(S) is space characters.
(*) is any character not mentioned higher in the list it occurs
EOF is end-of-file
'a' matches a string of characters a

XXX EOF is poorly handled right now
XXX need to say what actually happens with tokens

data
  &   "process entity"
  <   tag
  (*) data (append (*))


tag
  !   markup-declaration
  /   close-tag
  ?   pi-open
  (S) data (emit >(S) before switching)
  (*) start-tag-name (append (*) to name)

markup-declaration
  '--'      comment-start
  'DOCTYPE' (start tag not seen && doctype not seen) ? doctype-start : bogus-comment
  '[CDATA[' cdata-section
  (*)       bogus-comment

comment-start
  -   comment-dash
  (*) comment-start

comment-dash
  -   comment-end
  (*) comment-dash

comment-end
  >   data (emit comment)
  -   comment-dash
  (*) comment-start

doctype-start
  (S) doctype-before-root-name
  (*) bogus-comment

doctype-before-root-name
  (S) doctype-before-root-name
  (*) doctype-root-name

doctype-root-name
  [   doctype-internal-subset
  (S) doctype-root-name-after
  (*) doctype-root-name

doctype-root-name-after
  [   doctype-internal-subset
  >   data
  "   doctype-identifier-double-quoted
  '   doctype-identifier-single-quoted
  (*) doctype-root-name-after

doctype-identifier-double-quoted
  "   doctype-root-name-after
  (*) doctype-identifier-double-quoted

doctype-identifier-single-quoted
  '   doctype-root-name-after
  (*) doctype-identifier-single-quoted

doctype-internal-subset
  ]   doctype-internal-subset-after
  %   "process parameter entity"
  (S) doctype-internal-subset
  <   doctype-tag

doctype-tag
  !   doctype-markupdeclaration
  ?   doctype-pi-open
  (*) doctype-bogus-comment

doctype-markupdeclaration -- don't need ELEMENT
  'ATTLIST'  doctype-attlist
  'ENTITY'   doctype-entity
  'NOTATION' doctype-notation
  '--'       doctype-comment
  (*)        doctype-bogus-comment

doctype-attlist
  (S) doctype-attlist-before-name
  (*) doctype-bogus-comment

doctype-attlist-before-name
  (S) doctype-attlist-before-name
  (*) doctype-attlist-name

doctype-attlist-name
  (S) doctype-attlist-after-name
  (*) doctype-attlist-name

doctype-attlist-after-name
  (S) doctype-attlist-after-name
  >   doctype-internal-subset
  (*) doctype-attlist-attrname

doctype-attlist-attrname
  (S) doctype-attlist-after-attrname
  (*) doctype-attlist-attrname

doctype-attlist-after-attrname
  (S) doctype-attlist-after-attrname
  (*) doctype-attlist-attrtype

doctype-attlist-attrtype
  (S) doctype-attlist-after-attrtype
  (*) doctype-attlist-attrtype

doctype-attlist-after-attrtype
  (S) doctype-attlist-after-attrtype
  #   doctype-attlist-attrdecl-maybe
  (*) doctype-bogus-comment

doctype-attlist-attrdecl-maybe
  (S) doctype-bogus-comment
  (*) doctype-attlist-attrdecl

doctype-attlist-attrdecl
  (S) docxtype-attlist-after-attrdecl
  (*) doctype-attlist-attrdecl

doctype-attlist-after-attrdecl
  (S) doctype-attlist-after-attrdecl
  "   doctype-attlist-attrval-double-quoted
  '   doctype-attlist-attrval-single-quoted
  (*) doctype-bogus-comment

doctype-attlist-attrval-double-quoted
  "   doctype-attlist-after-name
  &   "process entity"
  %   "process parameter entity"
  (*) doctype-attlist-attrval-double-quoted

doctype-attlist-attrval-single-quoted
  '   doctype-attlist-after-name
  &   "process entity"
  %   "process parameter entity"
  (*) doctype-attlist-attrval-single-quoted

doctype-entity
  (S) doctype-entity-before-type
  (*) doctype-bogus-comment

doctype-entity-before-type
  (S) doctype-entity-before-type
  %   doctype-entity-parameter-maybe
  (*) doctype-entity-name

doctype-entity-parameter-maybe
  (S) doctype-entity-parameter
  (*) doctype-bogus-comment

doctype-entity-parameter
  (S) doctype-entity-parameter
  (*) doctype-entity-parameter-name

doctype-entity-parameter-name
  (S) doctype-entity-after-name
  (*) doctype-entity-parameter-name
  
doctype-entity-after-name
  (S) doctype-entity-after-name
  "   doctype-entity-val-double-quoted
  '   doctype-entity-val-single-quoted
  (*) doctype-entity-identifier

doctype-entity-val-double-quoted
  XXX what about & and %
  "   doctype-entity-after-val
  (*) doctype-entity-val-double-quoted

doctype-entity-val-single-quoted
  XXX what about & and %
  '   doctype-entity-after-val
  (*) doctype-entity-val-single-quoted

doctype-entity-after-val
  >   doctype-internal-subset
  (S) doctype-entity-after-val
  (*) doctype-entity-after-val (error)

doctype-entity-identifier
  >   doctype-internal-subset
  "   doctype-entity-identifier-double-quoted
  '   doctype-entity-identifier-single-quoted
  (*) doctype-entity-identifier

doctype-entity-identifier-double-quoted
  "   doctype-entity-identifier
  (*) doctype-entity-identifier-double-quoted

doctype-entity-identifier-single-quoted
  '   doctype-entity-identifier
  (*) doctype-entity-identifier-single-quoted

doctype-entity-name
  (S) doctype-entity-after-name
  (*) doctype-entity-name

doctype-notation
  (S) doctype-notation-identifier
  (*) doctype-bogus-comment

doctype-notation-identifier
  >   doctype-internal-subset
  "   doctype-notation-identifier-double-quoted
  '   doctype-notation-identifier-single-quoted
  (*) doctype-notation-identifier

doctype-notation-identifier-double-quoted
  "   doctype-notation-identifier
  (*) doctype-notation-identifier-double-quoted

doctype-notation-identifier-single-quoted
  '   doctype-notation-identifier
  (*) doctype-notation-identifier-single-quoted

doctype-comment
  XXX comment

doctype-pi-open
  XXX pi-open

doctype-bogus-comment
  >   doctype-internal-subset (emit comment)
  (*) doctype-bogus-comment

doctype-internal-subset-after
  >   data
  (*) doctype-internal-subset-after
  
cdata-section
  ]   (next two characters are ']>') ? data : cdata-section (reprocess)
  (*) cdata-section

bogus-comment
  >   data (emit comment)
  (*) bogus-comment

close-tag
  >   data (do nothing)
  (S) bogus-comment (reprocess (S))
  (*) close-tag-name (append (*))

close-tag-name
  (S) close-tag-name-end
  >   data (also emit close tag)
  (*) close-tag-name (append (*))

close-tag-name-end
  >   data
  (*) close-tag-name-end

pi-open
  (S) bogus comment
  (*) pi-name

pi-name
  (S) (pi-name known) ? pi-attributes-before : pi-normal
  ?   pi-maybe-close
  (*) pi-name (append (*))

pi-attributes-before
  XXX

pi-normal
  (S) pi-normal
  ?   pi-maybe-close
  (*) pi-data

pi-data
  ?   pi-maybe-close
  (*) pi-data

pi-maybe-close
  ?   pi-maybe-close
  >   data (emit pi)
  (*) pi-data

start-tag-name
  /   maybe-void-tag
  (S) tag-attribute-name-before
  (*) start-tag-name (append (*))

maybe-void-tag
  >   data (emit void tag)
  (*) tag-attribute-name-before (reprocess character)

tag-attribute-name-before
  (S) tag-attribute-name-before
  >   data (emit start tag)
  /   maybe-void-tag
  (*) tag-attribute-name

tag-attribute-name
  (S) tag-attribute-name-after
  =   tag-attribute-value-before
  >   data (emit start tag)
  /   maybe-void-tag
  (*) tag-attribute-name
  Important: When leaving tag-attribute-name drop duplicate attributes.

tag-attribute-name-after
  (S) tag-attribute-name-after
  =   tag-attribute-value-before
  >   data (emit start tag)
  /   maybe-void-tag
  (*) tag-attribute-name

tag-attribute-value-before
  "   tag-attribute-value-double-quoted
  '   tag-attribute-value-single-quoted
  &   tag-attribute-value-unquoted (reprocess)
  >   data (emit start tag)
  (*) tag-attribute-value

tag-attribute-value-double-quoted
  "   tag-attribute-name-before
  &   "process entity"
  (*) tag-attribute-value-double-quoted

tag-attribute-value-single-quoted
  '   tag-attribute-name-before
  &   "process entity"
  (*) tag-attribute-value-single-quoted

tag-attribute-value-unquoted
  (S) tag-attribute-name-before
  &   "process entity"
  >   data (emit start tag)
  (*) tag-attribute-value-unquoted