chore(roll): roll to new upstream RegExp / str escape logic (#2080)

playwright/_impl/_browser_context.py

@@ -566,7 +566,7 @@ def _on_dialog(self, dialog: Dialog) -> None:
             else:
                 asyncio.create_task(dialog.dismiss())
 
-    async def _on_page_error(self, error: Error, page: Optional[Page]) -> None:
+    def _on_page_error(self, error: Error, page: Optional[Page]) -> None:
         self.emit(BrowserContext.Events.WebError, WebError(self._loop, page, error))
         if page:
             page.emit(Page.Events.PageError, error)

playwright/_impl/_locator.py

@@ -52,7 +52,6 @@
 from playwright._impl._str_utils import (
     escape_for_attribute_selector,
     escape_for_text_selector,
-    escape_regex_flags,
 )
 
 if sys.version_info >= (3, 8):  # pragma: no cover
@@ -847,16 +846,12 @@ def set_test_id_attribute_name(attribute_name: str) -> None:
 def get_by_test_id_selector(
     test_id_attribute_name: str, test_id: Union[str, Pattern[str]]
 ) -> str:
-    if isinstance(test_id, Pattern):
-        return f"internal:testid=[{test_id_attribute_name}=/{test_id.pattern}/{escape_regex_flags(test_id)}]"
     return f"internal:testid=[{test_id_attribute_name}={escape_for_attribute_selector(test_id, True)}]"
 
 
 def get_by_attribute_text_selector(
     attr_name: str, text: Union[str, Pattern[str]], exact: bool = None
 ) -> str:
-    if isinstance(text, Pattern):
-        return f"internal:attr=[{attr_name}=/{text.pattern}/{escape_regex_flags(text)}]"
     return f"internal:attr=[{attr_name}={escape_for_attribute_selector(text, exact=exact)}]"
 
 
@@ -915,9 +910,7 @@ def get_by_role_selector(
         props.append(
             (
                 "name",
-                f"/{name.pattern}/{escape_regex_flags(name)}"
-                if isinstance(name, Pattern)
-                else escape_for_attribute_selector(name, exact),
+                escape_for_attribute_selector(name, exact=exact),
             )
         )
     if pressed is not None:

playwright/_impl/_str_utils.py

@@ -39,15 +39,31 @@ def escape_for_regex(text: str) -> str:
     return re.sub(r"[.*+?^>${}()|[\]\\]", "\\$&", text)
 
 
+def escape_regex_for_selector(text: Pattern) -> str:
+    # Even number of backslashes followed by the quote -> insert a backslash.
+    return (
+        "/"
+        + re.sub(r'(^|[^\\])(\\\\)*(["\'`])', r"\1\2\\\3", text.pattern).replace(
+            ">>", "\\>\\>"
+        )
+        + "/"
+        + escape_regex_flags(text)
+    )
+
+
 def escape_for_text_selector(
     text: Union[str, Pattern[str]], exact: bool = None, case_sensitive: bool = None
 ) -> str:
     if isinstance(text, Pattern):
-        return f"/{text.pattern}/{escape_regex_flags(text)}"
+        return escape_regex_for_selector(text)
     return json.dumps(text) + ("s" if exact else "i")
 
 
-def escape_for_attribute_selector(value: str, exact: bool = None) -> str:
+def escape_for_attribute_selector(
+    value: Union[str, Pattern], exact: bool = None
+) -> str:
+    if isinstance(value, Pattern):
+        return escape_regex_for_selector(value)
     # TODO: this should actually be
     #   cssEscape(value).replace(/\\ /g, ' ')
     # However, our attribute selectors do not conform to CSS parsing spec,

tests/async/test_selectors_get_by.py

@@ -12,6 +12,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import re
+
 from playwright.async_api import Page, expect
 
 
@@ -87,6 +89,42 @@ async def test_get_by_escaping(page: Page) -> None:
         0, timeout=500
     )
 
+    await page.set_content(
+        """<label for=target>foo &gt;&gt; bar</label><input id=target>"""
+    )
+    await page.eval_on_selector(
+        "input",
+        """input => {
+        input.setAttribute('placeholder', 'foo >> bar');
+        input.setAttribute('title', 'foo >> bar');
+        input.setAttribute('alt', 'foo >> bar');
+    }""",
+    )
+    assert await page.get_by_text("foo >> bar").text_content() == "foo >> bar"
+    await expect(page.locator("label")).to_have_text("foo >> bar")
+    await expect(page.get_by_text("foo >> bar")).to_have_text("foo >> bar")
+    assert (
+        await page.get_by_text(re.compile("foo >> bar")).text_content() == "foo >> bar"
+    )
+    await expect(page.get_by_label("foo >> bar")).to_have_attribute("id", "target")
+    await expect(page.get_by_label(re.compile("foo >> bar"))).to_have_attribute(
+        "id", "target"
+    )
+    await expect(page.get_by_placeholder("foo >> bar")).to_have_attribute(
+        "id", "target"
+    )
+    await expect(page.get_by_alt_text("foo >> bar")).to_have_attribute("id", "target")
+    await expect(page.get_by_title("foo >> bar")).to_have_attribute("id", "target")
+    await expect(page.get_by_placeholder(re.compile("foo >> bar"))).to_have_attribute(
+        "id", "target"
+    )
+    await expect(page.get_by_alt_text(re.compile("foo >> bar"))).to_have_attribute(
+        "id", "target"
+    )
+    await expect(page.get_by_title(re.compile("foo >> bar"))).to_have_attribute(
+        "id", "target"
+    )
+
 
 async def test_get_by_role_escaping(
     page: Page,

tests/async/test_selectors_text.py

@@ -1,6 +1,8 @@
 import re
 
-from playwright.async_api import Page, expect
+import pytest
+
+from playwright.async_api import Error, Page, expect
 
 
 async def test_has_text_and_internal_text_should_match_full_node_text_in_strict_mode(
@@ -33,3 +35,236 @@ async def test_has_text_and_internal_text_should_match_full_node_text_in_strict_
         "div1"
     )
     await expect(page.locator("div", has_text=re.compile("^hello$"))).to_have_id("div2")
+
+
+async def test_should_work(page: Page, server) -> None:
+    await page.set_content(
+        """
+        <div>yo</div><div>ya</div><div>\nye  </div>
+    """
+    )
+    assert await page.eval_on_selector("text=ya", "e => e.outerHTML") == "<div>ya</div>"
+    assert (
+        await page.eval_on_selector('text="ya"', "e => e.outerHTML") == "<div>ya</div>"
+    )
+    assert (
+        await page.eval_on_selector("text=/^[ay]+$/", "e => e.outerHTML")
+        == "<div>ya</div>"
+    )
+    assert (
+        await page.eval_on_selector("text=/Ya/i", "e => e.outerHTML") == "<div>ya</div>"
+    )
+    assert (
+        await page.eval_on_selector("text=ye", "e => e.outerHTML")
+        == "<div>\nye  </div>"
+    )
+    assert ">\nye  </div>" in await page.get_by_text("ye").evaluate("e => e.outerHTML")
+
+    await page.set_content(
+        """
+        <div> ye </div><div>ye</div>
+    """
+    )
+    assert (
+        await page.eval_on_selector('text="ye"', "e => e.outerHTML")
+        == "<div> ye </div>"
+    )
+    assert "> ye </div>" in await page.get_by_text("ye", exact=True).first.evaluate(
+        "e => e.outerHTML"
+    )
+
+    await page.set_content(
+        """
+        <div>yo</div><div>"ya</div><div> hello world! </div>
+    """
+    )
+    assert (
+        await page.eval_on_selector('text="\\"ya"', "e => e.outerHTML")
+        == '<div>"ya</div>'
+    )
+    assert (
+        await page.eval_on_selector("text=/hello/", "e => e.outerHTML")
+        == "<div> hello world! </div>"
+    )
+    assert (
+        await page.eval_on_selector("text=/^\\s*heLLo/i", "e => e.outerHTML")
+        == "<div> hello world! </div>"
+    )
+
+    await page.set_content(
+        """
+        <div>yo<div>ya</div>hey<div>hey</div></div>
+    """
+    )
+    assert (
+        await page.eval_on_selector("text=hey", "e => e.outerHTML") == "<div>hey</div>"
+    )
+    assert (
+        await page.eval_on_selector('text=yo>>text="ya"', "e => e.outerHTML")
+        == "<div>ya</div>"
+    )
+    assert (
+        await page.eval_on_selector('text=yo>> text="ya"', "e => e.outerHTML")
+        == "<div>ya</div>"
+    )
+    assert (
+        await page.eval_on_selector("text=yo >>text='ya'", "e => e.outerHTML")
+        == "<div>ya</div>"
+    )
+    assert (
+        await page.eval_on_selector("text=yo >> text='ya'", "e => e.outerHTML")
+        == "<div>ya</div>"
+    )
+    assert (
+        await page.eval_on_selector("'yo'>>\"ya\"", "e => e.outerHTML")
+        == "<div>ya</div>"
+    )
+    assert (
+        await page.eval_on_selector("\"yo\" >> 'ya'", "e => e.outerHTML")
+        == "<div>ya</div>"
+    )
+
+    await page.set_content(
+        """
+        <div>yo<span id="s1"></span></div><div>yo<span id="s2"></span><span id="s3"></span></div>
+        """
+    )
+    assert (
+        await page.eval_on_selector_all(
+            "text=yo", "es => es.map(e => e.outerHTML).join('\\n')"
+        )
+        == '<div>yo<span id="s1"></span></div>\n<div>yo<span id="s2"></span><span id="s3"></span></div>'
+    )
+
+    await page.set_content("<div>'</div><div>\"</div><div>\\</div><div>x</div>")
+    assert (
+        await page.eval_on_selector("text='\\''", "e => e.outerHTML") == "<div>'</div>"
+    )
+    assert (
+        await page.eval_on_selector("text='\"'", "e => e.outerHTML") == '<div>"</div>'
+    )
+    assert (
+        await page.eval_on_selector('text="\\""', "e => e.outerHTML") == '<div>"</div>'
+    )
+    assert (
+        await page.eval_on_selector('text="\'"', "e => e.outerHTML") == "<div>'</div>"
+    )
+    assert (
+        await page.eval_on_selector('text="\\x"', "e => e.outerHTML") == "<div>x</div>"
+    )
+    assert (
+        await page.eval_on_selector("text='\\x'", "e => e.outerHTML") == "<div>x</div>"
+    )
+    assert (
+        await page.eval_on_selector("text='\\\\'", "e => e.outerHTML")
+        == "<div>\\</div>"
+    )
+    assert (
+        await page.eval_on_selector('text="\\\\"', "e => e.outerHTML")
+        == "<div>\\</div>"
+    )
+    assert await page.eval_on_selector('text="', "e => e.outerHTML") == '<div>"</div>'
+    assert await page.eval_on_selector("text='", "e => e.outerHTML") == "<div>'</div>"
+    assert await page.eval_on_selector('"x"', "e => e.outerHTML") == "<div>x</div>"
+    assert await page.eval_on_selector("'x'", "e => e.outerHTML") == "<div>x</div>"
+    with pytest.raises(Error):
+        await page.query_selector_all('"')
+    with pytest.raises(Error):
+        await page.query_selector_all("'")
+
+    await page.set_content("<div> ' </div><div> \" </div>")
+    assert await page.eval_on_selector('text="', "e => e.outerHTML") == '<div> " </div>'
+    assert await page.eval_on_selector("text='", "e => e.outerHTML") == "<div> ' </div>"
+
+    await page.set_content("<div>Hi''&gt;&gt;foo=bar</div>")
+    assert (
+        await page.eval_on_selector("text=\"Hi''>>foo=bar\"", "e => e.outerHTML")
+        == "<div>Hi''&gt;&gt;foo=bar</div>"
+    )
+    await page.set_content("<div>Hi'\"&gt;&gt;foo=bar</div>")
+    assert (
+        await page.eval_on_selector('text="Hi\'\\">>foo=bar"', "e => e.outerHTML")
+        == "<div>Hi'\"&gt;&gt;foo=bar</div>"
+    )
+
+    await page.set_content("<div>Hi&gt;&gt;<span></span></div>")
+    assert (
+        await page.eval_on_selector('text="Hi>>">>span', "e => e.outerHTML")
+        == "<span></span>"
+    )
+    assert (
+        await page.eval_on_selector("text=/Hi\\>\\>/ >> span", "e => e.outerHTML")
+        == "<span></span>"
+    )
+
+    await page.set_content("<div>a<br>b</div><div>a</div>")
+    assert (
+        await page.eval_on_selector("text=a", "e => e.outerHTML") == "<div>a<br>b</div>"
+    )
+    assert (
+        await page.eval_on_selector("text=b", "e => e.outerHTML") == "<div>a<br>b</div>"
+    )
+    assert (
+        await page.eval_on_selector("text=ab", "e => e.outerHTML")
+        == "<div>a<br>b</div>"
+    )
+    assert await page.query_selector("text=abc") is None
+    assert await page.eval_on_selector_all("text=a", "els => els.length") == 2
+    assert await page.eval_on_selector_all("text=b", "els => els.length") == 1
+    assert await page.eval_on_selector_all("text=ab", "els => els.length") == 1
+    assert await page.eval_on_selector_all("text=abc", "els => els.length") == 0
+
+    await page.set_content("<div></div><span></span>")
+    await page.eval_on_selector(
+        "div",
+        """div => {
+        div.appendChild(document.createTextNode('hello'))
+        div.appendChild(document.createTextNode('world'))
+    }""",
+    )
+    await page.eval_on_selector(
+        "span",
+        """span => {
+        span.appendChild(document.createTextNode('hello'))
+        span.appendChild(document.createTextNode('world'))
+    }""",
+    )
+    assert (
+        await page.eval_on_selector("text=lowo", "e => e.outerHTML")
+        == "<div>helloworld</div>"
+    )
+    assert (
+        await page.eval_on_selector_all(
+            "text=lowo", "els => els.map(e => e.outerHTML).join('')"
+        )
+        == "<div>helloworld</div><span>helloworld</span>"
+    )
+
+    await page.set_content("<span>Sign&nbsp;in</span><span>Hello\n \nworld</span>")
+    assert (
+        await page.eval_on_selector("text=Sign in", "e => e.outerHTML")
+        == "<span>Sign&nbsp;in</span>"
+    )
+    assert len((await page.query_selector_all("text=Sign \tin"))) == 1
+    assert len((await page.query_selector_all('text="Sign in"'))) == 1
+    assert (
+        await page.eval_on_selector("text=lo wo", "e => e.outerHTML")
+        == "<span>Hello\n \nworld</span>"
+    )
+    assert (
+        await page.eval_on_selector('text="Hello world"', "e => e.outerHTML")
+        == "<span>Hello\n \nworld</span>"
+    )
+    assert await page.query_selector('text="lo wo"') is None
+    assert len((await page.query_selector_all("text=lo \nwo"))) == 1
+    assert len(await page.query_selector_all('text="lo \nwo"')) == 0
+
+    await page.set_content("<div>let's<span>hello</span></div>")
+    assert (
+        await page.eval_on_selector("text=/let's/i >> span", "e => e.outerHTML")
+        == "<span>hello</span>"
+    )
+    assert (
+        await page.eval_on_selector("text=/let\\'s/i >> span", "e => e.outerHTML")
+        == "<span>hello</span>"
+    )