From 82c52a7a1c2d1fcdad797fc73564d6f624650473 Mon Sep 17 00:00:00 2001 From: Chris Double Date: Thu, 12 Mar 2015 23:41:09 +1300 Subject: [PATCH 1/2] Implement view-source protocol. Fixes #4181. This follows the recommendation from issue #4181. A handler for 'view-source' delegates to the HTTP loader. In that loader I check for view-source, adjust the URL to be the URL to be viewed and modify the Content-Type header to be text/plain. This doesn't actually result in the source being viewed as rendering text/plain is not yet implemented. --- components/net/http_loader.rs | 21 +++++++++++++++++++-- components/net/resource_task.rs | 2 +- components/script/parse/html.rs | 14 ++++++++++++++ 3 files changed, 34 insertions(+), 3 deletions(-) diff --git a/components/net/http_loader.rs b/components/net/http_loader.rs index c3e6a5dfb3c..f949349faca 100644 --- a/components/net/http_loader.rs +++ b/components/net/http_loader.rs @@ -14,6 +14,7 @@ use hyper::client::Request; use hyper::header::{ContentLength, ContentType, Host, Location}; use hyper::HttpError; use hyper::method::Method; +use hyper::mime::{Mime, TopLevel, SubLevel}; use hyper::net::HttpConnector; use hyper::status::{StatusCode, StatusClass}; use std::error::Error; @@ -54,6 +55,18 @@ fn load(mut load_data: LoadData, start_chan: Sender, cooki let mut url = load_data.url.clone(); let mut redirected_to = HashSet::new(); + // If the URL is a view-source scheme then the scheme data contains the + // real URL that should be used for which the source is to be viewed. + // Change our existing URL to that and keep note that we are viewing + // the source rather than rendering the contents of the URL. + let viewing_source = if url.scheme == "view-source" { + let inner_url = load_data.url.non_relative_scheme_data().unwrap(); + url = Url::parse(inner_url).unwrap(); + true + } else { + false + }; + let senders = ResponseSenders { immediate_consumer: start_chan, eventual_consumer: load_data.consumer @@ -259,12 +272,16 @@ reason: \"certificate verify failed\" }]"; } } + let mut adjusted_headers = response.headers.clone(); + if viewing_source { + adjusted_headers.set(ContentType(Mime(TopLevel::Text, SubLevel::Plain, vec![]))); + } let mut metadata = Metadata::default(url); - metadata.set_content_type(match response.headers.get() { + metadata.set_content_type(match adjusted_headers.get() { Some(&ContentType(ref mime)) => Some(mime), None => None }); - metadata.headers = Some(response.headers.clone()); + metadata.headers = Some(adjusted_headers); metadata.status = Some(response.status_raw().clone()); let progress_chan = match start_sending_opt(senders, metadata) { diff --git a/components/net/resource_task.rs b/components/net/resource_task.rs index d0a65975d35..8ceb0d82306 100644 --- a/components/net/resource_task.rs +++ b/components/net/resource_task.rs @@ -333,7 +333,7 @@ impl ResourceManager { let loader = match load_data.url.scheme.as_slice() { "file" => from_factory(file_loader::factory), - "http" | "https" => http_loader::factory(self.resource_task.clone()), + "http" | "https" | "view-source" => http_loader::factory(self.resource_task.clone()), "data" => from_factory(data_loader::factory), "about" => from_factory(about_loader::factory), _ => { diff --git a/components/script/parse/html.rs b/components/script/parse/html.rs index 6da3dac1ec8..c4417459f96 100644 --- a/components/script/parse/html.rs +++ b/components/script/parse/html.rs @@ -190,6 +190,20 @@ pub fn parse_html(document: JSRef, let page = format!("", url.serialize()); parser.parse_chunk(page); }, + Some((ref t, ref st)) if t.as_slice().eq_ignore_ascii_case("text") && + st.as_slice().eq_ignore_ascii_case("plain") => { + // FIXME: When servo/html5ever#109 is fixed remove usage and + // replace with fix from that issue. + + // text/plain documents require setting the tokenizer into PLAINTEXT mode. + // This is done by using a <plaintext> element as the html5ever tokenizer + // provides no other way to change to that state. + // Spec for text/plain handling is: + // https://html.spec.whatwg.org/multipage/browsers.html#read-text + let page = format!("<pre>\u{000A}<plaintext>"); + parser.parse_chunk(page); + parse_progress(&parser, url, &load_response); + }, _ => { for msg in load_response.progress_port.iter() { match msg { From 618142fac7a580071f758333f91afc683e70f6d7 Mon Sep 17 00:00:00 2001 From: Chris Double <chris.double@double.co.nz> Date: Fri, 13 Mar 2015 16:42:43 +1300 Subject: [PATCH 2/2] Implement displaying of text/plain documents This is done by detecting the content type as text/plain and following the requirements from: https://html.spec.whatwg.org/multipage/browsers.html#read-text --- components/net/http_loader.rs | 18 +++++++---- components/script/parse/html.rs | 30 +++++++++++-------- components/script/script_task.rs | 11 ++++++- .../read-text/load-text-plain.html.ini | 6 ---- 4 files changed, 40 insertions(+), 25 deletions(-) diff --git a/components/net/http_loader.rs b/components/net/http_loader.rs index f949349faca..9b81de49f8c 100644 --- a/components/net/http_loader.rs +++ b/components/net/http_loader.rs @@ -55,6 +55,11 @@ fn load(mut load_data: LoadData, start_chan: Sender<TargetedLoadResponse>, cooki let mut url = load_data.url.clone(); let mut redirected_to = HashSet::new(); + let senders = ResponseSenders { + immediate_consumer: start_chan, + eventual_consumer: load_data.consumer + }; + // If the URL is a view-source scheme then the scheme data contains the // real URL that should be used for which the source is to be viewed. // Change our existing URL to that and keep note that we are viewing @@ -62,16 +67,19 @@ fn load(mut load_data: LoadData, start_chan: Sender<TargetedLoadResponse>, cooki let viewing_source = if url.scheme == "view-source" { let inner_url = load_data.url.non_relative_scheme_data().unwrap(); url = Url::parse(inner_url).unwrap(); + match url.scheme.as_slice() { + "http" | "https" => {} + _ => { + let s = format!("The {} scheme with view-source is not supported", url.scheme); + send_error(url, s, senders); + return; + } + }; true } else { false }; - let senders = ResponseSenders { - immediate_consumer: start_chan, - eventual_consumer: load_data.consumer - }; - // Loop to handle redirects. loop { iters = iters + 1; diff --git a/components/script/parse/html.rs b/components/script/parse/html.rs index c4417459f96..836eac6dfba 100644 --- a/components/script/parse/html.rs +++ b/components/script/parse/html.rs @@ -180,6 +180,22 @@ pub fn parse_html(document: JSRef<Document>, task_state::enter(IN_HTML_PARSER); } + fn parse_progress(parser: &JSRef<ServoHTMLParser>, url: &Url, load_response: &LoadResponse) { + for msg in load_response.progress_port.iter() { + match msg { + ProgressMsg::Payload(data) => { + // FIXME: use Vec<u8> (html5ever #34) + let data = UTF_8.decode(data.as_slice(), DecoderTrap::Replace).unwrap(); + parser.parse_chunk(data); + } + ProgressMsg::Done(Err(err)) => { + panic!("Failed to load page URL {}, error: {}", url.serialize(), err); + } + ProgressMsg::Done(Ok(())) => break, + } + } + }; + match input { HTMLInput::InputString(s) => { parser.parse_chunk(s); @@ -205,19 +221,7 @@ pub fn parse_html(document: JSRef<Document>, parse_progress(&parser, url, &load_response); }, _ => { - for msg in load_response.progress_port.iter() { - match msg { - ProgressMsg::Payload(data) => { - // FIXME: use Vec<u8> (html5ever #34) - let data = UTF_8.decode(data.as_slice(), DecoderTrap::Replace).unwrap(); - parser.parse_chunk(data); - } - ProgressMsg::Done(Err(err)) => { - panic!("Failed to load page URL {}, error: {}", url.serialize(), err); - } - ProgressMsg::Done(Ok(())) => break, - } - } + parse_progress(&parser, url, &load_response); } } } diff --git a/components/script/script_task.rs b/components/script/script_task.rs index 3041098e78c..8c8390822c4 100644 --- a/components/script/script_task.rs +++ b/components/script/script_task.rs @@ -82,6 +82,7 @@ use js; use url::Url; use libc; +use std::ascii::AsciiExt; use std::any::Any; use std::borrow::ToOwned; use std::cell::{Cell, RefCell}; @@ -984,10 +985,18 @@ impl ScriptTask { headers.get().map(|&LastModified(ref tm)| dom_last_modified(tm)) }); + let content_type = match response.metadata.content_type { + Some((ref t, ref st)) if t.as_slice().eq_ignore_ascii_case("text") && + st.as_slice().eq_ignore_ascii_case("plain") => { + Some("text/plain".to_owned()) + } + _ => None + }; + let document = Document::new(window.r(), Some(final_url.clone()), IsHTMLDocument::HTMLDocument, - None, + content_type, last_modified, DocumentSource::FromParser).root(); diff --git a/tests/wpt/metadata/html/browsers/browsing-the-web/read-text/load-text-plain.html.ini b/tests/wpt/metadata/html/browsers/browsing-the-web/read-text/load-text-plain.html.ini index cc9779501b8..ee851470532 100644 --- a/tests/wpt/metadata/html/browsers/browsing-the-web/read-text/load-text-plain.html.ini +++ b/tests/wpt/metadata/html/browsers/browsing-the-web/read-text/load-text-plain.html.ini @@ -1,11 +1,5 @@ [load-text-plain.html] type: testharness - [Checking document metadata for text file] - expected: FAIL - - [Checking DOM for text file] - expected: FAIL - [Checking contents for text file] expected: FAIL