BBC BASIC program line format
                      =============================

BBC BASIC programs are usually stored in a tokenised format. Most of the
tokens used by different versions of BBC BASIC are the same, but extended
tokens are different, and different versions have different line headers. 

The differences are between:
* Acorn/Wilson BASIC descended from the 6502 version, the 32000, ARM, PDP11,
  6809 and others
* Russell BASIC descended from the Z80 version, 80x86, Windows, SDL.

Acorn/Wilson format BASIC is stored as:
  <cr> {<linehi><linelo><len><text><cr>...} <ff>

Russell format BASIC is stored as:
       {<len><linelo><linehi><text><cr>...} <00><ff><ff>

BASIC can also be stored as text, as:
  {<text> [<cr>|<lf>|<cr><lf>|<lf><cr>]...}

Tokenised line numbers are stored in the same format in all BASICs. They are
encoded to make them easily found by RENUMBER, and to prevent them appearing
to be control characters or tokens by forcing them into the range &40-&7F.

  Byte 0 : 141
  Byte 1 :(((lsb AND &C0) DIV 4) OR ((msb AND &C0) DIV 16)) EOR &54
  Byte 2 :  (lsb AND &3F) OR &40
  Byte 3 :  (msb AND &3F) OR &40

6502 BASIC line numbers can go to a maximum of 32767 (&7FFF), other BASICs
can use the full range to 65279 (&FEFF).

It can be useful to be able to determine what format a BASIC file is stored
in. This can be done by examining the last few bytes of the program:

  <cr><00><ff><ff> - Russell format (Z80, 80x86, SDL)
  <xx><xx><cr><ff> - Wilson/Acorn format (6502, 32000, ARM, PDP11, 6809)
  <xx><xx><cr><lf> - text CR/LF terminated
  <xx><xx><lf><cr> - text LF/CR terminated
  <xx><xx><xx><lf> - text LF terminated (Brandy)
  <xx><xx><xx><cr> - text CR terminated
  <xx><xx><xx><xx> - unrecognised

A running program can tell what format the program is in by examining TOP-3:

  IF ?(TOP-3)=0 THEN Russell Format
  IF ?(TOP-3)>0 THEN Acorn format

The following code will examine the last four bytes of an open file and
determine what format it is:

    PTR#in%=EXT#in%-4:FOR A%=0 TO 3:buffer%?A%=BGET#in%:NEXT
    type%=0                                       :REM unknown
    IFbuffer%?3=&0D                      :type%=7 :REM text cr
    IFbuffer%?3=&0A                      :type%=6 :REM text lf
    IF(!buffer% AND &FFFF0000)=&0D0A0000 :type%=5 :REM text lfcr
    IF(!buffer% AND &FFFF0000)=&0A0D0000 :type%=4 :REM text crlf
    IF(!buffer% AND &FFFF0000)=&FF0D0000 :type%=2 :REM 6502 format
    IF!buffer%=&FFFF000D                 :type%=1 :REM 80/86 format

Here in% is the handle of the file that has been opened, buffer% is a
pointer to a four byte DIMensioned block of memory, A% is a temporary
variable and on exit type% contains the file type.

Note that in some circumstances it is legitimate to append data to the end
of a BBC BASIC program file. In that case this method of determining the
format will not work, since the last four bytes of the file will not be the
last four bytes of the program.