Redirect HTML resource bytes directly to parser thread (Blink side CL)

Source/core/html/parser/BackgroundHTMLParser.cpp

Index: Source/core/html/parser/BackgroundHTMLParser.cpp
diff --git a/Source/core/html/parser/BackgroundHTMLParser.cpp b/Source/core/html/parser/BackgroundHTMLParser.cpp
index 77d11543259836602909d8da77ebeed03c3d073e..98964214e32dd45f2045b0f76c3f05be0b29a8ec 100644
--- a/Source/core/html/parser/BackgroundHTMLParser.cpp
+++ b/Source/core/html/parser/BackgroundHTMLParser.cpp
@@ -27,7 +27,6 @@
 #include "core/html/parser/BackgroundHTMLParser.h"
 
 #include "core/html/parser/HTMLDocumentParser.h"
-#include "core/html/parser/HTMLParserThread.h"
 #include "core/html/parser/TextResourceDecoder.h"
 #include "core/html/parser/XSSAuditor.h"
 #include "wtf/MainThread.h"
@@ -94,11 +93,47 @@ BackgroundHTMLParser::BackgroundHTMLParser(PassRefPtr<WeakReference<BackgroundHT
     , m_xssAuditor(config->xssAuditor.release())
     , m_preloadScanner(config->preloadScanner.release())
     , m_decoder(config->decoder.release())
+    , m_resourceProvider(config->resourceProvider.release())
+    , m_receivingDataOnlyFromResourceProvider(false)
 {
+    // Tell the resource provider that we're ready to start receiving
+    // chunks of data directly from the parser thread.
+    if (m_resourceProvider)
+        m_resourceProvider->setBackgroundClient(this);
 }
 
 BackgroundHTMLParser::~BackgroundHTMLParser()
 {
+    if (m_resourceProvider) {
+        // Tell the resource provider that calling into this instance
+        // is no longer safe and any further data received should be
+        // ignored.
+        m_resourceProvider->setBackgroundClient(0);
+        // The resource provider needs to be destructed on the main thread,
+        // as the filter it installs on the I/O thread can only be removed

[abarth-chromium, 2014/02/03 21:17:36]

Blink doesn't know anything about I/O threads.

[oystein (OOO til 10th of July), 2014/02/05 00:05:00]

Removed that part of the comment; it was just a clarification, not relevant to
Blink.

+        // from the main thread...
+        // Note we can only call static functions on the HTMLDocumentParser at this point
+        // as the weakptr we have for it is likely invalidated now.
+        callOnMainThread(bind(&HTMLDocumentParser::destroyResourceProvider, m_resourceProvider.release()));
+    }
+}
+
+void BackgroundHTMLParser::didReceivedData(const char* data, size_t length)
+{
+    if (m_receivingDataOnlyFromResourceProvider && m_decoder) {
+        updateDocument(m_decoder->decode(data, length));
+        return;
+    }
+
+    // If the parser thread is not standalone yet, it means we may get
+    // further data packets from the main thread and will need to queue
+    // up any data coming directly from the parser thread until we're
+    // sure we're complete. This is also necessary if we've lost our
+    // decoder, until we get a new one passed from the main thread.
+
+    OwnPtr<Vector<char> > buffer = adoptPtr(new Vector<char>(length));
+    memcpy(buffer->data(), data, length);

[abarth-chromium, 2014/02/03 21:17:36]

buffer.append(data, length)

[oystein (OOO til 10th of July), 2014/02/05 00:05:00]

Done.

+    m_queuedData.append(buffer.release());
 }
 
 void BackgroundHTMLParser::append(const String& input)
@@ -110,12 +145,18 @@ void BackgroundHTMLParser::append(const String& input)
 
 void BackgroundHTMLParser::appendBytes(PassOwnPtr<Vector<char> > buffer)
 {
+    ASSERT(!m_receivingDataOnlyFromResourceProvider);
     updateDocument(m_decoder->decode(buffer->data(), buffer->size()));
 }
 
 void BackgroundHTMLParser::setDecoder(PassOwnPtr<TextResourceDecoder> decoder)
 {
     m_decoder = decoder;
+
+    // If our decoder got reset in mid-stream and got recreated, we may have
+    // some pending data here.
+    if (m_decoder)
+        flushQueuedData();
 }
 
 void BackgroundHTMLParser::flush()
@@ -129,7 +170,6 @@ void BackgroundHTMLParser::updateDocument(const String& decodedData)
 
     if (encodingData != m_lastSeenEncodingData) {
         m_lastSeenEncodingData = encodingData;
-
         m_xssAuditor->setEncoding(encodingData.encoding());
         callOnMainThread(bind(&HTMLDocumentParser::didReceiveEncodingDataFromBackgroundParser, m_parser, encodingData));
     }
@@ -170,6 +210,30 @@ void BackgroundHTMLParser::stop()
     delete this;
 }
 
+void BackgroundHTMLParser::flushQueuedData()
+{
+    // We only process queued chunks if we have a decoder and if we know
+    // we won't be receiving any more chunks for the main thread. Otherwise
+    // we hang on to the chunks for a bit longer.
+    if (!m_decoder || !m_receivingDataOnlyFromResourceProvider)
+        return;
+
+    for (Vector<OwnPtr<Vector<char> > >::iterator entry = m_queuedData.begin(); entry != m_queuedData.end(); ++entry)
+        updateDocument(m_decoder->decode((*entry)->data(), (*entry)->size()));

[abarth-chromium, 2014/02/03 21:17:36]

SharedBuffer has a better way of doing this sort of thing.

[oystein (OOO til 10th of July), 2014/02/05 00:05:00]

Done.

+
+    m_queuedData.clear();
+}
+
+void BackgroundHTMLParser::resourceFilterAdded()
+{
+    m_receivingDataOnlyFromResourceProvider = true;
+
+    // At this point we know for sure that no further data will
+    // be coming from the main thread, so we can process anything
+    // we've received directly on the parser thread in the meantime.
+    flushQueuedData();
+}
+
 void BackgroundHTMLParser::forcePlaintextForTextDocument()
 {
     // This is only used by the TextDocumentParser (a subclass of HTMLDocumentParser)