remove the old context tree (#268)

skyvern/webeye/browser_factory.py

@@ -136,13 +136,11 @@ class BrowserState:
         browser_context: BrowserContext | None = None,
         page: Page | None = None,
         browser_artifacts: BrowserArtifacts = BrowserArtifacts(),
-        new_context_tree: bool = False,
     ):
         self.pw = pw
         self.browser_context = browser_context
         self.page = page
         self.browser_artifacts = browser_artifacts
-        self.new_context_tree = new_context_tree
 
     async def _close_all_other_pages(self) -> None:
         if not self.browser_context or not self.page:

skyvern/webeye/browser_manager.py

@@ -25,7 +25,6 @@ class BrowserManager:
     async def _create_browser_state(
         proxy_location: ProxyLocation | None = None,
         url: str | None = None,
-        new_context_tree: bool = False,
         task_id: str | None = None,
     ) -> BrowserState:
         pw = await async_playwright().start()
@@ -40,7 +39,6 @@ class BrowserManager:
             browser_context=browser_context,
             page=None,
             browser_artifacts=browser_artifacts,
-            new_context_tree=new_context_tree,
         )
 
     async def get_or_create_for_task(self, task: Task) -> BrowserState:
@@ -55,9 +53,8 @@ class BrowserManager:
             self.pages[task.task_id] = self.pages[task.workflow_run_id]
             return self.pages[task.task_id]
 
-        new_ctx = True
+        LOG.info("Creating browser state for task", task_id=task.task_id)
-        LOG.info("Creating browser state for task", task_id=task.task_id, new_ctx=new_ctx)
+        browser_state = await self._create_browser_state(task.proxy_location, task.url, task.task_id)
-        browser_state = await self._create_browser_state(task.proxy_location, task.url, new_ctx, task.task_id)
 
         # The URL here is only used when creating a new page, and not when using an existing page.
         # This will make sure browser_state.page is not None.

skyvern/webeye/scraper/domUtils.js

@@ -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);

skyvern/webeye/scraper/scraper.py

@@ -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)"
+    js_script = "() => buildTreeFromBody()"
-    elements, element_tree = await page.evaluate(js_script, new_context_tree)
+    elements, element_tree = await page.evaluate(js_script)
     return elements, element_tree