type machine_state =
  | DataState
  | CharRefDataState
  | TagOpenState
  | CloseTagOpenState
  | TagNameState
  | BeforeAttributeNameState
  | AttributeNameState
  | AfterAttributeNameState
  | BeforeAttributeValueState
  | AttributeValueDoubleQuotedState
  | AttributeValueSingleQuotedState
  | AttributeValueUnquotedState
  | CharRefInAttributeValueState_DoubleQuoted (* split into three - saves the complexity of another state variable when exiting the state *)
  | CharRefInAttributeValueState_SingleQuoted
  | CharRefInAttributeValueState_Unquoted
  | AfterAttributeValueQuotedState
  | SelfClosingStartTagState
  | BogusCommentState
  | BogusCommentState_Continue (* makes things fit the state-machine model better *)
  | MarkupDeclarationOpenState
  | CommentStartState
  | CommentStartDashState
  | CommentState
  | CommentEndDashState
  | CommentEndState
  | DoctypeState
  | BeforeDoctypeNameState
  | DoctypeNameState
  | AfterDoctypeNameState
  | BeforeDoctypePublicIdentifierState
  | DoctypePublicIdentifierDoubleQuotedState
  | DoctypePublicIdentifierSingleQuotedState
  | AfterDoctypePublicIdentifierState
  | BeforeDoctypeSystemIdentifierState
  | DoctypeSystemIdentifierDoubleQuotedState
  | DoctypeSystemIdentifierSingleQuotedState
  | AfterDoctypeSystemIdentifierState
  | BogusDoctypeState

let string_of_machine_state = function
  | DataState -> "DataState"
  | CharRefDataState -> "CharRefDataState"
  | TagOpenState -> "TagOpenState"
  | CloseTagOpenState -> "CloseTagOpenState"
  | TagNameState -> "TagNameState"
  | BeforeAttributeNameState -> "BeforeAttributeNameState"
  | AttributeNameState -> "AttributeNameState"
  | AfterAttributeNameState -> "AfterAttributeNameState"
  | BeforeAttributeValueState -> "BeforeAttributeValueState"
  | AttributeValueDoubleQuotedState -> "AttributeValueDoubleQuotedState"
  | AttributeValueSingleQuotedState -> "AttributeValueSingleQuotedState"
  | AttributeValueUnquotedState -> "AttributeValueUnquotedState"
  | CharRefInAttributeValueState_DoubleQuoted -> "CharRefInAttributeValueState_DoubleQuoted"
  | CharRefInAttributeValueState_SingleQuoted -> "CharRefInAttributeValueState_SingleQuoted"
  | CharRefInAttributeValueState_Unquoted -> "CharRefInAttributeValueState_Unquoted"
  | AfterAttributeValueQuotedState -> "AfterAttributeValueQuotedState"
  | SelfClosingStartTagState -> "SelfClosingStartTagState"
  | BogusCommentState -> "BogusCommentState"
  | BogusCommentState_Continue -> "BogusCommentState_Continue"
  | MarkupDeclarationOpenState -> "MarkupDeclarationOpenState"
  | CommentStartState -> "CommentStartState"
  | CommentStartDashState -> "CommentStartDashState"
  | CommentState -> "CommentState"
  | CommentEndDashState -> "CommentEndDashState"
  | CommentEndState -> "CommentEndState"
  | DoctypeState -> "DoctypeState"
  | BeforeDoctypeNameState -> "BeforeDoctypeNameState"
  | DoctypeNameState -> "DoctypeNameState"
  | AfterDoctypeNameState -> "AfterDoctypeNameState"
  | BeforeDoctypePublicIdentifierState -> "BeforeDoctypePublicIdentifierState"
  | DoctypePublicIdentifierDoubleQuotedState -> "DoctypePublicIdentifierDoubleQuotedState"
  | DoctypePublicIdentifierSingleQuotedState -> "DoctypePublicIdentifierSingleQuotedState"
  | AfterDoctypePublicIdentifierState -> "AfterDoctypePublicIdentifierState"
  | BeforeDoctypeSystemIdentifierState -> "BeforeDoctypeSystemIdentifierState"
  | DoctypeSystemIdentifierDoubleQuotedState -> "DoctypeSystemIdentifierDoubleQuotedState"
  | DoctypeSystemIdentifierSingleQuotedState -> "DoctypeSystemIdentifierSingleQuotedState"
  | AfterDoctypeSystemIdentifierState -> "AfterDoctypeSystemIdentifierState"
  | BogusDoctypeState -> "BogusDoctypeState"

let prettyStateName s = (* strip the "State" and everything after it *)
    let n = string_of_machine_state s in
    let rec f i = if String.sub n i 5 = "State" then String.sub n 0 i else f (i+1) in
    f 0;;

let enumerate_machine_state = [
  DataState;
  CharRefDataState;
  TagOpenState;
  CloseTagOpenState;
  TagNameState;
  BeforeAttributeNameState;
  AttributeNameState;
  AfterAttributeNameState;
  BeforeAttributeValueState;
  AttributeValueDoubleQuotedState;
  AttributeValueSingleQuotedState;
  AttributeValueUnquotedState;
  CharRefInAttributeValueState_DoubleQuoted;
  CharRefInAttributeValueState_SingleQuoted;
  CharRefInAttributeValueState_Unquoted;
  AfterAttributeValueQuotedState;
  SelfClosingStartTagState;
  BogusCommentState;
  BogusCommentState_Continue;
  MarkupDeclarationOpenState;
  CommentStartState;
  CommentStartDashState;
  CommentState;
  CommentEndDashState;
  CommentEndState;
  DoctypeState;
  BeforeDoctypeNameState;
  DoctypeNameState;
  AfterDoctypeNameState;
  BeforeDoctypePublicIdentifierState;
  DoctypePublicIdentifierDoubleQuotedState;
  DoctypePublicIdentifierSingleQuotedState;
  AfterDoctypePublicIdentifierState;
  BeforeDoctypeSystemIdentifierState;
  DoctypeSystemIdentifierDoubleQuotedState;
  DoctypeSystemIdentifierSingleQuotedState;
  AfterDoctypeSystemIdentifierState;
  BogusDoctypeState;
]

type content_model = PCDATA | RCDATA | CDATA | PLAINTEXT
let string_of_content_model = function PCDATA -> "PCDATA" | RCDATA -> "RCDATA" | CDATA -> "CDATA" | PLAINTEXT -> "PLAINTEXT"
let enumerate_content_model = [PCDATA; RCDATA; CDATA; PLAINTEXT]