add native selectOptionFromDropdown

.changeset/swift-jokes-write.md

@@ -0,0 +1,5 @@
+---
+"@browserbasehq/stagehand": patch
+---
+
+improved handling for OS level dropdowns

evals/evals.config.json

@@ -382,6 +382,18 @@
     {
       "name": "shadow_dom",
       "categories": ["act"]
+    },
+    {
+      "name": "os_dropdown",
+      "categories": ["act"]
+    },
+    {
+      "name": "custom_dropdown",
+      "categories": ["act"]
+    },
+    {
+      "name": "hidden_input_dropdown",
+      "categories": ["act"]
     }
   ]
 }

evals/tasks/custom_dropdown.ts

@@ -0,0 +1,60 @@
+import { EvalFunction } from "@/types/evals";
+
+export const custom_dropdown: EvalFunction = async ({
+  debugUrl,
+  sessionUrl,
+  stagehand,
+  logger,
+}) => {
+  /**
+   * This eval is meant to test whether we do not incorrectly attempt
+   * the selectOptionFromDropdown method (defined in actHandlerUtils.ts) on a
+   * 'dropdown' that is not a <select> element.
+   *
+   * This kind of dropdown must be clicked to be expanded before being interacted
+   * with.
+   */
+
+  try {
+    const page = stagehand.page;
+    await page.goto(
+      "https://browserbase.github.io/stagehand-eval-sites/sites/expand-dropdown/",
+    );
+
+    await page.act("click the 'Select a Country' dropdown");
+
+    // we are expecting stagehand to click the dropdown to expand it,
+    // and therefore the available options should now be contained in the full
+    // a11y tree.
+
+    // to test, we'll grab the full a11y tree, and make sure it contains 'Canada'
+    const extraction = await page.extract();
+    const fullTree = extraction.page_text;
+
+    if (fullTree.includes("Canada")) {
+      return {
+        _success: true,
+        debugUrl,
+        sessionUrl,
+        logs: logger.getLogs(),
+      };
+    }
+    return {
+      _success: false,
+      message: "unable to expand the dropdown",
+      debugUrl,
+      sessionUrl,
+      logs: logger.getLogs(),
+    };
+  } catch (error) {
+    return {
+      _success: false,
+      message: `error attempting to select an option from the dropdown: ${error.message}`,
+      debugUrl,
+      sessionUrl,
+      logs: logger.getLogs(),
+    };
+  } finally {
+    await stagehand.close();
+  }
+};

evals/tasks/hidden_input_dropdown.ts

@@ -0,0 +1,60 @@
+import { EvalFunction } from "@/types/evals";
+
+export const hidden_input_dropdown: EvalFunction = async ({
+  debugUrl,
+  sessionUrl,
+  stagehand,
+  logger,
+}) => {
+  /**
+   * This eval is meant to test whether we do not incorrectly attempt
+   * the selectOptionFromDropdown method (defined in actHandlerUtils.ts) on a
+   * hidden input 'dropdown'.
+   *
+   * This kind of dropdown must be clicked to be expanded before being interacted
+   * with.
+   */
+
+  try {
+    const page = stagehand.page;
+    await page.goto(
+      "https://browserbase.github.io/stagehand-eval-sites/sites/hidden-input-dropdown/",
+    );
+
+    await page.act("click to expand the 'Favourite Colour' dropdown");
+
+    // we are expecting stagehand to click the dropdown to expand it,
+    // and therefore the available options should now be contained in the full
+    // a11y tree.
+
+    // to test, we'll grab the full a11y tree, and make sure it contains 'Green'
+    const extraction = await page.extract();
+    const fullTree = extraction.page_text;
+
+    if (fullTree.includes("Green")) {
+      return {
+        _success: true,
+        debugUrl,
+        sessionUrl,
+        logs: logger.getLogs(),
+      };
+    }
+    return {
+      _success: false,
+      message: "unable to expand the dropdown",
+      debugUrl,
+      sessionUrl,
+      logs: logger.getLogs(),
+    };
+  } catch (error) {
+    return {
+      _success: false,
+      message: `error attempting click to expand the dropdown: ${error.message}`,
+      debugUrl,
+      sessionUrl,
+      logs: logger.getLogs(),
+    };
+  } finally {
+    await stagehand.close();
+  }
+};

evals/tasks/os_dropdown.ts

@@ -0,0 +1,55 @@
+import { EvalFunction } from "@/types/evals";
+
+export const os_dropdown: EvalFunction = async ({
+  debugUrl,
+  sessionUrl,
+  stagehand,
+  logger,
+}) => {
+  /**
+   * This eval is meant to test whether we can correctly select an element
+   * from an OS level dropdown
+   */
+
+  try {
+    const page = stagehand.page;
+    await page.goto(
+      "https://browserbase.github.io/stagehand-eval-sites/sites/nested-dropdown/",
+    );
+
+    await page.act(
+      "choose 'Smog Check Technician' from the 'License Type' dropdown",
+    );
+    const selectedOption = await page
+      .locator(
+        "xpath=/html/body/form/div[1]/div[3]/article/div[2]/div[1]/select[2] >> option:checked",
+      )
+      .textContent();
+
+    if (selectedOption === "Smog Check Technician") {
+      return {
+        _success: true,
+        debugUrl,
+        sessionUrl,
+        logs: logger.getLogs(),
+      };
+    }
+    return {
+      _success: false,
+      message: "incorrect option selected from the dropdown",
+      debugUrl,
+      sessionUrl,
+      logs: logger.getLogs(),
+    };
+  } catch (error) {
+    return {
+      _success: false,
+      message: `error attempting to select an option from the dropdown: ${error.message}`,
+      debugUrl,
+      sessionUrl,
+      logs: logger.getLogs(),
+    };
+  } finally {
+    await stagehand.close();
+  }
+};

lib/a11y/utils.ts

@@ -303,6 +303,14 @@ async function cleanStructuralNodes(
     if (tagName) node.role = tagName;
   }
 
+  if (
+    node.role === "combobox" &&
+    node.encodedId !== undefined &&
+    tagNameMap[node.encodedId] === "select"
+  ) {
+    node.role = "select";
+  }
+
   // 5. drop redundant StaticText children
   const pruned = removeRedundantStaticTextChildren(node, cleanedChildren);
   if (!pruned.length && (node.role === "generic" || node.role === "none")) {

lib/handlers/handlerUtils/actHandlerUtils.ts

@@ -59,6 +59,7 @@ export const methodHandlerMap: Record<
   click: clickElement,
   nextChunk: scrollToNextChunk,
   prevChunk: scrollToPreviousChunk,
+  selectOptionFromDropdown: selectOption,
 };
 
 export async function scrollToNextChunk(ctx: MethodHandlerContext) {
@@ -349,6 +350,26 @@ export async function pressKey(ctx: MethodHandlerContext) {
   }
 }
 
+export async function selectOption(ctx: MethodHandlerContext) {
+  const { locator, xpath, args, logger } = ctx;
+  try {
+    const text = args[0]?.toString() || "";
+    await locator.selectOption(text, { timeout: 5000 });
+  } catch (e) {
+    logger({
+      category: "action",
+      message: "error selecting option",
+      level: 0,
+      auxiliary: {
+        error: { value: e.message, type: "string" },
+        trace: { value: e.stack, type: "string" },
+        xpath: { value: xpath, type: "string" },
+      },
+    });
+    throw new PlaywrightCommandException(e.message);
+  }
+}
+
 export async function clickElement(ctx: MethodHandlerContext) {
   const {
     locator,

lib/prompt.ts

@@ -159,7 +159,9 @@ export function buildActObservePrompt(
   ONLY return one action. If multiple actions are relevant, return the most relevant one. 
   If the user is asking to scroll to a position on the page, e.g., 'halfway' or 0.75, etc, you must return the argument formatted as the correct percentage, e.g., '50%' or '75%', etc.
   If the user is asking to scroll to the next chunk/previous chunk, choose the nextChunk/prevChunk method. No arguments are required here.
-  If the action implies a key press, e.g., 'press enter', 'press a', 'press space', etc., always choose the press method with the appropriate key as argument — e.g. 'a', 'Enter', 'Space'. Do not choose a click action on an on-screen keyboard. Capitalize the first character like 'Enter', 'Tab', 'Escape' only for special keys.`;
+  If the action implies a key press, e.g., 'press enter', 'press a', 'press space', etc., always choose the press method with the appropriate key as argument — e.g. 'a', 'Enter', 'Space'. Do not choose a click action on an on-screen keyboard. Capitalize the first character like 'Enter', 'Tab', 'Escape' only for special keys.
+  If the action implies choosing an option from a dropdown, AND the corresponding element is a 'select' element, choose the selectOptionFromDropdown method. The argument should be the text of the option to select.
+  If the action implies choosing an option from a dropdown, and the corresponding element is NOT a 'select' element, choose the click method.`;
 
   // Add variable names (not values) to the instruction if any
   if (variables && Object.keys(variables).length > 0) {

types/act.ts

@@ -39,6 +39,7 @@ export enum SupportedPlaywrightAction {
   SCROLL = "scrollTo",
   NEXT_CHUNK = "nextChunk",
   PREV_CHUNK = "prevChunk",
+  SELECT_OPTION_FROM_DROPDOWN = "selectOptionFromDropdown",
 }
 
 /**