chore: import upstream snapshot with attribution

libs/cua-bench/example_tasks/basic_gui_env/gui/index.html

@@ -0,0 +1,12 @@
+<div class="p-1">
+  <h1>Click the button</h1>
+  <button id="submit" class="btn" data-instruction="the button">Submit</button>
+  <script>
+    window.__submitted = false;
+    document.getElementById('submit').addEventListener('click', function () {
+      window.__submitted = true;
+      this.textContent = 'Submitted!';
+      this.disabled = true;
+    });
+  </script>
+</div>

libs/cua-bench/example_tasks/basic_gui_env/main.py

@@ -0,0 +1,71 @@
+from pathlib import Path
+
+import cua_bench as cb
+
+
+# Called once per batch
+@cb.tasks_config(split="train")
+def load():
+    os_types = ["linux"]  # ["macos", "win11", "win10"]
+    button_texts = ["Submit"]  # , "Click Me", "Click Here", "OK", "Cancel", "Close", "Start"]
+    return [
+        cb.Task(
+            description=f'Click the "{button_text}" button on the page.',
+            metadata={
+                "button_text": button_text,
+            },
+            computer={
+                "provider": "native",
+                "setup_config": {
+                    "os_type": os_type,
+                    "width": 512,
+                    "height": 512,
+                    "background": "#c0c0c0",
+                },
+            },
+        )
+        for os_type in os_types
+        for button_text in button_texts
+    ]
+
+
+# All code below will be running in a separate process per task
+
+pid = None
+
+
+# Called at start of task
+@cb.setup_task(split="train")
+async def start(task_cfg: cb.Task, session: cb.DesktopSession | cb.MobileSession):
+    global pid
+
+    pid = await session.launch_window(
+        html=(Path(__file__).parent / "gui/index.html").read_text(),
+        title="Simple Env",
+        width=256,
+        height=256,
+    )
+
+
+# Called at end of task
+@cb.evaluate_task(split="train")
+async def evaluate(task_cfg: cb.Task, session: cb.DesktopSession | cb.MobileSession) -> list[float]:
+    global pid
+
+    if pid is None:
+        return [0.0]
+
+    # Check if the submit button was clicked by checking a global flag
+    submitted = await session.execute_javascript(pid, "window.__submitted")
+    return [1.0] if submitted is True else [0.0]
+
+
+# Called after setup_task if run_solution is True
+@cb.solve_task(split="train")
+async def solve(task_cfg: cb.Task, session: cb.DesktopSession | cb.MobileSession):
+    global pid
+
+    if pid is None:
+        return
+
+    await session.click_element(pid, ".btn")

libs/cua-bench/example_tasks/basic_gui_env/pyproject.toml

@@ -0,0 +1,15 @@
+[project]
+name = "simple-env"
+version = "0.1.0"
+license = "MIT"
+authors = [
+    { name = "Dillon DuPont", email = "dillon@cua.ai" }
+]
+requires-python = ">=3.11"
+dependencies = []
+
+[tool.cua-bench]
+description = "Click a simple button."
+difficulty = "easy"
+category = "grounding"
+tags = ["grounding", "click"]