tags: remove NodeType and various sets

* NodeType is somewhat convenient, but adds a 1-word overhead to
  each node and makes object construction more error-prone. If needed,
  library users can still add it without us defining the enum.
* Now that we have atoms, the tagType function is useless.
* SpecialElements is only used in the specification's parser section,
  and is not even complete because it should also contain non-HTML
  tags. Moved to htmlparser.
* AllTagTypes can be expressed in a simpler way.

7 files changed, 95 insertions, 123 deletions

diff --git a/chame/htmlparser.nim b/chame/htmlparser.nim
index 279b9a1d..7d6591ba 100644
--- a/chame/htmlparser.nim
+++ b/chame/htmlparser.nim
@@ -996,6 +996,23 @@ proc findLastActiveFormattingAfterMarker(parser: var HTML5Parser,
       return i
   return -1
 
+#https://html.spec.whatwg.org/multipage/parsing.html#the-stack-of-open-elements
+const SpecialElements = {
+  TAG_ADDRESS, TAG_APPLET, TAG_AREA, TAG_ARTICLE, TAG_ASIDE, TAG_BASE,
+  TAG_BASEFONT, TAG_BGSOUND, TAG_BLOCKQUOTE, TAG_BODY, TAG_BR, TAG_BUTTON,
+  TAG_CAPTION, TAG_CENTER, TAG_COL, TAG_COLGROUP, TAG_DD, TAG_DETAILS, TAG_DIR,
+  TAG_DIV, TAG_DL, TAG_DT, TAG_EMBED, TAG_FIELDSET, TAG_FIGCAPTION, TAG_FIGURE,
+  TAG_FOOTER, TAG_FORM, TAG_FRAME, TAG_FRAMESET, TAG_H1, TAG_H2, TAG_H3, TAG_H4,
+  TAG_H5, TAG_H6, TAG_HEAD, TAG_HEADER, TAG_HGROUP, TAG_HR, TAG_HTML,
+  TAG_IFRAME, TAG_IMG, TAG_INPUT, TAG_KEYGEN, TAG_LI, TAG_LINK, TAG_LISTING,
+  TAG_MAIN, TAG_MARQUEE, TAG_MENU, TAG_META, TAG_NAV, TAG_NOEMBED, TAG_NOFRAMES,
+  TAG_NOSCRIPT, TAG_OBJECT, TAG_OL, TAG_P, TAG_PARAM, TAG_PLAINTEXT, TAG_PRE,
+  TAG_SCRIPT, TAG_SEARCH, TAG_SECTION, TAG_SELECT, TAG_SOURCE, TAG_STYLE,
+  TAG_SUMMARY, TAG_TABLE, TAG_TBODY, TAG_TD, TAG_TEMPLATE, TAG_TEXTAREA,
+  TAG_TFOOT, TAG_TH, TAG_THEAD, TAG_TITLE, TAG_TR, TAG_TRACK, TAG_UL, TAG_WBR,
+  TAG_XMP
+}
+
 proc isSpecialElement[Handle, Atom](parser: HTML5Parser[Handle, Atom],
     element: Handle): bool =
   let localName = parser.getLocalName(element)
@@ -1233,7 +1250,7 @@ macro match(token: Token, body: typed): untyped =
           assert s[i] in AsciiAlphaNumeric
           tagName &= s[i]
           inc i
-        let tt = int(tagType(tagName))
+        let tt = int(parseEnum[TagType](tagName))
         let tokt = if s[1] != '/': START_TAG else: END_TAG
         var found = false
         for i in 0..ofBranches[tokt].ofBranches.high:
@@ -1567,12 +1584,12 @@ proc processInHTMLContent[Handle, Atom](parser: var HTML5Parser[Handle, Atom],
 
     template parse_error_if_body_has_disallowed_open_elements =
       if parser.hasParseError():
-        const Disallowed = AllTagTypes - {
+        const Allowed = {
           TAG_DD, TAG_DT, TAG_LI, TAG_OPTGROUP, TAG_OPTION, TAG_P, TAG_RB,
           TAG_RP, TAG_RT, TAG_RTC, TAG_TBODY, TAG_TD, TAG_TFOOT, TAG_TH,
           TAG_THEAD, TAG_TR, TAG_BODY, TAG_HTML
         }
-        if parser.hasElement(Disallowed):
+        if parser.hasElement(AllTagTypes - Allowed):
           parse_error MISMATCHED_TAGS
 
     match token:
diff --git a/chame/minidom.nim b/chame/minidom.nim
index 1914a757..9a16fae5 100644
--- a/chame/minidom.nim
+++ b/chame/minidom.nim
@@ -75,7 +75,6 @@ type
   Attribute* = ParsedAttr[MAtom]
 
   Node* = ref object of RootObj
-    nodeType*: NodeType
     childList*: seq[Node]
     parentNode* {.cursor.}: Node
 
@@ -211,7 +210,6 @@ proc createElement(document: Document, localName: MAtom, namespace: Namespace):
     )
   else:
     Element()
-  element.nodeType = ELEMENT_NODE
   element.localName = localName
   element.namespace = namespace
   element.document = document
@@ -241,26 +239,38 @@ proc getTemplateContentImpl(builder: MiniDOMBuilder, handle: Node): Node =
   return HTMLTemplateElement(handle).content
 
 proc createCommentImpl(builder: MiniDOMBuilder, text: string): Node =
-  return Comment(nodeType: COMMENT_NODE, data: text.toValidUTF8())
+  return Comment(data: text.toValidUTF8())
 
 proc createDocumentTypeImpl(builder: MiniDOMBuilder, name, publicId,
     systemId: string): Node =
   return DocumentType(
-    nodeType: DOCUMENT_TYPE_NODE,
     name: name.toValidUTF8(),
     publicId: publicId.toValidUTF8(),
     systemId: systemId.toValidUTF8()
   )
 
-func countChildren(node: Node, nodeType: NodeType): int =
+func countElementChildren(node: Node): int =
   for child in node.childList:
-    if child.nodeType == nodeType:
+    if child of Element:
       inc result
 
-func hasChild(node: Node, nodeType: NodeType): bool =
+func hasTextChild(node: Node): bool =
   for child in node.childList:
-    if child.nodeType == nodeType:
+    if child of Text:
       return true
+  return false
+
+func hasElementChild(node: Node): bool =
+  for child in node.childList:
+    if child of Element:
+      return true
+  return false
+
+func hasDocumentTypeChild(node: Node): bool =
+  for child in node.childList:
+    if child of DocumentType:
+      return true
+  return false
 
 func isHostIncludingInclusiveAncestor(a, b: Node): bool =
   var b = b
@@ -269,61 +279,64 @@ func isHostIncludingInclusiveAncestor(a, b: Node): bool =
       return true
     b = b.parentNode
 
-func hasPreviousSibling(node: Node, nodeType: NodeType): bool =
+func hasPreviousElementSibling(node: Node): bool =
   for n in node.parentNode.childList:
     if n == node:
       break
-    if n.nodeType == nodeType:
+    if n of Element:
       return true
   return false
 
-func hasNextSibling(node: Node, nodeType: NodeType): bool =
+func hasNextDocumentTypeSibling(node: Node): bool =
   for i in countdown(node.parentNode.childList.len, 0):
     let n = node.parentNode.childList[i]
     if n == node:
       break
-    if n.nodeType == nodeType:
+    if n of DocumentType:
       return true
   return false
 
+func isValidParent(node: Node): bool =
+  return node of Element or node of Document or node of DocumentFragment
+
+func isValidChild(node: Node): bool =
+  return node.isValidParent or node of DocumentType or node of CharacterData
+
 # WARNING the ordering of the arguments in the standard is whack so this
 # doesn't match that
 func preInsertionValidity*(parent, node: Node, before: Node): bool =
-  if parent.nodeType notin {DOCUMENT_NODE, DOCUMENT_FRAGMENT_NODE, ELEMENT_NODE}:
+  if not parent.isValidParent:
     return false
   if node.isHostIncludingInclusiveAncestor(parent):
     return false
   if before != nil and before.parentNode != parent:
     return false
-  if node.nodeType notin {DOCUMENT_FRAGMENT_NODE, DOCUMENT_TYPE_NODE,
-      ELEMENT_NODE} + CharacterDataNodes:
+  if not node.isValidChild:
     return false
-  if node.nodeType == TEXT_NODE and parent.nodeType == DOCUMENT_NODE:
+  if node of Text and parent of Document:
     return false
-  if node.nodeType == DOCUMENT_TYPE_NODE and parent.nodeType != DOCUMENT_NODE:
+  if node of DocumentType and not (parent of Document):
     return false
-  if parent.nodeType == DOCUMENT_NODE:
-    case node.nodeType
-    of DOCUMENT_FRAGMENT_NODE:
-      let elems = node.countChildren(ELEMENT_NODE)
-      if elems > 1 or node.hasChild(TEXT_NODE):
+  if parent of Document:
+    if node of DocumentFragment:
+      let elems = node.countElementChildren()
+      if elems > 1 or node.hasTextChild():
         return false
-      elif elems == 1 and (parent.hasChild(ELEMENT_NODE) or
-          before != nil and (before.nodeType == DOCUMENT_TYPE_NODE or
-          before.hasNextSibling(DOCUMENT_TYPE_NODE))):
+      elif elems == 1 and (parent.hasElementChild() or
+          before != nil and
+          (before of DocumentType or before.hasNextDocumentTypeSibling())):
         return false
-    of ELEMENT_NODE:
-      if parent.hasChild(ELEMENT_NODE):
+    elif node of Element:
+      if parent.hasElementChild():
         return false
-      elif before != nil and (before.nodeType == DOCUMENT_TYPE_NODE or
-            before.hasNextSibling(DOCUMENT_TYPE_NODE)):
+      elif before != nil and (before of DocumentType or
+            before.hasNextDocumentTypeSibling()):
         return false
-    of DOCUMENT_TYPE_NODE:
-      if parent.hasChild(DOCUMENT_TYPE_NODE) or
-          before != nil and before.hasPreviousSibling(ELEMENT_NODE) or
-          before == nil and parent.hasChild(ELEMENT_NODE):
+    elif node of DocumentType:
+      if parent.hasDocumentTypeChild() or
+          before != nil and before.hasPreviousElementSibling() or
+          before == nil and parent.hasElementChild():
         return false
-    else: discard
   return true # no exception reached
 
 proc insertBefore(parent, child: Node, before: Option[Node]) =
@@ -355,10 +368,10 @@ proc insertTextImpl(builder: MiniDOMBuilder, parent: Node, text: string,
     parent.childList[^1]
   else:
     nil
-  if prevSibling != nil and prevSibling.nodeType == TEXT_NODE:
+  if prevSibling != nil and prevSibling of Text:
     Text(prevSibling).data &= text
   else:
-    let text = Text(nodeType: TEXT_NODE, data: text)
+    let text = Text(data: text)
     parent.insertBefore(text, option(before))
 
 proc removeImpl(builder: MiniDOMBuilder, child: Node) =
@@ -391,7 +404,7 @@ method setEncodingImpl(builder: MiniDOMBuilder, encoding: string):
   return SET_ENCODING_CONTINUE
 
 proc newMiniDOMBuilder*(stream: Stream, factory: MAtomFactory): MiniDOMBuilder =
-  let document = Document(nodeType: DOCUMENT_NODE, factory: factory)
+  let document = Document(factory: factory)
   let builder = MiniDOMBuilder(
     document: document,
     factory: factory,
@@ -440,7 +453,6 @@ proc parseHTMLFragment*(inputStream: Stream, element: Element,
     else: DATA
   let htmlAtom = builder.factory.tagTypeToAtom(TAG_HTML)
   let root = Element(
-    nodeType: ELEMENT_NODE,
     localName: htmlAtom,
     namespace: HTML,
     document: document
diff --git a/chame/minidom_cs.nim b/chame/minidom_cs.nim
index c1e1b6b3..9594d3be 100644
--- a/chame/minidom_cs.nim
+++ b/chame/minidom_cs.nim
@@ -50,7 +50,7 @@ method setEncodingImpl(builder: CharsetMiniDOMBuilder, encoding: string):
   return SET_ENCODING_STOP
 
 proc newCharsetMiniDOMBuilder(factory: MAtomFactory): CharsetMiniDOMBuilder =
-  let document = Document(nodeType: DOCUMENT_NODE, factory: factory)
+  let document = Document(factory: factory)
   let builder = CharsetMiniDOMBuilder(document: document, factory: factory)
   return builder
 
@@ -140,14 +140,14 @@ proc parseHTML*(inputStream: Stream, opts: HTML5ParserOpts[Node, MAtom],
       # A meta tag describing the charset has been found; force use of this
       # charset.
       inputStream.setPosition(0)
-      builder.document = Document(nodeType: DOCUMENT_NODE, factory: factory)
+      builder.document = Document(factory: factory)
       charsetStack.add(builder.charset)
       seekable = false
       continue
     if decoder.failed and seekable:
       # Retry with another charset.
       inputStream.setPosition(0)
-      builder.document = Document(nodeType: DOCUMENT_NODE, factory: factory)
+      builder.document = Document(factory: factory)
       continue
     break
   return builder.document
diff --git a/chame/tags.nim b/chame/tags.nim
index a9479e68..b8e45d09 100644
--- a/chame/tags.nim
+++ b/chame/tags.nim
@@ -1,20 +1,4 @@
-import std/tables
-
 type
-  NodeType* = enum
-    ELEMENT_NODE = 1,
-    ATTRIBUTE_NODE = 2,
-    TEXT_NODE = 3,
-    CDATA_SECTION_NODE = 4,
-    ENTITY_REFERENCE_NODE = 5,
-    ENTITY_NODE = 6
-    PROCESSING_INSTRUCTION_NODE = 7,
-    COMMENT_NODE = 8,
-    DOCUMENT_NODE = 9,
-    DOCUMENT_TYPE_NODE = 10,
-    DOCUMENT_FRAGMENT_NODE = 11,
-    NOTATION_NODE = 12
-
   TagType* = enum
     TAG_UNKNOWN = ""
     TAG_APPLET = "applet"
@@ -178,21 +162,7 @@ type
     PREFIX_XMLNS = "xmlns"
     PREFIX_UNKNOWN = ""
 
-func getTagTypeMap(): Table[string, TagType] =
-  for i in TagType:
-    result[$TagType(i)] = TagType(i)
-
-const tagTypeMap = getTagTypeMap()
-
-func tagType*(s: string): TagType =
-  if s in tagTypeMap:
-    return tagTypeMap[s]
-  return TAG_UNKNOWN
-
-const AllTagTypes* = (func(): set[TagType] =
-  for tag in TagType:
-    result.incl(tag)
-)()
+const AllTagTypes* = {TagType.low..TagType.high}
 
 const HTagTypes* = {
   TAG_H1, TAG_H2, TAG_H3, TAG_H4, TAG_H5, TAG_H6
@@ -200,32 +170,10 @@ const HTagTypes* = {
 
 # 4.10.2 Categories
 const FormAssociatedElements* = {
-  TAG_BUTTON, TAG_FIELDSET, TAG_INPUT, TAG_OBJECT, TAG_OUTPUT, TAG_SELECT, TAG_TEXTAREA, TAG_IMG
+  TAG_BUTTON, TAG_FIELDSET, TAG_INPUT, TAG_OBJECT, TAG_OUTPUT, TAG_SELECT,
+  TAG_TEXTAREA, TAG_IMG
 }
 
 const ListedElements* = {
   TAG_FIELDSET, TAG_INPUT, TAG_OBJECT, TAG_OUTPUT, TAG_SELECT, TAG_TEXTAREA
 }
-
-const CharacterDataNodes* = {
-  TEXT_NODE, CDATA_SECTION_NODE, PROCESSING_INSTRUCTION_NODE, COMMENT_NODE
-}
-
-#https://html.spec.whatwg.org/multipage/parsing.html#the-stack-of-open-elements
-#NOTE MathML not implemented
-#TODO SVG foreignObject, SVG desc, SVG title
-const SpecialElements* = {
- TAG_ADDRESS, TAG_APPLET, TAG_AREA, TAG_ARTICLE, TAG_ASIDE, TAG_BASE,
- TAG_BASEFONT, TAG_BGSOUND, TAG_BLOCKQUOTE, TAG_BODY, TAG_BR, TAG_BUTTON,
- TAG_CAPTION, TAG_CENTER, TAG_COL, TAG_COLGROUP, TAG_DD, TAG_DETAILS, TAG_DIR,
- TAG_DIV, TAG_DL, TAG_DT, TAG_EMBED, TAG_FIELDSET, TAG_FIGCAPTION, TAG_FIGURE,
- TAG_FOOTER, TAG_FORM, TAG_FRAME, TAG_FRAMESET, TAG_H1, TAG_H2, TAG_H3, TAG_H4,
- TAG_H5, TAG_H6, TAG_HEAD, TAG_HEADER, TAG_HGROUP, TAG_HR, TAG_HTML,
- TAG_IFRAME, TAG_IMG, TAG_INPUT, TAG_KEYGEN, TAG_LI, TAG_LINK, TAG_LISTING,
- TAG_MAIN, TAG_MARQUEE, TAG_MENU, TAG_META, TAG_NAV, TAG_NOEMBED, TAG_NOFRAMES,
- TAG_NOSCRIPT, TAG_OBJECT, TAG_OL, TAG_P, TAG_PARAM, TAG_PLAINTEXT, TAG_PRE,
- TAG_SCRIPT, TAG_SEARCH, TAG_SECTION, TAG_SELECT, TAG_SOURCE, TAG_STYLE,
- TAG_SUMMARY, TAG_TABLE, TAG_TBODY, TAG_TD, TAG_TEMPLATE, TAG_TEXTAREA,
- TAG_TFOOT, TAG_TH, TAG_THEAD, TAG_TITLE, TAG_TR, TAG_TRACK, TAG_UL, TAG_WBR,
- TAG_XMP
-}
diff --git a/tests/shared/tree_common.nim b/tests/shared/tree_common.nim
index c1b88295..6bbd10a2 100644
--- a/tests/shared/tree_common.nim
+++ b/tests/shared/tree_common.nim
@@ -101,7 +101,6 @@ proc parseTestFragment(ctx: var TCTestParser): TCFragment =
   of FT_MATHML: Namespace.MATHML
   of FT_HTML: Namespace.HTML
   let element = Element(
-    nodeType: ELEMENT_NODE,
     namespace: namespace,
     localName: ctx.factory.strToAtom(line)
   )
@@ -111,7 +110,7 @@ proc parseTestFragment(ctx: var TCTestParser): TCFragment =
   )
 
 proc parseDoctype(ctx: TCTestParser, s: string): DocumentType =
-  let doctype = DocumentType(nodeType: DOCUMENT_TYPE_NODE)
+  let doctype = DocumentType()
   var i = "<!DOCTYPE ".len
   while i < s.len and s[i] != ' ' and s[i] != '>':
     doctype.name &= s[i]
@@ -143,7 +142,7 @@ proc parseDoctype(ctx: TCTestParser, s: string): DocumentType =
   return doctype
 
 proc parseTestDocument(ctx: var TCTestParser): Document =
-  result = Document(nodeType: DOCUMENT_NODE, factory: ctx.factory)
+  result = Document(factory: ctx.factory)
   var stack: seq[Node]
   stack.add(result)
   template top: auto = stack[^1]
@@ -184,7 +183,7 @@ proc parseTestDocument(ctx: var TCTestParser): Document =
       let doctype = ctx.parseDoctype(str)
       top.childList.add(doctype)
     elif str.startsWith("<!-- "):
-      let comment = Comment(nodeType: COMMENT_NODE)
+      let comment = minidom.Comment()
       top.childList.add(comment)
       if not str.endsWith(" -->"):
         comment.data = str.substr("<!-- ".len) & "\n"
@@ -206,7 +205,6 @@ proc parseTestDocument(ctx: var TCTestParser): Document =
         HTMLTemplateElement()
       else:
         Element()
-      element.nodeType = ELEMENT_NODE
       element.localName = ctx.factory.strToAtom(nameStr)
       element.namespace = namespace
       element.document = result
@@ -219,7 +217,7 @@ proc parseTestDocument(ctx: var TCTestParser): Document =
       stack.add(fragment)
       indent += 2
     elif str[0] == '"':
-      let text = Text(nodeType: TEXT_NODE)
+      let text = Text()
       top.childList.add(text)
       if str[^1] != '"' or str.len == 1:
         text.data = str.substr(1) & "\n"
@@ -282,13 +280,12 @@ proc parseTests*(s: string, factory: MAtomFactory): seq[TCTest] =
       s &= $x & '\n'
 
 proc checkTest(nodein, nodep: Node) =
-  check nodein.nodeType == nodep.nodeType
   check nodein.childList.len == nodep.childList.len
   if nodein.childList.len != nodep.childList.len:
     echo nodein
     echo nodep
-  case nodein.nodeType
-  of ELEMENT_NODE:
+  if nodein of Element:
+    check nodep of Element
     let nodein = Element(nodein)
     let nodep = Element(nodep)
     check nodein.localName == nodep.localName
@@ -297,19 +294,18 @@ proc checkTest(nodein, nodep: Node) =
       echo "NODEIN", $nodein
       echo "NODEP", $nodep
     check nodein.attrs == nodep.attrs
-  of ATTRIBUTE_NODE, ENTITY_REFERENCE_NODE, ENTITY_NODE,
-      DOCUMENT_FRAGMENT_NODE, NOTATION_NODE:
+  elif nodein of DocumentFragment:
     assert false
-  of TEXT_NODE, CDATA_SECTION_NODE, COMMENT_NODE:
+  elif nodein of CharacterData:
+    check nodep of CharacterData
     check CharacterData(nodein).data == CharacterData(nodep).data
-  of PROCESSING_INSTRUCTION_NODE: assert false, "todo"
-  of DOCUMENT_TYPE_NODE:
+  elif nodein of DocumentType:
+    check nodep of DocumentType
     let nodein = DocumentType(nodein)
     let nodep = DocumentType(nodep)
     check nodein.name == nodep.name
     check nodein.publicId == nodep.publicId
     check nodein.systemId == nodep.systemId
-  of DOCUMENT_NODE: discard
   for i in 0 ..< nodein.childList.len:
     checkTest(nodein.childList[i], nodep.childList[i])
 
diff --git a/tests/test1.nim b/tests/test1.nim
index 0fde7216..d5d31837 100644
--- a/tests/test1.nim
+++ b/tests/test1.nim
@@ -32,8 +32,7 @@ func escapeText(s: string, attribute_mode = false): string =
       result &= c
 
 func `$`*(node: Node): string =
-  case node.nodeType
-  of ELEMENT_NODE:
+  if node of Element:
     let element = Element(node)
     var x = ""
     if element.namespace == Namespace.SVG:
@@ -47,17 +46,17 @@ func `$`*(node: Node): string =
     for node in element.childList:
       result &= $node
     result &= "</" & x & element.localNameStr & ">"
-  of TEXT_NODE:
+  elif node of Text:
     let text = Text(node)
     result = text.data.escapeText()
-  of COMMENT_NODE:
+  elif node of Comment:
     result = "<!-- " & Comment(node).data & "-->"
-  of PROCESSING_INSTRUCTION_NODE:
-    result = "" #TODO
-  of DOCUMENT_TYPE_NODE:
+  elif node of DocumentType:
     result = "<!DOCTYPE" & ' ' & DocumentType(node).name & ">"
+  elif node of Document:
+    result = "Node of Document"
   else:
-    result = "Node of " & $node.nodeType
+    assert false
 
 # This is, in fact, standards-compliant behavior.
 # Don't ask.
diff --git a/tests/tree.nim b/tests/tree.nim
index eee1f05f..527c7648 100644
--- a/tests/tree.nim
+++ b/tests/tree.nim
@@ -16,7 +16,7 @@ proc runTest(test: TCTest, factory: MAtomFactory, scripting, print: bool) =
     for child in childList:
       if ctx.preInsertionValidity(child, nil):
         ctx.childList.add(child)
-    Document(nodeType: DOCUMENT_NODE, childList: ctx.childList)
+    Document(childList: ctx.childList)
   if print:
     var ins = ""
     for x in test.document.childList: