Bug 1483458 - Change HTML parser to look at .py Atom files. r=hsivonen,heycam

--HG--
extra : amend_source : 9285251fc49d4f56b67557ef77e163d6acbc2b1e

parser/html/java/Makefile

@@ -37,13 +37,13 @@ translate:: translator \
 ; mkdir -p ../javasrc ; \
   java -jar translator.jar \
     htmlparser/src/nu/validator/htmlparser/impl \
-    .. ../../../xpcom/ds/nsGkAtomList.h
+    .. ../../../xpcom/ds/StaticAtoms.py ../../../xpcom/ds/HTMLAtoms.py
 
 translate_from_snapshot:: translator \
 ; mkdir -p ../javasrc ; \
   java -jar translator.jar \
     ../javasrc \
-    .. ../../../dom/base/nsGkAtomList.h
+    .. ../../../xpcom/ds/StaticAtoms.py ../../../xpcom/ds/HTMLAtoms.py
 
 named_characters:: translator \
 ; java -cp translator.jar \

parser/html/nsHtml5AttributeName.h

@@ -58,15 +58,12 @@ class nsHtml5AttributeName
 {
 public:
   static int32_t* ALL_NO_NS;
-
 private:
   static int32_t* XMLNS_NS;
   static int32_t* XML_NS;
   static int32_t* XLINK_NS;
-
 public:
   static nsAtom** ALL_NO_PREFIX;
-
 private:
   static nsAtom** XMLNS_PREFIX;
   static nsAtom** XLINK_PREFIX;
@@ -74,7 +71,6 @@ private:
   static nsAtom** SVG_DIFFERENT(nsAtom* name, nsAtom* camel);
   static nsAtom** MATH_DIFFERENT(nsAtom* name, nsAtom* camel);
   static nsAtom** COLONIFIED_LOCAL(nsAtom* name, nsAtom* suffix);
-
 public:
   static nsAtom** SAME_LOCAL(nsAtom* name);
   inline static int32_t levelOrderBinarySearch(jArray<int32_t, int32_t> data,
@@ -162,7 +158,6 @@ private:
   nsAtom** prefix;
   bool custom;
   nsHtml5AttributeName(int32_t* uri, nsAtom** local, nsAtom** prefix);
-
 public:
   nsHtml5AttributeName();
   inline bool isInterned() { return !custom; }
@@ -674,11 +669,9 @@ public:
   static nsHtml5AttributeName* ATTR_FY;
   static nsHtml5AttributeName* ATTR_RY;
   static nsHtml5AttributeName* ATTR_REFY;
-
 private:
   static nsHtml5AttributeName** ATTRIBUTE_NAMES;
   static staticJArray<int32_t, int32_t> ATTRIBUTE_HASHES;
-
 public:
   static void initializeStatics();
   static void releaseStatics();

parser/html/nsHtml5ElementName.h

@@ -80,7 +80,6 @@ private:
   nsAtom* camelCaseName;
   mozilla::dom::HTMLContentCreatorFunction htmlCreator;
   mozilla::dom::SVGContentCreatorFunction svgCreator;
-
 public:
   int32_t flags;
   inline nsAtom* getName() { return name; }
@@ -406,11 +405,9 @@ public:
   static nsHtml5ElementName* ELT_RUBY;
   static nsHtml5ElementName* ELT_SUMMARY;
   static nsHtml5ElementName* ELT_TBODY;
-
 private:
   static nsHtml5ElementName** ELEMENT_NAMES;
   static staticJArray<int32_t, int32_t> ELEMENT_HASHES;
-
 public:
   static void initializeStatics();
   static void releaseStatics();

parser/html/nsHtml5MetaScanner.h

@@ -120,17 +120,14 @@ private:
 
 protected:
   nsHtml5ByteReadable* readable;
-
 private:
   int32_t metaState;
   int32_t contentIndex;
   int32_t charsetIndex;
   int32_t httpEquivIndex;
   int32_t contentTypeIndex;
-
 protected:
   int32_t stateSave;
-
 private:
   int32_t strBufLen;
   autoJArray<char16_t, int32_t> strBuf;
@@ -138,14 +135,11 @@ private:
   nsHtml5String charset;
   int32_t httpEquivState;
   nsHtml5TreeBuilder* treeBuilder;
-
 public:
   explicit nsHtml5MetaScanner(nsHtml5TreeBuilder* tb);
   ~nsHtml5MetaScanner();
-
 protected:
   void stateLoop(int32_t state);
-
 private:
   void handleCharInAttributeValue(int32_t c);
   inline int32_t toAsciiLowerCase(int32_t c)
@@ -160,10 +154,8 @@ private:
   void handleAttributeValue();
   bool handleTag();
   bool handleTagInner();
-
 protected:
   bool tryCharset(nsHtml5String encoding);
-
 public:
   static void initializeStatics();
   static void releaseStatics();

parser/html/nsHtml5StackNode.h

@@ -66,11 +66,9 @@ public:
   int32_t ns;
   nsIContentHandle* node;
   nsHtml5HtmlAttributes* attributes;
-
 private:
   int32_t refcount;
   mozilla::dom::HTMLContentCreatorFunction htmlCreator;
-
 public:
   inline int32_t getFlags() { return flags; }
 
@@ -106,7 +104,6 @@ public:
 private:
   static int32_t prepareSvgFlags(int32_t flags);
   static int32_t prepareMathFlags(int32_t flags, bool markAsIntegrationPoint);
-
 public:
   ~nsHtml5StackNode();
   void dropAttributes();

parser/html/nsHtml5StateSnapshot.h

@@ -68,7 +68,6 @@ private:
   bool framesetOk;
   bool needToDropLF;
   bool quirks;
-
 public:
   nsHtml5StateSnapshot(
     jArray<nsHtml5StackNode*, int32_t> stack,

parser/html/nsHtml5Tokenizer.h

@@ -235,19 +235,15 @@ private:
   static staticJArray<char16_t, int32_t> NOEMBED_ARR;
   static staticJArray<char16_t, int32_t> NOSCRIPT_ARR;
   static staticJArray<char16_t, int32_t> NOFRAMES_ARR;
-
 protected:
   nsHtml5TreeBuilder* tokenHandler;
   nsHtml5StreamParser* encodingDeclarationHandler;
   bool lastCR;
   int32_t stateSave;
-
 private:
   int32_t returnStateSave;
-
 protected:
   int32_t index;
-
 private:
   bool forceQuirks;
   char16_t additional;
@@ -257,16 +253,12 @@ private:
   int32_t hi;
   int32_t candidate;
   int32_t charRefBufMark;
-
 protected:
   int32_t value;
-
 private:
   bool seenDigits;
-
 protected:
   int32_t cstart;
-
 private:
   nsHtml5String publicId;
   nsHtml5String systemId;
@@ -276,24 +268,18 @@ private:
   int32_t charRefBufLen;
   autoJArray<char16_t, int32_t> bmpChar;
   autoJArray<char16_t, int32_t> astralChar;
-
 protected:
   nsHtml5ElementName* endTagExpectation;
-
 private:
   jArray<char16_t, int32_t> endTagExpectationAsArray;
-
 protected:
   bool endTag;
-
 private:
   bool containsHyphen;
   nsHtml5ElementName* tagName;
   nsHtml5ElementName* nonInternedTagName;
-
 protected:
   nsHtml5AttributeName* attributeName;
-
 private:
   nsHtml5AttributeName* nonInternedAttributeName;
   nsAtom* doctypeName;
@@ -302,16 +288,13 @@ private:
   nsHtml5HtmlAttributes* attributes;
   bool newAttributesEachTime;
   bool shouldSuspend;
-
 protected:
   bool confident;
-
 private:
   int32_t line;
   int32_t attributeLine;
   nsHtml5AtomTable* interner;
   bool viewingXmlSource;
-
 public:
   nsHtml5Tokenizer(nsHtml5TreeBuilder* tokenHandler, bool viewingXmlSource);
   void setInterner(nsHtml5AtomTable* interner);
@@ -324,13 +307,11 @@ public:
 
 private:
   void endTagExpectationToArray();
-
 public:
   void setLineNumber(int32_t line);
   inline int32_t getLineNumber() { return line; }
 
   nsHtml5HtmlAttributes* emptyAttributes();
-
 private:
   inline void appendCharRefBuf(char16_t c)
   {
@@ -369,7 +350,6 @@ private:
 
 protected:
   nsHtml5String strBufToString();
-
 private:
   void strBufToDoctypeName();
   void emitStrBuf();
@@ -389,21 +369,17 @@ private:
   }
 
   void emitComment(int32_t provisionalHyphens, int32_t pos);
-
 protected:
   void flushChars(char16_t* buf, int32_t pos);
-
 private:
   void strBufToElementNameString();
   int32_t emitCurrentTagToken(bool selfClosing, int32_t pos);
   void attributeNameComplete();
   void addAttributeWithoutValue();
   void addAttributeWithValue();
-
 public:
   void start();
   bool tokenizeBuffer(nsHtml5UTF16Buffer* buffer);
-
 private:
   template<class P>
   int32_t stateLoop(int32_t state,
@@ -455,23 +431,18 @@ private:
   void bogusDoctype();
   void bogusDoctypeWithoutQuirks();
   void handleNcrValue(int32_t returnState);
-
 public:
   void eof();
-
 private:
   void emitDoctypeToken(int32_t pos);
-
 protected:
   inline char16_t checkChar(char16_t* buf, int32_t pos) { return buf[pos]; }
 
 public:
   bool internalEncodingDeclaration(nsHtml5String internalCharset);
-
 private:
   void emitOrAppendTwo(const char16_t* val, int32_t returnState);
   void emitOrAppendOne(const char16_t* val, int32_t returnState);
-
 public:
   void end();
   void requestSuspension();

parser/html/nsHtml5TreeBuilder.h

@@ -71,7 +71,6 @@ class nsHtml5TreeBuilder : public nsAHtml5TreeBuilderState
 {
 private:
   static char16_t REPLACEMENT_CHARACTER[];
-
 public:
   static const int32_t OTHER = 0;
 
@@ -288,10 +287,8 @@ private:
   int32_t mode;
   int32_t originalMode;
   bool framesetOk;
-
 protected:
   nsHtml5Tokenizer* tokenizer;
-
 private:
   bool scriptingEnabled;
   bool needToDropLF;
@@ -311,11 +308,9 @@ private:
   nsIContentHandle* formPointer;
   nsIContentHandle* headPointer;
   nsIContentHandle* deepTreeSurrogateParent;
-
 protected:
   autoJArray<char16_t, int32_t> charBuffer;
   int32_t charBufferLen;
-
 private:
   bool quirks;
   bool isSrcdocDocument;
@@ -362,17 +357,14 @@ private:
   bool isTemplateContents();
   bool isTemplateModeStackEmpty();
   bool isSpecialParentInForeign(nsHtml5StackNode* stackNode);
-
 public:
   static nsHtml5String extractCharsetFromContent(nsHtml5String attributeValue,
                                                  nsHtml5TreeBuilder* tb);
 
 private:
   void checkMetaCharset(nsHtml5HtmlAttributes* attributes);
-
 public:
   void endTag(nsHtml5ElementName* elementName);
-
 private:
   void endTagTemplateInHead();
   int32_t findLastInTableScopeOrRootTemplateTbodyTheadTfoot();
@@ -435,10 +427,8 @@ private:
   void addAttributesToHtml(nsHtml5HtmlAttributes* attributes);
   void pushHeadPointerOntoStack();
   void reconstructTheActiveFormattingElements();
-
 public:
   void notifyUnusedStackNode(int32_t idxInStackNodes);
-
 private:
   nsHtml5StackNode* getUnusedStackNode();
   nsHtml5StackNode* createStackNode(
@@ -521,7 +511,6 @@ private:
   void appendVoidInputToCurrent(nsHtml5HtmlAttributes* attributes,
                                 nsIContentHandle* form);
   void appendVoidFormToCurrent(nsHtml5HtmlAttributes* attributes);
-
 protected:
   void accumulateCharacters(const char16_t* buf, int32_t start, int32_t length);
   void requestSuspension();
@@ -579,14 +568,12 @@ protected:
                                nsHtml5String systemIdentifier);
   void elementPushed(int32_t ns, nsAtom* name, nsIContentHandle* node);
   void elementPopped(int32_t ns, nsAtom* name, nsIContentHandle* node);
-
 public:
   inline bool cdataSectionAllowed() { return isInForeign(); }
 
 private:
   bool isInForeign();
   bool isInForeignButNotHtmlOrMathTextIntegrationPoint();
-
 public:
   void setFragmentContext(nsAtom* context,
                           int32_t ns,
@@ -595,16 +582,13 @@ public:
 
 protected:
   nsIContentHandle* currentNode();
-
 public:
   bool isScriptingEnabled();
   void setScriptingEnabled(bool scriptingEnabled);
   void setIsSrcdocDocument(bool isSrcdocDocument);
   void flushCharacters();
-
 private:
   bool charBufferContainsNonWhitespace();
-
 public:
   nsAHtml5TreeBuilderState* newSnapshot();
   bool snapshotMatches(nsAHtml5TreeBuilderState* snapshot);

parser/html/nsHtml5UTF16Buffer.h

@@ -60,7 +60,6 @@ private:
   char16_t* buffer;
   int32_t start;
   int32_t end;
-
 public:
   int32_t getStart();
   void setStart(int32_t start);

xpcom/ds/StaticAtoms.py

@@ -17,8 +17,13 @@ import sys
 # The first argument is the atom's identifier.
 # The second argument is the atom's string value.
 #
+# Please keep the Atom() definitions on one line as this is parsed by the
+#   htmlparser: parser/html/java/htmlparser
+# Please keep "START ATOMS" and "END ATOMS" comments as the parser uses them.
+#
 # It is not possible to conditionally define static atoms with #ifdef etc.
 STATIC_ATOMS = [
+    # START ATOMS
     # --------------------------------------------------------------------------
     # Generic atoms
     # --------------------------------------------------------------------------
@@ -75,8 +80,7 @@ STATIC_ATOMS = [
     Atom("allowpopupstoescapesandbox", "allow-popups-to-escape-sandbox"),
     Atom("allowpopups", "allow-popups"),
     Atom("allowpresentation", "allow-presentation"),
-    Atom("allowstorageaccessbyuseractivatetion",
-         "allow-storage-access-by-user-activation"),
+    Atom("allowstorageaccessbyuseractivatetion", "allow-storage-access-by-user-activation"),
     Atom("allowsameorigin", "allow-same-origin"),
     Atom("allowscripts", "allow-scripts"),
     Atom("allowscriptstoclose", "allowscriptstoclose"),
@@ -2316,6 +2320,7 @@ STATIC_ATOMS = [
     InheritingAnonBoxAtom("AnonBox_mozSVGOuterSVGAnonChild", ":-moz-svg-outer-svg-anon-child"),
     InheritingAnonBoxAtom("AnonBox_mozSVGForeignContent", ":-moz-svg-foreign-content"),
     InheritingAnonBoxAtom("AnonBox_mozSVGText", ":-moz-svg-text"),
+    # END ATOMS
 ] + HTML_PARSER_ATOMS