1 files changed, 301 insertions, 0 deletions
diff --git a/parser/html/nsHtml5TreeBuilderHSupplement.h b/parser/html/nsHtml5TreeBuilderHSupplement.h
new file mode 100644
index 0000000000..813654cb24
--- /dev/null
+++ b/parser/html/nsHtml5TreeBuilderHSupplement.h
@@ -0,0 +1,301 @@
+/* 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/. */
+
+#define NS_HTML5_TREE_BUILDER_HANDLE_ARRAY_LENGTH 512
+private:
+using Encoding = mozilla::Encoding;
+template <typename T>
+using NotNull = mozilla::NotNull<T>;
+
+nsHtml5OplessBuilder* mBuilder;
+// !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
+// If mBuilder is not null, the tree op machinery is not in use and
+// the fields below aren't in use, either.
+// !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
+nsHtml5Highlighter* mViewSource;
+mozilla::ImportScanner mImportScanner;
+nsTArray<nsHtml5TreeOperation> mOpQueue;
+nsTArray<nsHtml5SpeculativeLoad> mSpeculativeLoadQueue;
+nsAHtml5TreeOpSink* mOpSink;
+mozilla::UniquePtr<nsIContent*[]> mHandles;
+int32_t mHandlesUsed;
+nsTArray<mozilla::UniquePtr<nsIContent*[]>> mOldHandles;
+nsHtml5TreeOpStage* mSpeculativeLoadStage;
+nsresult mBroken;
+// Controls whether the current HTML script goes through the more complex
+// path that accommodates the possibility of the script becoming a
+// parser-blocking script and the possibility of the script inserting
+// content into this parse using document.write (as it is observable from
+// the Web).
+//
+// Notably, in some cases scripts that do NOT NEED the more complex path
+// BREAK the parse if they incorrectly go onto the complex path as their
+// other handling doesn't necessarily take care of the responsibilities
+// associated with the more complex path. See comments in
+// `nsHtml5TreeBuilder::createElement` in the CppSupplement for details.
+bool mCurrentHtmlScriptCannotDocumentWriteOrBlock;
+bool mPreventScriptExecution;
+/**
+ * Whether to actually generate speculative load operations that actually
+ * represent speculative loads as opposed to other operations traveling
+ * in the same queue. True for normal loads and false for XHR, plain text,
+ * and View Source. Needed, because we can't just null-check
+ * mSpeculativeLoadStage, since it is used for transferring encoding
+ * information even in the XHR/plain text/View Source cases.
+ */
+bool mGenerateSpeculativeLoads;
+
+bool mHasSeenImportMap;
+#ifdef DEBUG
+bool mActive;
+#endif
+
+// DocumentModeHandler
+/**
+ * Tree builder uses this to report quirkiness of the document
+ */
+void documentMode(nsHtml5DocumentMode m);
+
+nsIContentHandle* getDocumentFragmentForTemplate(nsIContentHandle* aTemplate);
+void setDocumentFragmentForTemplate(nsIContentHandle* aTemplate,
+                                    nsIContentHandle* aFragment);
+
+nsIContentHandle* getShadowRootFromHost(nsIContentHandle* aHost,
+                                        nsIContentHandle* aTemplateNode,
+                                        nsHtml5String aShadowRootMode,
+                                        bool aShadowRootDelegatesFocus);
+
+nsIContentHandle* getFormPointerForContext(nsIContentHandle* aContext);
+
+/**
+ * Using nsIContent** instead of nsIContent* is the parser deals with DOM
+ * nodes in a way that works off the main thread. Non-main-thread code
+ * can't refcount or otherwise touch nsIContent objects in any way.
+ * Yet, the off-the-main-thread code needs to have a way to hold onto a
+ * particular node and repeatedly operate on the same node.
+ *
+ * The way this works is that the off-the-main-thread code has an
+ * nsIContent** for each DOM node and a given nsIContent** is only ever
+ * actually dereferenced into an actual nsIContent* on the main thread.
+ * When the off-the-main-thread code requests a new node, it gets an
+ * nsIContent** immediately and a tree op is enqueued for later allocating
+ * an actual nsIContent object and writing a pointer to it into the memory
+ * location pointed to by the nsIContent**.
+ *
+ * Since tree ops are in a queue, the node creating tree op will always
+ * run before tree ops that try to further operate on the node that the
+ * nsIContent** is a handle to.
+ *
+ * On-the-main-thread parts of the parser use nsIContent* instead of
+ * nsIContent**. Since both cases share the same parser core, the parser
+ * core casts both to nsIContentHandle*.
+ */
+nsIContentHandle* AllocateContentHandle();
+
+void accumulateCharactersForced(const char16_t* aBuf, int32_t aStart,
+                                int32_t aLength) {
+  accumulateCharacters(aBuf, aStart, aLength);
+}
+
+void MarkAsBrokenAndRequestSuspensionWithBuilder(nsresult aRv) {
+  mBuilder->MarkAsBroken(aRv);
+  requestSuspension();
+}
+
+void MarkAsBrokenAndRequestSuspensionWithoutBuilder(nsresult aRv) {
+  MarkAsBroken(aRv);
+  requestSuspension();
+}
+
+void MarkAsBrokenFromPortability(nsresult aRv);
+
+public:
+explicit nsHtml5TreeBuilder(nsHtml5OplessBuilder* aBuilder);
+
+nsHtml5TreeBuilder(nsAHtml5TreeOpSink* aOpSink, nsHtml5TreeOpStage* aStage,
+                   bool aGenerateSpeculativeLoads);
+
+~nsHtml5TreeBuilder();
+
+bool WantsLineAndColumn() {
+  // Perhaps just checking mBuilder would be sufficient.
+  // For createContextualFragment, we have non-null mBuilder and
+  // false for mPreventScriptExecution. However, do the line and
+  // column that get attached to script elements make any sense
+  // anyway in that case?
+  return !(mBuilder && mPreventScriptExecution);
+}
+
+void StartPlainTextViewSource(const nsAutoString& aTitle);
+
+void StartPlainText();
+
+void StartPlainTextBody();
+
+bool HasScriptThatMayDocumentWriteOrBlock();
+
+void SetOpSink(nsAHtml5TreeOpSink* aOpSink) { mOpSink = aOpSink; }
+
+void ClearOps() { mOpQueue.Clear(); }
+
+/**
+ * Flushes tree ops.
+ * @return Ok(true) if there were ops to flush, Ok(false)
+ *         if there were no ops to flush and Err() on OOM.
+ */
+mozilla::Result<bool, nsresult> Flush(bool aDiscretionary = false);
+
+void FlushLoads();
+
+/**
+ * Sets the document charset via the speculation queue.
+ *
+ * @param aCommitEncodingSpeculation true iff the main thread should
+ *                           treat the first speculation as an
+ *                           encoding speculation.
+ */
+void SetDocumentCharset(NotNull<const Encoding*> aEncoding,
+                        nsCharsetSource aCharsetSource,
+                        bool aCommitEncodingSpeculation);
+
+/**
+ * Updates the charset source via the op queue.
+ */
+void UpdateCharsetSource(nsCharsetSource aCharsetSource);
+
+void StreamEnded();
+
+void NeedsCharsetSwitchTo(NotNull<const Encoding*> aEncoding, int32_t aSource,
+                          int32_t aLineNumber);
+
+void MaybeComplainAboutCharset(const char* aMsgId, bool aError,
+                               int32_t aLineNumber);
+
+void TryToEnableEncodingMenu();
+
+void AddSnapshotToScript(nsAHtml5TreeBuilderState* aSnapshot, int32_t aLine);
+
+void DropHandles();
+
+void SetPreventScriptExecution(bool aPrevent) {
+  mPreventScriptExecution = aPrevent;
+}
+
+bool HasBuilder() { return mBuilder; }
+
+/**
+ * Makes sure the buffers are large enough to be able to tokenize aLength
+ * UTF-16 code units before having to make the buffers larger.
+ *
+ * @param aLength the number of UTF-16 code units to be tokenized before the
+ *                next call to this method.
+ * @return true if successful; false if out of memory
+ */
+bool EnsureBufferSpace(int32_t aLength);
+
+void EnableViewSource(nsHtml5Highlighter* aHighlighter);
+
+void errDeepTree();
+
+void errStrayStartTag(nsAtom* aName);
+
+void errStrayEndTag(nsAtom* aName);
+
+void errUnclosedElements(int32_t aIndex, nsAtom* aName);
+
+void errUnclosedElementsImplied(int32_t aIndex, nsAtom* aName);
+
+void errUnclosedElementsCell(int32_t aIndex);
+
+void errStrayDoctype();
+
+void errAlmostStandardsDoctype();
+
+void errQuirkyDoctype();
+
+void errNonSpaceInTrailer();
+
+void errNonSpaceAfterFrameset();
+
+void errNonSpaceInFrameset();
+
+void errNonSpaceAfterBody();
+
+void errNonSpaceInColgroupInFragment();
+
+void errNonSpaceInNoscriptInHead();
+
+void errFooBetweenHeadAndBody(nsAtom* aName);
+
+void errStartTagWithoutDoctype();
+
+void errNoSelectInTableScope();
+
+void errStartSelectWhereEndSelectExpected();
+
+void errStartTagWithSelectOpen(nsAtom* aName);
+
+void errBadStartTagInNoscriptInHead(nsAtom* aName);
+
+void errImage();
+
+void errIsindex();
+
+void errFooSeenWhenFooOpen(nsAtom* aName);
+
+void errHeadingWhenHeadingOpen();
+
+void errFramesetStart();
+
+void errNoCellToClose();
+
+void errStartTagInTable(nsAtom* aName);
+
+void errFormWhenFormOpen();
+
+void errTableSeenWhileTableOpen();
+
+void errStartTagInTableBody(nsAtom* aName);
+
+void errEndTagSeenWithoutDoctype();
+
+void errEndTagAfterBody();
+
+void errEndTagSeenWithSelectOpen(nsAtom* aName);
+
+void errGarbageInColgroup();
+
+void errEndTagBr();
+
+void errNoElementToCloseButEndTagSeen(nsAtom* aName);
+
+void errHtmlStartTagInForeignContext(nsAtom* aName);
+
+void errNoTableRowToClose();
+
+void errNonSpaceInTable();
+
+void errUnclosedChildrenInRuby();
+
+void errStartTagSeenWithoutRuby(nsAtom* aName);
+
+void errSelfClosing();
+
+void errNoCheckUnclosedElementsOnStack();
+
+void errEndTagDidNotMatchCurrentOpenElement(nsAtom* aName, nsAtom* aOther);
+
+void errEndTagViolatesNestingRules(nsAtom* aName);
+
+void errEndWithUnclosedElements(nsAtom* aName);
+
+void errListUnclosedStartTags(int32_t aIgnored);
+
+void MarkAsBroken(nsresult aRv);
+
+/**
+ * Checks if this parser is broken. Returns a non-NS_OK (i.e. non-0)
+ * value if broken.
+ */
+nsresult IsBroken() { return mBroken; }