Auto merge of #14361 - nox:write, r=jdm

Implement document.write (fixes #3704)

<!-- Reviewable:start -->
This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/servo/14361)
<!-- Reviewable:end -->

components/script/dom/document.rs

@@ -142,12 +142,6 @@ pub enum IsHTMLDocument {
     NonHTMLDocument,
 }
 
-#[derive(PartialEq)]
-enum ParserBlockedByScript {
-    Blocked,
-    Unblocked,
-}
-
 #[derive(JSTraceable, HeapSizeOf)]
 #[must_root]
 pub struct StylesheetInDocument {
@@ -287,6 +281,8 @@ pub struct Document {
     /// https://w3c.github.io/uievents/#event-type-dblclick
     #[ignore_heap_size_of = "Defined in std"]
     last_click_info: DOMRefCell<Option<(Instant, Point2D<f32>)>>,
+    /// https://html.spec.whatwg.org/multipage/#ignore-destructive-writes-counter
+    ignore_destructive_writes_counter: Cell<u32>,
 }
 
 #[derive(JSTraceable, HeapSizeOf)]
@@ -378,15 +374,16 @@ impl Document {
         self.trigger_mozbrowser_event(MozBrowserEvent::SecurityChange(https_state));
     }
 
+    // https://html.spec.whatwg.org/multipage/#active-document
+    pub fn is_active(&self) -> bool {
+        self.browsing_context().map_or(false, |context| {
+            self == &*context.active_document()
+        })
+    }
+
     // https://html.spec.whatwg.org/multipage/#fully-active
     pub fn is_fully_active(&self) -> bool {
-        let browsing_context = match self.browsing_context() {
-            Some(browsing_context) => browsing_context,
-            None => return false,
-        };
-        let active_document = browsing_context.active_document();
-
-        if self != &*active_document {
+        if !self.is_active() {
             return false;
         }
         // FIXME: It should also check whether the browser context is top-level or not
@@ -1545,15 +1542,13 @@ impl Document {
             self.process_asap_scripts();
         }
 
-        if self.maybe_execute_parser_blocking_script() == ParserBlockedByScript::Blocked {
-            return;
-        }
-
-        // A finished resource load can potentially unblock parsing. In that case, resume the
-        // parser so its loop can find out.
         if let Some(parser) = self.get_current_parser() {
-            if parser.is_suspended() {
-                parser.resume();
+            if let Some(script) = self.pending_parsing_blocking_script.get() {
+                if self.script_blocking_stylesheets_count.get() > 0 || !script.is_ready_to_be_executed() {
+                    return;
+                }
+                self.pending_parsing_blocking_script.set(None);
+                parser.resume_with_pending_parsing_blocking_script(&script);
             }
         } else if self.reflow_timeout.get().is_none() {
             // If we don't have a parser, and the reflow timer has been reset, explicitly
@@ -1576,23 +1571,6 @@ impl Document {
         }
     }
 
-    /// If document parsing is blocked on a script, and that script is ready to run,
-    /// execute it.
-    /// https://html.spec.whatwg.org/multipage/#ready-to-be-parser-executed
-    fn maybe_execute_parser_blocking_script(&self) -> ParserBlockedByScript {
-        let script = match self.pending_parsing_blocking_script.get() {
-            None => return ParserBlockedByScript::Unblocked,
-            Some(script) => script,
-        };
-
-        if self.script_blocking_stylesheets_count.get() == 0 && script.is_ready_to_be_executed() {
-            self.pending_parsing_blocking_script.set(None);
-            script.execute();
-            return ParserBlockedByScript::Unblocked;
-        }
-        ParserBlockedByScript::Blocked
-    }
-
     /// https://html.spec.whatwg.org/multipage/#the-end step 3
     pub fn process_deferred_scripts(&self) {
         if self.ready_state.get() != DocumentReadyState::Interactive {
@@ -1901,6 +1879,7 @@ impl Document {
             referrer_policy: Cell::new(referrer_policy),
             target_element: MutNullableHeap::new(None),
             last_click_info: DOMRefCell::new(None),
+            ignore_destructive_writes_counter: Default::default(),
         }
     }
 
@@ -2077,6 +2056,16 @@ impl Document {
                            ReflowQueryType::NoQuery,
                            ReflowReason::ElementStateChanged);
     }
+
+    pub fn incr_ignore_destructive_writes_counter(&self) {
+        self.ignore_destructive_writes_counter.set(
+            self.ignore_destructive_writes_counter.get() + 1);
+    }
+
+    pub fn decr_ignore_destructive_writes_counter(&self) {
+        self.ignore_destructive_writes_counter.set(
+            self.ignore_destructive_writes_counter.get() - 1);
+    }
 }
 
 
@@ -3043,6 +3032,55 @@ impl DocumentMethods for Document {
         elements
     }
 
+    // https://html.spec.whatwg.org/multipage/#dom-document-write
+    fn Write(&self, text: Vec<DOMString>) -> ErrorResult {
+        if !self.is_html_document() {
+            // Step 1.
+            return Err(Error::InvalidState);
+        }
+
+        // Step 2.
+        // TODO: handle throw-on-dynamic-markup-insertion counter.
+
+        if !self.is_active() {
+            // Step 3.
+            return Ok(());
+        }
+
+        let parser = self.get_current_parser();
+        let parser = match parser.as_ref() {
+            Some(parser) if parser.script_nesting_level() > 0 => parser,
+            _ => {
+                // Either there is no parser, which means the parsing ended;
+                // or script nesting level is 0, which means the method was
+                // called from outside a parser-executed script.
+                if self.ignore_destructive_writes_counter.get() > 0 {
+                    // Step 4.
+                    // TODO: handle ignore-opens-during-unload counter.
+                    return Ok(());
+                }
+                // Step 5.
+                // TODO: call document.open().
+                return Err(Error::InvalidState);
+            }
+        };
+
+        // Step 7.
+        // TODO: handle reload override buffer.
+
+        // Steps 6-8.
+        parser.write(text);
+
+        // Step 9.
+        Ok(())
+    }
+
+    // https://html.spec.whatwg.org/multipage/#dom-document-writeln
+    fn Writeln(&self, mut text: Vec<DOMString>) -> ErrorResult {
+        text.push("\n".into());
+        self.Write(text)
+    }
+
     // https://html.spec.whatwg.org/multipage/#documentandelementeventhandlers
     document_and_element_event_handlers!();
 }

components/script/dom/htmlscriptelement.rs

@@ -274,12 +274,10 @@ fn fetch_a_classic_script(script: &HTMLScriptElement,
 
 impl HTMLScriptElement {
     /// https://html.spec.whatwg.org/multipage/#prepare-a-script
-    ///
-    /// Returns true if tokenization should continue, false otherwise.
-    pub fn prepare(&self) -> bool {
+    pub fn prepare(&self) {
         // Step 1.
         if self.already_started.get() {
-            return true;
+            return;
         }
 
         // Step 2.
@@ -297,17 +295,17 @@ impl HTMLScriptElement {
         // Step 4.
         let text = self.Text();
         if text.is_empty() && !element.has_attribute(&local_name!("src")) {
-            return true;
+            return;
         }
 
         // Step 5.
         if !self.upcast::<Node>().is_in_doc() {
-            return true;
+            return;
         }
 
         // Step 6.
         if !self.is_javascript() {
-            return true;
+            return;
         }
 
         // Step 7.
@@ -322,12 +320,12 @@ impl HTMLScriptElement {
         // Step 9.
         let doc = document_from_node(self);
         if self.parser_inserted.get() && &*self.parser_document != &*doc {
-            return true;
+            return;
         }
 
         // Step 10.
         if !doc.is_scripting_enabled() {
-            return true;
+            return;
         }
 
         // TODO(#4577): Step 11: CSP.
@@ -340,13 +338,13 @@ impl HTMLScriptElement {
                 let for_value = for_attribute.value().to_ascii_lowercase();
                 let for_value = for_value.trim_matches(HTML_SPACE_CHARACTERS);
                 if for_value != "window" {
-                    return true;
+                    return;
                 }
 
                 let event_value = event_attribute.value().to_ascii_lowercase();
                 let event_value = event_value.trim_matches(HTML_SPACE_CHARACTERS);
                 if event_value != "onload" && event_value != "onload()" {
-                    return true;
+                    return;
                 }
             },
             (_, _) => (),
@@ -381,7 +379,7 @@ impl HTMLScriptElement {
                 // Step 18.2.
                 if src.is_empty() {
                     self.queue_error_event();
-                    return true;
+                    return;
                 }
 
                 // Step 18.4-18.5.
@@ -389,7 +387,7 @@ impl HTMLScriptElement {
                     Err(_) => {
                         warn!("error parsing URL for script {}", &**src);
                         self.queue_error_event();
-                        return true;
+                        return;
                     }
                     Ok(url) => url,
                 };
@@ -412,7 +410,6 @@ impl HTMLScriptElement {
            !async {
             doc.add_deferred_script(self);
             // Second part implemented in Document::process_deferred_scripts.
-            return true;
         // Step 20.b: classic, has src, was parser-inserted, is not async.
         } else if is_external &&
                   was_parser_inserted &&
@@ -432,7 +429,7 @@ impl HTMLScriptElement {
         // Step 20.e: doesn't have src, was parser-inserted, is blocked on stylesheet.
         } else if !is_external &&
                   was_parser_inserted &&
-                  // TODO: check for script nesting levels.
+                  doc.get_current_parser().map_or(false, |parser| parser.script_nesting_level() <= 1) &&
                   doc.get_script_blocking_stylesheets_count() > 0 {
             doc.set_pending_parsing_blocking_script(Some(self));
             *self.load.borrow_mut() = Some(Ok(ScriptOrigin::internal(text, base_url)));
@@ -443,16 +440,7 @@ impl HTMLScriptElement {
             self.ready_to_be_parser_executed.set(true);
             *self.load.borrow_mut() = Some(Ok(ScriptOrigin::internal(text, base_url)));
             self.execute();
-            return true;
         }
-
-        // TODO: make this suspension happen automatically.
-        if was_parser_inserted {
-            if let Some(parser) = doc.get_current_parser() {
-                parser.suspend();
-            }
-        }
-        false
     }
 
     pub fn is_ready_to_be_executed(&self) -> bool {
@@ -481,19 +469,20 @@ impl HTMLScriptElement {
             Ok(script) => script,
         };
 
-        if script.external {
-            debug!("loading external script, url = {}", script.url);
-        }
-
         // TODO(#12446): beforescriptexecute.
         if self.dispatch_before_script_execute_event() == EventStatus::Canceled {
             return;
         }
 
         // Step 3.
-        // TODO: If the script is from an external file, then increment the
-        // ignore-destructive-writes counter of the script element's node
-        // document. Let neutralised doc be that Document.
+        let neutralized_doc = if script.external {
+            debug!("loading external script, url = {}", script.url);
+            let doc = document_from_node(self);
+            doc.incr_ignore_destructive_writes_counter();
+            Some(doc)
+        } else {
+            None
+        };
 
         // Step 4.
         let document = document_from_node(self);
@@ -512,8 +501,9 @@ impl HTMLScriptElement {
         document.set_current_script(old_script.r());
 
         // Step 7.
-        // TODO: Decrement the ignore-destructive-writes counter of neutralised
-        // doc, if it was incremented in the earlier step.
+        if let Some(doc) = neutralized_doc {
+            doc.decr_ignore_destructive_writes_counter();
+        }
 
         // TODO(#12446): afterscriptexecute.
         self.dispatch_after_script_execute_event();

components/script/dom/servoparser/mod.rs

@@ -33,12 +33,25 @@ use profile_traits::time::{TimerMetadataReflowType, ProfilerCategory, profile};
 use script_thread::ScriptThread;
 use servo_url::ServoUrl;
 use std::cell::Cell;
+use std::mem;
 use util::resource_files::read_resource_file;
 
 mod html;
 mod xml;
 
 #[dom_struct]
+/// The parser maintains two input streams: one for input from script through
+/// document.write(), and one for input from network.
+///
+/// There is no concrete representation of the insertion point, instead it
+/// always points to just before the next character from the network input,
+/// with all of the script input before itself.
+///
+/// ```text
+///     ... script input ... | ... network input ...
+///                          ^
+///                 insertion point
+/// ```
 pub struct ServoParser {
     reflector: Reflector,
     /// The document associated with this parser.
@@ -46,15 +59,20 @@ pub struct ServoParser {
     /// The pipeline associated with this parse, unavailable if this parse
     /// does not correspond to a page load.
     pipeline: Option<PipelineId>,
-    /// Input chunks received but not yet passed to the parser.
+    /// Input received from network.
     #[ignore_heap_size_of = "Defined in html5ever"]
-    pending_input: DOMRefCell<BufferQueue>,
+    network_input: DOMRefCell<BufferQueue>,
+    /// Input received from script. Used only to support document.write().
+    #[ignore_heap_size_of = "Defined in html5ever"]
+    script_input: DOMRefCell<BufferQueue>,
     /// The tokenizer of this parser.
     tokenizer: DOMRefCell<Tokenizer>,
     /// Whether to expect any further input from the associated network request.
     last_chunk_received: Cell<bool>,
     /// Whether this parser should avoid passing any further data to the tokenizer.
     suspended: Cell<bool>,
+    /// https://html.spec.whatwg.org/multipage/#script-nesting-level
+    script_nesting_level: Cell<usize>,
 }
 
 #[derive(PartialEq)]
@@ -134,6 +152,84 @@ impl ServoParser {
         parser.parse_chunk(String::from(input));
     }
 
+    pub fn script_nesting_level(&self) -> usize {
+        self.script_nesting_level.get()
+    }
+
+    /// Corresponds to the latter part of the "Otherwise" branch of the 'An end
+    /// tag whose tag name is "script"' of
+    /// https://html.spec.whatwg.org/multipage/#parsing-main-incdata
+    ///
+    /// This first moves everything from the script input to the beginning of
+    /// the network input, effectively resetting the insertion point to just
+    /// before the next character to be consumed.
+    ///
+    ///
+    /// ```text
+    ///     | ... script input ... network input ...
+    ///     ^
+    ///     insertion point
+    /// ```
+    pub fn resume_with_pending_parsing_blocking_script(&self, script: &HTMLScriptElement) {
+        assert!(self.suspended.get());
+        self.suspended.set(false);
+
+        mem::swap(&mut *self.script_input.borrow_mut(), &mut *self.network_input.borrow_mut());
+        while let Some(chunk) = self.script_input.borrow_mut().pop_front() {
+            self.network_input.borrow_mut().push_back(chunk);
+        }
+
+        let script_nesting_level = self.script_nesting_level.get();
+        assert_eq!(script_nesting_level, 0);
+
+        self.script_nesting_level.set(script_nesting_level + 1);
+        script.execute();
+        self.script_nesting_level.set(script_nesting_level);
+
+        if !self.suspended.get() {
+            self.parse_sync();
+        }
+    }
+
+    /// Steps 6-8 of https://html.spec.whatwg.org/multipage/#document.write()
+    pub fn write(&self, text: Vec<DOMString>) {
+        assert!(self.script_nesting_level.get() > 0);
+
+        if self.document.get_pending_parsing_blocking_script().is_some() {
+            // There is already a pending parsing blocking script so the
+            // parser is suspended, we just append everything to the
+            // script input and abort these steps.
+            for chunk in text {
+                self.script_input.borrow_mut().push_back(String::from(chunk).into());
+            }
+            return;
+        }
+
+        // There is no pending parsing blocking script, so all previous calls
+        // to document.write() should have seen their entire input tokenized
+        // and process, with nothing pushed to the parser script input.
+        assert!(self.script_input.borrow().is_empty());
+
+        let mut input = BufferQueue::new();
+        for chunk in text {
+            input.push_back(String::from(chunk).into());
+        }
+
+        self.tokenize(|tokenizer| tokenizer.feed(&mut input));
+
+        if self.suspended.get() {
+            // Parser got suspended, insert remaining input at end of
+            // script input, following anything written by scripts executed
+            // reentrantly during this call.
+            while let Some(chunk) = input.pop_front() {
+                self.script_input.borrow_mut().push_back(chunk);
+            }
+            return;
+        }
+
+        assert!(input.is_empty());
+    }
+
     #[allow(unrooted_must_root)]
     fn new_inherited(
             document: &Document,
@@ -145,10 +241,12 @@ impl ServoParser {
             reflector: Reflector::new(),
             document: JS::from_ref(document),
             pipeline: pipeline,
-            pending_input: DOMRefCell::new(BufferQueue::new()),
+            network_input: DOMRefCell::new(BufferQueue::new()),
+            script_input: DOMRefCell::new(BufferQueue::new()),
             tokenizer: DOMRefCell::new(tokenizer),
             last_chunk_received: Cell::new(last_chunk_state == LastChunkState::Received),
             suspended: Default::default(),
+            script_nesting_level: Default::default(),
         }
     }
 
@@ -165,106 +263,87 @@ impl ServoParser {
             ServoParserBinding::Wrap)
     }
 
-    pub fn document(&self) -> &Document {
-        &self.document
-    }
-
-    pub fn pipeline(&self) -> Option<PipelineId> {
-        self.pipeline
-    }
-
-    fn has_pending_input(&self) -> bool {
-        !self.pending_input.borrow().is_empty()
-    }
-
     fn push_input_chunk(&self, chunk: String) {
-        self.pending_input.borrow_mut().push_back(chunk.into());
-    }
-
-    fn last_chunk_received(&self) -> bool {
-        self.last_chunk_received.get()
-    }
-
-    fn mark_last_chunk_received(&self) {
-        self.last_chunk_received.set(true)
-    }
-
-    fn set_plaintext_state(&self) {
-        self.tokenizer.borrow_mut().set_plaintext_state()
-    }
-
-    pub fn suspend(&self) {
-        assert!(!self.suspended.get());
-        self.suspended.set(true);
-    }
-
-    pub fn resume(&self) {
-        assert!(self.suspended.get());
-        self.suspended.set(false);
-        self.parse_sync();
-    }
-
-    pub fn is_suspended(&self) -> bool {
-        self.suspended.get()
+        self.network_input.borrow_mut().push_back(chunk.into());
     }
 
     fn parse_sync(&self) {
         let metadata = TimerMetadata {
-            url: self.document().url().as_str().into(),
+            url: self.document.url().as_str().into(),
             iframe: TimerMetadataFrameType::RootWindow,
             incremental: TimerMetadataReflowType::FirstReflow,
         };
         let profiler_category = self.tokenizer.borrow().profiler_category();
         profile(profiler_category,
                 Some(metadata),
-                self.document().window().upcast::<GlobalScope>().time_profiler_chan().clone(),
+                self.document.window().upcast::<GlobalScope>().time_profiler_chan().clone(),
                 || self.do_parse_sync())
     }
 
     fn do_parse_sync(&self) {
+        assert!(self.script_input.borrow().is_empty());
+
         // This parser will continue to parse while there is either pending input or
         // the parser remains unsuspended.
-        loop {
-            self.document().reflow_if_reflow_timer_expired();
-            if let Err(script) = self.tokenizer.borrow_mut().feed(&mut *self.pending_input.borrow_mut()) {
-                if script.prepare() {
-                    continue;
-                }
-            }
 
-            // Document parsing is blocked on an external resource.
-            if self.suspended.get() {
-                return;
-            }
+        self.tokenize(|tokenizer| tokenizer.feed(&mut *self.network_input.borrow_mut()));
 
-            if !self.has_pending_input() {
-                break;
-            }
+        if self.suspended.get() {
+            return;
         }
 
-        if self.last_chunk_received() {
+        assert!(self.network_input.borrow().is_empty());
+
+        if self.last_chunk_received.get() {
             self.finish();
         }
     }
 
     fn parse_chunk(&self, input: String) {
-        self.document().set_current_parser(Some(self));
+        self.document.set_current_parser(Some(self));
         self.push_input_chunk(input);
-        if !self.is_suspended() {
+        if !self.suspended.get() {
             self.parse_sync();
         }
     }
 
+    fn tokenize<F>(&self, mut feed: F)
+        where F: FnMut(&mut Tokenizer) -> Result<(), Root<HTMLScriptElement>>
+    {
+        loop {
+            assert!(!self.suspended.get());
+
+            self.document.reflow_if_reflow_timer_expired();
+            let script = match feed(&mut *self.tokenizer.borrow_mut()) {
+                Ok(()) => return,
+                Err(script) => script,
+            };
+
+            let script_nesting_level = self.script_nesting_level.get();
+
+            self.script_nesting_level.set(script_nesting_level + 1);
+            script.prepare();
+            self.script_nesting_level.set(script_nesting_level);
+
+            if self.document.get_pending_parsing_blocking_script().is_some() {
+                self.suspended.set(true);
+                return;
+            }
+        }
+    }
+
     fn finish(&self) {
         assert!(!self.suspended.get());
-        assert!(!self.has_pending_input());
+        assert!(self.last_chunk_received.get());
+        assert!(self.script_input.borrow().is_empty());
+        assert!(self.network_input.borrow().is_empty());
 
         self.tokenizer.borrow_mut().end();
         debug!("finished parsing");
 
-        self.document().set_current_parser(None);
+        self.document.set_current_parser(None);
 
-        if let Some(pipeline) = self.pipeline() {
+        if let Some(pipeline) = self.pipeline {
             ScriptThread::parsing_complete(pipeline);
         }
     }
@@ -372,7 +451,7 @@ impl FetchResponseListener for ParserContext {
                 parser.push_input_chunk(page);
                 parser.parse_sync();
 
-                let doc = parser.document();
+                let doc = &parser.document;
                 let doc_body = Root::upcast::<Node>(doc.GetBody().unwrap());
                 let img = HTMLImageElement::new(local_name!("img"), None, doc);
                 img.SetSrc(DOMString::from(self.url.to_string()));
@@ -384,7 +463,7 @@ impl FetchResponseListener for ParserContext {
                 let page = "<pre>\n".into();
                 parser.push_input_chunk(page);
                 parser.parse_sync();
-                parser.set_plaintext_state();
+                parser.tokenizer.borrow_mut().set_plaintext_state();
             },
             Some(ContentType(Mime(TopLevel::Text, SubLevel::Html, _))) => { // Handle text/html
                 if let Some(reason) = ssl_error {
@@ -449,11 +528,11 @@ impl FetchResponseListener for ParserContext {
             debug!("Failed to load page URL {}, error: {:?}", self.url, err);
         }
 
-        parser.document()
+        parser.document
             .finish_load(LoadType::PageSource(self.url.clone()));
 
-        parser.mark_last_chunk_received();
-        if !parser.is_suspended() {
+        parser.last_chunk_received.set(true);
+        if !parser.suspended.get() {
             parser.parse_sync();
         }
     }

components/script/dom/webidls/Document.webidl

@@ -114,8 +114,10 @@ partial /*sealed*/ interface Document {
   // Document open(optional DOMString type = "text/html", optional DOMString replace = "");
   // WindowProxy open(DOMString url, DOMString name, DOMString features, optional boolean replace = false);
   // void close();
-  // void write(DOMString... text);
-  // void writeln(DOMString... text);
+  [Throws]
+  void write(DOMString... text);
+  [Throws]
+  void writeln(DOMString... text);
 
   // user interaction
   readonly attribute Window?/*Proxy?*/ defaultView;