gecko-dev/servo/components/net_traits/lib.rs

533 строки
17 KiB
Rust
Исходник Обычный вид История

/* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
#![feature(box_syntax)]
#![feature(slice_patterns)]
#![feature(step_by)]
#![deny(unsafe_code)]
extern crate cookie as cookie_rs;
extern crate heapsize;
#[macro_use]
extern crate heapsize_derive;
extern crate hyper;
extern crate hyper_serde;
extern crate image as piston_image;
extern crate ipc_channel;
#[allow(unused_extern_crates)]
#[macro_use]
extern crate lazy_static;
#[macro_use]
extern crate log;
extern crate msg;
extern crate num_traits;
extern crate serde;
#[macro_use]
extern crate serde_derive;
servo: Merge #14592 - Remove the util crate (from asajeffrey:util-goodbye); r=mbrubeck <!-- Please describe your changes on the following line: --> This PR removes the `util` crate. * Replaced the `spawn_named` and `clamp` functions by appropriate uses of `std::thread::Builder::spawn`, `std::cmp::min` and `std::cmp::max`. * Moved `opts`, `prefs` and `resource_files` into a new `config` crate. * Moved `remutex` and `geometry` into their own crates. --- <!-- Thank you for contributing to Servo! Please replace each `[ ]` by `[X]` when the step is complete, and replace `__` with appropriate data: --> - [X] `./mach build -d` does not report any errors - [X] `./mach test-tidy` does not report any errors - [X] These changes do not require tests because they are refactorings <!-- Pull requests that do not address these steps are welcome, but they will require additional verification as part of the review process. --> Source-Repo: https://github.com/servo/servo Source-Revision: 4eb653817f87e5fb47de34356f558eb76ecbca9f --HG-- rename : servo/components/util/Cargo.toml => servo/components/config/Cargo.toml rename : servo/components/util/basedir.rs => servo/components/config/basedir.rs rename : servo/components/util/lib.rs => servo/components/config/lib.rs rename : servo/components/util/opts.rs => servo/components/config/opts.rs rename : servo/components/util/prefs.rs => servo/components/config/prefs.rs rename : servo/components/util/resource_files.rs => servo/components/config/resource_files.rs rename : servo/components/util/geometry.rs => servo/components/geometry/lib.rs rename : servo/components/util/remutex.rs => servo/components/remutex/lib.rs rename : servo/tests/unit/util/lib.rs => servo/tests/unit/servo_config/lib.rs rename : servo/tests/unit/util/opts.rs => servo/tests/unit/servo_config/opts.rs rename : servo/tests/unit/util/prefs.rs => servo/tests/unit/servo_config/prefs.rs rename : servo/tests/unit/util/remutex.rs => servo/tests/unit/servo_remutex/lib.rs
2016-12-15 03:48:42 +03:00
extern crate servo_config;
extern crate servo_url;
extern crate url;
extern crate uuid;
extern crate webrender_traits;
extern crate websocket;
use cookie_rs::Cookie;
use filemanager_thread::FileManagerThreadMsg;
use heapsize::HeapSizeOf;
use hyper::header::{ContentType, Headers};
use hyper::http::RawStatus;
use hyper::mime::{Attr, Mime};
use hyper_serde::Serde;
servo: Merge #6586 - script: Make the resource task communication use IPC channels (from pcwalton:resource-task-ipc); r=jdm This change makes Servo use serialized messages over IPC channels for resource loading. The goal is to make it easier to make Servo multiprocess in the future. This patch does not make Servo multiprocess now; there are many other channels that need to be changed to IPC before that can happen. It does introduce a dependency on https://github.com/serde-rs/serde and https://github.com/pcwalton/ipc-channel for the first time. At the moment, `ipc-channel` uses JSON for serialization. This is because serde does not yet have official support for bincode. When serde gains support for bincode, I'll switch to that. For now, however, the JSON encoding and decoding will constitute a significant performance regression in resource loading. To avoid having to send boxed `AsyncResponseTarget` trait objects across process boundaries, this series of commits changes `AsyncResponseTarget` to wrap a sender only. It is then the client's responsibility to spawn a thread to proxy calls from that sender to the consumer of the resource data. This only had to be done in a few places. In the future, we may want to collapse those threads into one per process to reduce overhead. (It is impossible to continue to use `AsyncResponseTarget` as a boxed trait object across processes, regardless of how much work is done on `ipc-channel`. Vtables are fundamentally incompatible with IPC across mutually untrusting processes.) In general, I was pretty pleased with how this turned out. The main changes are adding serialization functionality to various objects that `serde` does not know how to serialize natively—the most complicated being Hyper objects—and reworking `AsyncResponseTarget`. The overall structure of the code is unchanged, and other than `AsyncResponseTarget` no functionality was lost in moving to serialization and IPC. r? @jdm Source-Repo: https://github.com/servo/servo Source-Revision: 2eb122f394651232abd683fc576a5c4288bf277f
2015-08-01 01:06:36 +03:00
use ipc_channel::ipc::{self, IpcReceiver, IpcSender};
use ipc_channel::router::ROUTER;
use request::{Request, RequestInit};
use response::{HttpsState, Response};
use servo_url::ServoUrl;
use std::io::Error as IOError;
use storage_thread::StorageThreadMsg;
use websocket::header;
pub mod blob_url_store;
pub mod filemanager_thread;
pub mod hosts;
servo: Merge #9201 - task -> thread (from ajnirp:8512-task-thread); r=jdm for #8512 Source-Repo: https://github.com/servo/servo Source-Revision: d3e2f94f2024f4735f836588ed11303a0abafdf8 --HG-- rename : servo/components/canvas/canvas_paint_task.rs => servo/components/canvas/canvas_paint_thread.rs rename : servo/components/canvas/webgl_paint_task.rs => servo/components/canvas/webgl_paint_thread.rs rename : servo/components/compositing/compositor_task.rs => servo/components/compositing/compositor_thread.rs rename : servo/components/gfx/font_cache_task.rs => servo/components/gfx/font_cache_thread.rs rename : servo/components/gfx/paint_task.rs => servo/components/gfx/paint_thread.rs rename : servo/components/layout/layout_task.rs => servo/components/layout/layout_thread.rs rename : servo/components/net/image_cache_task.rs => servo/components/net/image_cache_thread.rs rename : servo/components/net/resource_task.rs => servo/components/net/resource_thread.rs rename : servo/components/net/storage_task.rs => servo/components/net/storage_thread.rs rename : servo/components/net_traits/image_cache_task.rs => servo/components/net_traits/image_cache_thread.rs rename : servo/components/net_traits/storage_task.rs => servo/components/net_traits/storage_thread.rs rename : servo/components/script/script_task.rs => servo/components/script/script_thread.rs rename : servo/components/util/task.rs => servo/components/util/thread.rs rename : servo/tests/unit/gfx/font_cache_task.rs => servo/tests/unit/gfx/font_cache_thread.rs rename : servo/tests/unit/net/resource_task.rs => servo/tests/unit/net/resource_thread.rs rename : servo/tests/unit/util/task.rs => servo/tests/unit/util/thread.rs
2016-01-10 13:19:04 +03:00
pub mod image_cache_thread;
pub mod net_error_list;
pub mod pub_domains;
pub mod request;
pub mod response;
servo: Merge #9201 - task -> thread (from ajnirp:8512-task-thread); r=jdm for #8512 Source-Repo: https://github.com/servo/servo Source-Revision: d3e2f94f2024f4735f836588ed11303a0abafdf8 --HG-- rename : servo/components/canvas/canvas_paint_task.rs => servo/components/canvas/canvas_paint_thread.rs rename : servo/components/canvas/webgl_paint_task.rs => servo/components/canvas/webgl_paint_thread.rs rename : servo/components/compositing/compositor_task.rs => servo/components/compositing/compositor_thread.rs rename : servo/components/gfx/font_cache_task.rs => servo/components/gfx/font_cache_thread.rs rename : servo/components/gfx/paint_task.rs => servo/components/gfx/paint_thread.rs rename : servo/components/layout/layout_task.rs => servo/components/layout/layout_thread.rs rename : servo/components/net/image_cache_task.rs => servo/components/net/image_cache_thread.rs rename : servo/components/net/resource_task.rs => servo/components/net/resource_thread.rs rename : servo/components/net/storage_task.rs => servo/components/net/storage_thread.rs rename : servo/components/net_traits/image_cache_task.rs => servo/components/net_traits/image_cache_thread.rs rename : servo/components/net_traits/storage_task.rs => servo/components/net_traits/storage_thread.rs rename : servo/components/script/script_task.rs => servo/components/script/script_thread.rs rename : servo/components/util/task.rs => servo/components/util/thread.rs rename : servo/tests/unit/gfx/font_cache_task.rs => servo/tests/unit/gfx/font_cache_thread.rs rename : servo/tests/unit/net/resource_task.rs => servo/tests/unit/net/resource_thread.rs rename : servo/tests/unit/util/task.rs => servo/tests/unit/util/thread.rs
2016-01-10 13:19:04 +03:00
pub mod storage_thread;
/// Image handling.
///
/// It may be surprising that this goes in the network crate as opposed to the graphics crate.
/// However, image handling is generally very integrated with the network stack (especially where
/// caching is involved) and as a result it must live in here.
pub mod image {
pub mod base;
}
/// A loading context, for context-specific sniffing, as defined in
/// https://mimesniff.spec.whatwg.org/#context-specific-sniffing
#[derive(Clone, Deserialize, Serialize, HeapSizeOf)]
pub enum LoadContext {
Browsing,
Image,
AudioVideo,
Plugin,
Style,
Script,
Font,
TextTrack,
CacheManifest,
}
#[derive(Clone, Debug, Deserialize, Serialize, HeapSizeOf)]
pub struct CustomResponse {
#[ignore_heap_size_of = "Defined in hyper"]
#[serde(deserialize_with = "::hyper_serde::deserialize",
serialize_with = "::hyper_serde::serialize")]
pub headers: Headers,
#[ignore_heap_size_of = "Defined in hyper"]
#[serde(deserialize_with = "::hyper_serde::deserialize",
serialize_with = "::hyper_serde::serialize")]
pub raw_status: RawStatus,
pub body: Vec<u8>,
}
impl CustomResponse {
pub fn new(headers: Headers, raw_status: RawStatus, body: Vec<u8>) -> CustomResponse {
CustomResponse {
headers: headers,
raw_status: raw_status,
body: body,
}
}
}
servo: Merge #11727 - Integrate service worker manager thread (from creativcoder:swmanager); r=jdm <!-- Please describe your changes on the following line: --> --- <!-- Thank you for contributing to Servo! Please replace each `[ ]` by `[X]` when the step is complete, and replace `__` with appropriate data: --> - [X] `./mach build -d` does not report any errors - [X] `./mach test-tidy` does not report any errors - [X] These changes are part of #11091 <!-- Either: --> - [X] There are tests for these changes at my [gh-pages](https://github.com/creativcoder/gsoc16/tree/gh-pages) branch to test the instantiation of service workers by their manager, but will need to discuss how that would integrate into master. Changes: - Introduces a `ServiceWorkerManager`, which maintains an map of registered service workers as well as a map of active workers keyed by their `scope_url`. - Adds the initialization of ServiceWorkerManager, at the `script::init()`, which makes it available as a single entity listening for requests from different script threads. - Adds a timeout thread in `serviceworkerglobalscope`, which terminates the workers, after a timeout of 60 secs, thereby removing it from the active workers list. - Adds the matching of scope urls, in longest prefix way rather than path structural way, according to [spec](https://slightlyoff.github.io/ServiceWorker/spec/service_worker/#scope-match-algorithm). - Make ServiceWorkerManager, the holder of network sender, instead of script thread, so it can send `CustomResponse`. <!-- Pull requests that do not address these steps are welcome, but they will require additional verification as part of the review process. --> Source-Repo: https://github.com/servo/servo Source-Revision: 513811f6b40d522bc425c2588320b889614f2973
2016-07-18 19:50:59 +03:00
#[derive(Clone, Deserialize, Serialize)]
pub struct CustomResponseMediator {
pub response_chan: IpcSender<Option<CustomResponse>>,
pub load_url: ServoUrl,
}
/// [Policies](https://w3c.github.io/webappsec-referrer-policy/#referrer-policy-states)
/// for providing a referrer header for a request
#[derive(Clone, Copy, Debug, Deserialize, HeapSizeOf, Serialize)]
pub enum ReferrerPolicy {
/// "no-referrer"
NoReferrer,
/// "no-referrer-when-downgrade"
NoReferrerWhenDowngrade,
/// "origin"
Origin,
/// "same-origin"
SameOrigin,
/// "origin-when-cross-origin"
OriginWhenCrossOrigin,
/// "unsafe-url"
UnsafeUrl,
servo: Merge #14059 - Network Security : Implement StrictOrigin and StrictOriginWhenCrossOr… (from mrnayak:refPolicy); r=nox This pull request contains commit implementing initial steps for Improving Network Security project. As part of initial steps referer policy enums for strict-origin and strict-origin-when-cross-origin have been added to [hyper](https://github.com/hyperium/hyper/pull/943). Unit tests and additional logic has been added to handle these policies. Since enum changes are available on hyper version 0.9.11. We had to update hyper version to 0.9.11. Hyper 0.9.11 depends on num_cpus 1.1.0. To avoid different version of num_cpus. We have updated rayon version from 0.4.0 to 0.4.3. Cargo.toml of util, style, geckolib, stylo component has been updated to use num_cpus version 1.1.0 instead of 0.2.2. --- <!-- Thank you for contributing to Servo! Please replace each `[ ]` by `[X]` when the step is complete, and replace `__` with appropriate data: --> - [ X] `./mach build -d` does not report any errors - [ X] `./mach test-tidy` does not report any errors - [ ] These changes fix #__ (github issue number if applicable). <!-- Either: --> - [ ] There are tests for these changes OR - [ ] These changes do not require tests because _____ <!-- Pull requests that do not address these steps are welcome, but they will require additional verification as part of the review process. --> …igin Referer policy strict-origin and strict-origin-when-cross-origin changes have been implemented. Relevant unit test cases have been added. Enum for RefererPolicy has been added to hyper codebase and v 0.9.11 of hyper contains these changes. This commit also contains changes related to upgrade of hyper from v0.9.10 to v0.9.11. Other dependencies changed are rayon, utils, num_cpus. Source-Repo: https://github.com/servo/servo Source-Revision: dd34b2a3355dc7fa23d118888359d70f8b445db8
2016-11-07 13:37:35 +03:00
/// "strict-origin"
StrictOrigin,
/// "strict-origin-when-cross-origin"
StrictOriginWhenCrossOrigin,
}
#[derive(Deserialize, Serialize)]
pub enum FetchResponseMsg {
// todo: should have fields for transmitted/total bytes
ProcessRequestBody,
ProcessRequestEOF,
// todo: send more info about the response (or perhaps the entire Response)
ProcessResponse(Result<FetchMetadata, NetworkError>),
ProcessResponseChunk(Vec<u8>),
ProcessResponseEOF(Result<(), NetworkError>),
}
pub trait FetchTaskTarget {
/// https://fetch.spec.whatwg.org/#process-request-body
///
/// Fired when a chunk of the request body is transmitted
fn process_request_body(&mut self, request: &Request);
/// https://fetch.spec.whatwg.org/#process-request-end-of-file
///
/// Fired when the entire request finishes being transmitted
fn process_request_eof(&mut self, request: &Request);
/// https://fetch.spec.whatwg.org/#process-response
///
/// Fired when headers are received
fn process_response(&mut self, response: &Response);
/// Fired when a chunk of response content is received
fn process_response_chunk(&mut self, chunk: Vec<u8>);
/// https://fetch.spec.whatwg.org/#process-response-end-of-file
///
/// Fired when the response is fully fetched
fn process_response_eof(&mut self, response: &Response);
}
#[derive(Serialize, Deserialize)]
pub enum FilteredMetadata {
Opaque,
Transparent(Metadata),
}
#[derive(Serialize, Deserialize)]
pub enum FetchMetadata {
Unfiltered(Metadata),
Filtered {
filtered: FilteredMetadata,
unsafe_: Metadata,
},
}
pub trait FetchResponseListener {
fn process_request_body(&mut self);
fn process_request_eof(&mut self);
fn process_response(&mut self, metadata: Result<FetchMetadata, NetworkError>);
fn process_response_chunk(&mut self, chunk: Vec<u8>);
fn process_response_eof(&mut self, response: Result<(), NetworkError>);
}
impl FetchTaskTarget for IpcSender<FetchResponseMsg> {
fn process_request_body(&mut self, _: &Request) {
let _ = self.send(FetchResponseMsg::ProcessRequestBody);
}
fn process_request_eof(&mut self, _: &Request) {
let _ = self.send(FetchResponseMsg::ProcessRequestEOF);
}
fn process_response(&mut self, response: &Response) {
let _ = self.send(FetchResponseMsg::ProcessResponse(response.metadata()));
}
fn process_response_chunk(&mut self, chunk: Vec<u8>) {
let _ = self.send(FetchResponseMsg::ProcessResponseChunk(chunk));
}
fn process_response_eof(&mut self, response: &Response) {
if response.is_network_error() {
// todo: finer grained errors
let _ =
self.send(FetchResponseMsg::ProcessResponseEOF(Err(NetworkError::Internal("Network error".into()))));
} else {
let _ = self.send(FetchResponseMsg::ProcessResponseEOF(Ok(())));
}
}
}
pub trait Action<Listener> {
fn process(self, listener: &mut Listener);
}
impl<T: FetchResponseListener> Action<T> for FetchResponseMsg {
/// Execute the default action on a provided listener.
fn process(self, listener: &mut T) {
match self {
FetchResponseMsg::ProcessRequestBody => listener.process_request_body(),
FetchResponseMsg::ProcessRequestEOF => listener.process_request_eof(),
FetchResponseMsg::ProcessResponse(meta) => listener.process_response(meta),
FetchResponseMsg::ProcessResponseChunk(data) => listener.process_response_chunk(data),
FetchResponseMsg::ProcessResponseEOF(data) => listener.process_response_eof(data),
}
}
}
servo: Merge #9201 - task -> thread (from ajnirp:8512-task-thread); r=jdm for #8512 Source-Repo: https://github.com/servo/servo Source-Revision: d3e2f94f2024f4735f836588ed11303a0abafdf8 --HG-- rename : servo/components/canvas/canvas_paint_task.rs => servo/components/canvas/canvas_paint_thread.rs rename : servo/components/canvas/webgl_paint_task.rs => servo/components/canvas/webgl_paint_thread.rs rename : servo/components/compositing/compositor_task.rs => servo/components/compositing/compositor_thread.rs rename : servo/components/gfx/font_cache_task.rs => servo/components/gfx/font_cache_thread.rs rename : servo/components/gfx/paint_task.rs => servo/components/gfx/paint_thread.rs rename : servo/components/layout/layout_task.rs => servo/components/layout/layout_thread.rs rename : servo/components/net/image_cache_task.rs => servo/components/net/image_cache_thread.rs rename : servo/components/net/resource_task.rs => servo/components/net/resource_thread.rs rename : servo/components/net/storage_task.rs => servo/components/net/storage_thread.rs rename : servo/components/net_traits/image_cache_task.rs => servo/components/net_traits/image_cache_thread.rs rename : servo/components/net_traits/storage_task.rs => servo/components/net_traits/storage_thread.rs rename : servo/components/script/script_task.rs => servo/components/script/script_thread.rs rename : servo/components/util/task.rs => servo/components/util/thread.rs rename : servo/tests/unit/gfx/font_cache_task.rs => servo/tests/unit/gfx/font_cache_thread.rs rename : servo/tests/unit/net/resource_task.rs => servo/tests/unit/net/resource_thread.rs rename : servo/tests/unit/util/task.rs => servo/tests/unit/util/thread.rs
2016-01-10 13:19:04 +03:00
/// Handle to a resource thread
pub type CoreResourceThread = IpcSender<CoreResourceMsg>;
pub type IpcSendResult = Result<(), IOError>;
/// Abstraction of the ability to send a particular type of message,
/// used by net_traits::ResourceThreads to ease the use its IpcSender sub-fields
/// XXX: If this trait will be used more in future, some auto derive might be appealing
pub trait IpcSend<T>
where T: serde::Serialize + serde::Deserialize,
{
/// send message T
fn send(&self, T) -> IpcSendResult;
/// get underlying sender
fn sender(&self) -> IpcSender<T>;
}
// FIXME: Originally we will construct an Arc<ResourceThread> from ResourceThread
// in script_thread to avoid some performance pitfall. Now we decide to deal with
// the "Arc" hack implicitly in future.
// See discussion: http://logs.glob.uno/?c=mozilla%23servo&s=16+May+2016&e=16+May+2016#c430412
// See also: https://github.com/servo/servo/blob/735480/components/script/script_thread.rs#L313
#[derive(Clone, Serialize, Deserialize)]
pub struct ResourceThreads {
core_thread: CoreResourceThread,
storage_thread: IpcSender<StorageThreadMsg>,
}
impl ResourceThreads {
pub fn new(c: CoreResourceThread, s: IpcSender<StorageThreadMsg>) -> ResourceThreads {
ResourceThreads {
core_thread: c,
storage_thread: s,
}
}
}
impl IpcSend<CoreResourceMsg> for ResourceThreads {
fn send(&self, msg: CoreResourceMsg) -> IpcSendResult {
self.core_thread.send(msg)
}
fn sender(&self) -> IpcSender<CoreResourceMsg> {
self.core_thread.clone()
}
}
impl IpcSend<StorageThreadMsg> for ResourceThreads {
fn send(&self, msg: StorageThreadMsg) -> IpcSendResult {
self.storage_thread.send(msg)
}
fn sender(&self) -> IpcSender<StorageThreadMsg> {
self.storage_thread.clone()
}
}
// Ignore the sub-fields
impl HeapSizeOf for ResourceThreads {
fn heap_size_of_children(&self) -> usize {
0
}
}
servo: Merge #6586 - script: Make the resource task communication use IPC channels (from pcwalton:resource-task-ipc); r=jdm This change makes Servo use serialized messages over IPC channels for resource loading. The goal is to make it easier to make Servo multiprocess in the future. This patch does not make Servo multiprocess now; there are many other channels that need to be changed to IPC before that can happen. It does introduce a dependency on https://github.com/serde-rs/serde and https://github.com/pcwalton/ipc-channel for the first time. At the moment, `ipc-channel` uses JSON for serialization. This is because serde does not yet have official support for bincode. When serde gains support for bincode, I'll switch to that. For now, however, the JSON encoding and decoding will constitute a significant performance regression in resource loading. To avoid having to send boxed `AsyncResponseTarget` trait objects across process boundaries, this series of commits changes `AsyncResponseTarget` to wrap a sender only. It is then the client's responsibility to spawn a thread to proxy calls from that sender to the consumer of the resource data. This only had to be done in a few places. In the future, we may want to collapse those threads into one per process to reduce overhead. (It is impossible to continue to use `AsyncResponseTarget` as a boxed trait object across processes, regardless of how much work is done on `ipc-channel`. Vtables are fundamentally incompatible with IPC across mutually untrusting processes.) In general, I was pretty pleased with how this turned out. The main changes are adding serialization functionality to various objects that `serde` does not know how to serialize natively—the most complicated being Hyper objects—and reworking `AsyncResponseTarget`. The overall structure of the code is unchanged, and other than `AsyncResponseTarget` no functionality was lost in moving to serialization and IPC. r? @jdm Source-Repo: https://github.com/servo/servo Source-Revision: 2eb122f394651232abd683fc576a5c4288bf277f
2015-08-01 01:06:36 +03:00
#[derive(PartialEq, Copy, Clone, Deserialize, Serialize)]
pub enum IncludeSubdomains {
Included,
NotIncluded,
}
#[derive(HeapSizeOf, Deserialize, Serialize)]
pub enum MessageData {
Text(String),
Binary(Vec<u8>),
}
#[derive(Deserialize, Serialize)]
pub enum WebSocketDomAction {
SendMessage(MessageData),
Close(Option<u16>, Option<String>),
}
#[derive(Deserialize, Serialize)]
pub enum WebSocketNetworkEvent {
ConnectionEstablished(#[serde(deserialize_with = "::hyper_serde::deserialize",
serialize_with = "::hyper_serde::serialize")]
header::Headers,
Vec<String>),
MessageReceived(MessageData),
Close(Option<u16>, String),
Fail,
}
#[derive(Deserialize, Serialize)]
pub struct WebSocketCommunicate {
pub event_sender: IpcSender<WebSocketNetworkEvent>,
pub action_receiver: IpcReceiver<WebSocketDomAction>,
}
#[derive(Deserialize, Serialize)]
pub struct WebSocketConnectData {
pub resource_url: ServoUrl,
pub origin: String,
pub protocols: Vec<String>,
}
servo: Merge #6586 - script: Make the resource task communication use IPC channels (from pcwalton:resource-task-ipc); r=jdm This change makes Servo use serialized messages over IPC channels for resource loading. The goal is to make it easier to make Servo multiprocess in the future. This patch does not make Servo multiprocess now; there are many other channels that need to be changed to IPC before that can happen. It does introduce a dependency on https://github.com/serde-rs/serde and https://github.com/pcwalton/ipc-channel for the first time. At the moment, `ipc-channel` uses JSON for serialization. This is because serde does not yet have official support for bincode. When serde gains support for bincode, I'll switch to that. For now, however, the JSON encoding and decoding will constitute a significant performance regression in resource loading. To avoid having to send boxed `AsyncResponseTarget` trait objects across process boundaries, this series of commits changes `AsyncResponseTarget` to wrap a sender only. It is then the client's responsibility to spawn a thread to proxy calls from that sender to the consumer of the resource data. This only had to be done in a few places. In the future, we may want to collapse those threads into one per process to reduce overhead. (It is impossible to continue to use `AsyncResponseTarget` as a boxed trait object across processes, regardless of how much work is done on `ipc-channel`. Vtables are fundamentally incompatible with IPC across mutually untrusting processes.) In general, I was pretty pleased with how this turned out. The main changes are adding serialization functionality to various objects that `serde` does not know how to serialize natively—the most complicated being Hyper objects—and reworking `AsyncResponseTarget`. The overall structure of the code is unchanged, and other than `AsyncResponseTarget` no functionality was lost in moving to serialization and IPC. r? @jdm Source-Repo: https://github.com/servo/servo Source-Revision: 2eb122f394651232abd683fc576a5c4288bf277f
2015-08-01 01:06:36 +03:00
#[derive(Deserialize, Serialize)]
pub enum CoreResourceMsg {
Fetch(RequestInit, IpcSender<FetchResponseMsg>),
/// Try to make a websocket connection to a URL.
WebsocketConnect(WebSocketCommunicate, WebSocketConnectData),
/// Store a cookie for a given originating URL
SetCookieForUrl(ServoUrl,
#[serde(deserialize_with = "::hyper_serde::deserialize",
serialize_with = "::hyper_serde::serialize")]
Cookie,
CookieSource),
/// Store cookies for a given originating URL
SetCookiesForUrl(ServoUrl, Vec<Serde<Cookie>>, CookieSource),
/// Retrieve the stored cookies for a given URL
GetCookiesForUrl(ServoUrl, IpcSender<Option<String>>, CookieSource),
/// Get a cookie by name for a given originating URL
GetCookiesDataForUrl(ServoUrl, IpcSender<Vec<Serde<Cookie>>>, CookieSource),
/// Cancel a network request corresponding to a given `ResourceId`
Cancel(ResourceId),
/// Synchronization message solely for knowing the state of the ResourceChannelManager loop
Synchronize(IpcSender<()>),
servo: Merge #11727 - Integrate service worker manager thread (from creativcoder:swmanager); r=jdm <!-- Please describe your changes on the following line: --> --- <!-- Thank you for contributing to Servo! Please replace each `[ ]` by `[X]` when the step is complete, and replace `__` with appropriate data: --> - [X] `./mach build -d` does not report any errors - [X] `./mach test-tidy` does not report any errors - [X] These changes are part of #11091 <!-- Either: --> - [X] There are tests for these changes at my [gh-pages](https://github.com/creativcoder/gsoc16/tree/gh-pages) branch to test the instantiation of service workers by their manager, but will need to discuss how that would integrate into master. Changes: - Introduces a `ServiceWorkerManager`, which maintains an map of registered service workers as well as a map of active workers keyed by their `scope_url`. - Adds the initialization of ServiceWorkerManager, at the `script::init()`, which makes it available as a single entity listening for requests from different script threads. - Adds a timeout thread in `serviceworkerglobalscope`, which terminates the workers, after a timeout of 60 secs, thereby removing it from the active workers list. - Adds the matching of scope urls, in longest prefix way rather than path structural way, according to [spec](https://slightlyoff.github.io/ServiceWorker/spec/service_worker/#scope-match-algorithm). - Make ServiceWorkerManager, the holder of network sender, instead of script thread, so it can send `CustomResponse`. <!-- Pull requests that do not address these steps are welcome, but they will require additional verification as part of the review process. --> Source-Repo: https://github.com/servo/servo Source-Revision: 513811f6b40d522bc425c2588320b889614f2973
2016-07-18 19:50:59 +03:00
/// Send the network sender in constellation to CoreResourceThread
NetworkMediator(IpcSender<CustomResponseMediator>),
/// Message forwarded to file manager's handler
ToFileManager(FileManagerThreadMsg),
/// Break the load handler loop, send a reply when done cleaning up local resources
/// and exit
Exit(IpcSender<()>),
}
/// Instruct the resource thread to make a new request.
pub fn fetch_async<F>(request: RequestInit, core_resource_thread: &CoreResourceThread, f: F)
where F: Fn(FetchResponseMsg) + Send + 'static,
{
let (action_sender, action_receiver) = ipc::channel().unwrap();
ROUTER.add_route(action_receiver.to_opaque(),
box move |message| f(message.to().unwrap()));
core_resource_thread.send(CoreResourceMsg::Fetch(request, action_sender)).unwrap();
}
#[derive(Clone, Deserialize, Serialize, HeapSizeOf)]
pub struct ResourceCorsData {
/// CORS Preflight flag
pub preflight: bool,
/// Origin of CORS Request
pub origin: ServoUrl,
}
/// Metadata about a loaded resource, such as is obtained from HTTP headers.
#[derive(Clone, Deserialize, Serialize, HeapSizeOf)]
pub struct Metadata {
/// Final URL after redirects.
pub final_url: ServoUrl,
#[ignore_heap_size_of = "Defined in hyper"]
/// MIME type / subtype.
pub content_type: Option<Serde<ContentType>>,
/// Character set.
pub charset: Option<String>,
#[ignore_heap_size_of = "Defined in hyper"]
/// Headers
pub headers: Option<Serde<Headers>>,
/// HTTP Status
servo: Merge #13058 - Response API (from malisas:malisa-responseAPI); r=Manishearth,jdm <!-- Please describe your changes on the following line: --> This PR adds the [dom::Response](https://fetch.spec.whatwg.org/#response-class) implementation and addresses #11896. The relevant passing tests` expectations have been updated. In order to allow non-UTF-8-encoded status messages, `net_traits::response::Response`'s `raw_status` field has been changed from type [`Option<RawStatus>`](https://doc.servo.org/hyper/http/struct.RawStatus.html) to type `Option<(u16, Vec<u8>)>`. As a result, a few other files which rely on the `raw_status` field were affected and updated. TODOs: - The `body` and `trailer` methods. Relies on implementation of `ReadableStream` and `Promise`s. - Similarly, replace the dummy constructor `_body: Option<USVString>` argument with `body: ResponseBodyInit`. - Currently, whenever `r's response's header list` or `r's Headers object` are mentioned, I always modify the `headers_reflector` field (of type dom::Headers, or `r's Headers object`) and not the corresponding hyper::Headers list in the net_traits::Response field. A completely accurate interpretation of the spec might consider making both of these lists the same thing via a reference. [Discussion](https://github.com/whatwg/fetch/issues/358) was [had](https://github.com/servo/servo/pull/12884). --- <!-- Thank you for contributing to Servo! Please replace each `[ ]` by `[X]` when the step is complete, and replace `__` with appropriate data: --> - [X] `./mach build -d` does not report any errors - [X] `./mach test-tidy` does not report any errors - [ ] These changes fix #__ (github issue number if applicable). <!-- Either: --> - [X] There are tests for these changes OR - [ ] These changes do not require tests because _____ <!-- Pull requests that do not address these steps are welcome, but they will require additional verification as part of the review process. --> Source-Repo: https://github.com/servo/servo Source-Revision: 5a5a76cc5db830d2e622d4e0924837383b64dfa2
2016-09-09 02:58:05 +03:00
pub status: Option<(u16, Vec<u8>)>,
/// Is successful HTTPS connection
pub https_state: HttpsState,
/// Referrer Url
pub referrer: Option<ServoUrl>,
}
impl Metadata {
/// Metadata with defaults for everything optional.
pub fn default(url: ServoUrl) -> Self {
Metadata {
final_url: url,
content_type: None,
charset: None,
headers: None,
// https://fetch.spec.whatwg.org/#concept-response-status-message
servo: Merge #13058 - Response API (from malisas:malisa-responseAPI); r=Manishearth,jdm <!-- Please describe your changes on the following line: --> This PR adds the [dom::Response](https://fetch.spec.whatwg.org/#response-class) implementation and addresses #11896. The relevant passing tests` expectations have been updated. In order to allow non-UTF-8-encoded status messages, `net_traits::response::Response`'s `raw_status` field has been changed from type [`Option<RawStatus>`](https://doc.servo.org/hyper/http/struct.RawStatus.html) to type `Option<(u16, Vec<u8>)>`. As a result, a few other files which rely on the `raw_status` field were affected and updated. TODOs: - The `body` and `trailer` methods. Relies on implementation of `ReadableStream` and `Promise`s. - Similarly, replace the dummy constructor `_body: Option<USVString>` argument with `body: ResponseBodyInit`. - Currently, whenever `r's response's header list` or `r's Headers object` are mentioned, I always modify the `headers_reflector` field (of type dom::Headers, or `r's Headers object`) and not the corresponding hyper::Headers list in the net_traits::Response field. A completely accurate interpretation of the spec might consider making both of these lists the same thing via a reference. [Discussion](https://github.com/whatwg/fetch/issues/358) was [had](https://github.com/servo/servo/pull/12884). --- <!-- Thank you for contributing to Servo! Please replace each `[ ]` by `[X]` when the step is complete, and replace `__` with appropriate data: --> - [X] `./mach build -d` does not report any errors - [X] `./mach test-tidy` does not report any errors - [ ] These changes fix #__ (github issue number if applicable). <!-- Either: --> - [X] There are tests for these changes OR - [ ] These changes do not require tests because _____ <!-- Pull requests that do not address these steps are welcome, but they will require additional verification as part of the review process. --> Source-Repo: https://github.com/servo/servo Source-Revision: 5a5a76cc5db830d2e622d4e0924837383b64dfa2
2016-09-09 02:58:05 +03:00
status: Some((200, b"OK".to_vec())),
https_state: HttpsState::None,
referrer: None,
}
}
/// Extract the parts of a Mime that we care about.
pub fn set_content_type(&mut self, content_type: Option<&Mime>) {
if self.headers.is_none() {
self.headers = Some(Serde(Headers::new()));
}
if let Some(mime) = content_type {
self.headers.as_mut().unwrap().set(ContentType(mime.clone()));
self.content_type = Some(Serde(ContentType(mime.clone())));
let Mime(_, _, ref parameters) = *mime;
for &(ref k, ref v) in parameters {
if Attr::Charset == *k {
self.charset = Some(v.to_string());
}
}
}
}
}
/// The creator of a given cookie
servo: Merge #6586 - script: Make the resource task communication use IPC channels (from pcwalton:resource-task-ipc); r=jdm This change makes Servo use serialized messages over IPC channels for resource loading. The goal is to make it easier to make Servo multiprocess in the future. This patch does not make Servo multiprocess now; there are many other channels that need to be changed to IPC before that can happen. It does introduce a dependency on https://github.com/serde-rs/serde and https://github.com/pcwalton/ipc-channel for the first time. At the moment, `ipc-channel` uses JSON for serialization. This is because serde does not yet have official support for bincode. When serde gains support for bincode, I'll switch to that. For now, however, the JSON encoding and decoding will constitute a significant performance regression in resource loading. To avoid having to send boxed `AsyncResponseTarget` trait objects across process boundaries, this series of commits changes `AsyncResponseTarget` to wrap a sender only. It is then the client's responsibility to spawn a thread to proxy calls from that sender to the consumer of the resource data. This only had to be done in a few places. In the future, we may want to collapse those threads into one per process to reduce overhead. (It is impossible to continue to use `AsyncResponseTarget` as a boxed trait object across processes, regardless of how much work is done on `ipc-channel`. Vtables are fundamentally incompatible with IPC across mutually untrusting processes.) In general, I was pretty pleased with how this turned out. The main changes are adding serialization functionality to various objects that `serde` does not know how to serialize natively—the most complicated being Hyper objects—and reworking `AsyncResponseTarget`. The overall structure of the code is unchanged, and other than `AsyncResponseTarget` no functionality was lost in moving to serialization and IPC. r? @jdm Source-Repo: https://github.com/servo/servo Source-Revision: 2eb122f394651232abd683fc576a5c4288bf277f
2015-08-01 01:06:36 +03:00
#[derive(PartialEq, Copy, Clone, Deserialize, Serialize)]
pub enum CookieSource {
/// An HTTP API
HTTP,
/// A non-HTTP API
NonHTTP,
}
/// Convenience function for synchronously loading a whole resource.
pub fn load_whole_resource(request: RequestInit,
core_resource_thread: &CoreResourceThread)
-> Result<(Metadata, Vec<u8>), NetworkError> {
let (action_sender, action_receiver) = ipc::channel().unwrap();
core_resource_thread.send(CoreResourceMsg::Fetch(request, action_sender)).unwrap();
let mut buf = vec![];
let mut metadata = None;
loop {
match action_receiver.recv().unwrap() {
FetchResponseMsg::ProcessRequestBody |
FetchResponseMsg::ProcessRequestEOF => (),
FetchResponseMsg::ProcessResponse(Ok(m)) => {
metadata = Some(match m {
FetchMetadata::Unfiltered(m) => m,
FetchMetadata::Filtered { unsafe_, .. } => unsafe_,
})
},
FetchResponseMsg::ProcessResponseChunk(data) => buf.extend_from_slice(&data),
FetchResponseMsg::ProcessResponseEOF(Ok(())) => return Ok((metadata.unwrap(), buf)),
FetchResponseMsg::ProcessResponse(Err(e)) |
FetchResponseMsg::ProcessResponseEOF(Err(e)) => return Err(e),
}
}
}
/// Defensively unwraps the protocol string from the response object's protocol
pub fn unwrap_websocket_protocol(wsp: Option<&header::WebSocketProtocol>) -> Option<&str> {
wsp.and_then(|protocol_list| protocol_list.get(0).map(|protocol| protocol.as_ref()))
}
/// An unique identifier to keep track of each load message in the resource handler
#[derive(Clone, PartialEq, Eq, Copy, Hash, Debug, Deserialize, Serialize, HeapSizeOf)]
pub struct ResourceId(pub u32);
/// Network errors that have to be exported out of the loaders
#[derive(Clone, PartialEq, Eq, Debug, Deserialize, Serialize, HeapSizeOf)]
pub enum NetworkError {
/// Could be any of the internal errors, like unsupported scheme, connection errors, etc.
Internal(String),
LoadCancelled,
/// SSL validation error that has to be handled in the HTML parser
SslValidation(ServoUrl, String),
}
/// Normalize `slice`, as defined by
/// [the Fetch Spec](https://fetch.spec.whatwg.org/#concept-header-value-normalize).
pub fn trim_http_whitespace(mut slice: &[u8]) -> &[u8] {
const HTTP_WS_BYTES: &'static [u8] = b"\x09\x0A\x0D\x20";
loop {
match slice.split_first() {
Some((first, remainder)) if HTTP_WS_BYTES.contains(first) => slice = remainder,
_ => break,
}
}
loop {
match slice.split_last() {
Some((last, remainder)) if HTTP_WS_BYTES.contains(last) => slice = remainder,
_ => break,
}
}
slice
}