Bug 1770480 - [wdspec] Add tests for script evaluation inside sandboxes for "script.evaluate" and "script.callFunction" commands. r=webdriver-reviewers,jdescottes

Depends on D153440

Differential Revision: https://phabricator.services.mozilla.com/D153442

testing/web-platform/meta/webdriver/tests/bidi/script/call_function/sandbox.py.ini

@@ -0,0 +1,28 @@
+[sandbox.py]
+  [test_sandbox]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782
+
+  [test_sandbox_with_empty_name]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782
+
+  [test_switch_sandboxes]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782
+
+  [test_sandbox_with_side_effects]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782
+
+  [test_arguments]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782
+
+  [test_exception_details[True\]]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782
+
+  [test_exception_details[False\]]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782

testing/web-platform/meta/webdriver/tests/bidi/script/evaluate/sandbox.py.ini

@@ -0,0 +1,24 @@
+[sandbox.py]
+  [test_sandbox]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782
+
+  [test_sandbox_with_empty_name]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782
+
+  [test_switch_sandboxes]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782
+
+  [test_sandbox_with_side_effects]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782
+
+  [test_exception_details[True\]]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782
+
+  [test_exception_details[False\]]
+    disabled:
+      if os == "android": https://bugzilla.mozilla.org/show_bug.cgi?id=1506782

testing/web-platform/tests/webdriver/tests/bidi/script/call_function/sandbox.py

@@ -0,0 +1,180 @@
+import pytest
+
+from webdriver.bidi.modules.script import ContextTarget, ScriptEvaluateResultException
+
+from ... import any_int, any_string, recursive_compare
+from .. import any_stack_trace
+
+
+@pytest.mark.asyncio
+async def test_sandbox(bidi_session, new_tab):
+    # Make changes in window
+    await bidi_session.script.call_function(
+        function_declaration="() => { window.foo = 1; }",
+        target=ContextTarget(new_tab["context"]),
+        await_promise=True,
+    )
+
+    # Check that changes are not present in sandbox
+    result_in_sandbox = await bidi_session.script.call_function(
+        function_declaration="() => window.foo",
+        target=ContextTarget(new_tab["context"], "sandbox"),
+        await_promise=True,
+    )
+    assert result_in_sandbox == {"type": "undefined"}
+
+    # Make changes in sandbox
+    await bidi_session.script.call_function(
+        function_declaration="() => { window.bar = 2; }",
+        target=ContextTarget(new_tab["context"], "sandbox"),
+        await_promise=True,
+    )
+
+    # Make sure that changes are present in sandbox
+    result_in_sandbox = await bidi_session.script.call_function(
+        function_declaration="() => window.bar",
+        target=ContextTarget(new_tab["context"], "sandbox"),
+        await_promise=True,
+    )
+    assert result_in_sandbox == {"type": "number", "value": 2}
+
+    # Make sure that changes didn't leak from sandbox
+    result_in_window = await bidi_session.script.call_function(
+        function_declaration="() => window.bar",
+        target=ContextTarget(new_tab["context"]),
+        await_promise=True,
+    )
+    assert result_in_window == {"type": "undefined"}
+
+
+@pytest.mark.asyncio
+async def test_sandbox_with_empty_name(bidi_session, new_tab):
+    # BiDi specification doesn't have restrictions of a sandbox name,
+    # that's why we want to make sure that it works with an empty name
+    await bidi_session.script.call_function(
+        function_declaration="() => window.foo = 'bar'",
+        target=ContextTarget(new_tab["context"], ""),
+        await_promise=True,
+    )
+
+    # Make sure that we can find the sandbox with the empty name
+    result = await bidi_session.script.call_function(
+        function_declaration="() => window.foo",
+        target=ContextTarget(new_tab["context"], ""),
+        await_promise=True,
+    )
+    assert result == {"type": "string", "value": "bar"}
+
+    # Make sure that changes didn't leak from sandbox
+    result = await bidi_session.script.evaluate(
+        expression="window.foo",
+        target=ContextTarget(new_tab["context"]),
+        await_promise=True,
+    )
+    assert result == {"type": "undefined"}
+
+
+@pytest.mark.asyncio
+async def test_switch_sandboxes(bidi_session, new_tab):
+    # Test that sandboxes are retained when switching between them
+    await bidi_session.script.call_function(
+        function_declaration="() => { window.foo = 1; }",
+        target=ContextTarget(new_tab["context"], "sandbox_1"),
+        await_promise=True,
+    )
+    await bidi_session.script.call_function(
+        function_declaration="() => { window.foo = 2; }",
+        target=ContextTarget(new_tab["context"], "sandbox_2"),
+        await_promise=True,
+    )
+
+    result_in_sandbox_1 = await bidi_session.script.call_function(
+        function_declaration="() => window.foo",
+        target=ContextTarget(new_tab["context"], "sandbox_1"),
+        await_promise=True,
+    )
+    assert result_in_sandbox_1 == {"type": "number", "value": 1}
+
+    result_in_sandbox_2 = await bidi_session.script.call_function(
+        function_declaration="() => window.foo",
+        target=ContextTarget(new_tab["context"], "sandbox_2"),
+        await_promise=True,
+    )
+    assert result_in_sandbox_2 == {"type": "number", "value": 2}
+
+
+@pytest.mark.asyncio
+async def test_sandbox_with_side_effects(bidi_session, new_tab):
+    # Make sure changing the node in sandbox will affect the other sandbox as well
+    await bidi_session.script.call_function(
+        function_declaration="() => document.querySelector('body').textContent = 'foo'",
+        target=ContextTarget(new_tab["context"], "sandbox_1"),
+        await_promise=True,
+    )
+    expected_value = {"type": "string", "value": "foo"}
+
+    result_in_sandbox_1 = await bidi_session.script.call_function(
+        function_declaration="() => document.querySelector('body').textContent",
+        target=ContextTarget(new_tab["context"], "sandbox_1"),
+        await_promise=True,
+    )
+    assert result_in_sandbox_1 == expected_value
+
+    result_in_sandbox_2 = await bidi_session.script.call_function(
+        function_declaration="() => document.querySelector('body').textContent",
+        target=ContextTarget(new_tab["context"], "sandbox_2"),
+        await_promise=True,
+    )
+    assert result_in_sandbox_2 == expected_value
+
+
+@pytest.mark.asyncio
+async def test_arguments(bidi_session, new_tab):
+    argument = {
+        "type": "set",
+        "value": [
+            {"type": "string", "value": "foobar"},
+        ],
+    }
+    result = await bidi_session.script.call_function(
+        function_declaration=f"""(arg) => {{
+            if(! (arg instanceof Set))
+                throw Error("Argument type should be Set, but was "+
+                    Object.prototype.toString.call(arg));
+            return arg;
+        }}""",
+        arguments=[argument],
+        await_promise=False,
+        target=ContextTarget(new_tab["context"], "sandbox"),
+    )
+
+    recursive_compare(argument, result)
+
+
+@pytest.mark.asyncio
+@pytest.mark.parametrize("await_promise", [True, False])
+async def test_exception_details(bidi_session, new_tab, await_promise):
+    function_declaration = "()=>{{ throw 1 }}"
+    if await_promise:
+        function_declaration = "async" + function_declaration
+
+    with pytest.raises(ScriptEvaluateResultException) as exception:
+        await bidi_session.script.call_function(
+            function_declaration=function_declaration,
+            await_promise=await_promise,
+            target=ContextTarget(new_tab["context"], "sandbox"),
+        )
+
+    recursive_compare(
+        {
+            "realm": any_string,
+            "exceptionDetails": {
+                "columnNumber": any_int,
+                "exception": {"type": "number", "value": 1},
+                "lineNumber": any_int,
+                "stackTrace": any_stack_trace,
+                "text": any_string,
+            },
+        },
+        exception.value.result,
+    )

testing/web-platform/tests/webdriver/tests/bidi/script/evaluate/sandbox.py

@@ -0,0 +1,158 @@
+import pytest
+
+from webdriver.bidi.modules.script import ContextTarget, ScriptEvaluateResultException
+
+from ... import any_int, any_string, recursive_compare
+from .. import any_stack_trace
+
+
+@pytest.mark.asyncio
+async def test_sandbox(bidi_session, new_tab):
+    # Make changes in window
+    await bidi_session.script.evaluate(
+        expression="window.foo = 1",
+        target=ContextTarget(new_tab["context"]),
+        await_promise=True,
+    )
+
+    # Check that changes are not present in sandbox
+    result_in_sandbox = await bidi_session.script.evaluate(
+        expression="window.foo",
+        target=ContextTarget(new_tab["context"], "sandbox"),
+        await_promise=True,
+    )
+    assert result_in_sandbox == {"type": "undefined"}
+
+    # Make changes in sandbox
+    await bidi_session.script.evaluate(
+        expression="window.bar = 1",
+        target=ContextTarget(new_tab["context"], "sandbox"),
+        await_promise=True,
+    )
+
+    # Make sure that changes are present in sandbox
+    result_in_sandbox = await bidi_session.script.evaluate(
+        expression="window.bar",
+        target=ContextTarget(new_tab["context"], "sandbox"),
+        await_promise=True,
+    )
+    assert result_in_sandbox == {"type": "number", "value": 1}
+
+    # Make sure that changes didn't leak from sandbox
+    result = await bidi_session.script.evaluate(
+        expression="window.bar",
+        target=ContextTarget(new_tab["context"]),
+        await_promise=True,
+    )
+    assert result == {"type": "undefined"}
+
+
+@pytest.mark.asyncio
+async def test_sandbox_with_empty_name(bidi_session, new_tab):
+    # BiDi specification doesn't have restrictions of a sandbox name,
+    # that's why we want to make sure that it works with an empty name
+    await bidi_session.script.evaluate(
+        expression="window.foo = 'bar'",
+        target=ContextTarget(new_tab["context"], ""),
+        await_promise=True,
+    )
+
+    # Make sure that we can find the sandbox with the empty name
+    result = await bidi_session.script.evaluate(
+        expression="window.foo",
+        target=ContextTarget(new_tab["context"], ""),
+        await_promise=True,
+    )
+    assert result == {"type": "string", "value": "bar"}
+
+    # Make sure that changes didn't leak from sandbox
+    result = await bidi_session.script.evaluate(
+        expression="window.foo",
+        target=ContextTarget(new_tab["context"]),
+        await_promise=True,
+    )
+    assert result == {"type": "undefined"}
+
+
+@pytest.mark.asyncio
+async def test_switch_sandboxes(bidi_session, new_tab):
+    # Test that sandboxes are retained when switching between them
+    await bidi_session.script.evaluate(
+        expression="window.foo = 1",
+        target=ContextTarget(new_tab["context"], "sandbox_1"),
+        await_promise=True,
+    )
+    await bidi_session.script.evaluate(
+        expression="window.foo = 2",
+        target=ContextTarget(new_tab["context"], "sandbox_2"),
+        await_promise=True,
+    )
+
+    result_in_sandbox_1 = await bidi_session.script.evaluate(
+        expression="window.foo",
+        target=ContextTarget(new_tab["context"], "sandbox_1"),
+        await_promise=True,
+    )
+    assert result_in_sandbox_1 == {"type": "number", "value": 1}
+
+    result_in_sandbox_2 = await bidi_session.script.evaluate(
+        expression="window.foo",
+        target=ContextTarget(new_tab["context"], "sandbox_2"),
+        await_promise=True,
+    )
+    assert result_in_sandbox_2 == {"type": "number", "value": 2}
+
+
+@pytest.mark.asyncio
+async def test_sandbox_with_side_effects(bidi_session, new_tab):
+    # Make sure changing the node in sandbox will affect the other sandbox as well
+    await bidi_session.script.evaluate(
+        expression="document.querySelector('body').textContent = 'foo'",
+        target=ContextTarget(new_tab["context"], "sandbox_1"),
+        await_promise=True,
+    )
+    expected_value = {"type": "string", "value": "foo"}
+
+    result_in_sandbox_1 = await bidi_session.script.evaluate(
+        expression="document.querySelector('body').textContent",
+        target=ContextTarget(new_tab["context"], "sandbox_1"),
+        await_promise=True,
+    )
+    assert result_in_sandbox_1 == expected_value
+
+    result_in_sandbox_2 = await bidi_session.script.evaluate(
+        expression="document.querySelector('body').textContent",
+        target=ContextTarget(new_tab["context"], "sandbox_2"),
+        await_promise=True,
+    )
+    assert result_in_sandbox_2 == expected_value
+
+
+@pytest.mark.asyncio
+@pytest.mark.parametrize("await_promise", [True, False])
+async def test_exception_details(bidi_session, new_tab, await_promise):
+    if await_promise:
+        expression = "Promise.reject(1)"
+    else:
+        expression = "throw 1"
+
+    with pytest.raises(ScriptEvaluateResultException) as exception:
+        await bidi_session.script.evaluate(
+            expression=expression,
+            target=ContextTarget(new_tab["context"], "sandbox"),
+            await_promise=await_promise,
+        )
+
+    recursive_compare(
+        {
+            "realm": any_string,
+            "exceptionDetails": {
+                "columnNumber": any_int,
+                "exception": {"type": "number", "value": 1},
+                "lineNumber": any_int,
+                "stackTrace": any_stack_trace,
+                "text": any_string,
+            },
+        },
+        exception.value.result,
+    )