compiler.bas

'Macro for a temp string that's actually a global null string with the header
'modified to point ot the middle of an existing string.
#define SetConst(_sSrc,_iStart,_iLen) *cptr(fbstr ptr, @sGlobalTemp) = type( strptr(_sSrc)+(_iStart) , _iLen , _iLen )
Dim Shared as String sGlobalTemp
type fbstr
  pzData as zstring ptr
  iLen as integer
  iSize as integer
end type
#if 0
float MoveCount = 0
float MoveNum = 32
Start:
    If MoveCount = MoveNum Goto TurnRobot

    Add MoveCount, 1
    BIOS.Move(1)
    GOTO Start

TurnRobot:
    BIOS.Turn(1)
    Let MoveCount = 0
    Goto Start
#endif
#include once "windows.bi"
#include once "../include/opcode.bi"
#include once "HashTable.bi"
Declare Function ProcessToken() as integer
Declare Sub EndOfLine()
Dim Shared as String sSrcBuff
Dim as String sSourcefile = "test1.txt" 'Command
dim as integer iFileSz

If Open (sSourceFile for binary access read as #1) Then
    Print "Error opening source file!"
    Sleep: System
End If

iFileSz = LOF(1)
sSrcBuff = Space(iFileSz)
Get #1,, sSrcBuff
close #1

initHashTable

enum TokenTypes
  tkUnknown
  tkLabel
  tkVariable
  tkNamespace
  tkNumber
end enum

' State variabes... (if there's multiple threads, must be a UDT)
dim shared as integer iTokenBegin = -1, iTokenEnd = -1, iTokenType = tkUnknown
dim as integer iNumParseMode = 0
dim shared as integer iNumberParse, iDecimalParse
dim shared as integer iNumberDigits, iDecimalDigits
dim shared as uByte iTokenHash

color 14
print "iTokenBegin", "iTokenEnd", "difference", "sGlobalTemp"
print String(79, "-")
color 7

' =================================================
' ========= Parse File based on the state =========
' =================================================
for iPosi as integer = 0 to iFileSz-1

  var iChar = sSrcBuff[iPosi] 'maybe it wont be needed

  if iTokenType = tkNumber then 'chars that can end numbers...
    select case as const iChar
    case asc("0") to asc("9"),asc(".")
      rem ok so still a number...
    case 32,9,13,10 'Space/Tab/cr/lf (i guess all operators)
      iTokenType = tkNumber
      if ProcessToken()=0 then exit for
    case else
      'bad number...
    end select
  end if

  select case as const iChar
  case 32,9 'Space/Tab
    if iTokenBegin >= 0 orelse iTokenType<>tkUnknown then
      iTokenEnd = iPosi-1: if ProcessToken()=0 then exit for
    end if
    ' ignore blanks

  case asc("0") to asc("9")
    if iTokenBegin<0 then
      'not part of a token so must be a number/expression/error
      'if number parsing... then insert next digit into number
      iTokenType = tkNumber
      if iNumParseMode = 0 then 'whole number
        iNumberParse = (iNumberParse*10)+iChar-asc("0")
        iNumberDigits += 1
      else
        iDecimalParse = (iDecimalParse*10)+iChar-asc("0")
        iDecimalDigits += 1
      end if
    else
      iTokenHash += iChar 'add to the hash of the token
    end if

  case asc("A") to asc("Z")
    iChar += (asc("a")-asc("A"))  'lowercase convert
    sSrcBuff[iPosi] = iChar  'and update on string?
    if iTokenBegin<0 then 'wasnt reading a token
      iTokenBegin = iPosi 'so now token start here
    end if
    iTokenHash += iChar 'add to the hash of the token

  case asc("a") to asc("z"), asc("_")
    iTokenHash += iChar 'add to the hash of the token

  case 13                    'possible newline?
    iTokenEnd = iPosi-1
    if sSrcBuff[iPosi+1] = 10 then
      iPosi += 1 'newline so skipping the LF too
      EndOfLine()
    end if
    if ProcessToken()=0 then exit for

  case 10                    'newline!
    iTokenEnd = iPosi-1
    EndOfLine()
    if ProcessToken()=0 then exit for

  case asc(":")              'label?
    iTokenType = tkLabel
    iTokenEnd = iPosi-1: if ProcessToken()=0 then exit for

  case asc(".")              'decimal/namespace
    if iTokenBegin < 0 then 'number :)
      if iNumParseMode then 'already decimal or something?
        'Invalid number? @_@
      else
        iNumParseMode = 1 'decimal
        iTokenType = tkNumber
      end if
    else
      iTokenType = tkNamespace
      iTokenEnd = iPosi-1: if ProcessToken()=0 then exit for
    end if

  case else
    'probabily this will work as a warning to add new char
    'since this case never would be entered
    'unless there is a a syntax error
    iTokenEnd = iPosi-1: if ProcessToken()=0 then exit for
    'account for unexpected characters?
  end select

next iPosi

Sleep

Function ProcessToken() as integer
    'if iTokenBegin = -1 orElse iTokenEnd-iTokenBegin = -2 Then return 1

    var pTokenBegin = StrPtr(sSrcBuff)+iTokenBegin
    var pTokenEnd = pTokenBegin+iTokenEnd

    SetConst(sSrcBuff, iTokenBegin, iTokenEnd-iTokenBegin+1)
    print iTokenBegin, iTokenEnd, iTokenEnd-iTokenBegin, sGlobalTemp
    'print iTokenBegin, iTokenEnd, iTokenType, Hex(iTokenHash)
    'Search for hash in hashtable of keywords
    for N as integer = 0 to MAX_CONFLICTS
        var iOffset = HashTable(iTokenHash, N)
        if iOffset = INVALID_OFFSET_VALUE then exit for 'not found in table
        if memcmp(pTokenBegin, @Keywords(iOffset), (pTokenEnd-pTokenBegin)+1)=0 then
            print "match found!": exit for 'Found a match!
        end if
    next N

    iTokenBegin = -1: iTokenEnd = -1: iTokenType = tkUnknown: iTokenHash = 0
    return 42
end function

sub EndOfLine()
    'Print "Hello World"
end sub