omlish 0.0.0.dev447__py3-none-any.whl → 0.0.0.dev484__py3-none-any.whl

omlish/http/clients/default.py

@@ -1,60 +1,239 @@
+import abc
+import contextlib
 import typing as ta
 
+from ... import lang
 from ..headers import CanHttpHeaders
+from .asyncs import AsyncHttpClient
+from .base import HttpClientContext
 from .base import HttpRequest
 from .base import HttpResponse
 from .sync import HttpClient
-from .urllib import UrllibHttpClient
+
+
+with lang.auto_proxy_import(globals()):
+    from . import httpx as _httpx
+    from . import urllib as _urllib
+
+
+C = ta.TypeVar('C')
+R = ta.TypeVar('R')
+
+
+##
+
+
+class _DefaultRequester(lang.Abstract, ta.Generic[C, R]):
+    def __call__(
+            self,
+            url: str,
+            method: str | None = None,
+            *,
+            headers: CanHttpHeaders | None = None,
+            data: bytes | str | None = None,
+
+            timeout_s: float | None = None,
+
+            context: HttpClientContext | None = None,
+            check: bool = False,
+            client: C | None = None,  # noqa
+
+            **kwargs: ta.Any,
+    ) -> R:
+        request = HttpRequest(  # noqa
+            url,
+            method=method,
+
+            headers=headers,
+            data=data,
+
+            timeout_s=timeout_s,
+
+            **kwargs,
+        )
+
+        return self._do(
+            request,
+            context=context,
+            check=check,
+            client=client,
+        )
+
+    @abc.abstractmethod
+    def _do(
+            self,
+            request: HttpRequest,  # noqa
+            *,
+            context: HttpClientContext | None = None,
+            check: bool = False,
+            client: C | None = None,  # noqa
+    ) -> R:
+        raise NotImplementedError
 
 
 ##
 
 
 def _default_client() -> HttpClient:
-    return UrllibHttpClient()
+    return _urllib.UrllibHttpClient()
 
 
 def client() -> HttpClient:
     return _default_client()
 
 
-def request(
-        url: str,
-        method: str | None = None,
-        *,
-        headers: CanHttpHeaders | None = None,
-        data: bytes | str | None = None,
+@contextlib.contextmanager
+def manage_client(client: HttpClient | None = None) -> ta.Generator[HttpClient]:  # noqa
+    if client is not None:
+        yield client
+
+    else:
+        with _default_client() as client:  # noqa
+            yield client
+
+
+#
+
+
+class _BaseSyncDefaultRequester(_DefaultRequester[HttpClient, R], lang.Abstract, ta.Generic[R]):
+    def _do(
+            self,
+            request: HttpRequest,  # noqa
+            *,
+            context: HttpClientContext | None = None,
+            check: bool = False,
+            client: HttpClient | None = None,  # noqa
+    ) -> R:
+        if context is None:
+            context = HttpClientContext()
+
+        if client is not None:
+            return self._do_(
+                client,
+                context,
+                request,
+                check=check,
+            )
+
+        else:
+            with _default_client() as client:  # noqa
+                return self._do_(
+                    client,
+                    context,
+                    request,
+                    check=check,
+                )
+
+    @abc.abstractmethod
+    def _do_(
+            self,
+            client: HttpClient,  # noqa
+            context: HttpClientContext,
+            request: HttpRequest,  # noqa
+            *,
+            check: bool = False,  # noqa
+    ) -> R:
+        raise NotImplementedError
+
+
+class _SyncDefaultRequester(_BaseSyncDefaultRequester[HttpResponse]):
+    def _do_(
+            self,
+            client: HttpClient,  # noqa
+            context: HttpClientContext,
+            request: HttpRequest,  # noqa
+            *,
+            check: bool = False,  # noqa
+    ) -> HttpResponse:
+        return client.request(
+            request,
+            context=context,
+            check=check,
+        )
+
+
+request = _SyncDefaultRequester()
 
-        timeout_s: float | None = None,
 
-        check: bool = False,
+##
 
-        client: HttpClient | None = None,  # noqa
 
-        **kwargs: ta.Any,
-) -> HttpResponse:
-    req = HttpRequest(
-        url,
-        method=method,
+def _default_async_client() -> AsyncHttpClient:
+    return _httpx.HttpxAsyncHttpClient()
 
-        headers=headers,
-        data=data,
 
-        timeout_s=timeout_s,
+def async_client() -> AsyncHttpClient:
+    return _default_async_client()
 
-        **kwargs,
-    )
 
-    def do(cli: HttpClient) -> HttpResponse:
-        return cli.request(
-            req,
+@contextlib.asynccontextmanager
+async def manage_async_client(client: AsyncHttpClient | None = None) -> ta.AsyncGenerator[AsyncHttpClient]:  # noqa
+    if client is not None:
+        yield client
 
+    else:
+        async with _default_async_client() as client:  # noqa
+            yield client
+
+
+#
+
+
+class _BaseAsyncDefaultRequester(_DefaultRequester[AsyncHttpClient, ta.Awaitable[R]], lang.Abstract, ta.Generic[R]):
+    async def _do(
+            self,
+            request: HttpRequest,  # noqa
+            *,
+            context: HttpClientContext | None = None,
+            check: bool = False,
+            client: AsyncHttpClient | None = None,  # noqa
+    ) -> R:
+        if context is None:
+            context = HttpClientContext()
+
+        if client is not None:
+            return await self._do_(
+                client,
+                context,
+                request,
+                check=check,
+            )
+
+        else:
+            async with _default_async_client() as client:  # noqa
+                return await self._do_(
+                    client,
+                    context,
+                    request,
+                    check=check,
+                )
+
+    @abc.abstractmethod
+    def _do_(
+            self,
+            client: AsyncHttpClient,  # noqa
+            context: HttpClientContext,
+            request: HttpRequest,  # noqa
+            *,
+            check: bool = False,  # noqa
+    ) -> ta.Awaitable[R]:
+        raise NotImplementedError
+
+
+class _AsyncDefaultRequester(_BaseAsyncDefaultRequester[HttpResponse]):
+    async def _do_(
+            self,
+            client: AsyncHttpClient,  # noqa
+            context: HttpClientContext,
+            request: HttpRequest,  # noqa
+            *,
+            check: bool = False,
+    ) -> HttpResponse:  # noqa
+        return await client.request(
+            request,
+            context=context,
             check=check,
         )
 
-    if client is not None:
-        return do(client)
 
-    else:
-        with _default_client() as cli:
-            return do(cli)
+async_request = _AsyncDefaultRequester()

omlish/http/clients/executor.py

@@ -0,0 +1,56 @@
+# ruff: noqa: UP043 UP045
+# @omlish-lite
+import dataclasses as dc
+import typing as ta
+
+from .asyncs import AsyncHttpClient
+from .asyncs import AsyncStreamHttpResponse
+from .base import HttpClientContext
+from .base import HttpRequest
+from .sync import HttpClient
+from .sync import StreamHttpResponse
+
+
+##
+
+
+class ExecutorAsyncHttpClient(AsyncHttpClient):
+    def __init__(
+            self,
+            run_in_executor: ta.Callable[..., ta.Awaitable],
+            client: HttpClient,
+    ) -> None:
+        super().__init__()
+
+        self._run_in_executor = run_in_executor
+        self._client = client
+
+    @dc.dataclass(frozen=True)
+    class _StreamAdapter:
+        owner: 'ExecutorAsyncHttpClient'
+        resp: StreamHttpResponse
+
+        async def read1(self, n: int = -1, /) -> bytes:
+            return await self.owner._run_in_executor(self.resp.stream.read1, n)  # noqa
+
+        async def read(self, n: int = -1, /) -> bytes:
+            return await self.owner._run_in_executor(self.resp.stream.read, n)  # noqa
+
+        async def readall(self) -> bytes:
+            return await self.owner._run_in_executor(self.resp.stream.readall)  # noqa
+
+        async def close(self) -> None:
+            return await self.owner._run_in_executor(self.resp.close)  # noqa
+
+    async def _stream_request(self, ctx: HttpClientContext, req: HttpRequest) -> AsyncStreamHttpResponse:
+        resp: StreamHttpResponse = await self._run_in_executor(lambda: self._client.stream_request(req, context=ctx))
+        return AsyncStreamHttpResponse(
+            status=resp.status,
+            headers=resp.headers,
+            request=req,
+            underlying=resp,
+            **(dict(  # type: ignore
+                _stream=(adapter := self._StreamAdapter(self, resp)),
+                _closer=adapter.close,
+            ) if resp.has_data else {}),
+        )

omlish/http/clients/httpx.py

@@ -2,12 +2,17 @@
 TODO:
  - standardize following redirects
 """
+import contextlib
 import functools
 import typing as ta
 
 from ... import dataclasses as dc
 from ... import lang
+from ...io.buffers import ReadableListBuffer
 from ..headers import HttpHeaders
+from .asyncs import AsyncHttpClient
+from .asyncs import AsyncStreamHttpResponse
+from .base import HttpClientContext
 from .base import HttpClientError
 from .base import HttpRequest
 from .sync import HttpClient
@@ -28,16 +33,13 @@ class HttpxHttpClient(HttpClient):
     class _StreamAdapter:
         it: ta.Iterator[bytes]
 
-        def read(self, /, n: int = -1) -> bytes:
-            if n < 0:
-                return b''.join(self.it)
-            else:
-                try:
-                    return next(self.it)
-                except StopIteration:
-                    return b''
+        def read1(self, n: int = -1, /) -> bytes:
+            try:
+                return next(self.it)
+            except StopIteration:
+                return b''
 
-    def _stream_request(self, req: HttpRequest) -> StreamHttpResponse:
+    def _stream_request(self, ctx: HttpClientContext, req: HttpRequest) -> StreamHttpResponse:
         try:
             resp_cm = httpx.stream(
                 method=req.method_or_default,
@@ -59,7 +61,7 @@ class HttpxHttpClient(HttpClient):
                 headers=HttpHeaders(resp.headers.raw),
                 request=req,
                 underlying=resp,
-                stream=self._StreamAdapter(resp.iter_bytes()),
+                _stream=ReadableListBuffer().new_buffered_reader(self._StreamAdapter(resp.iter_bytes())),
                 _closer=resp_close,  # type: ignore
             )
 
@@ -67,6 +69,65 @@ class HttpxHttpClient(HttpClient):
             resp_close()
             raise HttpClientError from e
 
-        except Exception:
+        except BaseException:
             resp_close()
             raise
+
+
+##
+
+
+class HttpxAsyncHttpClient(AsyncHttpClient):
+    @dc.dataclass(frozen=True)
+    class _StreamAdapter:
+        it: ta.AsyncIterator[bytes]
+
+        async def read1(self, n: int = -1, /) -> bytes:
+            try:
+                return await anext(self.it)
+            except StopAsyncIteration:
+                return b''
+
+    async def _stream_request(self, ctx: HttpClientContext, req: HttpRequest) -> AsyncStreamHttpResponse:
+        es = contextlib.AsyncExitStack()
+
+        try:
+            client = await es.enter_async_context(httpx.AsyncClient())
+
+            resp = await es.enter_async_context(client.stream(
+                method=req.method_or_default,
+                url=req.url,
+                headers=req.headers_ or None,  # type: ignore
+                content=req.data,
+                timeout=req.timeout_s,
+            ))
+
+            it = resp.aiter_bytes()
+
+            # FIXME:
+            #  this has a tendency to raise `RuntimeError: async generator ignored GeneratorExit` when all of the
+            #  following conditions are met:
+            #   - stopped iterating midway through
+            #   - shutting down the event loop
+            #   - debugging under pycharm / pydevd
+            #   - running under asyncio
+            #  it does not seem to happen unless all of these conditions are met. see:
+            #   https://gist.github.com/wrmsr/a0578ee5d5371b53804cfb56aeb84cdf .
+            es.push_async_callback(it.aclose)  # type: ignore[attr-defined]
+
+            return AsyncStreamHttpResponse(
+                status=resp.status_code,
+                headers=HttpHeaders(resp.headers.raw),
+                request=req,
+                underlying=resp,
+                _stream=ReadableListBuffer().new_async_buffered_reader(self._StreamAdapter(it)),
+                _closer=es.aclose,
+            )
+
+        except httpx.HTTPError as e:
+            await es.aclose()
+            raise HttpClientError from e
+
+        except BaseException:
+            await es.aclose()
+            raise

omlish/http/clients/middleware.py

@@ -0,0 +1,181 @@
+# ruff: noqa: UP007 UP043 UP045
+# @omlish-lite
+"""
+TODO:
+ - redirect
+  - referrer header?
+  - non-forwarded headers, host check, etc lol
+ - 'check' kw becomes StatusCheckingMiddleware?
+"""
+import dataclasses as dc
+import typing as ta
+import urllib.parse
+
+from ...lite.abstract import Abstract
+from ...lite.check import check
+from ..urls import parsed_url_replace
+from .asyncs import AsyncHttpClient
+from .asyncs import AsyncStreamHttpResponse
+from .base import BaseHttpClient
+from .base import BaseHttpResponse
+from .base import HttpClientContext
+from .base import HttpClientError
+from .base import HttpRequest
+from .sync import HttpClient
+from .sync import StreamHttpResponse
+from .sync import close_http_client_response
+
+
+BaseHttpClientT = ta.TypeVar('BaseHttpClientT', bound=BaseHttpClient)
+
+
+##
+
+
+class HttpClientMiddleware(Abstract):
+    def process_request(
+            self,
+            ctx: HttpClientContext,
+            req: HttpRequest,
+    ) -> HttpRequest:
+        return req
+
+    def process_response(
+            self,
+            ctx: HttpClientContext,
+            req: HttpRequest,
+            resp: BaseHttpResponse,
+    ) -> ta.Union[BaseHttpResponse, HttpRequest]:
+        return resp
+
+
+class AbstractMiddlewareHttpClient(Abstract, ta.Generic[BaseHttpClientT]):
+    def __init__(
+            self,
+            client: BaseHttpClientT,
+            middlewares: ta.Iterable[HttpClientMiddleware],
+    ) -> None:
+        super().__init__()
+
+        self._client = client
+        self._middlewares = list(middlewares)
+
+    def _process_request(
+            self,
+            ctx: HttpClientContext,
+            req: HttpRequest,
+    ) -> HttpRequest:
+        for mw in self._middlewares:
+            req = mw.process_request(ctx, req)
+        return req
+
+    def _process_response(
+            self,
+            ctx: HttpClientContext,
+            req: HttpRequest,
+            resp: BaseHttpResponse,
+    ) -> ta.Union[BaseHttpResponse, HttpRequest]:
+        for mw in self._middlewares:
+            nxt = mw.process_response(ctx, req, resp)
+            if isinstance(nxt, HttpRequest):
+                return nxt
+            else:
+                resp = nxt
+        return resp
+
+
+#
+
+
+class MiddlewareHttpClient(AbstractMiddlewareHttpClient[HttpClient], HttpClient):
+    def _stream_request(self, ctx: HttpClientContext, req: HttpRequest) -> StreamHttpResponse:
+        while True:
+            req = self._process_request(ctx, req)
+
+            resp = self._client.stream_request(req, context=ctx)
+
+            try:
+                out = self._process_response(ctx, req, resp)
+
+                if isinstance(out, HttpRequest):
+                    close_http_client_response(resp)
+                    req = out
+                    continue
+
+                elif isinstance(out, StreamHttpResponse):
+                    return out
+
+                else:
+                    raise TypeError(out)  # noqa
+
+            except Exception:
+                close_http_client_response(resp)
+                raise
+
+        raise RuntimeError
+
+
+class MiddlewareAsyncHttpClient(AbstractMiddlewareHttpClient[AsyncHttpClient], AsyncHttpClient):
+    def _stream_request(self, ctx: HttpClientContext, req: HttpRequest) -> ta.Awaitable[AsyncStreamHttpResponse]:
+        return self._client.stream_request(self._process_request(ctx, req))
+
+
+##
+
+
+class TooManyRedirectsHttpClientError(HttpClientError):
+    pass
+
+
+class RedirectHandlingHttpClientMiddleware(HttpClientMiddleware):
+    DEFAULT_MAX_REDIRECTS: ta.ClassVar[int] = 5
+
+    def __init__(
+            self,
+            *,
+            max_redirects: ta.Optional[int] = None,
+    ) -> None:
+        super().__init__()
+
+        if max_redirects is None:
+            max_redirects = self.DEFAULT_MAX_REDIRECTS
+        self._max_redirects = max_redirects
+
+    @dc.dataclass()
+    class _State:
+        num_redirects: int = 0
+
+    def _get_state(self, ctx: HttpClientContext) -> _State:
+        try:
+            return ctx._dct[self._State]  # noqa
+        except KeyError:
+            ret = ctx._dct[self._State] = self._State()  # noqa
+            return ret
+
+    def process_response(
+            self,
+            ctx: HttpClientContext,
+            req: HttpRequest,
+            resp: BaseHttpResponse,
+    ) -> ta.Union[BaseHttpResponse, HttpRequest]:  # noqa
+        if resp.status == 302:
+            st = self._get_state(ctx)
+            if st.num_redirects >= self._max_redirects:
+                raise TooManyRedirectsHttpClientError
+            st.num_redirects += 1
+
+            rd_url = check.not_none(resp.headers).single_str_dct['location']
+
+            rd_purl = urllib.parse.urlparse(rd_url)
+            if not rd_purl.netloc:
+                rq_purl = urllib.parse.urlparse(req.url)
+                rd_purl = parsed_url_replace(
+                    rd_purl,
+                    scheme=rq_purl.scheme,
+                    netloc=rq_purl.netloc,
+                )
+                rd_url = urllib.parse.urlunparse(rd_purl)
+
+            return dc.replace(req, url=rd_url)
+
+        return resp