#            NimYAML - YAML implementation in Nim
#        (c) Copyright 2015-2023 Felix Krause
#
#    See the file "copying.txt", included in this
#    distribution, for details about the copyright.

## =================
## Module yaml/hints
## =================
##
## The hints API enables you to guess the type of YAML scalars.

import macros
import private/internal

type
  TypeHint* = enum
    ## A type hint can be computed from scalar content and tells you what
    ## NimYAML thinks the scalar's type is. It is generated by
    ## `guessType <#guessType,string>`_ The first matching RegEx
    ## in the following table will be the type hint of a scalar string.
    ##
    ## You can use it to determine the type of YAML scalars that have a '?'
    ## non-specific tag, but using this feature is completely optional.
    ##
    ## See also: https://yaml.org/spec/1.2.2/#103-core-schema
    ##
    ## ================== =========================
    ## Name               RegEx
    ## ================== =========================
    ## ``yTypeInteger``   ``[-+]? [0-9]+``
    ## ``yTypeFloat``     ``[-+]? ( \. [0-9]+ | [0-9]+ ( \. [0-9]* )? ) ( [eE] [-+]? [0-9]+ )?``
    ## ``yTypeFloatInf``  ``[-+]? ( \.inf | \.Inf | \.INF )``
    ## ``yTypeFloatNaN``  ``\.nan | \.NaN | \.NAN``
    ## ``yTypeBoolTrue``  ``true | True | TRUE``
    ## ``yTypeBoolFalse`` ``false | False | FALSE``
    ## ``yTypeNull``      ``null | Null | NULL | ~``
    ## ``yTypeTimestamp`` see `here <http://yaml.org/type/timestamp.html>`_.
    ## ``yTypeUnknown``   ``*``
    ## ================== =========================
    yTypeInteger, yTypeFloat, yTypeFloatInf, yTypeFloatNaN, yTypeBoolTrue,
    yTypeBoolFalse, yTypeNull, yTypeUnknown, yTypeTimestamp

  YamlTypeHintState = enum
    ythInitial,
    ythF, ythFA, ythFAL, ythFALS, ythFALSE,
    ythN, ythNU, ythNUL, ythNULL,
    ythT, ythTR, ythTRU, ythTRUE,

    ythPoint, ythPointI, ythPointIN, ythPointINF,
              ythPointN, ythPointNA, ythPointNAN,

    ythLowerF, ythLowerFA, ythLowerFAL, ythLowerFALS,
    ythLowerN, ythLowerNU, ythLowerNUL,
    ythLowerT, ythLowerTR, ythLowerTRU,

    ythPointLowerI, ythPointLowerIN,
    ythPointLowerN, ythPointLowerNA,

    ythMinus, ythPlus, ythInt1, ythInt2, ythInt3, ythInt4, ythInt,
    ythDecimal, ythNumE, ythNumEPlusMinus, ythExponent,

    ythYearMinus, ythMonth1, ythMonth2, ythMonthMinus, ythMonthMinusNoYmd,
    ythDay1, ythDay1NoYmd, ythDay2, ythDay2NoYmd,
    ythAfterDayT, ythAfterDaySpace, ythHour1, ythHour2, ythHourColon,
    ythMinute1, ythMinute2, ythMinuteColon, ythSecond1, ythSecond2, ythFraction,
    ythAfterTimeSpace, ythAfterTimeZ, ythAfterTimePlusMinus, ythTzHour1,
    ythTzHour2, ythTzHourColon, ythTzMinute1, ythTzMinute2

macro typeHintStateMachine(c: untyped, content: varargs[untyped]) =
  yAssert content.kind == nnkArgList
  result = newNimNode(nnkCaseStmt, content).add(copyNimNode(c))
  for branch in content.children:
    yAssert branch.kind == nnkOfBranch
    var
      charBranch = newNimNode(nnkOfBranch, branch)
      i = 0
      stateBranches = newNimNode(nnkCaseStmt, branch).add(
          newIdentNode("typeHintState"))
    while branch[i].kind != nnkStmtList:
      charBranch.add(copyNimTree(branch[i]))
      inc(i)
    for rule in branch[i].children:
      yAssert rule.kind == nnkInfix
      yAssert rule[0].strVal == "=>"
      var stateBranch = newNimNode(nnkOfBranch, rule)
      case rule[1].kind
      of nnkBracket:
        for item in rule[1].children: stateBranch.add(item)
      of nnkIdent: stateBranch.add(rule[1])
      else: internalError("Invalid rule kind: " & $rule[1].kind)
      if rule[2].kind == nnkNilLit:
        stateBranch.add(newStmtList(newNimNode(nnkDiscardStmt).add(
                        newEmptyNode())))
      else:
        stateBranch.add(newStmtList(newAssignment(
                        newIdentNode("typeHintState"), copyNimTree(rule[2]))))
      stateBranches.add(stateBranch)
    stateBranches.add(newNimNode(nnkElse).add(newStmtList(
        newNimNode(nnkReturnStmt).add(newIdentNode("yTypeUnknown")))))
    charBranch.add(newStmtList(stateBranches))
    result.add(charBranch)
  result.add(newNimNode(nnkElse).add(newStmtList(
             newNimNode(nnkReturnStmt).add(newIdentNode("yTypeUnknown")))))

template advanceTypeHint(ch: char) {.dirty.} =
  typeHintStateMachine ch:
  of '~': ythInitial => ythNULL
  of '.':
    [ythInt1, ythInt2, ythInt3, ythInt4, ythInt] => ythDecimal
    [ythInitial, ythMinus, ythPlus] => ythPoint
    ythSecond2                      => ythFraction
  of '+':
    ythInitial                => ythPlus
    ythNumE                   => ythNumEPlusMinus
    [ythFraction, ythSecond2] => ythAfterTimePlusMinus
  of '-':
    ythInitial                => ythMinus
    ythNumE                   => ythNumEPlusMinus
    ythInt4                   => ythYearMinus
    ythMonth1                 => ythMonthMinusNoYmd
    ythMonth2                 => ythMonthMinus
    [ythFraction, ythSecond2] => ythAfterTimePlusMinus
  of '_':
    [ythInt1, ythInt2, ythInt3, ythInt4] => ythInt
    [ythInt, ythDecimal]      => nil
  of ':':
    [ythHour1, ythHour2]      => ythHourColon
    ythMinute2                => ythMinuteColon
    [ythTzHour1, ythTzHour2]  => ythTzHourColon
  of '0'..'9':
    ythInitial                        => ythInt1
    ythInt1                           => ythInt2
    ythInt2                           => ythInt3
    ythInt3                           => ythInt4
    [ythInt4, ythMinus, ythPlus]      => ythInt
    [ythNumE, ythNumEPlusMinus]       => ythExponent
    ythYearMinus                      => ythMonth1
    ythMonth1                         => ythMonth2
    ythMonthMinus                     => ythDay1
    ythMonthMinusNoYmd                => ythDay1NoYmd
    ythDay1                           => ythDay2
    ythDay1NoYmd                      => ythDay2NoYmd
    [ythAfterDaySpace, ythAfterDayT]  => ythHour1
    ythHour1                          => ythHour2
    ythHourColon                      => ythMinute1
    ythMinute1                        => ythMinute2
    ythMinuteColon                    => ythSecond1
    ythSecond1                        => ythSecond2
    ythAfterTimePlusMinus             => ythTzHour1
    ythTzHour1                        => ythTzHour2
    ythTzHourColon                    => ythTzMinute1
    ythTzMinute1                      => ythTzMinute2
    ythPoint                          => ythDecimal
    [ythInt, ythDecimal, ythExponent, ythFraction] => nil
  of 'a':
    [ythF, ythLowerF] => ythLowerFA
    ythPointN         => ythPointNA
    ythPointLowerN    => ythPointLowerNA
  of 'A':
    ythF      => ythFA
    ythPointN => ythPointNA
  of 'e':
    [ythInt, ythDecimal,
      ythInt1, ythInt2, ythInt3, ythInt4] => ythNumE
    ythLowerFALS => ythFALSE
    ythLowerTRU  => ythTRUE
  of 'E':
    [ythInt, ythDecimal,
      ythInt1, ythInt2, ythInt3, ythInt4] => ythNumE
    ythFALS => ythFALSE
    ythTRU  => ythTRUE
  of 'f':
    ythInitial      => ythLowerF
    ythPointLowerIN => ythPointINF
  of 'F':
    ythInitial => ythF
    ythPointIN => ythPointINF
  of 'i': ythPoint => ythPointLowerI
  of 'I': ythPoint => ythPointI
  of 'l':
    ythLowerNU  => ythLowerNUL
    ythLowerNUL => ythNULL
    ythLowerFA  => ythLowerFAL
  of 'L':
    ythNU  => ythNUL
    ythNUL => ythNULL
    ythFA  => ythFAL
  of 'n':
    ythInitial      => ythLowerN
    ythPoint        => ythPointLowerN
    [ythPointI, ythPointLowerI] => ythPointLowerIN
    ythPointLowerNA => ythPointNAN
  of 'N':
    ythInitial => ythN
    ythPoint   => ythPointN
    ythPointI  => ythPointIN
    ythPointNA => ythPointNAN
  of 'r': [ythT, ythLowerT] => ythLowerTR
  of 'R': ythT => ythTR
  of 's':
    ythLowerFAL => ythLowerFALS
  of 'S':
    ythFAL => ythFALS
  of 't':
    ythInitial         => ythLowerT
    [ythDay1, ythDay2, ythDay1NoYmd, ythDay2NoYmd] => ythAfterDayT
  of 'T':
    ythInitial         => ythT
    [ythDay1, ythDay2, ythDay1NoYmd, ythDay2NoYmd] => ythAfterDayT
  of 'u':
    [ythN, ythLowerN]  => ythLowerNU
    ythLowerTR         => ythLowerTRU
  of 'U':
    ythN  => ythNU
    ythTR => ythTRU
  of 'Z': [ythSecond2, ythFraction, ythAfterTimeSpace] => ythAfterTimeZ
  of ' ', '\t':
    [ythSecond2, ythFraction] => ythAfterTimeSpace
    [ythDay1, ythDay2, ythDay1NoYmd, ythDay2NoYmd] => ythAfterDaySpace
    [ythAfterTimeSpace, ythAfterDaySpace] => nil

proc guessType*(scalar: string): TypeHint {.raises: [].} =
  ## Parse scalar string according to the RegEx table documented at
  ## `TypeHint <#TypeHind>`_.
  var typeHintState: YamlTypeHintState = ythInitial
  for c in scalar: advanceTypeHint(c)
  case typeHintState
  of ythNULL, ythInitial: result = yTypeNull
  of ythTRUE: result = yTypeBoolTrue
  of ythFALSE: result = yTypeBoolFalse
  of ythInt1, ythInt2, ythInt3, ythInt4, ythInt: result = yTypeInteger
  of ythDecimal, ythExponent: result = yTypeFloat
  of ythPointINF: result = yTypeFloatInf
  of ythPointNAN: result = yTypeFloatNaN
  of ythDay2, ythSecond2, ythFraction, ythAfterTimeZ, ythTzHour1, ythTzHour2,
     ythTzMinute1, ythTzMinute2: result = yTypeTimestamp
  else: result = yTypeUnknown