chore: port routeFromHar, roll to 1.23.0 driver (#1384)

This is part 5/n of the 1.23 port.

Relates #1308, #1374, #1376, #1382, #1383

Ports:

  - [x] microsoft/playwright@259c8d6 (feat: Page.routeFromHar (#14870))
  - [x] microsoft/playwright@79378dd (fix: add pw:api logging to har router (#14903))
  - [x] microsoft/playwright@030e7d2 (chore(har): allow replaying from zip har (#14962))
  - [x] microsoft/playwright@ed6b14f (fix(har): restart redirected navigation (#14939))
  - [x] microsoft/playwright@e5372c3 (chore: move har router into local utils (#14967))
  - [x] microsoft/playwright@920f1d5 (chore: allow routing by uncompressed har (#14987))
  - [x] microsoft/playwright@eb87966 (feat(har): disambiguate requests by post data (#14993))
  - [x] microsoft/playwright@6af6fab (fix(har): internal redirect in renderer-initiated navigations (#15000))
  - [x] microsoft/playwright@9525bed (feat(har): re-add routeFromHAR (#15024))

.pre-commit-config.yaml

@@ -6,6 +6,7 @@ repos:
     hooks:
       - id: trailing-whitespace
       - id: end-of-file-fixer
+        exclude: tests/assets/har-sha1-main-response.txt
       - id: check-yaml
       - id: check-toml
       - id: requirements-txt-fixer

README.md

@@ -4,7 +4,7 @@ Playwright is a Python library to automate [Chromium](https://www.chromium.org/H
 
 |          | Linux | macOS | Windows |
 |   :---   | :---: | :---: | :---:   |
-| Chromium <!-- GEN:chromium-version -->103.0.5060.53<!-- GEN:stop --> | ✅ | ✅ | ✅ |
+| Chromium <!-- GEN:chromium-version -->104.0.5112.20<!-- GEN:stop --> | ✅ | ✅ | ✅ |
 | WebKit <!-- GEN:webkit-version -->15.4<!-- GEN:stop --> | ✅ | ✅ | ✅ |
 | Firefox <!-- GEN:firefox-version -->100.0.2<!-- GEN:stop --> | ✅ | ✅ | ✅ |
 

playwright/_impl/_browser_context.py

@@ -35,7 +35,9 @@
 from playwright._impl._event_context_manager import EventContextManagerImpl
 from playwright._impl._fetch import APIRequestContext
 from playwright._impl._frame import Frame
+from playwright._impl._har_router import HarRouter
 from playwright._impl._helper import (
+    RouteFromHarNotFoundPolicy,
     RouteHandler,
     RouteHandlerCallback,
     TimeoutSettings,
@@ -292,6 +294,20 @@ async def unroute(
         if len(self._routes) == 0:
             await self._disable_interception()
 
+    async def route_from_har(
+        self,
+        har: Union[Path, str],
+        url: URLMatch = None,
+        not_found: RouteFromHarNotFoundPolicy = None,
+    ) -> None:
+        router = await HarRouter.create(
+            local_utils=self._connection.local_utils,
+            file=str(har),
+            not_found_action=not_found or "abort",
+            url_matcher=url,
+        )
+        await router.add_context_route(self)
+
     async def _disable_interception(self) -> None:
         await self._channel.send("setNetworkInterceptionEnabled", dict(enabled=False))
 

playwright/_impl/_har_router.py

@@ -0,0 +1,104 @@
+import asyncio
+import base64
+from typing import TYPE_CHECKING, Optional, cast
+
+from playwright._impl._api_structures import HeadersArray
+from playwright._impl._helper import (
+    HarLookupResult,
+    RouteFromHarNotFoundPolicy,
+    URLMatch,
+)
+from playwright._impl._local_utils import LocalUtils
+
+if TYPE_CHECKING:  # pragma: no cover
+    from playwright._impl._browser_context import BrowserContext
+    from playwright._impl._network import Route
+    from playwright._impl._page import Page
+
+
+class HarRouter:
+    def __init__(
+        self,
+        local_utils: LocalUtils,
+        har_id: str,
+        not_found_action: RouteFromHarNotFoundPolicy,
+        url_matcher: Optional[URLMatch] = None,
+    ) -> None:
+        self._local_utils: LocalUtils = local_utils
+        self._har_id: str = har_id
+        self._not_found_action: RouteFromHarNotFoundPolicy = not_found_action
+        self._options_url_match: Optional[URLMatch] = url_matcher
+
+    @staticmethod
+    async def create(
+        local_utils: LocalUtils,
+        file: str,
+        not_found_action: RouteFromHarNotFoundPolicy,
+        url_matcher: Optional[URLMatch] = None,
+    ) -> "HarRouter":
+        har_id = await local_utils._channel.send("harOpen", {"file": file})
+        return HarRouter(
+            local_utils=local_utils,
+            har_id=har_id,
+            not_found_action=not_found_action,
+            url_matcher=url_matcher,
+        )
+
+    async def _handle(self, route: "Route") -> None:
+        request = route.request
+        response: HarLookupResult = await self._local_utils.har_lookup(
+            harId=self._har_id,
+            url=request.url,
+            method=request.method,
+            headers=await request.headers_array(),
+            postData=request.post_data_buffer,
+            isNavigationRequest=request.is_navigation_request(),
+        )
+        action = response["action"]
+        if action == "redirect":
+            redirect_url = response["redirectURL"]
+            assert redirect_url
+            await route._redirected_navigation_request(redirect_url)
+            return
+
+        if action == "fulfill":
+            body = response["body"]
+            assert body is not None
+            await route.fulfill(
+                status=response.get("status"),
+                headers={
+                    v["name"]: v["value"]
+                    for v in cast(HeadersArray, response.get("headers", []))
+                },
+                body=base64.b64decode(body),
+            )
+            return
+
+        if action == "error":
+            pass
+        # Report the error, but fall through to the default handler.
+
+        if self._not_found_action == "abort":
+            await route.abort()
+            return
+
+        await route.fallback()
+
+    async def add_context_route(self, context: "BrowserContext") -> None:
+        await context.route(
+            url=self._options_url_match or "**/*",
+            handler=lambda route, _: asyncio.create_task(self._handle(route)),
+        )
+        context.once("close", lambda _: self._dispose())
+
+    async def add_page_route(self, page: "Page") -> None:
+        await page.route(
+            url=self._options_url_match or "**/*",
+            handler=lambda route, _: asyncio.create_task(self._handle(route)),
+        )
+        page.once("close", lambda _: self._dispose())
+
+    def _dispose(self) -> None:
+        asyncio.create_task(
+            self._local_utils._channel.send("harClose", {"harId": self._har_id})
+        )

playwright/_impl/_helper.py

@@ -49,6 +49,7 @@
 
 
 if TYPE_CHECKING:  # pragma: no cover
+    from playwright._impl._api_structures import HeadersArray
     from playwright._impl._network import Request, Response, Route
 
 URLMatch = Union[str, Pattern, Callable[[str], bool]]
@@ -67,6 +68,7 @@
 ServiceWorkersPolicy = Literal["allow", "block"]
 HarMode = Literal["full", "minimal"]
 HarContentPolicy = Literal["attach", "embed", "omit"]
+RouteFromHarNotFoundPolicy = Literal["abort", "fallback"]
 
 
 class ErrorPayload(TypedDict, total=False):
@@ -135,6 +137,15 @@ def matches(self, url: str) -> bool:
         return False
 
 
+class HarLookupResult(TypedDict, total=False):
+    action: Literal["error", "redirect", "fulfill", "noentry"]
+    message: Optional[str]
+    redirectURL: Optional[str]
+    status: Optional[int]
+    headers: Optional["HeadersArray"]
+    body: Optional[str]
+
+
 class TimeoutSettings:
     def __init__(self, parent: Optional["TimeoutSettings"]) -> None:
         self._parent = parent

playwright/_impl/_local_utils.py

@@ -12,10 +12,12 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from typing import Dict, List
+import base64
+from typing import Dict, List, Optional, cast
 
-from playwright._impl._api_structures import NameValue
+from playwright._impl._api_structures import HeadersArray, NameValue
 from playwright._impl._connection import ChannelOwner
+from playwright._impl._helper import HarLookupResult, locals_to_params
 
 
 class LocalUtils(ChannelOwner):
@@ -26,3 +28,28 @@ def __init__(
 
     async def zip(self, zip_file: str, entries: List[NameValue]) -> None:
         await self._channel.send("zip", {"zipFile": zip_file, "entries": entries})
+
+    async def har_open(self, file: str) -> None:
+        params = locals_to_params(locals())
+        await self._channel.send("harOpen", params)
+
+    async def har_lookup(
+        self,
+        harId: str,
+        url: str,
+        method: str,
+        headers: HeadersArray,
+        isNavigationRequest: bool,
+        postData: Optional[bytes] = None,
+    ) -> HarLookupResult:
+        params = locals_to_params(locals())
+        if "postData" in params:
+            params["postData"] = base64.b64encode(params["postData"]).decode()
+        return cast(
+            HarLookupResult,
+            await self._channel.send_return_as_dict("harLookup", params),
+        )
+
+    async def har_close(self, harId: str) -> None:
+        params = locals_to_params(locals())
+        await self._channel.send("harClose", params)

playwright/_impl/_network.py

@@ -355,7 +355,6 @@ async def continue_route() -> None:
 
         return continue_route()
 
-    # FIXME: Port corresponding tests, and call this method
     async def _redirected_navigation_request(self, url: str) -> None:
         self._check_not_handled()
         await self._race_with_page_close(

playwright/_impl/_page.py

@@ -53,13 +53,15 @@
 from playwright._impl._event_context_manager import EventContextManagerImpl
 from playwright._impl._file_chooser import FileChooser
 from playwright._impl._frame import Frame
+from playwright._impl._har_router import HarRouter
 from playwright._impl._helper import (
     ColorScheme,
     DocumentLoadState,
     ForcedColors,
     KeyboardModifier,
     MouseButton,
     ReducedMotion,
+    RouteFromHarNotFoundPolicy,
     RouteHandler,
     RouteHandlerCallback,
     TimeoutSettings,
@@ -600,6 +602,20 @@ async def unroute(
         if len(self._routes) == 0:
             await self._disable_interception()
 
+    async def route_from_har(
+        self,
+        har: Union[Path, str],
+        url: URLMatch = None,
+        not_found: RouteFromHarNotFoundPolicy = None,
+    ) -> None:
+        router = await HarRouter.create(
+            local_utils=self._connection.local_utils,
+            file=str(har),
+            not_found_action=not_found or "abort",
+            url_matcher=url,
+        )
+        await router.add_page_route(self)
+
     async def _disable_interception(self) -> None:
         await self._channel.send("setNetworkInterceptionEnabled", dict(enabled=False))
 

playwright/async_api/_generated.py

@@ -268,7 +268,9 @@ def timing(self) -> ResourceTiming:
     def headers(self) -> typing.Dict[str, str]:
         """Request.headers
 
-        **DEPRECATED** Incomplete list of headers as seen by the rendering engine. Use `request.all_headers()` instead.
+        An object with the request HTTP headers. The header names are lower-cased. Note that this method does not return
+        security-related headers, including cookie-related ones. You can use `request.all_headers()` for complete list of
+        headers that include `cookie` information.
 
         Returns
         -------
@@ -411,7 +413,9 @@ def status_text(self) -> str:
     def headers(self) -> typing.Dict[str, str]:
         """Response.headers
 
-        **DEPRECATED** Incomplete list of headers as seen by the rendering engine. Use `response.all_headers()` instead.
+        An object with the response HTTP headers. The header names are lower-cased. Note that this method does not return
+        security-related headers, including cookie-related ones. You can use `response.all_headers()` for complete list
+        of headers that include `cookie` information.
 
         Returns
         -------
@@ -7736,6 +7740,43 @@ async def unroute(
             )
         )
 
+    async def route_from_har(
+        self,
+        har: typing.Union[pathlib.Path, str],
+        *,
+        url: typing.Union[str, typing.Pattern, typing.Callable[[str], bool]] = None,
+        not_found: Literal["abort", "fallback"] = None
+    ) -> NoneType:
+        """Page.route_from_har
+
+        If specified the network requests that are made in the page will be served from the HAR file. Read more about
+        [Replaying from HAR](https://playwright.dev/python/docs/network#replaying-from-har).
+
+        Playwright will not serve requests intercepted by Service Worker from the HAR file. See
+        [this](https://github.com/microsoft/playwright/issues/1090) issue. We recommend disabling Service Workers when using
+        request interception by setting `Browser.newContext.serviceWorkers` to `'block'`.
+
+        Parameters
+        ----------
+        har : Union[pathlib.Path, str]
+            Path to a [HAR](http://www.softwareishard.com/blog/har-12-spec) file with prerecorded network data. If `path` is a
+            relative path, then it is resolved relative to the current working directory.
+        url : Union[Callable[[str], bool], Pattern, str, NoneType]
+            A glob pattern, regular expression or predicate to match the request URL. Only requests with URL matching the pattern
+            will be surved from the HAR file. If not specified, all requests are served from the HAR file.
+        not_found : Union["abort", "fallback", NoneType]
+            - If set to 'abort' any request not found in the HAR file will be aborted.
+            - If set to 'fallback' missing requests will be sent to the network.
+
+            Defaults to abort.
+        """
+
+        return mapping.from_maybe_impl(
+            await self._impl_obj.route_from_har(
+                har=har, url=self._wrap_handler(url), not_found=not_found
+            )
+        )
+
     async def screenshot(
         self,
         *,
@@ -10432,6 +10473,43 @@ async def unroute(
             )
         )
 
+    async def route_from_har(
+        self,
+        har: typing.Union[pathlib.Path, str],
+        *,
+        url: typing.Union[str, typing.Pattern, typing.Callable[[str], bool]] = None,
+        not_found: Literal["abort", "fallback"] = None
+    ) -> NoneType:
+        """BrowserContext.route_from_har
+
+        If specified the network requests that are made in the context will be served from the HAR file. Read more about
+        [Replaying from HAR](https://playwright.dev/python/docs/network#replaying-from-har).
+
+        Playwright will not serve requests intercepted by Service Worker from the HAR file. See
+        [this](https://github.com/microsoft/playwright/issues/1090) issue. We recommend disabling Service Workers when using
+        request interception by setting `Browser.newContext.serviceWorkers` to `'block'`.
+
+        Parameters
+        ----------
+        har : Union[pathlib.Path, str]
+            Path to a [HAR](http://www.softwareishard.com/blog/har-12-spec) file with prerecorded network data. If `path` is a
+            relative path, then it is resolved relative to the current working directory.
+        url : Union[Callable[[str], bool], Pattern, str, NoneType]
+            A glob pattern, regular expression or predicate to match the request URL. Only requests with URL matching the pattern
+            will be surved from the HAR file. If not specified, all requests are served from the HAR file.
+        not_found : Union["abort", "fallback", NoneType]
+            - If set to 'abort' any request not found in the HAR file will be aborted.
+            - If set to 'fallback' falls through to the next route handler in the handler chain.
+
+            Defaults to abort.
+        """
+
+        return mapping.from_maybe_impl(
+            await self._impl_obj.route_from_har(
+                har=har, url=self._wrap_handler(url), not_found=not_found
+            )
+        )
+
     def expect_event(
         self, event: str, predicate: typing.Callable = None, *, timeout: float = None
     ) -> AsyncEventContextManager: