In this programming pearl Albert Nijhof shows how to extract strings form a sequence of strings.
The idea here is to represent the sequence of strings as a single string.
1 \ Day of the week -- an-10jan2022 2 \ From number to name, not using the case statement 3 4 \ 0 1 2 3 4 5 6 5 \ sun mon tue wed thu fri sat 6 7 \ ===== For constant string length ===== 8 : .DAY ( n -- ) \ n in [0,6] 9 7 umin 10 s" SunMonTueWedThuFriSat???" drop 11 swap 3 * + 3 type space ; 12 13 \ 3 .day -> wed 14 \ 8 .day -> ??? 15 16 \ ===== For variable string length ===== 17 : .DAY ( n -- ) \ n in [0,6] 18 7 umin 19 s" 3Sun 3Mon 4Tues 6Wednes 5Thurs 3Fri 5Satur 1?" drop 20 swap 0 21 ?do begin count bl = until 22 loop count [char] 0 - type 23 ." day " ; 24 25 \ 3 .day -> Wednesday 26 \ 9 .day -> ?day 27 28 \ ===== For longer strings ===== ( :septiembre ) 29 : M, ( adr len -- ) 0 ?do count c, loop drop ; 30 decimal create (MESES) 31 char " parse 5enero 7febrero 5marzo 5abril 4mayo 5junio 5julio " m, 32 char " parse 6agosto :septiembre 7octubre 9noviembre 9diciembre 1?" m, 33 align 34 : .MES ( n -- ) \ n in [1,12] 35 1- 12 umin (meses) swap 0 36 ?do begin count bl = until 37 loop count [char] 0 - type space ; 38 39 \ 9 .mes -> septiembre 40 \ 0 .mes -> ? 41 42 \ <><>
First let's assume all element strings of the sequence have the same length.
Then the start address of the string n in the sequence can be calculated by
«address of string n» = «start-address of sequence» + n * «fixed length of element strings»
In Forth that looks like the definition of
.DAY in line 8-11. The fixed length here is 3. The sequence has 8 elements.
7 umin (line 9) is a bounds check, that maps every value outside the range 0 to 7 to the value 7 (negative numbers considered as unsigned numbers are greater than 7 thus will map to 7).
S" DROP (line 10) puts
c-addr on the stack, the start address of the string. Line 11 then calculates the address of the substring, provides the length (3), and displays it.
If the element strings in the sequence are not all of the same length, then their individual lengths need to be stored as well. The lookup then has to traverse the sequence on a string by string basis to find the nth string.
The definition of
.DAY in line 17 to 23 shows how to encode the length of the element strings and traverse the sequence.
The range check in line 18 is as described above.
The represented sequence (line 19) now has a length byte in front of each element string. Character '1' represents the length 1, '2' the length 2 and so on. Lengths greater 9 would be represented by characters :(10) ;(11) <(12) =(13) >(14) ?(15) @(16) etc. according to the ASCII character encoding but that is not necessary in this example.
Line 21 ?terates over the sequence to the nth element string. It assumes that not only is their individual length embedded but also that they are separated by a single space character: The
BEGIN count bl = UNTIL loop skips over the characters until a space is encountered.
COUNT ( c-addr1 -- c-addr2 u ) extracts the next character. That functionality is sometimes called
c@+ and can be defined as
: c@+ ( c-addr1 -- c-addr2 u ) DUP 1 + SWAP c@ ; assuming a byte addressed machine (i.e. 1 CHARS = 1).
?DO LOOP iterates n times so it leaves the address of the nth element string. Memory at that address holds the encoded length that is fetched (line 22,
COUNT) and transformed to the actual length by subtracting the character value of '0'. Having address and length of the element string
type displays it. The trailing „day “ line 23 completes the weekday (every weekday in english ends in „day“).
If n is out of range 0 to 6 then
.DAY works as follows:
n is mapped to 7 (line 18), the last element string is found („1?“, line 21), its length extracted (1) and „?“ displayed (line 22).
.DAY eventually prints „?day “.
In the english version displaying days of the week the encoded sequence (line 19) has a length of 45 characters. Short enough to be represented in a single line of source code.
If however the string to be stored in memory is larger than a single line, it might be better to construct it in a different way.
Line 30-33 show how to do it. They assume that a word
M, ( adr len --) (see comment in line 29) is available that lays down a string character by character at
HERE in the dictionary. It moves
HERE forward so that it continues to point to available dictionary space.
Line 30 gives the name
(MESES) to the sequence (start address of the string representing the sequence).
Line 31 and 32 use
M, to store two parts of the string. Both lines parse a quote („) terminated string and then lay it down in the dictionary. For systems that require aligned dictionary addresses the
ALIGN in line 33 pads the dictionary so
HERE will be a cell aligned address.
The definition of
.MES (line 34-37) is similar to
.DAY (line 17-23) only that indices n for months go from 1 to 12 (not 0 to 7 as before). Also Spanish month names do not have a common ending, so the element string is all that is displayed (no „day “).
It is possible to store sequences of strings as single strings in memory and it is easy to extract the individual element strings:
If all element strings have fixed length you can do address calculation, otherwise you can traverse these sequences and find the appropriate strings.
Lengths can be encoded as characters.
Longer strings can be constructed step by step in memory.