remove the old context tree (#268)

This commit is contained in:
LawyZheng
2024-05-08 10:16:30 +08:00
committed by GitHub
parent bce6326eef
commit 8d87e71891
4 changed files with 10 additions and 20 deletions

View File

@@ -544,7 +544,7 @@ function getListboxOptions(element) {
return selectOptions;
}
function buildTreeFromBody(new_ctx = false) {
function buildTreeFromBody() {
var elements = [];
var resultArray = [];
@@ -634,7 +634,6 @@ function buildTreeFromBody(new_ctx = false) {
}
if (
new_ctx &&
checkRequiredFromStyle(element) &&
!attrs["required"] &&
!attrs["aria-required"]
@@ -712,7 +711,7 @@ function buildTreeFromBody(new_ctx = false) {
elements[interactableParentId].children.push(elementObj);
}
// options already added to the select.options, no need to add options anymore
if (new_ctx && elementObj.options && elementObj.options.length > 0) {
if (elementObj.options && elementObj.options.length > 0) {
return elementObj;
}
// Recursively process the children of the element
@@ -744,7 +743,7 @@ function buildTreeFromBody(new_ctx = false) {
if (parentEle) {
if (
targetParentElements.has(parentEle.tagName.toLowerCase()) ||
(new_ctx && checkParentClass(parentEle.className.toLowerCase()))
checkParentClass(parentEle.className.toLowerCase())
) {
targetContextualParent = parentEle;
}
@@ -939,7 +938,7 @@ function buildTreeFromBody(new_ctx = false) {
element.context = context;
}
if (new_ctx && checkStringIncludeRequire(context)) {
if (checkStringIncludeRequire(context)) {
if (
!element.attributes["required"] &&
!element.attributes["aria-required"]
@@ -949,10 +948,6 @@ function buildTreeFromBody(new_ctx = false) {
}
}
if (!new_ctx) {
return [elements, resultArray];
}
resultArray = removeOrphanNode(resultArray);
resultArray.forEach((root) => {
trimDuplicatedText(root);

View File

@@ -184,7 +184,7 @@ async def scrape_web_unsafe(
await remove_bounding_boxes(page)
await scroll_to_top(page, drow_boxes=False)
elements, element_tree = await get_interactable_element_tree(page, browser_state.new_context_tree)
elements, element_tree = await get_interactable_element_tree(page)
element_tree = cleanup_elements(copy.deepcopy(element_tree))
_build_element_links(elements)
@@ -211,15 +211,15 @@ async def scrape_web_unsafe(
)
async def get_interactable_element_tree(page: Page, new_context_tree: bool) -> tuple[list[dict], list[dict]]:
async def get_interactable_element_tree(page: Page) -> tuple[list[dict], list[dict]]:
"""
Get the element tree of the page, including all the elements that are interactable.
:param page: Page instance to get the element tree from.
:return: Tuple containing the element tree and a map of element IDs to elements.
"""
await page.evaluate(JS_FUNCTION_DEFS)
js_script = "(new_ctx) => buildTreeFromBody(new_ctx)"
elements, element_tree = await page.evaluate(js_script, new_context_tree)
js_script = "() => buildTreeFromBody()"
elements, element_tree = await page.evaluate(js_script)
return elements, element_tree