Split and join strings

[luke]

Given a string of words separated by spaces (or any other character), splits it into an array of the words. I've no doubt many people have written a version of this over the years and no doubt there's a million ways to do it, but I thought I'd put mine here so we have at least one version. There's also a join function that does the opposite array -> single string.

Code is hopefully reasonably self explanatory with comments and a little demo. Note, this is akin to Python/JavaScript split/join, PHP explode/implode.

Code: QB64: [Select]

1. redim words$(0)
2.  
3. original$ = "The rain   in Spain  "
4. print "Original string: "; original$
5. print
6.  
7. split original$, " ", words$()
8.  
9. print "Words:"
10. for i = lbound(words$) to ubound(words$)
11.     print words$(i)
12. next i
13. print
14.  
15. print "Joined with commas: ";join$(words$(), ",")
16.  
17. 'Split in$ into pieces, chopping at every occurrence of delimiter$. Multiple consecutive occurrences
18. 'of delimiter$ are treated as a single instance. The chopped pieces are stored in result$().
19. '
20. 'delimiter$ must be one character long.
21. 'result$() must have been REDIMmed previously.
22. sub split(in$, delimiter$, result$())
23.     redim result$(-1)
24.     start = 1
25.     do
26.         while mid$(in$, start, 1) = delimiter$
27.             start = start + 1
28.             if start > len(in$) then exit sub
29.         wend
30.         finish = instr(start, in$, delimiter$)
31.         if finish = 0 then finish = len(in$) + 1
32.         redim _preserve result$(0 to ubound(result$) + 1)
33.         result$(ubound(result$)) = mid$(in$, start, finish - start)
34.         start = finish + 1
35.     loop while start <= len(in$)
36. end sub
37.  
38. 'Combine all elements of in$() into a single string with delimiter$ separating the elements.
39. function join$(in$(), delimiter$)
40.     result$ = in$(lbound(in$))
41.     for i = lbound(in$) + 1 to ubound(in$)
42.         result$ = result$ + delimiter$ + in$(i)
43.     next i
44.     join$ = result$
45. end function

[RhoSigma]

Here are my two cents,

words/components can be seperated by arbitrary number of whitespace (TAB/SPACE) and "quoted" sections are taken as one word/component. It also is satisfied with any defined dynamic array, as it REDIMs it as needed.

EDIT:
Old code removed, goto https://qb64forum.alephc.xyz/index.php?topic=4142 for the latest update.

[bplus]

And my 2 cents:
My Split sub handles more than one char delimiters, comes in handy for Splitting a file into an array of lines with the delimiter set as Chr($13)+chr$(10) for a .bas for instance. It also handles the space delimiter specially by removing all double spaces before splitting.

Oh, also, I like to use long strings for arrays. For that you need to be able to leave empty spots open for strings eg see the days of week example.

Code: QB64: [Select]

1. 'split test.bas for qb64 bplus 2018-05-07
2. ' I think I want to replace my inefficient Wrd function
3.  
4. '2018-08-25 reworked for space delimiters and more variable declares
5. '2019-02-15 add Luke's version to compare
6. ntests = 5
7. DIM a(ntests - 1) AS STRING, d(ntests - 1) AS STRING
8.  
9. a(0) = ""
10. d(0) = " "
11. a(1) = " test test    test " 'good no error!
12. d(1) = " "
13. a(2) = " test"
14. d(2) = " "
15. a(3) = "3d,z6d,z1 10 #d,z5"
16. d(3) = ",z"
17. a(4) = "Monday, , Wednesday, THursday, Friday, , Sunday"
18. d(4) = ", "
19.  
20. FOR test = 0 TO ntests - 1
21.     PRINT: PRINT "splitting {"; a(test); "} with delimeter {"; d(test); "}"
22.     REDIM myarr(0) AS STRING '<<<<< REDIM forces the creation of a dynamic/resizable array
23.     Split a(test), d(test), myarr()
24.     amax = UBOUND(myarr)
25.     FOR i = 0 TO amax
26.         PRINT i; ":"; myarr(i)
27.     NEXT i
28.     INPUT "press enter for next test... "; wate$
29. NEXT
30.  
31. ' how about a quick file reader test?
32. PRINT: INPUT "Press enter for file test, any other + enter quits! "; wate$
33. IF LEN(wate$) THEN END
34. CLS
35.  
36. 'other wise continue
37. OPEN "Split test.bas" FOR BINARY AS #1 '<<< this file name!!!
38. ftext$ = SPACE$(LOF(1))
39. GET #1, , ftext$
40. CLOSE #1
41. Split ftext$, CHR$(13) + CHR$(10), myarr()
42. FOR i = 0 TO UBOUND(myarr)
43.     PRINT myarr(i)
44.     IF i MOD 20 = 19 THEN PRINT: INPUT "press enter for more "; wate$
45. NEXT
46. PRINT "the end"
47. END ' end program
48.  
49.  
50. 'the space delimiter is such a special case perhaps I should develope a single split for that alone?
51.  
52.  
53. 'notes: REDIM the array(0) to be loaded before calling Split '<<<<<<<<<<<<<<<<<<<<<<< IMPORTANT!!!!
54. SUB Split (mystr AS STRING, delim AS STRING, arr() AS STRING)
55.     ' bplus modifications of Galleon fix of Bulrush Split reply #13
56.     ' http://xmaxw.[abandoned, outdated and now likely malicious qb64 dot net website - don’t go there]/forum/index.php?topic=1612.0
57.     ' this sub further developed and tested here: \test\Strings\Split test.bas
58.     DIM copy AS STRING, p AS LONG, curpos AS LONG, arrpos AS LONG, dpos AS LONG
59.  
60.     copy = mystr 'make copy since we are messing with mystr when the delimiter is a space
61.  
62.     'special case if delim is space, probably want to remove all excess space
63.     IF delim = " " THEN
64.         copy = RTRIM$(LTRIM$(copy))
65.         p = INSTR(copy, "  ")
66.         WHILE p > 0
67.             copy = MID$(copy, 1, p - 1) + MID$(copy, p + 1)
68.             p = INSTR(copy, "  ")
69.         WEND
70.     END IF
71.     curpos = 1
72.     arrpos = 0
73.     dpos = INSTR(curpos, copy, delim)
74.     DO UNTIL dpos = 0
75.         arr(arrpos) = MID$(copy, curpos, dpos - curpos)
76.         arrpos = arrpos + 1
77.         REDIM _PRESERVE arr(arrpos + 1) AS STRING
78.         curpos = dpos + LEN(delim)
79.         dpos = INSTR(curpos, copy, delim)
80.     LOOP
81.     arr(arrpos) = MID$(copy, curpos)
82.     REDIM _PRESERVE arr(arrpos) AS STRING 'need this line? YES to get the ubound correct
83. END SUB
84.  
85.  
86. ' Luke 2019-02-15
87. 'Split in$ into pieces, chopping at every occurrence of delimiter$. Multiple consecutive occurrences
88. 'of delimiter$ are treated as a single instance. The chopped pieces are stored in result$().
89. '
90. 'delimiter$ must be one character long.
91. 'result$() must have been REDIMmed previously.
92. SUB Lsplit (in$, delimiter$, result$())
93.     REDIM result$(-1)
94.     start = 1
95.     DO
96.         WHILE MID$(in$, start, 1) = delimiter$
97.             start = start + 1
98.             IF start > LEN(in$) THEN EXIT SUB
99.         WEND
100.         finish = INSTR(start, in$, delimiter$)
101.         IF finish = 0 THEN finish = LEN(in$) + 1
102.         REDIM _PRESERVE result$(0 TO UBOUND(result$) + 1)
103.         result$(UBOUND(result$)) = MID$(in$, start, finish - start)
104.         start = finish + 1
105.     LOOP WHILE start <= LEN(in$)
106. END SUB
107.  
108. 'Combine all elements of in$() into a single string with delimiter$ separating the elements.
109. FUNCTION join$ (in$(), delimiter$)
110.     result$ = in$(LBOUND(in$))
111.     FOR i = LBOUND(in$) + 1 TO UBOUND(in$)
112.         result$ = result$ + delimiter$ + in$(i)
113.     NEXT i
114.     join$ = result$
115. END FUNCTION
116.  
117.  

I would like to test RhoSigma's but looks like a couple of specific delimiters are used.

I think I had a reason not to redim the arr inside the Split sub, can't recall it now. It does seem more convenient to handle it inside the sub.

Append: you only have to redim once outside the Split sub before the call to it, to let QB64 know you are using that name as a dynamic array.

[RhoSigma]

I would like to test RhoSigma's but looks like a couple of specific delimiters are used.

??? - I hope you doesn't speak about the CHR$(34) in the example lines, it's a quote (").

The ParseLine&() function is a rather simple (yet incomplete) command line parsing function, whitespace separates options/words, whitespace in this case means any number of TABs and/or SPACEs, and quoted parts are taken as is (ie. incl. whitespace) as you usually would need it on a command line for filenames which contain spaces or any chars with special meanings in the command interpreter suche as %()<>& etc..
My function does otherwise not allow for any user defined delimiters.

[SMcNeill]

Here's a split routine which I'll toss into the mess as well:

Code: QB64: [Select]

1. CONST ntests = 5
2. DIM a(ntests - 1) AS STRING, d(ntests - 1) AS STRING
3. REDIM results(0) AS STRING
4.  
5. a(0) = ""
6. d(0) = " "
7. a(1) = " test test    test " 'good no error!
8. d(1) = " "
9. a(2) = " test"
10. d(2) = " "
11. a(3) = "3d,z6d,z1 10 #d,z5"
12. d(3) = ",z"
13. a(4) = "Monday, , Wednesday, THursday, Friday, , Sunday"
14. d(4) = ", "
15.  
16. FOR i = 0 TO ntests - 1
17.     PRINT "Splitting: "; a(i)
18.     SteveSplit a(i), d(i), results()
19.     FOR j = 1 TO UBOUND(results)
20.         PRINT j, results(j)
21.     NEXT
22.     SLEEP
23. NEXT
24.  
25. SUB SteveSplit (text$, delimiter$, storage_array() AS STRING)
26.     STATIC count AS LONG
27.     count = count + 1
28.     u = UBOUND(storage_array)
29.     IF count > u THEN REDIM _PRESERVE storage_array(u + 1000) AS STRING
30.     i = INSTR(text$, delimiter$)
31.     IF i THEN
32.         storage_array(count) = LEFT$(text$, i - 1)
33.         SteveSplit MID$(text$, i + LEN(delimiter$)), delimiter$, storage_array()
34.     ELSE
35.         storage_array(count) = text$
36.         REDIM _PRESERVE storage_array(count) AS STRING
37.         count = 0
38.     END IF
39. END SUB

I was even nice and named it "SteveSplit" so folks can test for speed and compare results if they wish verses the other split routines.

One thing to note:  This *doesn't* strip off any extra leading/trailing spaces.  Why would it, if they're delimiters in our data?

Let's say we have the data of the following: 

1,2,,,5,6

When we INPUT it from a file, we get data of:
"1"
"2"
""
""
"5"
"6"

Those commas are valid delimiters of null strings.  If we're using a space as a delimiter, then shouldn't it also follow the same behavior of the comma?  If the user doesn't want to include/process null strings, then let them ignore them elsewhere in their code.  As far as our data is concerned, they're valid split points, in my opinion.

[SMcNeill]

And a timed comparison of the three routines:

Code: QB64: [Select]

1. CONST ntests = 5
2. DIM a(ntests - 1) AS STRING, d(ntests - 1) AS STRING
3. REDIM results1(0) AS STRING
4. REDIM results2(0) AS STRING
5. REDIM results3(0) AS STRING
6.  
7. CONST Limit = 1000000
8.  
9. a(0) = ""
10. d(0) = " "
11. a(1) = " test test    test " 'good no error!
12. d(1) = " "
13. a(2) = " test"
14. d(2) = " "
15. a(3) = "3d,z6d,z1 10 #d,z5"
16. d(3) = ",z"
17. a(4) = "Monday, , Wednesday, THursday, Friday, , Sunday"
18. d(4) = ", "
19.  
20. FOR i = 0 TO ntests - 1
21.     t# = TIMER
22.     FOR j = 1 TO Limit 'repeat the process multiple times so we can time it.
23.         SteveSplit a(i), d(i), results1()
24.     NEXT
25.     t1# = TIMER
26.     FOR j = 1 TO Limit 'repeat the process multiple times so we can time it.
27.         REDIM results2(0) AS STRING
28.         Split a(i), d(i), results2()
29.     NEXT
30.     t2# = TIMER
31.     FOR j = 1 TO Limit 'repeat the process multiple times so we can time it.
32.         Lsplit a(i), d(i), results3()
33.     NEXT
34.     t3# = TIMER
35.     PRINT "TEST #"; i; " -- Splitting: "; CHR$(34); a(i); CHR$(34); " with "; CHR$(34); d(i); CHR$(34)
36.     PRINT USING "###.####     ###.####     ###.####"; t1# - t#, t2# - t1#, t3# - t2#
37.  
38.     FOR j = 1 TO UBOUND(results1)
39.         PRINT j, CHR$(34); results1(j); CHR$(34),
40.         IF j <= UBOUND(results2) + 1 THEN PRINT CHR$(34); results2(j - 1); CHR$(34),
41.         IF j <= UBOUND(results3) + 1 THEN PRINT CHR$(34); results3(j - 1); CHR$(34),
42.         PRINT
43.     NEXT
44.     SLEEP
45.  
46. NEXT
47.  
48.  
49.  
50. 'notes: REDIM the array(0) to be loaded before calling Split '<<<<<<<<<<<<<<<<<<<<<<< IMPORTANT!!!!
51. SUB Split (mystr AS STRING, delim AS STRING, arr() AS STRING)
52.     ' bplus modifications of Galleon fix of Bulrush Split reply #13
53.     ' http://xmaxw.[abandoned, outdated and now likely malicious qb64 dot net website - don’t go there]/forum/index.php?topic=1612.0
54.     ' this sub further developed and tested here: \test\Strings\Split test.bas
55.     DIM copy AS STRING, p AS LONG, curpos AS LONG, arrpos AS LONG, dpos AS LONG
56.  
57.     copy = mystr 'make copy since we are messing with mystr when the delimiter is a space
58.  
59.     'special case if delim is space, probably want to remove all excess space
60.     IF delim = " " THEN
61.         copy = RTRIM$(LTRIM$(copy))
62.         p = INSTR(copy, "  ")
63.         WHILE p > 0
64.             copy = MID$(copy, 1, p - 1) + MID$(copy, p + 1)
65.             p = INSTR(copy, "  ")
66.         WEND
67.     END IF
68.     curpos = 1
69.     arrpos = 0
70.     dpos = INSTR(curpos, copy, delim)
71.     DO UNTIL dpos = 0
72.         arr(arrpos) = MID$(copy, curpos, dpos - curpos)
73.         arrpos = arrpos + 1
74.         REDIM _PRESERVE arr(arrpos + 1) AS STRING
75.         curpos = dpos + LEN(delim)
76.         dpos = INSTR(curpos, copy, delim)
77.     LOOP
78.     arr(arrpos) = MID$(copy, curpos)
79.     REDIM _PRESERVE arr(arrpos) AS STRING 'need this line? YES to get the ubound correct
80. END SUB
81.  
82.  
83. ' Luke 2019-02-15
84. 'Split in$ into pieces, chopping at every occurrence of delimiter$. Multiple consecutive occurrences
85. 'of delimiter$ are treated as a single instance. The chopped pieces are stored in result$().
86. '
87. 'delimiter$ must be one character long.
88. 'result$() must have been REDIMmed previously.
89. SUB Lsplit (in$, delimiter$, result$())
90.     REDIM result$(-1)
91.     start = 1
92.     DO
93.         WHILE MID$(in$, start, 1) = delimiter$
94.             start = start + 1
95.             IF start > LEN(in$) THEN EXIT SUB
96.         WEND
97.         finish = INSTR(start, in$, delimiter$)
98.         IF finish = 0 THEN finish = LEN(in$) + 1
99.         REDIM _PRESERVE result$(0 TO UBOUND(result$) + 1)
100.         result$(UBOUND(result$)) = MID$(in$, start, finish - start)
101.         start = finish + 1
102.     LOOP WHILE start <= LEN(in$)
103. END SUB
104.  
105. 'Combine all elements of in$() into a single string with delimiter$ separating the elements.
106. FUNCTION join$ (in$(), delimiter$)
107.     result$ = in$(LBOUND(in$))
108.     FOR i = LBOUND(in$) + 1 TO UBOUND(in$)
109.         result$ = result$ + delimiter$ + in$(i)
110.     NEXT i
111.     join$ = result$
112. END FUNCTION
113.  
114.  
115.  
116.  
117.  
118.  
119.  
120.  
121.  
122.  
123.  
124.  
125.  
126.  
127.  
128.  
129.  
130.  
131. SUB SteveSplit (text$, delimiter$, storage_array() AS STRING)
132.     STATIC count AS LONG
133.     count = count + 1
134.     u = UBOUND(storage_array)
135.     IF count > u THEN REDIM _PRESERVE storage_array(u + 1000) AS STRING
136.     i = INSTR(text$, delimiter$)
137.     IF i THEN
138.         storage_array(count) = LEFT$(text$, i - 1)
139.         SteveSplit MID$(text$, i + LEN(delimiter$)), delimiter$, storage_array()
140.     ELSE
141.         storage_array(count) = text$
142.         REDIM _PRESERVE storage_array(count) AS STRING
143.         count = 0
144.     END IF
145. END SUB

On my machine, the test speeds are as follows for:

               SteveSplit , Split (bplus), LSplit (Luke)
TEST #0: 0.1650 , 0.4395, 0.3296
TEST #1: 2.3076, 2.3076, 1.5386
TEST #2: 0.4399, 0.5493, 0.5493
TEST #3: 1.0439, 1.2637, 1.5381 (False results)
TEST #4: 1.9229, 2.1431, 2.6924 (False results)

In tests 0, 2, 3, 4, SteveSplit ran fastest. 
In test 1, LSplit was the fastest routine.

The difference in the speeds in Test #1 is a general philosophy of *how* we behave with the act of splitting.  For my routine, we generate several null strings, delimited by spaces.  For everyone else, the extra spaces are simply ignored and removed.

Luke's routine is set to only use a single character as a string delimiter, so for test 3 and 4, it produces false results as we have a 2-character delimiter.

******************

As to WHY my routine tosses out those extra null strings, it's to simply uniformly answer the question of, "How would we behave if if used a FIND AND REPLACE to change all the spaces to commas?"

Instead of a(1) = " test test    test ", what results would we expect if a(1) = ",test,test,,,,test," and we used a comma as a delimiter instead of a space?  Would we not then count the characters between each comma as as null-string?

I'd think so, and if two commas side-by-side designate a null-string result between them, then I feel like two spaces side-by-side should do the same for space delimited data.  This concept also allows me to perfectly reproduce the original data when using a join function, without automatically losing those spaces which may (or may not) have been important for formatting or data purposes.

If I don't want the extra spaces, all I need to do is ignore those null-strings. 

[bplus]

Steve:

The difference in the speeds in Test #1 is a general philosophy of *how* we behave with the act of splitting.  For my routine, we generate several null strings, delimited by spaces.  For everyone else, the extra spaces are simply ignored and removed.

Luke's routine is set to only use a single character as a string delimiter, so for test 3 and 4, it produces false results as we have a 2-character delimiter.

******************

As to WHY my routine tosses out those extra null strings, it's to simply uniformly answer the question of, "How would we behave if if used a FIND AND REPLACE to change all the spaces to commas?"

Instead of a(1) = " test test    test ", what results would we expect if a(1) = ",test,test,,,,test," and we used a comma as a delimiter instead of a space?  Would we not then count the characters between each comma as as null-string?

I'd think so, and if two commas side-by-side designate a null-string result between them, then I feel like two spaces side-by-side should do the same for space delimited data.  This concept also allows me to perfectly reproduce the original data when using a join function, without automatically losing those spaces which may (or may not) have been important for formatting or data purposes.

If I don't want the extra spaces, all I need to do is ignore those null-strings. 

Well in defense of my routine, you have the best of both worlds:
1. If you want to ignore spaces and just use any amount of spaces to separate items then use a space delimiter,
2. otherwise use any other delimiter of any length!

There is a real need for option #1 like when use string a number of numbers together, you won't need to worry about trimming them which would otherwise load the array with unwanted null strings. Having to ignore unwanted null strings would likely pose a needless burden on an app that depends upon the order placement of the items. I have in mind my simple little interpreter that used no punctuation only spaces as a delimiter.

Once you have a good general Split routine working, it would be a snap to modify and optimize for the particular application.
Likely it would need one delimiter or 1 character only delimiters or it may go the n-spaces route. Piece of cake to drop the unneeded general options to speed the thing up for the app. For this reason, Steve your method of REDIM _PRESERVE in large batches of say 1000, instead of at every new item, that is a good idea that I will incorporate in my general Split routine.

[SMcNeill]

Generally, if I want to strip out extra spaces, I’ll just run the string through a find-replace routine to change “  “ to “ “ before running the split routine.  It’s a fast process and saves time overall to do the whole process at once, instead of checking/trimming every loop inside the split routine.

And if you notice, my little routine only resizes the array larger when needed, and since REDIM _PRESERVE is a slow process, it’s better to oversized it by 1000 (or more) elements and then resize down when finished, than it is to resize it to the proper size for each word.  It’s very rare that I’ll ever REDIM _PRESERVE Array(Limit + 1).  REDIM _PRESERVE Array(Limit + 1000) is generally the smallest increment I’ll use in a routine, and then I’ll resize to free unneeded memory when done.  ;)

[bplus]

Generally, if I want to strip out extra spaces, I’ll just run the string through a find-replace routine to change “  “ to “ “ before running the split routine.  It’s a fast process and saves time overall to do the whole process at once, instead of checking/trimming every loop inside the split routine.

And if you notice, my little routine only resizes the array larger when needed, and since REDIM _PRESERVE is a slow process, it’s better to oversized it by 1000 (or more) elements and then resize down when finished, than it is to resize it to the proper size for each word.  It’s very rare that I’ll ever REDIM _PRESERVE Array(Limit + 1).  REDIM _PRESERVE Array(Limit + 1000) is generally the smallest increment I’ll use in a routine, and then I’ll resize to free unneeded memory when done.  ;)

Well Steve, I did strip out all extra spaces (if more than one) before running through the Split code (only if the delimiter was a space) and I did notice increasing the array by chucks instead of by 1 and already mentioned I was going to use that time saver tip!

One more interesting point about your method is that it is recursive which is a favorite technique I like to see employed. But I have heard that recursive techniques aren't as efficient ultimately as non recursive ones ie for every recursive routine there exists a more efficient non recursive one. (The proof of that might be interesting!) If true, then there is a faster version still needing to be revealed. :)

[Pete]

I have had to trim extra spaces a lot in html parsing. I made something simple years ago that I still use for that. It's so simple, I'll just code it here...

Code: QB64: [Select]

1. WIDTH 120, 25
2. a$ = "   This   is a test      of   eliminating     multiple         spaces    in    a text    line.   "
3. PRINT a$
4. DO UNTIL INSTR(a$, "  ") = 0
5.     a$ = MID$(a$, 1, INSTR(a$, "  ") - 1) + MID$(a$, INSTR(a$, "  ") + 1)
6. LOOP
7. a$ = RTRIM$(LTRIM$(a$))
8. PRINT a$
9.  

I probably should have read previous posts before putting it up. It may have no relevance, but at least it prints text with an apolitical message!

Replacement, splitting, concatenating are all very useful. Good luck with these additions, as I assume this is more stuff for the tool box forum.

Pete

[bplus]

Ha, ha, ha! Steve pulled a fast one! ;)

With such trivial tests his Split shines but check out a test with 10,000 items to Split and the new Split1000 sub:

Code: QB64: [Select]

1. 'split test.bas for qb64 bplus 2018-05-07
2. ' directly below is Steve's Timed test orig tests commented out
3. ' 2019-02-17 modified by B+ with a new Split1000 sub and a SERIOUS String to Split!
4.  
5. '=================================================================== steve;s speed test
6. CONST ntests = 6
7. DIM a(ntests - 1) AS STRING, d(ntests - 1) AS STRING
8. REDIM results1(0) AS STRING
9. REDIM results2(0) AS STRING
10. REDIM results3(0) AS STRING
11.  
12. CONST Limit = 100
13.  
14. 'trivial tests just to test accuracy of split
15. a(0) = ""
16. d(0) = " "
17. a(1) = " test test    test " 'good no error!
18. d(1) = " "
19. a(2) = " test"
20. d(2) = " "
21. a(3) = "3d,z6d,z1 10 #d,z5"
22. d(3) = ",z"
23. a(4) = "Monday, , Wednesday, THursday, Friday, , Sunday"
24. d(4) = ", "
25.  
26. 'lets get a serious test in here!! test a 10,000 random number string
27. FOR i = 1 TO 10000
28.     s$ = s$ + STR$(RND)
29. NEXT
30. a(5) = s$
31. d(5) = " "
32.  
33. FOR i = 0 TO ntests - 1
34.     CLS
35.     t# = TIMER
36.     FOR j = 1 TO Limit 'repeat the process multiple times so we can time it.
37.         SteveSplit a(i), d(i), results1()
38.     NEXT
39.     t1# = TIMER
40.     FOR j = 1 TO Limit 'repeat the process multiple times so we can time it.
41.         REDIM results2(0) AS STRING
42.         Split1000 a(i), d(i), results2()
43.     NEXT
44.     t2# = TIMER
45.     FOR j = 1 TO Limit 'repeat the process multiple times so we can time it.
46.         Lsplit a(i), d(i), results3()
47.     NEXT
48.     t3# = TIMER
49.     PRINT "TEST #"; i; " -- Splitting: "; CHR$(34); MID$(a(i), 1, 80); CHR$(34); " with "; CHR$(34); d(i); CHR$(34)
50.     PRINT: PRINT "Test names:", "SteveSplit", "Split1000", "Lsplit"
51.     PRINT "Times:",
52.     PRINT USING "###.####     ###.####     ###.####"; t1# - t#, t2# - t1#, t3# - t2#
53.     PRINT: PRINT "First Items in Results arrays (up to 10):"
54.     FOR j = 1 TO 10
55.         p = 0
56.         IF j <= UBOUND(results1) THEN p = 1: PRINT j, CHR$(34); MID$(results1(j), 1, 15); CHR$(34),
57.         IF j <= UBOUND(results2) + 1 THEN p = 1: PRINT CHR$(34); results2(j - 1); CHR$(34),
58.         IF j <= UBOUND(results3) + 1 THEN p = 1: PRINT CHR$(34); results3(j - 1); CHR$(34),
59.         IF p THEN PRINT
60.     NEXT
61.     PRINT: INPUT "Press enter for next test... "; wate$
62.  
63. NEXT
64.  
65.  
66. '' ================================= My Old Split test Code
67. 'the space delimiter is such a special case perhaps I should develope a single split for that alone?
68. ''2018-08-25 reworked for space delimiters and more variable declares
69. ''2019-02-15 add Luke's version to compare
70. 'ntests = 5
71. 'DIM a(ntests - 1) AS STRING, d(ntests - 1) AS STRING
72.  
73. 'a(0) = ""
74. 'd(0) = " "
75. 'a(1) = " test test    test " 'good no error!
76. 'd(1) = " "
77. 'a(2) = " test"
78. 'd(2) = " "
79. 'a(3) = "3d,z6d,z1 10 #d,z5"
80. 'd(3) = ",z"
81. 'a(4) = "Monday, , Wednesday, THursday, Friday, , Sunday"
82. 'd(4) = ", "
83. 'REDIM myarr(0) AS STRING '<<<<< REDIM forces the creation of a dynamic/resizable array
84. 'FOR test = 0 TO ntests - 1
85. '    PRINT: PRINT "splitting {"; a(test); "} with delimeter {"; d(test); "}"
86. '    Split1000 a(test), d(test), myarr()
87. '    amax = UBOUND(myarr)
88. '    FOR i = 0 TO amax
89. '        PRINT i; ":"; myarr(i)
90. '    NEXT i
91. '    INPUT "press enter for next test... "; wate$
92. 'NEXT
93.  
94. '' how about a quick file reader test?
95. 'PRINT: INPUT "Press enter for file test, any other + enter quits! "; wate$
96. 'IF LEN(wate$) THEN END
97. 'CLS
98.  
99. ''other wise continue
100. 'OPEN "Split test.bas" FOR BINARY AS #1 '<<< this file name!!!
101. 'ftext$ = SPACE$(LOF(1))
102. 'GET #1, , ftext$
103. 'CLOSE #1
104. 'Split ftext$, CHR$(13) + CHR$(10), myarr()
105. 'FOR i = 0 TO UBOUND(myarr)
106. '    PRINT myarr(i)
107. '    IF i MOD 20 = 19 THEN PRINT: INPUT "press enter for more "; wate$
108. 'NEXT
109. 'PRINT "the end"
110. 'END ' end program
111.  
112.  
113.  
114. '
115. 'notes: REDIM the array(0) to be loaded before calling Split '<<<<<<<<<<<<<<<<<<<<<<< IMPORTANT!!!!
116. SUB Split1000 (mystr AS STRING, delim AS STRING, arr() AS STRING)
117.     ' bplus modifications of Galleon fix of Bulrush Split reply #13
118.     ' http://xmaxw.[abandoned, outdated and now likely malicious qb64 dot net website - don’t go there]/forum/index.php?topic=1612.0
119.     ' this sub further developed and tested here: \test\Strings\Split test.bas
120.     DIM copy AS STRING, p AS LONG, curpos AS LONG, arrpos AS LONG, dpos AS LONG
121.  
122.     copy = mystr 'make copy since we are messing with mystr when the delimiter is a space
123.  
124.     'special case if delim is space, probably want to remove all excess space
125.     IF delim = " " THEN
126.         copy = RTRIM$(LTRIM$(copy))
127.         p = INSTR(copy, "  ")
128.         WHILE p > 0
129.             copy = MID$(copy, 1, p - 1) + MID$(copy, p + 1)
130.             p = INSTR(copy, "  ")
131.         WEND
132.     END IF
133.     curpos = 1
134.     arrpos = 0
135.     dpos = INSTR(curpos, copy, delim)
136.     DO UNTIL dpos = 0
137.         arr(arrpos) = MID$(copy, curpos, dpos - curpos)
138.         arrpos = arrpos + 1
139.         IF arrpos > UBOUND(arr) THEN REDIM _PRESERVE arr(UBOUND(arr) + 1000) AS STRING
140.         curpos = dpos + LEN(delim)
141.         dpos = INSTR(curpos, copy, delim)
142.     LOOP
143.     arr(arrpos) = MID$(copy, curpos)
144.     REDIM _PRESERVE arr(arrpos) AS STRING 'need this line? YES to get the ubound correct
145. END SUB
146.  
147.  
148. 'notes: REDIM the array(0) to be loaded before calling Split '<<<<<<<<<<<<<<<<<<<<<<< IMPORTANT!!!!
149. SUB Split (mystr AS STRING, delim AS STRING, arr() AS STRING)
150.     ' bplus modifications of Galleon fix of Bulrush Split reply #13
151.     ' http://xmaxw.[abandoned, outdated and now likely malicious qb64 dot net website - don’t go there]/forum/index.php?topic=1612.0
152.     ' this sub further developed and tested here: \test\Strings\Split test.bas
153.     DIM copy AS STRING, p AS LONG, curpos AS LONG, arrpos AS LONG, dpos AS LONG
154.  
155.     copy = mystr 'make copy since we are messing with mystr when the delimiter is a space
156.  
157.     'special case if delim is space, probably want to remove all excess space
158.     IF delim = " " THEN
159.         copy = RTRIM$(LTRIM$(copy))
160.         p = INSTR(copy, "  ")
161.         WHILE p > 0
162.             copy = MID$(copy, 1, p - 1) + MID$(copy, p + 1)
163.             p = INSTR(copy, "  ")
164.         WEND
165.     END IF
166.     curpos = 1
167.     arrpos = 0
168.     dpos = INSTR(curpos, copy, delim)
169.     DO UNTIL dpos = 0
170.         arr(arrpos) = MID$(copy, curpos, dpos - curpos)
171.         arrpos = arrpos + 1
172.         REDIM _PRESERVE arr(arrpos + 1000) AS STRING
173.         curpos = dpos + LEN(delim)
174.         dpos = INSTR(curpos, copy, delim)
175.     LOOP
176.     arr(arrpos) = MID$(copy, curpos)
177.     REDIM _PRESERVE arr(arrpos) AS STRING 'need this line? YES to get the ubound correct
178. END SUB
179.  
180.  
181. ' Luke 2019-02-15
182. 'Split in$ into pieces, chopping at every occurrence of delimiter$. Multiple consecutive occurrences
183. 'of delimiter$ are treated as a single instance. The chopped pieces are stored in result$().
184. '
185. 'delimiter$ must be one character long.
186. 'result$() must have been REDIMmed previously.
187. SUB Lsplit (in$, delimiter$, result$())
188.     REDIM result$(-1)
189.     start = 1
190.     DO
191.         WHILE MID$(in$, start, 1) = delimiter$
192.             start = start + 1
193.             IF start > LEN(in$) THEN EXIT SUB
194.         WEND
195.         finish = INSTR(start, in$, delimiter$)
196.         IF finish = 0 THEN finish = LEN(in$) + 1
197.         REDIM _PRESERVE result$(0 TO UBOUND(result$) + 1)
198.         result$(UBOUND(result$)) = MID$(in$, start, finish - start)
199.         start = finish + 1
200.     LOOP WHILE start <= LEN(in$)
201. END SUB
202.  
203. 'Combine all elements of in$() into a single string with delimiter$ separating the elements.
204. FUNCTION join$ (in$(), delimiter$)
205.     result$ = in$(LBOUND(in$))
206.     FOR i = LBOUND(in$) + 1 TO UBOUND(in$)
207.         result$ = result$ + delimiter$ + in$(i)
208.     NEXT i
209.     join$ = result$
210. END FUNCTION
211.  
212. SUB SteveSplit (text$, delimiter$, storage_array() AS STRING)
213.     STATIC count AS LONG
214.     count = count + 1
215.     u = UBOUND(storage_array)
216.     IF count > u THEN REDIM _PRESERVE storage_array(u + 1000) AS STRING
217.     i = INSTR(text$, delimiter$)
218.     IF i THEN
219.         storage_array(count) = LEFT$(text$, i - 1)
220.         SteveSplit MID$(text$, i + LEN(delimiter$)), delimiter$, storage_array()
221.     ELSE
222.         storage_array(count) = text$
223.         REDIM _PRESERVE storage_array(count) AS STRING
224.         count = 0
225.     END IF
226. END SUB
227.  
228.  

I knew his recursive method was curse worthy. ;-)))

[SMcNeill]

25 seconds doesn’t seem right at all.  I’ll do some digging to see what’s up later and report back.  ;)

[SMcNeill]

Try this non-recursive version and see how it performs for you:

Code: QB64: [Select]

1. 'split test.bas for qb64 bplus 2018-05-07
2. ' directly below is Steve's Timed test orig tests commented out
3. ' 2019-02-17 modified by B+ with a new Split1000 sub and a SERIOUS String to Split!
4.  
5. '=================================================================== steve;s speed test
6. CONST ntests = 6
7. DIM a(ntests - 1) AS STRING, d(ntests - 1) AS STRING
8. REDIM results1(0) AS STRING
9. REDIM results2(0) AS STRING
10. REDIM results3(0) AS STRING
11.  
12. CONST Limit = 100
13. CONST NoNull = -1
14.  
15. 'trivial tests just to test accuracy of split
16. a(0) = ""
17. d(0) = " "
18. a(1) = " test test    test " 'good no error!
19. d(1) = " "
20. a(2) = " test"
21. d(2) = " "
22. a(3) = "3d,z6d,z1 10 #d,z5"
23. d(3) = ",z"
24. a(4) = "Monday, , Wednesday, THursday, Friday, , Sunday"
25. d(4) = ", "
26.  
27. 'lets get a serious test in here!! test a 10,000 random number string
28. FOR i = 1 TO 10000
29.     s$ = s$ + STR$(RND)
30. NEXT
31. a(5) = s$
32. d(5) = " "
33.  
34. FOR i = 0 TO ntests - 1
35.     CLS
36.     t# = TIMER
37.     FOR j = 1 TO Limit 'repeat the process multiple times so we can time it.
38.         SteveSplit2 a(i), d(i), results1(), NoNull
39.     NEXT
40.     t1# = TIMER
41.     FOR j = 1 TO Limit 'repeat the process multiple times so we can time it.
42.         REDIM results2(0) AS STRING
43.         Split1000 a(i), d(i), results2()
44.     NEXT
45.     t2# = TIMER
46.     FOR j = 1 TO Limit 'repeat the process multiple times so we can time it.
47.         Lsplit a(i), d(i), results3()
48.     NEXT
49.     t3# = TIMER
50.     PRINT "TEST #"; i; " -- Splitting: "; CHR$(34); MID$(a(i), 1, 80); CHR$(34); " with "; CHR$(34); d(i); CHR$(34)
51.     PRINT: PRINT "Test names:", "SteveSplit", "Split1000", "Lsplit"
52.     PRINT "Times:",
53.     PRINT USING "###.####     ###.####     ###.####"; t1# - t#, t2# - t1#, t3# - t2#
54.     PRINT: PRINT "First Items in Results arrays (up to 10):"
55.     FOR j = 1 TO 10
56.         p = 0
57.         PRINT j,
58.         IF j <= UBOUND(results1) THEN PRINT CHR$(34); MID$(results1(j), 1, 15); CHR$(34), ELSE PRINT ,
59.         IF j <= UBOUND(results2) + 1 THEN PRINT CHR$(34); results2(j - 1); CHR$(34), ELSE PRINT ,
60.         IF j <= UBOUND(results3) + 1 THEN PRINT CHR$(34); results3(j - 1); CHR$(34),
61.         PRINT
62.     NEXT
63.     PRINT: INPUT "Press enter for next test... "; wate$
64.  
65. NEXT
66.  
67.  
68.  
69. '
70. 'notes: REDIM the array(0) to be loaded before calling Split '<<<<<<<<<<<<<<<<<<<<<<< IMPORTANT!!!!
71. SUB Split1000 (mystr AS STRING, delim AS STRING, arr() AS STRING)
72.     ' bplus modifications of Galleon fix of Bulrush Split reply #13
73.     ' http://xmaxw.[abandoned, outdated and now likely malicious qb64 dot net website - don’t go there]/forum/index.php?topic=1612.0
74.     ' this sub further developed and tested here: \test\Strings\Split test.bas
75.     DIM copy AS STRING, p AS LONG, curpos AS LONG, arrpos AS LONG, dpos AS LONG
76.  
77.     copy = mystr 'make copy since we are messing with mystr when the delimiter is a space
78.  
79.     'special case if delim is space, probably want to remove all excess space
80.     IF delim = " " THEN
81.         copy = RTRIM$(LTRIM$(copy))
82.         p = INSTR(copy, "  ")
83.         WHILE p > 0
84.             copy = MID$(copy, 1, p - 1) + MID$(copy, p + 1)
85.             p = INSTR(copy, "  ")
86.         WEND
87.     END IF
88.     curpos = 1
89.     arrpos = 0
90.     dpos = INSTR(curpos, copy, delim)
91.     DO UNTIL dpos = 0
92.         arr(arrpos) = MID$(copy, curpos, dpos - curpos)
93.         arrpos = arrpos + 1
94.         IF arrpos > UBOUND(arr) THEN REDIM _PRESERVE arr(UBOUND(arr) + 1000) AS STRING
95.         curpos = dpos + LEN(delim)
96.         dpos = INSTR(curpos, copy, delim)
97.     LOOP
98.     arr(arrpos) = MID$(copy, curpos)
99.     REDIM _PRESERVE arr(arrpos) AS STRING 'need this line? YES to get the ubound correct
100. END SUB
101.  
102.  
103. 'notes: REDIM the array(0) to be loaded before calling Split '<<<<<<<<<<<<<<<<<<<<<<< IMPORTANT!!!!
104. SUB Split (mystr AS STRING, delim AS STRING, arr() AS STRING)
105.     ' bplus modifications of Galleon fix of Bulrush Split reply #13
106.     ' http://xmaxw.[abandoned, outdated and now likely malicious qb64 dot net website - don’t go there]/forum/index.php?topic=1612.0
107.     ' this sub further developed and tested here: \test\Strings\Split test.bas
108.     DIM copy AS STRING, p AS LONG, curpos AS LONG, arrpos AS LONG, dpos AS LONG
109.  
110.     copy = mystr 'make copy since we are messing with mystr when the delimiter is a space
111.  
112.     'special case if delim is space, probably want to remove all excess space
113.     IF delim = " " THEN
114.         copy = RTRIM$(LTRIM$(copy))
115.         p = INSTR(copy, "  ")
116.         WHILE p > 0
117.             copy = MID$(copy, 1, p - 1) + MID$(copy, p + 1)
118.             p = INSTR(copy, "  ")
119.         WEND
120.     END IF
121.     curpos = 1
122.     arrpos = 0
123.     dpos = INSTR(curpos, copy, delim)
124.     DO UNTIL dpos = 0
125.         arr(arrpos) = MID$(copy, curpos, dpos - curpos)
126.         arrpos = arrpos + 1
127.         REDIM _PRESERVE arr(arrpos + 1000) AS STRING
128.         curpos = dpos + LEN(delim)
129.         dpos = INSTR(curpos, copy, delim)
130.     LOOP
131.     arr(arrpos) = MID$(copy, curpos)
132.     REDIM _PRESERVE arr(arrpos) AS STRING 'need this line? YES to get the ubound correct
133. END SUB
134.  
135.  
136. ' Luke 2019-02-15
137. 'Split in$ into pieces, chopping at every occurrence of delimiter$. Multiple consecutive occurrences
138. 'of delimiter$ are treated as a single instance. The chopped pieces are stored in result$().
139. '
140. 'delimiter$ must be one character long.
141. 'result$() must have been REDIMmed previously.
142. SUB Lsplit (in$, delimiter$, result$())
143.     REDIM result$(-1)
144.     start = 1
145.     DO
146.         WHILE MID$(in$, start, 1) = delimiter$
147.             start = start + 1
148.             IF start > LEN(in$) THEN EXIT SUB
149.         WEND
150.         finish = INSTR(start, in$, delimiter$)
151.         IF finish = 0 THEN finish = LEN(in$) + 1
152.         REDIM _PRESERVE result$(0 TO UBOUND(result$) + 1)
153.         result$(UBOUND(result$)) = MID$(in$, start, finish - start)
154.         start = finish + 1
155.     LOOP WHILE start <= LEN(in$)
156. END SUB
157.  
158. 'Combine all elements of in$() into a single string with delimiter$ separating the elements.
159. FUNCTION join$ (in$(), delimiter$)
160.     result$ = in$(LBOUND(in$))
161.     FOR i = LBOUND(in$) + 1 TO UBOUND(in$)
162.         result$ = result$ + delimiter$ + in$(i)
163.     NEXT i
164.     join$ = result$
165. END FUNCTION
166.  
167. SUB SteveSplit (text$, delimiter$, storage_array() AS STRING)
168.     STATIC count AS LONG
169.     count = count + 1
170.     u = UBOUND(storage_array)
171.     IF count > u THEN REDIM _PRESERVE storage_array(u + 1000) AS STRING
172.     i = INSTR(text$, delimiter$)
173.     IF i THEN
174.         storage_array(count) = LEFT$(text$, i - 1)
175.         SteveSplit MID$(text$, i + LEN(delimiter$)), delimiter$, storage_array()
176.     ELSE
177.         storage_array(count) = text$
178.         REDIM _PRESERVE storage_array(count) AS STRING
179.         count = 0
180.     END IF
181. END SUB
182.  
183. SUB SteveSplit2 (text$, delimiter$, storage_array() AS STRING, Options AS INTEGER)
184.     IF Options AND 1 THEN text$ = LTRIM$(text$)
185.     IF Options AND 2 THEN text$ = RTRIM$(text$)
186.     count = 1: oldi = 1
187.     l = LEN(delimiter$)
188.     u = UBOUND(storage_array)
189.     IF u < 1 THEN REDIM _PRESERVE storage_array(1000) AS STRING
190.     DO
191.         i = INSTR(oldi, text$, delimiter$)
192.         IF i THEN
193.             length = i - oldi
194.             u = UBOUND(storage_array)
195.             storage_array(count) = MID$(text$, oldi, length)
196.             IF (Options AND 4) AND (LEN(storage_array(count)) = 0) THEN
197.                 count = count - 1 'remove null-strings.
198.             END IF
199.             oldi = i + l
200.             i = oldi
201.             count = count + 1
202.             IF count > u THEN REDIM _PRESERVE storage_array(u + 1000) AS STRING
203.         END IF
204.     LOOP UNTIL i = 0
205.     storage_array(count) = MID$(text$, oldi)
206.     REDIM _PRESERVE storage_array(count) AS STRING
207. END SUB
208.  

Added feature:  This now has a simple flag which you can use to make null$ a non-acceptable result for your storage_array, so it'll behave exactly as the other routines.  Just change CONST NoNull = -1  to 0 and see how it toggles between the two methods.

(And both are the fastest splitters yet, so now it might not sound so bad to say, "I pulled a fast one."  ;D )

((And, if you like keeping the null-strings inside a split routine, as I generally do, you can remove those IF checks and make it even speedier. ))

[bplus]

Performs great Steve!

They are now running neck and neck on the big test and I agree without the IF check for options, SteveSplit2 should be faster yet.
So that means I might have another fix to speed up Split1000 because for my code, the delimiter check for space is a sort of Option check done before running the main part of the Split code.

[bplus]

One tiny mod to Split1000 and another test that shows it out paces SteveSplit2 when delimiter isn't a space (which is what slows down Split1000 specially clear now that the limit is increased), probably because of the extra decision about Options in SteveSplit2.

Code: QB64: [Select]

1. 'split test.bas for qb64 bplus 2018-05-07
2. ' directly below is Steve's Timed test orig tests commented out
3. ' 2019-02-17 modified by B+ with a new Split1000 sub and a SERIOUS String to Split!
4.  
5. '=================================================================== steve;s speed test
6. CONST ntests = 7
7. DIM a(ntests - 1) AS STRING, d(ntests - 1) AS STRING
8. REDIM results1(0) AS STRING
9. REDIM results2(0) AS STRING
10. REDIM results3(0) AS STRING
11.  
12. CONST Limit = 1000
13. CONST NoNull = -1
14.  
15. 'trivial tests just to test accuracy of split
16. a(0) = ""
17. d(0) = " "
18. a(1) = " test test    test " 'good no error!
19. d(1) = " "
20. a(2) = " test"
21. d(2) = " "
22. a(3) = "3d,z6d,z1 10 #d,z5"
23. d(3) = ",z"
24. a(4) = "Monday, , Wednesday, THursday, Friday, , Sunday"
25. d(4) = ", "
26.  
27. 'lets get a serious test in here!! test a 10,000 random number string
28. FOR i = 1 TO 10000
29.     s$ = s$ + STR$(RND)
30. NEXT
31. a(5) = s$
32. d(5) = " "
33. FOR i = 1 TO 10000
34.     IF i = 1 THEN s$ = STR$(RND * 1000 \ 1) ELSE s$ = s$ + "," + STR$(RND * 1000 \ 1)
35. NEXT
36. a(6) = s$
37. d(6) = ", "
38.  
39.  
40. FOR i = 0 TO ntests - 1
41.     CLS
42.     t# = TIMER
43.     FOR j = 1 TO Limit 'repeat the process multiple times so we can time it.
44.         SteveSplit2 a(i), d(i), results1(), NoNull
45.     NEXT
46.     t1# = TIMER
47.     FOR j = 1 TO Limit 'repeat the process multiple times so we can time it.
48.         REDIM results2(0) AS STRING
49.         Split1000 a(i), d(i), results2()
50.     NEXT
51.     t2# = TIMER
52.     FOR j = 1 TO Limit 'repeat the process multiple times so we can time it.
53.         Lsplit a(i), d(i), results3()
54.     NEXT
55.     t3# = TIMER
56.     PRINT "TEST #"; i; " -- Splitting: "; CHR$(34); MID$(a(i), 1, 80); CHR$(34); " with "; CHR$(34); d(i); CHR$(34)
57.     PRINT: PRINT "Test names:", "SteveSplit2", "Split1000", "Lsplit"
58.     PRINT "Times:",
59.     PRINT USING "###.####     ###.####     ###.####"; t1# - t#, t2# - t1#, t3# - t2#
60.     PRINT: PRINT "First Items in Results arrays (up to 10):"
61.     FOR j = 1 TO 10
62.         PRINT j,
63.         IF j <= UBOUND(results1) THEN PRINT CHR$(34); MID$(results1(j), 1, 15); CHR$(34), ELSE PRINT "      ",
64.         IF j <= UBOUND(results2) + 1 THEN PRINT CHR$(34); results2(j - 1); CHR$(34), ELSE PRINT "      ",
65.         IF j <= UBOUND(results3) + 1 THEN PRINT CHR$(34); results3(j - 1); CHR$(34), ELSE PRINT "      ",
66.         PRINT
67.     NEXT
68.     PRINT: INPUT "Press enter for next test... "; wate$
69.  
70. NEXT
71.  
72.  
73. '' ================================= My Old Split test Code
74. 'the space delimiter is such a special case perhaps I should develope a single split for that alone?
75. ''2018-08-25 reworked for space delimiters and more variable declares
76. ''2019-02-15 add Luke's version to compare
77. 'ntests = 5
78. 'DIM a(ntests - 1) AS STRING, d(ntests - 1) AS STRING
79.  
80. 'a(0) = ""
81. 'd(0) = " "
82. 'a(1) = " test test    test " 'good no error!
83. 'd(1) = " "
84. 'a(2) = " test"
85. 'd(2) = " "
86. 'a(3) = "3d,z6d,z1 10 #d,z5"
87. 'd(3) = ",z"
88. 'a(4) = "Monday, , Wednesday, THursday, Friday, , Sunday"
89. 'd(4) = ", "
90. 'REDIM myarr(0) AS STRING '<<<<< REDIM forces the creation of a dynamic/resizable array
91. 'FOR test = 0 TO ntests - 1
92. '    PRINT: PRINT "splitting {"; a(test); "} with delimeter {"; d(test); "}"
93. '    Split1000 a(test), d(test), myarr()
94. '    amax = UBOUND(myarr)
95. '    FOR i = 0 TO amax
96. '        PRINT i; ":"; myarr(i)
97. '    NEXT i
98. '    INPUT "press enter for next test... "; wate$
99. 'NEXT
100.  
101. '' how about a quick file reader test?
102. 'PRINT: INPUT "Press enter for file test, any other + enter quits! "; wate$
103. 'IF LEN(wate$) THEN END
104. 'CLS
105.  
106. ''other wise continue
107. 'OPEN "Split test.bas" FOR BINARY AS #1 '<<< this file name!!!
108. 'ftext$ = SPACE$(LOF(1))
109. 'GET #1, , ftext$
110. 'CLOSE #1
111. 'Split ftext$, CHR$(13) + CHR$(10), myarr()
112. 'FOR i = 0 TO UBOUND(myarr)
113. '    PRINT myarr(i)
114. '    IF i MOD 20 = 19 THEN PRINT: INPUT "press enter for more "; wate$
115. 'NEXT
116. 'PRINT "the end"
117. 'END ' end program
118.  
119. '
120. 'notes: REDIM the array(0) to be loaded before calling Split '<<<<<<<<<<<<<<<<<<<<<<< IMPORTANT!!!!
121. SUB Split1000 (mystr AS STRING, delim AS STRING, arr() AS STRING)
122.     ' bplus modifications of Galleon fix of Bulrush Split reply #13
123.     ' http://xmaxw.[abandoned, outdated and now likely malicious qb64 dot net website - don’t go there]/forum/index.php?topic=1612.0
124.     ' this sub further developed and tested here: \test\Strings\Split test.bas
125.     DIM copy AS STRING, p AS LONG, curpos AS LONG, arrpos AS LONG, dpos AS LONG
126.  
127.     copy = mystr 'make copy since we are messing with mystr when the delimiter is a space
128.  
129.     'special case if delim is space, probably want to remove all excess space
130.     IF delim = " " THEN
131.         copy = RTRIM$(LTRIM$(copy))
132.         p = INSTR(copy, "  ")
133.         WHILE p > 0
134.             copy = MID$(copy, 1, p - 1) + MID$(copy, p + 1)
135.             p = INSTR(copy, "  ")
136.         WEND
137.     END IF
138.     curpos = 1
139.     arrpos = 0
140.     LD = LEN(delim) 'mod
141.     dpos = INSTR(curpos, copy, delim)
142.     DO UNTIL dpos = 0
143.         arr(arrpos) = MID$(copy, curpos, dpos - curpos)
144.         arrpos = arrpos + 1
145.         IF arrpos > UBOUND(arr) THEN REDIM _PRESERVE arr(UBOUND(arr) + 1000) AS STRING
146.         curpos = dpos + LD
147.         dpos = INSTR(curpos, copy, delim)
148.     LOOP
149.     arr(arrpos) = MID$(copy, curpos)
150.     REDIM _PRESERVE arr(arrpos) AS STRING 'need this line? YES to get the ubound correct
151. END SUB
152.  
153.  
154. 'notes: REDIM the array(0) to be loaded before calling Split '<<<<<<<<<<<<<<<<<<<<<<< IMPORTANT!!!!
155. SUB Split (mystr AS STRING, delim AS STRING, arr() AS STRING)
156.     ' bplus modifications of Galleon fix of Bulrush Split reply #13
157.     ' http://xmaxw.[abandoned, outdated and now likely malicious qb64 dot net website - don’t go there]/forum/index.php?topic=1612.0
158.     ' this sub further developed and tested here: \test\Strings\Split test.bas
159.     DIM copy AS STRING, p AS LONG, curpos AS LONG, arrpos AS LONG, dpos AS LONG
160.  
161.     copy = mystr 'make copy since we are messing with mystr when the delimiter is a space
162.  
163.     'special case if delim is space, probably want to remove all excess space
164.     IF delim = " " THEN
165.         copy = RTRIM$(LTRIM$(copy))
166.         p = INSTR(copy, "  ")
167.         WHILE p > 0
168.             copy = MID$(copy, 1, p - 1) + MID$(copy, p + 1)
169.             p = INSTR(copy, "  ")
170.         WEND
171.     END IF
172.     curpos = 1
173.     arrpos = 0
174.     dpos = INSTR(curpos, copy, delim)
175.     DO UNTIL dpos = 0
176.         arr(arrpos) = MID$(copy, curpos, dpos - curpos)
177.         arrpos = arrpos + 1
178.         REDIM _PRESERVE arr(arrpos + 1000) AS STRING
179.         curpos = dpos + LEN(delim)
180.         dpos = INSTR(curpos, copy, delim)
181.     LOOP
182.     arr(arrpos) = MID$(copy, curpos)
183.     REDIM _PRESERVE arr(arrpos) AS STRING 'need this line? YES to get the ubound correct
184. END SUB
185.  
186.  
187. ' Luke 2019-02-15
188. 'Split in$ into pieces, chopping at every occurrence of delimiter$. Multiple consecutive occurrences
189. 'of delimiter$ are treated as a single instance. The chopped pieces are stored in result$().
190. '
191. 'delimiter$ must be one character long.
192. 'result$() must have been REDIMmed previously.
193. SUB Lsplit (in$, delimiter$, result$())
194.     REDIM result$(-1)
195.     start = 1
196.     DO
197.         WHILE MID$(in$, start, 1) = delimiter$
198.             start = start + 1
199.             IF start > LEN(in$) THEN EXIT SUB
200.         WEND
201.         finish = INSTR(start, in$, delimiter$)
202.         IF finish = 0 THEN finish = LEN(in$) + 1
203.         REDIM _PRESERVE result$(0 TO UBOUND(result$) + 1)
204.         result$(UBOUND(result$)) = MID$(in$, start, finish - start)
205.         start = finish + 1
206.     LOOP WHILE start <= LEN(in$)
207. END SUB
208.  
209. 'Combine all elements of in$() into a single string with delimiter$ separating the elements.
210. FUNCTION join$ (in$(), delimiter$)
211.     result$ = in$(LBOUND(in$))
212.     FOR i = LBOUND(in$) + 1 TO UBOUND(in$)
213.         result$ = result$ + delimiter$ + in$(i)
214.     NEXT i
215.     join$ = result$
216. END FUNCTION
217.  
218. SUB SteveSplit2 (text$, delimiter$, storage_array() AS STRING, Options AS INTEGER)
219.     IF Options AND 1 THEN text$ = LTRIM$(text$)
220.     IF Options AND 2 THEN text$ = RTRIM$(text$)
221.     count = 1: oldi = 1
222.     l = LEN(delimiter$)
223.     u = UBOUND(storage_array)
224.     IF u < 1 THEN REDIM _PRESERVE storage_array(1000) AS STRING
225.     DO
226.         i = INSTR(oldi, text$, delimiter$)
227.         IF i THEN
228.             length = i - oldi
229.             u = UBOUND(storage_array)
230.             storage_array(count) = MID$(text$, oldi, length)
231.             IF (Options AND 4) AND (LEN(storage_array(count)) = 0) THEN
232.                 count = count - 1 'remove null-strings.
233.             END IF
234.             oldi = i + l
235.             i = oldi
236.             count = count + 1
237.             IF count > u THEN REDIM _PRESERVE storage_array(u + 1000) AS STRING
238.         END IF
239.     LOOP UNTIL i = 0
240.     storage_array(count) = MID$(text$, oldi)
241.     REDIM _PRESERVE storage_array(count) AS STRING
242. END SUB
243.  
244. SUB SteveSplit (text$, delimiter$, storage_array() AS STRING)
245.     STATIC count AS LONG
246.     count = count + 1
247.     u = UBOUND(storage_array)
248.     IF count > u THEN REDIM _PRESERVE storage_array(u + 1000) AS STRING
249.     i = INSTR(text$, delimiter$)
250.     IF i THEN
251.         storage_array(count) = LEFT$(text$, i - 1)
252.         SteveSplit MID$(text$, i + LEN(delimiter$)), delimiter$, storage_array()
253.     ELSE
254.         storage_array(count) = text$
255.         REDIM _PRESERVE storage_array(count) AS STRING
256.         count = 0
257.     END IF
258. END SUB
259.  
260.