Sinclair BASIC tokenized file

From Just Solve the File Format Problem
Revision as of 14:23, 10 February 2020 by Shram86 (Talk | contribs)

Jump to: navigation, search
File Format
Name Sinclair BASIC tokenized file
Ontology

Sinclair BASIC is a dialect of the BASIC programming language created by Nine Tiles Networks Ltd and used in the 8-bit home computers from Sinclair Research and Timex Sinclair.

The original 4KB version was developed for the Sinclair ZX80, followed by an 8KB version for the ZX81 and 16KB version for ZX Spectrum.

Some unusual features of the Sinclair BASIC:

  • There were keys on the keyboard for each BASIC keyword. For example, pressing P caused the entire command PRINT to appear. Some commands needed multiple keypresses to enter, For example, BEEP was keyed by pressing CAPS SHIFT plus SYMBOL SHIFT to access extended mode, keeping SYMBOL SHIFT held down and pressing Z.
  • When programs were SAVEd, the file written to disk or tape contained all of BASIC's internal state information, including the values of any defined basic variables, as well as the BASIC tokens.

Contents

BASIC File Layout

On a ZX81, a saved BASIC file is a snapshot of the computer memory from memory location 16393 through to the end of the variable table. There is no header.

Address Name Description
16393 VERSN 0 Identifies ZX81 BASIC in saved programs.
16394 E_PPC Number of current line (with program cursor).
16396 D_FILE Pointer to the start of the 'Display file', i.e. what is being displayed on screen
16398 DF_CC Address of PRINT position in display file. Can be poked so that PRINT output is sent elsewhere.
16400 VARS Pointer to start of BASIC Variable table
16402 DEST Address of variable in assignment.
16404 E_LINE Pointer to line currently being entered
16406 CH_ADD Address of the next character to be interpreted: the character after the argument of PEEK, or the NEWLINE at the end of a POKE statement.
16408 X_PTR Address of the character preceding the marker.
16410 STKBOT pointer to start (bottom) of stack
16412 STKEND pointer to end (top) of stack
16414 BERG Calculator's b register.
16415 MEM Address of area used for calculator's memory. (Usually MEMBOT, but not always.)
16417 not used
16418 DF_SZ The number of lines (including one blank line) in the lower part of the screen.
16419 S_TOP The number of the top program line in automatic listings.
16421 LAST_K Shows which keys pressed.
16423 Debounce status of keyboard.
16424 MARGIN Number of blank lines above or below picture: 55 in Britain, 31 in America.
16425 NXTLIN Address of next program line to be executed.
16427 OLDPPC Line number of which CONT jumps.
16429 FLAGX Various flags.
16430 STRLEN Length of string type destination in assignment.
16432 T_ADDR Address of next item in syntax table (very unlikely to be useful).
16434 SEED The seed for RND. This is the variable that is set by RAND.
16436 FRAMES Counts the frames displayed on the television. Bit 15 is 1. Bits 0 to 14 are decremented for each frame set to the television. This can be used for timing, but PAUSE also uses it. PAUSE resets to 0 bit 15, & puts in bits 0 to 14 the length of the pause. When these have been counted down to zero, the pause stops. If the pause stops because of a key depression, bit 15 is set to 1 again.
16438 COORDS x-coordinate of last point PLOTted.
16439 y-coordinate of last point PLOTted.
16440 PR_CC Less significant byte of address of next position for LPRINT to print as (in PRBUFF).
16441 S_POSN Column number for PRINT position.
16442 Line number for PRINT position.
16443 CDFLAG Various flags. Bit 7 is on (1) during compute & display mode.
16444 PRBUFF Printer buffer (33rd character is NEWLINE).
16477 MEMBOT Calculator's memory area; used to store numbers that cannot conveniently be put on the calculator stack.
16507 not used
16509 First BASIC line.

BASIC lines

Each BASIC line is stored as:

  • 2 byte line number (in big-endian format)
  • 2 byte length of text including NEWLINE (in little endian format, length "excludes" the line number and length, i.e. to skip between lines you add "length of text" +4 bytes.
  • text (BASIC tokens)
  • NEWLINE (0x76 on ZX80/81, 0x0D on Spectrum)

When a numeric constant is included in the text of a BASIC line, an ASCII string displaying the constant value will be inserted, followed by one of two tokens (indicating floating-point or integral) and then the number in 5-byte numeric format.

BASIC Variables Table

Following the last BASIC line comes the variables table. Each entry in this table is of varying length and format.

The first byte in each entry is the variable name, of which the upper 3 bits indicate the variable type.

Most types of variables can only have a one-character name: A to Z for numerics, A$ to Z$ for strings. Numeric variables can have a multi-character names beginning with A-Z and continuing with A-Z or 0-9, e.g. F0O or BAR. Names are case-insensitive and whitespace insensitive (hello world is the same variable as HeLlOwOrL d). Numeric variables and FOR-NEXT control variables share the same namespace, but no other types do. Numeric variable A, string variable A$, numeric array A(10) and string array A$(10) can all coexist under the name "A"[1]

First byte Format Examples
011nnnnn numeric variable
  • 1 byte: variable name (0x61 to 0x7A)
  • 5 bytes: numeric value
  • 61 00 00 01 00 00LET a=1
  • 69 00 FF 02 00 00LET i=-65534
  • 6A 00 00 FE FF 00LET j=65534
  • 62 93 3C 61 4E 00LET b=12345678
    • 12345678 = 0xBC614E
  • 63 93 BC 61 4E 00LET c=-12345678
  • 65 82 2D F8 4C ADLET e=2.71828
101nnnnn numeric variable with multi-character name
  • 1 byte: first character of variable name (0xA1 to 0xBA)
  • n bytes: remainder of name (A-Z, a-z, 0-9 allowed, final character has high bit set)
  • 5 bytes: numeric value
  • B0 69 E5 82 49 4A F0 41LET pie=3.1452
    • "p"=0x10 ORed with 0xA0
    • "i" is regular ASCII (0x69)
    • "e" is regular ASCII (0x65) ORed with 0x80 to indicate the end of the variable name
010nnnnn string variable
  • 1 byte: variable name (0x41 to 0x5A)
  • 2 bytes: little-endian string length (n)
  • n bytes: string value (the character set is mostly ASCII)[2]
  • 41 04 00 54 45 53 54LET a$="TEST"
100nnnnn numeric array
  • 1 byte: array name (0x81 to 0x9A)
  • 2 bytes: little-endian size of data to follow in bytes, so you can easily skip to the next variable without computing full array size
  • 1 byte: number of dimensions in the array (1-255?)
  • 2 bytes: little-endian valid range of first dimension e.g. 0A 00 is valid range 1-10
  • ... 2 bytes more for every other dimension to declare the rest of the dimension ranges
  • ... array values (each a 5-byte number) in C-style order, iterating rightmost index first and the leftmost index last, e.g. DIM a(2,3,4) will store values in this order: a(1,1,1), a(1,1,2), a(1,1,3), (1,1,4), a(1,2,1), a(1,2,2), a(1,2,3), a(1,2,4), a(1,3,1), a(1,3,2), a(1,3,3), a(1,3,4), a(2,1,1), ...
  • 81 1C 00 01 05 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00DIM a(5)
  • 81 19 00 02 02 00 02 00 00 00 01 00 00 00 00 02 00 00 00 00 03 00 00 00 00 04 00 00DIM a(2,2); LET a(1,1)=1; LET a(1,2)=2; LET a(2,1)=3; LET a(2,2)=4
110nnnnn character array
  • 1 byte: array name (0xC1 to 0xDA)
  • 2 bytes: little-endian size of data to follow in bytes, so you can easily skip to the next variable without computing full array size
  • 1 byte: number of dimensions in the array (1-255?)
  • 2 bytes: little-endian valid range of first dimension e.g. 0A 00 is valid range 1-10
  • ... 2 bytes more for every other dimension to declare the rest of the dimension ranges
  • ... array values (each a 1-byte character) in C-style order, iterating rightmost index first and the leftmost index last

A single-dimensional character array acts like a string, but has a fixed length. If it is set to a shorter string, the remaining space in the array will be padded with spaces (0x20). In general, an n dimensional character array where n>1 acts like an n-1 dimensional string array, e.g. DIM a$(5,10); LET a$(1)="FOO"; LET a$(2)="BAR" is valid, and sets a$(1,1) to a$(1,10) to "FOO       ", and sets a$(2,1) to a$(2,10) to "BAR       "

  • C8 12 00 01 0F 00 68 65 6C 6C 6F 2C 20 77 6F 72 6C 64 20 20 20DIM h$(15); LET h$="hello, world"
  • D7 0F 00 02 02 00 05 00 68 65 6C 6C 6F 77 6F 72 6C 64DIM w$(2,5); LET w$(1)="hello"; LET w$(2)="world"
111nnnnnn control variable of a FOR-NEXT loop.
  • 1 byte: variable name (0xE1 to 0xFA)
  • 5 bytes: current value (number)
  • 5 bytes: loop end limit (number)
  • 5 bytes: loop step increment (number)
  • 2 bytes: little-endian line number where the FOR loop was started
  • 1 byte: statement number within line where the FOR loop was started
  • E9 00 00 01 00 00 00 00 05 00 00 00 00 01 00 00 FF FE 02FOR i=1 TO 5 (before loop has run)
  • E9 00 00 06 00 00 00 00 05 00 00 00 00 01 00 00 FF FE 02FOR i=1 TO 5 (after loop has run)
  • E9 00 00 01 00 00 00 00 0A 00 00 00 00 02 00 00 FF FE 02FOR i=1 TO 10 STEP 2
  • E9 81 0C CC CC CD 83 76 66 66 66 7D 4C CC CC CC FE FF 02FOR i=1.1 TO 7.7 STEP .1

5-byte numeric format

Numbers have one of two formats:[3] integers between -65535 and +65535 (inclusive) are in an "integral" format, while all other numbers are in "floating point" format. The value 0 can't be represented by the floating point format, so is always stored as an integral. The token Ox7E before the number indicates floating-point, and 0x0E indicates an integral number.

With "integral" format:

  • 1 byte: always 0
  • 1 byte: 0 if the number is positive or -1 (0xFF) if the number is negative
  • 2 bytes: little-endian unsigned integer from 0 to 65535. Subtract 65536 from this if number is flagged as negative
  • 1 byte: always 0

With "floating point" format:

  • 1 byte: exponent + 128 (0 → e=-128, 255 → e=127)
  • 4 bytes: big-endian mantissa

The number has to be normalised so that its most significant mantissa bit is always 1. This bit is then assumed to be 1 and is overwritten with a sign bit: cleared to 0 for positive numbers and set to 1 for negative numbers

Floating point examples
Representation Value
7F 7F FF FF FF 0.5
7E 7F FF FF FF 0.25
7D 7F FF FF FF 0.125
91 00 00 00 00 65536
A0 07 65 43 21 2271560481 (0x87654321)
7F FF FF FF FF -0.5
7E FF FF FF FF -0.25
7D FF FF FF FF -0.125
91 80 00 00 00 -65536
A0 87 65 43 21 -2271560481 (-0x87654321)
Integral examples
Representation Value
00 FF 01 00 00 -65535
00 FF 02 00 00 -65534
00 FF FE FF 00 -2
00 FF FF FF 00 -1
00 00 00 00 00 0
00 00 01 00 00 1
00 00 02 00 00 2
00 00 FE FF 00 65534
00 00 FF FF 00 65535

ZX81 BASIC Tokens

Token (Decimal) Description
0
11 "
12 £
13 $
14 :
15 ?
16 (
17 )
18 >
19 <
20 =
21 +
22 -
23 *
24 /
25 ;
26 ,
27 .
28 0
29 1
30 2
31 3
32 4
33 5
34 6
35 7
36 8
37 9
38 A
39 B
40 C
41 D
42 E
43 F
44 G
45 H
46 I
47 J
48 K
49 L
50 M
51 N
52 O
53 P
54 Q
55 R
56 S
57 T
58 U
59 V
60 W
61 X
62 Y
63 Z
64 RND
65 INKEY$
66 PI
112 <cursor up>
113 <cursor down>
114 <cursor left>
115 <cursor right>
116 GRAPHICS
117 EDIT
118 NEWLINE
119 RUBOUT
120 /
121 FUNCTION
127 cursor
128
139 "
140 £
141 $
142 :
143 ?
144 (
145 )
146 >
147 <
148 =
149 +
150 -
151 *
152 /
153 ;
154 -
155 .
156 0
157 1
158 2
159 3
160 4
161 5
162 6
163 7
164 8
165 9
166 A
167 B
168 C
169 D
170 E
171 F
172 G
173 H
174 I
175 J
176 K
177 L
178 M
179 N
180 O
181 P
182 Q
183 R
184 S
185 T
186 U
187 V
188 W
189 X
190 Y
191 Z
192 ""
193 AT
194 TAB
195 ?
196 CODE
197 VAL
198 LEN
199 SIN
200 COS
201 TAN
202 ASN
203 ACS
204 ATN
205 LN
206 EXP
207 INT
208 SQR
209 SGN
210 ABS
211 PEEK
212 USR
213 STR$
214 CHR$
215 NOT
216 **
217 OR
218 AND
219 <=
220 >=
221 <>
222 THEN
223 TO
224 STEP
225 LPRINT
226 LLIST
227 STOP
228 SLOW
229 FAST
230 NEW
231 SCROLL
232 CONT
233 DIM
234 REM
235 FOR
236 GOTO
237 GOSUB
238 INPUT
239 LOAD
240 LIST
241 LET
242 PAUSE
243 NEXT
244 POKE
245 PRINT
246 PLOT
247 RUN
248 SAVE
249 RAND
250 IF
251 CLS
252 UNPLOT
253 CLEAR
254 RETURN
255 COPY

Links and references

  1. http://www.worldofspectrum.org/ZXBasicManual/zxmanchap7.html
  2. http://www.worldofspectrum.org/ZXBasicManual/zxmanappa.html
  3. http://www.worldofspectrum.org/ZXBasicManual/zxmanchap24.html
Personal tools
Namespaces

Variants
Actions
Navigation
Toolbox