handle iframe traversal
if (element.tagName === 'IFRAME') {
try {
const iframeDoc = element.contentDocument || element.contentWindow.document;
nextElements.push(...iframeDoc.querySelectorAll(part));
} catch (e) {
console.warn('Cannot access iframe content:', e);
continue;
}
} else {
// Regular DOM elements
if (element.querySelectorAll) {
nextElements.push(...element.querySelectorAll(part));
}
// Shadow DOM elements
if (element.shadowRoot) {
nextElements.push(...element.shadowRoot.querySelectorAll(part));
}
// Check children's shadow roots
const children = Array.from(element.children || []);
for (const child of children) {
if (child.shadowRoot) {
nextElements.push(...child.shadowRoot.querySelectorAll(part));
}
}
}
}
currentElements = nextElements;
}
return currentElements;
};
// Enhanced table processing helper functions with shadow DOM support
// Enhanced value extraction with context awareness
function extractValue(element, attribute) {
if (!element) return null;
// Check for shadow root first
if (element.shadowRoot) {
const shadowContent = element.shadowRoot.textContent;
if (shadowContent && shadowContent.trim()) {
return shadowContent.trim();
}
}
if (attribute === 'innerText') {
return element.innerText.trim();
} else if (attribute === 'innerHTML') {
return element.innerHTML.trim();
} else if (attribute === 'src' || attribute === 'href') {
const attrValue = element.getAttribute(attribute);
return attrValue ? new URL(attrValue, window.location.origin).href : null;
}
return element.getAttribute(attribute);
if (!element) return null;
// Get context-aware base URL
const baseURL = element.ownerDocument?.location?.href || window.location.origin;
// Check shadow root first
if (element.shadowRoot) {
const shadowContent = element.shadowRoot.textContent;
if (shadowContent?.trim()) {
return shadowContent.trim();
}
}
if (attribute === 'innerText') {
return element.innerText.trim();
} else if (attribute === 'innerHTML') {
return element.innerHTML.trim();
} else if (attribute === 'src' || attribute === 'href') {
const attrValue = element.getAttribute(attribute);
return attrValue ? new URL(attrValue, baseURL).href : null;
}
return element.getAttribute(attribute);
}
// Enhanced table ancestor finding with context support
function findTableAncestor(element) {
let currentElement = element;
const MAX_DEPTH = 5;
let depth = 0;
while (currentElement && depth < MAX_DEPTH) {
// Check if current element is in shadow DOM
if (currentElement.getRootNode() instanceof ShadowRoot) {
currentElement = currentElement.getRootNode().host;
continue;
}
if (currentElement.tagName === 'TD') {
return { type: 'TD', element: currentElement };
} else if (currentElement.tagName === 'TR') {
return { type: 'TR', element: currentElement };
}
currentElement = currentElement.parentElement;
depth++;
}
return null;
let currentElement = element;
const MAX_DEPTH = 5;
let depth = 0;
while (currentElement && depth < MAX_DEPTH) {
// Handle shadow DOM
if (currentElement.getRootNode() instanceof ShadowRoot) {
currentElement = currentElement.getRootNode().host;
continue;
}
if (currentElement.tagName === 'TD') {
return { type: 'TD', element: currentElement };
} else if (currentElement.tagName === 'TR') {
return { type: 'TR', element: currentElement };
}
// Handle iframe crossing
if (currentElement.tagName === 'IFRAME') {
try {
currentElement = currentElement.contentDocument.body;
} catch (e) {
return null;
}
} else {
currentElement = currentElement.parentElement;
}
depth++;
}
return null;
}
// Helper function to get cell index
function getCellIndex(td) {
let index = 0;
let sibling = td;
// Handle shadow DOM case
if (td.getRootNode() instanceof ShadowRoot) {
const shadowRoot = td.getRootNode();
const allCells = Array.from(shadowRoot.querySelectorAll('td'));
return allCells.indexOf(td);
}
while (sibling = sibling.previousElementSibling) {
index++;
}
return index;
if (td.getRootNode() instanceof ShadowRoot) {
const shadowRoot = td.getRootNode();
const allCells = Array.from(shadowRoot.querySelectorAll('td'));
return allCells.indexOf(td);
}
let index = 0;
let sibling = td;
while (sibling = sibling.previousElementSibling) {
index++;
}
return index;
}
// Helper function to check for TH elements
function hasThElement(row, tableFields) {
for (const [label, { selector }] of Object.entries(tableFields)) {
const element = queryShadowDOM(row, selector);
if (element) {
let current = element;
while (current && current !== row) {
// Check if we're in shadow DOM
if (current.getRootNode() instanceof ShadowRoot) {
current = current.getRootNode().host;
continue;
}
if (current.tagName === 'TH') {
return true;
}
current = current.parentElement;
}
}
}
return false;
for (const [_, { selector }] of Object.entries(tableFields)) {
const element = queryElement(row, selector);
if (element) {
let current = element;
while (current && current !== row) {
if (current.getRootNode() instanceof ShadowRoot) {
current = current.getRootNode().host;
continue;
}
if (current.tagName === 'TH') return true;
if (current.tagName === 'IFRAME') {
try {
current = current.contentDocument.body;
} catch (e) {
break;
}
} else {
current = current.parentElement;
}
}
}
}
return false;
}
// Helper function to filter rows
function filterRowsBasedOnTag(rows, tableFields) {
for (const row of rows) {
if (hasThElement(row, tableFields)) {
@@ -520,7 +600,7 @@ function scrapableHeuristics(maxCountPerPage = 50, minArea = 20000, scrolls = 3,
});
}
// Class similarity functions remain unchanged
// Class similarity comparison functions
function calculateClassSimilarity(classList1, classList2) {
const set1 = new Set(classList1);
const set2 = new Set(classList2);
@@ -529,189 +609,237 @@ function scrapableHeuristics(maxCountPerPage = 50, minArea = 20000, scrolls = 3,
return intersection.size / union.size;
}
// Enhanced similar elements finding with context support
function findSimilarElements(baseElement, similarityThreshold = 0.7) {
const baseClasses = Array.from(baseElement.classList);
if (baseClasses.length === 0) return [];
const potentialElements = document.getElementsByTagName(baseElement.tagName);
return Array.from(potentialElements).filter(element => {
if (element === baseElement) return false;
const similarity = calculateClassSimilarity(
baseClasses,
Array.from(element.classList)
);
return similarity >= similarityThreshold;
});
const baseClasses = Array.from(baseElement.classList);
if (baseClasses.length === 0) return [];
const allElements = [];
// Get elements from main document
allElements.push(...document.getElementsByTagName(baseElement.tagName));
// Get elements from shadow DOM
if (baseElement.getRootNode() instanceof ShadowRoot) {
const shadowHost = baseElement.getRootNode().host;
allElements.push(...shadowHost.getElementsByTagName(baseElement.tagName));
}
// Get elements from iframes
const iframes = document.getElementsByTagName('iframe');
for (const iframe of iframes) {
try {
const iframeDoc = iframe.contentDocument || iframe.contentWindow.document;
allElements.push(...iframeDoc.getElementsByTagName(baseElement.tagName));
} catch (e) {
console.warn('Cannot access iframe content:', e);
}
}
return allElements.filter(element => {
if (element === baseElement) return false;
const similarity = calculateClassSimilarity(
baseClasses,
Array.from(element.classList)
);
return similarity >= similarityThreshold;
});
}
// Main scraping logic with shadow DOM support
let containers = queryShadowDOMAll(document, listSelector);
// Main scraping logic with context support
let containers = queryElementAll(document, listSelector);
containers = Array.from(containers);
if (containers.length === 0) return [];
if (limit > 1 && containers.length === 1) {
const baseContainer = containers[0];
const similarContainers = findSimilarElements(baseContainer);
if (similarContainers.length > 0) {
const newContainers = similarContainers.filter(container =>
!container.matches(listSelector)
);
containers = [...containers, ...newContainers];
}
const baseContainer = containers[0];
const similarContainers = findSimilarElements(baseContainer);
if (similarContainers.length > 0) {
const newContainers = similarContainers.filter(container =>
!container.matches(listSelector)
);
containers = [...containers, ...newContainers];
}
}
const containerFields = containers.map(() => ({
tableFields: {},
nonTableFields: {}
tableFields: {},
nonTableFields: {}
}));
// Classify fields
containers.forEach((container, containerIndex) => {
for (const [label, field] of Object.entries(fields)) {
const sampleElement = queryShadowDOM(container, field.selector);
if (sampleElement) {
const ancestor = findTableAncestor(sampleElement);
if (ancestor) {
containerFields[containerIndex].tableFields[label] = {
...field,
tableContext: ancestor.type,
cellIndex: ancestor.type === 'TD' ? getCellIndex(ancestor.element) : -1
};
} else {
containerFields[containerIndex].nonTableFields[label] = field;
}
for (const [label, field] of Object.entries(fields)) {
const sampleElement = queryElement(container, field.selector);
if (sampleElement) {
const ancestor = findTableAncestor(sampleElement);
if (ancestor) {
containerFields[containerIndex].tableFields[label] = {
...field,
tableContext: ancestor.type,
cellIndex: ancestor.type === 'TD' ? getCellIndex(ancestor.element) : -1
};
} else {
containerFields[containerIndex].nonTableFields[label] = field;
}
} else {
containerFields[containerIndex].nonTableFields[label] = field;
}
}
});
const tableData = [];
const nonTableData = [];
// Process table data with shadow DOM support
// Process table data with both iframe and shadow DOM support
for (let containerIndex = 0; containerIndex < containers.length; containerIndex++) {
const container = containers[containerIndex];
const { tableFields } = containerFields[containerIndex];
const container = containers[containerIndex];
const { tableFields } = containerFields[containerIndex];
if (Object.keys(tableFields).length > 0) {
const firstField = Object.values(tableFields)[0];
const firstElement = queryShadowDOM(container, firstField.selector);
let tableContext = firstElement;
// Find table context including shadow DOM
while (tableContext && tableContext.tagName !== 'TABLE' && tableContext !== container) {
if (tableContext.getRootNode() instanceof ShadowRoot) {
tableContext = tableContext.getRootNode().host;
} else {
tableContext = tableContext.parentElement;
}
}
if (Object.keys(tableFields).length > 0) {
const firstField = Object.values(tableFields)[0];
const firstElement = queryElement(container, firstField.selector);
let tableContext = firstElement;
// Find table context including both iframe and shadow DOM
while (tableContext && tableContext.tagName !== 'TABLE' && tableContext !== container) {
if (tableContext.getRootNode() instanceof ShadowRoot) {
tableContext = tableContext.getRootNode().host;
continue;
}
if (tableContext.tagName === 'IFRAME') {
try {
tableContext = tableContext.contentDocument.body;
} catch (e) {
break;
}
} else {
tableContext = tableContext.parentElement;
}
}
if (tableContext) {
// Get rows from both regular DOM and shadow DOM
const rows = [];
if (tableContext.shadowRoot) {
rows.push(...tableContext.shadowRoot.getElementsByTagName('TR'));
}
rows.push(...tableContext.getElementsByTagName('TR'));
const processedRows = filterRowsBasedOnTag(rows, tableFields);
for (let rowIndex = 0; rowIndex < Math.min(processedRows.length, limit); rowIndex++) {
const record = {};
const currentRow = processedRows[rowIndex];
for (const [label, { selector, attribute, cellIndex }] of Object.entries(tableFields)) {
let element = null;
if (cellIndex >= 0) {
let td = currentRow.children[cellIndex];
// Check shadow DOM for td
if (!td && currentRow.shadowRoot) {
const shadowCells = currentRow.shadowRoot.children;
if (shadowCells && shadowCells.length > cellIndex) {
td = shadowCells[cellIndex];
}
}
if (td) {
element = queryShadowDOM(td, selector);
if (!element && selector.split(">").pop().includes('td:nth-child')) {
element = td;
}
if (tableContext) {
// Get rows from all contexts
const rows = [];
// Get rows from regular DOM
rows.push(...tableContext.getElementsByTagName('TR'));
// Get rows from shadow DOM
if (tableContext.shadowRoot) {
rows.push(...tableContext.shadowRoot.getElementsByTagName('TR'));
}
// Get rows from iframes
if (tableContext.tagName === 'IFRAME') {
try {
const iframeDoc = tableContext.contentDocument || tableContext.contentWindow.document;
rows.push(...iframeDoc.getElementsByTagName('TR'));
} catch (e) {
console.warn('Cannot access iframe rows:', e);
}
}
const processedRows = filterRowsBasedOnTag(rows, tableFields);
for (let rowIndex = 0; rowIndex < Math.min(processedRows.length, limit); rowIndex++) {
const record = {};
const currentRow = processedRows[rowIndex];
for (const [label, { selector, attribute, cellIndex }] of Object.entries(tableFields)) {
let element = null;
if (cellIndex >= 0) {
// Get TD element considering both contexts
let td = currentRow.children[cellIndex];
// Check shadow DOM for td
if (!td && currentRow.shadowRoot) {
const shadowCells = currentRow.shadowRoot.children;
if (shadowCells && shadowCells.length > cellIndex) {
td = shadowCells[cellIndex];
}
}
if (td) {
element = queryElement(td, selector);
if (!element && selector.split(/(?:>>|:>>)/).pop().includes('td:nth-child')) {
element = td;
}
if (!element) {
const tagOnlySelector = selector.split('.')[0];
element = queryShadowDOM(td, tagOnlySelector);
}
if (!element) {
let currentElement = td;
while (currentElement && currentElement.children.length > 0) {
let foundContentChild = false;
for (const child of currentElement.children) {
if (extractValue(child, attribute)) {
currentElement = child;
foundContentChild = true;
break;
}
}
if (!foundContentChild) break;
}
element = currentElement;
}
}
} else {
element = queryShadowDOM(currentRow, selector);
}
if (element) {
record[label] = extractValue(element, attribute);
}
}
if (!element) {
const tagOnlySelector = selector.split('.')[0];
element = queryElement(td, tagOnlySelector);
}
if (!element) {
let currentElement = td;
while (currentElement && currentElement.children.length > 0) {
let foundContentChild = false;
for (const child of currentElement.children) {
if (extractValue(child, attribute)) {
currentElement = child;
foundContentChild = true;
break;
}
}
if (!foundContentChild) break;
}
element = currentElement;
}
}
} else {
element = queryElement(currentRow, selector);
}
if (element) {
record[label] = extractValue(element, attribute);
}
}
if (Object.keys(record).length > 0) {
tableData.push(record);
}
}
}
}
if (Object.keys(record).length > 0) {
tableData.push(record);
}
}
}
}
}
// Non-table data scraping remains unchanged
// Process non-table data with both contexts support
for (let containerIndex = 0; containerIndex < containers.length; containerIndex++) {
if (nonTableData.length >= limit) break;
if (nonTableData.length >= limit) break;
const container = containers[containerIndex];
const { nonTableFields } = containerFields[containerIndex];
const container = containers[containerIndex];
const { nonTableFields } = containerFields[containerIndex];
if (Object.keys(nonTableFields).length > 0) {
const record = {};
if (Object.keys(nonTableFields).length > 0) {
const record = {};
for (const [label, { selector, attribute }] of Object.entries(nonTableFields)) {
const relativeSelector = selector.split('>>').slice(-1)[0];
const element = queryShadowDOM(container, relativeSelector);
if (element) {
record[label] = extractValue(element, attribute);
}
}
if (Object.keys(record).length > 0) {
nonTableData.push(record);
}
}
for (const [label, { selector, attribute }] of Object.entries(nonTableFields)) {
// Get the last part of the selector after any context delimiter
const relativeSelector = selector.split(/(?:>>|:>>)/).slice(-1)[0];
const element = queryElement(container, relativeSelector);
if (element) {
record[label] = extractValue(element, attribute);
}
}
if (Object.keys(record).length > 0) {
nonTableData.push(record);
}
}
}
// Merge and limit the results
const scrapedData = [...tableData, ...nonTableData];
return scrapedData;
};
};
/**
* Gets all children of the elements matching the listSelector,

View File

@@ -680,11 +680,25 @@ export default class Interpreter extends EventEmitter {
return workflow;
}
private removeSpecialSelectors(workflow: Workflow) {
for (let actionId = workflow.length - 1; actionId >= 0; actionId--) {
const step = workflow[actionId];
if (step.where && Array.isArray(step.where.selectors)) {
// Filter out if selector has EITHER ":>>" OR ">>"
step.where.selectors = step.where.selectors.filter(selector =>
!(selector.includes(':>>') || selector.includes('>>'))
);
}
}
return workflow;
}
private async runLoop(p: Page, workflow: Workflow) {
let workflowCopy: Workflow = JSON.parse(JSON.stringify(workflow));
// remove shadow selectors
workflowCopy = this.removeShadowSelectors(workflowCopy);
workflowCopy = this.removeSpecialSelectors(workflowCopy);
// apply ad-blocker to the current page
try {

View File

@@ -129,6 +129,12 @@ export interface BaseActionInfo {
hasOnlyText: boolean;
}
interface IframeSelector {
full: string;
isIframe: boolean;
}
interface ShadowSelector {
full: string;
mode: string;
@@ -148,7 +154,8 @@ export interface Selectors {
hrefSelector: string|null;
accessibilitySelector: string|null;
formSelector: string|null;
shadowSelector: ShadowSelector | null;
iframeSelector: IframeSelector|null;
shadowSelector: ShadowSelector|null;
}
/**

File diff suppressed because it is too large Load Diff

View File

@@ -13,6 +13,10 @@ export const getBestSelectorForAction = (action: Action) => {
case ActionType.DragAndDrop: {
const selectors = action.selectors;
if (selectors?.iframeSelector?.full) {
return selectors.iframeSelector.full;
if (selectors?.shadowSelector?.full) {
return selectors.shadowSelector.full;
}

View File

@@ -12,6 +12,7 @@ import { useTranslation } from 'react-i18next';
interface ElementInfo {
tagName: string;
hasOnlyText?: boolean;
isIframeContent?: boolean;
isShadowRoot?: boolean;
innerText?: string;
url?: string;
@@ -117,24 +118,47 @@ export const BrowserWindow = () => {
}, [screenShot, canvasRef, socket, screencastHandler]);
const highlighterHandler = useCallback((data: { rect: DOMRect, selector: string, elementInfo: ElementInfo | null, childSelectors?: string[] }) => {
console.log("LIST SELECTOR", listSelector);
console.log("DATA SELECTOR", data.selector);
console.log("CHILD SELECTORS", data.childSelectors);
if (getList === true) {
if (listSelector) {
socket?.emit('listSelector', { selector: listSelector });
const hasValidChildSelectors = Array.isArray(data.childSelectors) && data.childSelectors.length > 0;
if (limitMode) {
setHighlighterData(null);
} else if (paginationMode) {
// only set highlighterData if type is not empty, 'none', 'scrollDown', or 'scrollUp'
// Only set highlighterData if type is not empty, 'none', 'scrollDown', or 'scrollUp'
if (paginationType !== '' && !['none', 'scrollDown', 'scrollUp'].includes(paginationType)) {
setHighlighterData(data);
} else {
setHighlighterData(null);
}
} else if (data.childSelectors && data.childSelectors.includes(data.selector)) {
// highlight only valid child elements within the listSelector
// Highlight only valid child elements within the listSelector
setHighlighterData(data);
} else if (data.elementInfo?.isIframeContent && data.childSelectors) {
// Handle pure iframe elements - similar to previous shadow DOM logic but using iframe syntax
// Check if the selector matches any iframe child selectors
const isIframeChild = data.childSelectors.some(childSelector =>
data.selector.includes(':>>') && // Iframe uses :>> for traversal
childSelector.split(':>>').some(part =>
data.selector.includes(part.trim())
)
);
setHighlighterData(isIframeChild ? data : null);
} else if (data.selector.includes(':>>') && hasValidChildSelectors) {
// Handle mixed DOM cases with iframes
// Split the selector into parts and check each against child selectors
const selectorParts = data.selector.split(':>>').map(part => part.trim());
const isValidMixedSelector = selectorParts.some(part =>
// We know data.childSelectors is defined due to hasValidChildSelectors check
data.childSelectors!.some(childSelector =>
childSelector.includes(part)
)
);
setHighlighterData(isValidMixedSelector ? data : null);
} else if (data.elementInfo?.isShadowRoot && data.childSelectors) {
// New case: Handle pure Shadow DOM elements
// Check if the selector matches any shadow root child selectors
@@ -145,7 +169,7 @@ export const BrowserWindow = () => {
)
);
setHighlighterData(isShadowChild ? data : null);
} else if (data.selector.includes('>>') && hasValidChildSelectors) {
} else if (data.selector.includes('>>') && hasValidChildSelectors) {
// New case: Handle mixed DOM cases
// Split the selector into parts and check each against child selectors
const selectorParts = data.selector.split('>>').map(part => part.trim());
@@ -156,19 +180,18 @@ export const BrowserWindow = () => {
)
);
setHighlighterData(isValidMixedSelector ? data : null);
}
else {
} else {
// if !valid child in normal mode, clear the highlighter
setHighlighterData(null);
}
} else {
// set highlighterData for the initial listSelector selection
}
} else {
// Set highlighterData for the initial listSelector selection
setHighlighterData(data);
}
} else {
// for non-list steps
}
} else {
// For non-list steps
setHighlighterData(data);
}
}
}, [highlighterData, getList, socket, listSelector, paginationMode, paginationType, captureStage]);