Move the networking code in parse_html into ScriptTask::load.

This doesn't really have anything to do with parsing HTML, and fits in better
with the code in ScriptTask::load. In particular, all changes to Page's url
now go through ScriptTask methods.

components/script/parse/html.rs

@@ -16,72 +16,26 @@ use dom::node::{Node, NodeHelpers, TrustedNodeAddress};
 use dom::servohtmlparser;
 use dom::servohtmlparser::ServoHTMLParser;
 use dom::text::Text;
-use page::Page;
 use parse::Parser;
 
 use encoding::all::UTF_8;
 use encoding::types::{Encoding, DecodeReplace};
 
-use servo_net::resource_task::{Load, LoadData, Payload, Done, ResourceTask};
+use servo_net::resource_task::{Payload, Done, LoadResponse};
-use servo_msg::constellation_msg::LoadData as MsgLoadData;
 use servo_util::task_state;
 use servo_util::task_state::IN_HTML_PARSER;
 use std::ascii::AsciiExt;
-use std::comm::channel;
-use std::fmt::{mod, Show};
 use std::str::MaybeOwned;
 use url::Url;
-use time::{Tm, strptime};
 use html5ever::Attribute;
 use html5ever::tree_builder::{TreeSink, QuirksMode, NodeOrText, AppendNode, AppendText};
 use string_cache::QualName;
-use hyper::header::{Header, HeaderFormat};
-use hyper::header::common::util as header_util;
 
 pub enum HTMLInput {
     InputString(String),
     InputUrl(Url),
 }
 
-//FIXME(seanmonstar): uplift to Hyper
-#[deriving(Clone)]
-struct LastModified(pub Tm);
-
-impl Header for LastModified {
-    #[inline]
-    fn header_name(_: Option<LastModified>) -> &'static str {
-        "Last-Modified"
-    }
-
-    // Parses an RFC 2616 compliant date/time string,
-    fn parse_header(raw: &[Vec<u8>]) -> Option<LastModified> {
-        header_util::from_one_raw_str(raw).and_then(|s: String| {
-            let s = s.as_slice();
-            strptime(s, "%a, %d %b %Y %T %Z").or_else(|_| {
-                strptime(s, "%A, %d-%b-%y %T %Z")
-            }).or_else(|_| {
-                strptime(s, "%c")
-            }).ok().map(|tm| LastModified(tm))
-        })
-    }
-}
-
-impl HeaderFormat for LastModified {
-    // a localized date/time string in a format suitable
-    // for document.lastModified.
-    fn fmt_header(&self, f: &mut fmt::Formatter) -> fmt::Result {
-        let LastModified(ref tm) = *self;
-        match tm.tm_gmtoff {
-            0 => tm.rfc822().fmt(f),
-            _ => tm.to_utc().rfc822().fmt(f)
-        }
-    }
-}
-
-fn dom_last_modified(tm: &Tm) -> String {
-    tm.to_local().strftime("%m/%d/%Y %H:%M:%S").unwrap()
-}
-
 trait SinkHelpers {
     fn get_or_create(&self, child: NodeOrText<TrustedNodeAddress>) -> Temporary<Node>;
 }
@@ -207,52 +161,10 @@ impl<'a> TreeSink<TrustedNodeAddress> for servohtmlparser::Sink {
     }
 }
 
-// The url from msg_load_data is ignored here
+pub fn parse_html(document: JSRef<Document>,
-pub fn parse_html(page: &Page,
-                  document: JSRef<Document>,
                   input: HTMLInput,
-                  resource_task: ResourceTask,
+                  base_url: Option<Url>,
-                  msg_load_data: MsgLoadData) {
+                  load_response: Option<LoadResponse>) {
-    let (base_url, load_response) = match input {
-        InputUrl(ref url) => {
-            // Wait for the LoadResponse so that the parser knows the final URL.
-            let (input_chan, input_port) = channel();
-            resource_task.send(Load(LoadData {
-                url: url.clone(),
-                method: msg_load_data.method,
-                headers: msg_load_data.headers,
-                data: msg_load_data.data,
-                cors: None,
-                consumer: input_chan,
-            }));
-
-            let load_response = input_port.recv();
-
-            load_response.metadata.headers.as_ref().map(|headers| {
-                headers.get().map(|&LastModified(ref tm)| {
-                    document.set_last_modified(dom_last_modified(tm));
-                });
-            });
-
-            let base_url = load_response.metadata.final_url.clone();
-
-            {
-                // Store the final URL before we start parsing, so that DOM routines
-                // (e.g. HTMLImageElement::update_image) can resolve relative URLs
-                // correctly.
-                *page.mut_url() = Some((base_url.clone(), true));
-            }
-
-            (Some(base_url), Some(load_response))
-        },
-        InputString(_) => {
-            match *page.url() {
-                Some((ref page_url, _)) => (Some(page_url.clone()), None),
-                None => (None, None),
-            }
-        },
-    };
-
     let parser = ServoHTMLParser::new(base_url.clone(), document).root();
     let parser: JSRef<ServoHTMLParser> = *parser;
 

components/script/script_task.rs

@@ -51,7 +51,8 @@ use servo_msg::constellation_msg::{KeyModifiers, SUPER, SHIFT, CONTROL, ALT, Rep
 use servo_msg::constellation_msg::{Released};
 use servo_msg::constellation_msg;
 use servo_net::image_cache_task::ImageCacheTask;
-use servo_net::resource_task::ResourceTask;
+use servo_net::resource_task::{ResourceTask, Load};
+use servo_net::resource_task::LoadData as NetLoadData;
 use servo_net::storage_task::StorageTask;
 use servo_util::geometry::to_frac_px;
 use servo_util::smallvec::{SmallVec1, SmallVec};
@@ -59,6 +60,8 @@ use servo_util::task::spawn_named_with_send_on_failure;
 use servo_util::task_state;
 
 use geom::point::Point2D;
+use hyper::header::{Header, HeaderFormat};
+use hyper::header::common::util as header_util;
 use js::jsapi::{JS_SetWrapObjectCallbacks, JS_SetGCZeal, JS_DEFAULT_ZEAL_FREQ, JS_GC};
 use js::jsapi::{JSContext, JSRuntime, JSTracer};
 use js::jsapi::{JS_SetGCParameter, JSGC_MAX_BYTES};
@@ -71,9 +74,11 @@ use libc::size_t;
 use std::any::{Any, AnyRefExt};
 use std::collections::HashSet;
 use std::comm::{channel, Sender, Receiver, Select};
+use std::fmt::{mod, Show};
 use std::mem::replace;
 use std::rc::Rc;
 use std::u32;
+use time::{Tm, strptime};
 
 local_data_key!(pub StackRoots: *const RootCollection)
 
@@ -748,7 +753,47 @@ impl ScriptTask {
             InputString(strval.unwrap_or("".to_string()))
         };
 
-        parse_html(&*page, *document, parser_input, self.resource_task.clone(), load_data);
+        let (base_url, load_response) = match parser_input {
+            InputUrl(ref url) => {
+                // Wait for the LoadResponse so that the parser knows the final URL.
+                let (input_chan, input_port) = channel();
+                self.resource_task.send(Load(NetLoadData {
+                    url: url.clone(),
+                    method: load_data.method,
+                    headers: load_data.headers,
+                    data: load_data.data,
+                    cors: None,
+                    consumer: input_chan,
+                }));
+
+                let load_response = input_port.recv();
+
+                load_response.metadata.headers.as_ref().map(|headers| {
+                    headers.get().map(|&LastModified(ref tm)| {
+                        document.set_last_modified(dom_last_modified(tm));
+                    });
+                });
+
+                let base_url = load_response.metadata.final_url.clone();
+
+                {
+                    // Store the final URL before we start parsing, so that DOM routines
+                    // (e.g. HTMLImageElement::update_image) can resolve relative URLs
+                    // correctly.
+                    *page.mut_url() = Some((base_url.clone(), true));
+                }
+
+                (Some(base_url), Some(load_response))
+            },
+            InputString(_) => {
+                match *page.url() {
+                    Some((ref page_url, _)) => (Some(page_url.clone()), None),
+                    None => (None, None),
+                }
+            },
+        };
+
+        parse_html(*document, parser_input, base_url, load_response);
         url = page.get_url().clone();
 
         document.set_ready_state(DocumentReadyStateValues::Interactive);
@@ -1159,3 +1204,42 @@ pub fn get_page(page: &Rc<Page>, pipeline_id: PipelineId) -> Rc<Page> {
         message for a layout channel that is not associated with this script task.\
          This is a bug.")
 }
+
+//FIXME(seanmonstar): uplift to Hyper
+#[deriving(Clone)]
+struct LastModified(pub Tm);
+
+impl Header for LastModified {
+    #[inline]
+    fn header_name(_: Option<LastModified>) -> &'static str {
+        "Last-Modified"
+    }
+
+    // Parses an RFC 2616 compliant date/time string,
+    fn parse_header(raw: &[Vec<u8>]) -> Option<LastModified> {
+        header_util::from_one_raw_str(raw).and_then(|s: String| {
+            let s = s.as_slice();
+            strptime(s, "%a, %d %b %Y %T %Z").or_else(|_| {
+                strptime(s, "%A, %d-%b-%y %T %Z")
+            }).or_else(|_| {
+                strptime(s, "%c")
+            }).ok().map(|tm| LastModified(tm))
+        })
+    }
+}
+
+impl HeaderFormat for LastModified {
+    // a localized date/time string in a format suitable
+    // for document.lastModified.
+    fn fmt_header(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        let LastModified(ref tm) = *self;
+        match tm.tm_gmtoff {
+            0 => tm.rfc822().fmt(f),
+            _ => tm.to_utc().rfc822().fmt(f)
+        }
+    }
+}
+
+fn dom_last_modified(tm: &Tm) -> String {
+    tm.to_local().strftime("%m/%d/%Y %H:%M:%S").unwrap()
+}