using instr to find an expression multiple times in array

[dkl]

Sorry Gentlemen,

I have asked this question before and search all my posts but cannot find the relevant one. I want to find each position of "name" in the file below and using that info and then extract the next bit of info as seen below

I used the following code to get the info below....... pos = instr(inf$,"actor"):if pos <> 0 then actor1$ = after$(inf$,"[{"):actor2$ = upto$(actor1$,"}]") (NOTE: I am aware that after$ and upto$ can be combined, but I always get the syntax wrong, hence what you see there!!)

this makes the array$ - actor2$

"@type":"Person","url":"/name/nm0915208/","name":"Naomi Watts"},{"@type":"Person","url":"/name/nm0000705/","name":"Robin Wright"},{"@type":"Person","url":"/name/nm1882152/","name":"Xavier Samuel"

Now I want to extract the info from actor2$.The info below will do the trick, but it only finds the first instance.

pos = instr(actor2$,chr$(34);"name";chr$(34);":";chr$(34))

if pos <> 0 then actor3$ = after$(actor2$,chr$(34);"name";chr$(34);":";chr$(34)):actor4$ = upto$(actor3$,chr$(34);"}")

I have tried a for/next loop and a do/loop and I'm aware that I have to 'move pos on' in order to continue the search.

In a past example I saw pos = pos + 1, but that doesn't seem to work here.

Some help would be appreciated please.

On another note, is there a simple 'rule of thumb' when adding chr$(34) to array$ with multiple quotation symbols?

[Walt Decker]


I = 1
[DO.LOOP]

I = INSTR(I, Source$, Match$)

IF I = 0 THEN GOTO [END.LOOP]
IF I THEN
 DoSomething
 I = I + LEN(Match$)
END IF

GOTO [DO.LOOP]

[END.LOOP]


[dkl]

Thanks Walt, that's what I meant (the 'I' variable should be at the end, not the beginning), but not to worry I realised that!
I got the code to print out the position of each search. However, I'm still flummaxed as to how once I find each position with 'INSTR' after then telling it to store some info from then array$, how to store info for the rest of the array$.
I still keep getting the same piece of info saved, I'm not getting the rest of the array$. I want to find "NAME" and then store the name immediately after. all I get is the first name in the array$ continually!

[Walt Decker]

How about posting a couple of actual data sets so they can be analyzed. I am certain that I, or someone with more knowledge than I, can come up with something.

Ok, I see what you are getting at.

this makes the array$ - actor2$
"@type":"Person","url":"/name/nm0915208/","name":"Naomi Watts"},{"@type":"Person","url":"/name/nm0000705/","name":"Robin Wright"},{"@type":"Person","url":"/name/nm1882152/","name":"Xavier Samuel"

Actor2$ appears to be a single string rather than an array of strings.  Why not make it an actual array of strings then parse each element from back to front for the first colon.  Then extract the name from that?

[Walt Decker]

Using your data for Actor2$:

'
'this makes the array$ - actor2$
DQ$  = CHR$(34)
CM$ = ","
C$   = ":"

Actor3$ = ""
Actor2$ = DQ$ + "@type" + C$+ DQ$ + "Person" + DQ$ + CM$ + DQ$ + "url" + DQ$
Actor2$ = Actor2$ + C$ + DQ$ + "/name/nm0915208/" + DQ$ + CM$ + DQ$ + "name" + DQ$
Actor2$ = Actor2$ + C$ + DQ$ + "Naomi Watts" + DQ$ + "}"
Actor2$ = Actor2$ + "," + "{" + DQ$ + "@type" + ":" + DQ$ + "Person" + DQ$
Actor2$ = Actor2$ + "," + DQ$ + "url" + DQ$ + ":" + DQ$ + "/name/nm0000705/" + DQ$
Actor2$ = Actor2$ + "," + DQ$ + "name" + DQ$ + ":" + DQ$ + "Robin Wright" + DQ$
Actor2$ = Actor2$ + "}" + "," + "{" + DQ$ + "@type" + DQ$ + ":" + DQ$
Actor2$ = Actor2$ + DQ$ + "Person" + DQ$ + "," + DQ$ + "url" + DQ$
Actor2$ = Actor2$ + ":" + DQ$ + "/name/nm1882152/" + DQ$ + ","
Actor2$ = Actor2$ + DQ$ + "name" + DQ$ + ":" + DQ$ + "Xavier Samuel" + DQ$

'"@type":"Person","url":"/name/nm0915208/","name":"Naomi Watts"},{"@type":"Person","url":"/name/nm0000705/","name":"Robin Wright"},{"@type":"Person","url":"/name/nm1882152/","name":"Xavier Samuel"
print Actor2$

Actor2$ = REMCHAR$(Actor2$, DQ$)

print Actor2$
FOR I = LEN(Actor2$) TO 1 STEP -1
 IF MID$(Actor2$, I, 1) = "{" THEN
   J = INSTR(Actor2$, "}", I)
   IF J = 0 THEN
     Actor3$ = MID$(Actor2$, I)
     PRINT Actor3$

     FOR K = LEN(Actor3$) TO 1 STEP -1
       IF MID$(Actor3$, K, 1) = ":" THEN
         Actor4$ = Actor4$ + "|" + MID$(Actor3$, K + 1)
         PRINT Actor4$
         EXIT FOR
       END IF
     NEXT K
   ELSE
     Actor3$ = MID$(Actor2$, I, J - I)
     FOR K = LEN(Actor3$) TO 1 STEP -1
       IF MID$(Actor3$, K, 1) = ":" THEN
         Actor4$ = Actor4$ + "|" + MID$(Actor3$, K + 1)
         PRINT Actor4$
         EXIT FOR
       END IF
     NEXT K
   END IF
 END IF
NEXT I

J = INSTR(Actor2$, "}")
FOR I = J - 1 TO 1 STEP -1
 IF MID$(Actor2$, I, 1) = ":" THEN
   Actor4$ = Actor4$ + "|" + MID$(Actor2$, I + 1, J - I - 1)
   PRINT Actor4$
   EXIT FOR
 END IF
NEXT I
'

It might be a bit easier if your data started with "{" and ended with "}".

[Rod]

As Walt says the data needs cleaned up and needs an ending } else more code to catch that. Here is a version with te pesky " removed to make the demo simpler.

a$="@type:Person,url:/name/nm0915208/,name:Naomi Watts},{@type:Person,url:/name/nm0000705/,name:Robin Wright},{@type:Person,url:/name/nm1882152/,name:Xavier Samuel}"
tagstart$="name:"
tagend$="}"
'find first item
pos=instr(a$,tagstart$,1)
while pos
   'now find end of item
   posend=instr(a$,tagend$,pos+1)
   'print the item
   print mid$(a$,pos+5,posend-pos-5)
   'move the pointer on
   pos=posend
   'now find the next item if it exists
   pos=instr(a$,tagstart$,pos)
wend


[tsh73]

It really looks like (part of) Python code to me (just learning).
See?
a=({"@type":"Person","url":"/name/nm0915208/","name":"Naomi Watts"},
  {"@type":"Person","url":"/name/nm0000705/","name":"Robin Wright"},
  {"@type":"Person","url":"/name/nm1882152/","name":"Xavier Samuel"}
  )

for i in range(3):
print (a[i]["name"])

produces
Naomi Watts
Robin Wright
Xavier Samuel

Just looked - probably this is JSON format
Anyway, both of these could explain why so much strange delimiters.

---

EDIT: Full JSON is probably hard to do with all that nesting
But subset - just set of pairs "name:value", without nesting - is surely doable

---

Since it is standard format (or subset),
and essentially just set of pairs "name=value",
probably some LB code exists/could be adapted (some object browsers by Carl etc) to work with it on higher level? Read/parse, create such strings.
-
Like, (pseudocode)
(pretending quotes set right / data read from a file)
obj$='{"@type":"Person","url":"/name/nm0000705/","name":"Robin Wright"}'
print obj.objType(obj$)
print obj.value(obj$, "name")
print obj.hasKey(bj$, "lastName")
print obj.keys(obj$)


[tsh73]

Based on Walt data generator.

EDIT realized all functions I mentioned in previous post.

Results
"@type":"Person","url":"/name/nm0915208/","name":"Naomi Watts"},{"@type":"Person","url":"/name/nm0000705/","name":"Robin Wright"},{"@type":"Person","url":"/name/nm1882152/","name":"Xavier Samuel"

Number of objects 3

1             "@type":"Person","url":"/name/nm0915208/","name":"Naomi Watts"
Naomi Watts
@type         >Person<
missingKey    ><

2             "@type":"Person","url":"/name/nm0000705/","name":"Robin Wright"
Robin Wright
@type         >Person<
missingKey    ><

3             "@type":"Person","url":"/name/nm1882152/","name":"Xavier Samuel"
Xavier Samuel
@type         >Person<
missingKey    ><

Object 2
"@type":"Person","url":"/name/nm0000705/","name":"Robin Wright"

its' type
Person

Does it have key
@type         >1<
missingKey    >0<
url           >1<

its' keys
@type|url|name


'this makes the array$ - actor2$
global DQ$  'used everythere
DQ$  = CHR$(34)
CM$ = ","
C$   = ":"

Actor3$ = ""
Actor2$ = DQ$ + "@type" + DQ$ + C$+ DQ$ + "Person" + DQ$ + CM$ + DQ$ + "url" + DQ$
Actor2$ = Actor2$ + C$ + DQ$ + "/name/nm0915208/" + DQ$ + CM$ + DQ$ + "name" + DQ$
Actor2$ = Actor2$ + C$ + DQ$ + "Naomi Watts" + DQ$ + "}"
Actor2$ = Actor2$ + "," + "{" + DQ$ + "@type" + DQ$ + ":" + DQ$ + "Person" + DQ$
Actor2$ = Actor2$ + "," + DQ$ + "url" + DQ$ + ":" + DQ$ + "/name/nm0000705/" + DQ$
Actor2$ = Actor2$ + "," + DQ$ + "name" + DQ$ + ":" + DQ$ + "Robin Wright" + DQ$
Actor2$ = Actor2$ + "}" + "," + "{" + DQ$ + "@type" + DQ$ + ":"
Actor2$ = Actor2$ + DQ$ + "Person" + DQ$ + "," + DQ$ + "url" + DQ$
Actor2$ = Actor2$ + ":" + DQ$ + "/name/nm1882152/" + DQ$ + ","
Actor2$ = Actor2$ + DQ$ + "name" + DQ$ + ":" + DQ$ + "Xavier Samuel" + DQ$

'"@type":"Person","url":"/name/nm0915208/","name":"Naomi Watts"},{"@type":"Person","url":"/name/nm0000705/","name":"Robin Wright"},{"@type":"Person","url":"/name/nm1882152/","name":"Xavier Samuel"
print Actor2$
print
'Actor2$ = trim$(Actor2$)
'if left$(Actor2$,1)<>"{" then Actor2$="{"+Actor2$
'if right$(Actor2$,1)<>"}" then Actor2$=Actor2$+"}"
'word$ eats delimiter, so first/last {} not needed anyway

'getting separate "objects" delimited with "},{"
N=nWords(Actor2$, "},{")
print "Number of objects ";N
dim obj$(N) 'so I have it later, just to sho off names$ function
print

w$=""
i=0
while 1
   i=i+1
   w$ = word$(Actor2$, i, "},{")
   if w$="" then exit while
   obj$(i)=w$
   print i, w$
   'now get value by name
   print objValue$(w$, "name")

   toFind$="@type": print toFind$, ">";objValue$(w$, toFind$);"<"
   toFind$="missingKey": print toFind$, ">";objValue$(w$, toFind$);"<"   'just ""
   print
wend

print "Object ";2
print obj$(2)
obj$ = obj$(2)

print
print "its' type"
print objType$(obj$)

print
print "Does it have key"
   toFind$="@type":        print toFind$, ">";objHasKey(obj$, toFind$);"<"
   toFind$="missingKey":   print toFind$, ">";objHasKey(obj$, toFind$);"<"   'just ""
   toFind$="url":          print toFind$, ">";objHasKey(obj$, toFind$);"<"   'just ""

print
print "its' keys"
print objKeys$(obj$(2))

end
'------------------------------------------------------
function nWords(s$, delim$)
   i=0
   while 1
       i=i+1
       w$ = word$(s$, i, delim$)
       if w$="" then exit while
   wend
   nWords = i-1
end function

function stripDQ$(a$)
   if left$(a$, 1)=DQ$ then a$ = mid$(a$,2)
   if right$(a$, 1)=DQ$ then a$ = left$(a$,len(a$)-1)
   stripDQ$ = a$
end function

function objType$(obj$)
    objType$ = objValue$(obj$, "@type")
end function

function objHasKey(obj$, key$)
   objHasKey = instr("|"+objKeys$(obj$)+"|", "|"+key$+"|")>0
end function

function objKeys$(obj$)
   'get pairs
   'srip DQ
   'get aName
   w$=""
   i=0
   while 1
       i=i+1
       pair$ = word$(obj$, i, DQ$;",";DQ$)
       if pair$="" then exit while
       pair$=stripDQ$(pair$)

       aName$ = word$(pair$, 1, DQ$;":";DQ$)   'here we actually break no-quote, like ("Id":123)
       objKeys$=objKeys$+"|"+aName$
   wend
   objKeys$=mid$(objKeys$, 2)
end function

function objValue$(obj$, name$)
   name$ = stripDQ$(name$)
   'get pairs
   'srip DQ
   'get aName
   'if match, return aVla
   w$=""
   i=0
   while 1
       i=i+1
       pair$ = word$(obj$, i, DQ$;",";DQ$)
       if pair$="" then exit while
       pair$=stripDQ$(pair$)

       aName$ = word$(pair$, 1, DQ$;":";DQ$)   'here we actually break no-quote, like ("Id":123)
       if aName$ = name$ then
           aVal$ = word$(pair$, 2, DQ$;":";DQ$)
           objValue$ = aVal$
           exit function
       end if
   wend

end function


[Walt Decker]

Here is a function that will take any delimiter and fill an array with the fields defined by that delimiter.

If the case of the delimiter is unimportant set the last parameter in the argument to zero as in the example.

'
'this makes the array$ - actor2$
DQ$  = CHR$(34)
CM$ = ","
C$   = ":"

Ubnd = -1

Actor3$ = ""
Actor2$ = DQ$ + "@type" + C$+ DQ$ + "Person" + DQ$ + CM$ + DQ$ + "url" + DQ$
Actor2$ = Actor2$ + C$ + DQ$ + "/name/nm0915208/" + DQ$ + CM$ + DQ$ + "name" + DQ$
Actor2$ = Actor2$ + C$ + DQ$ + "Naomi Watts" + DQ$ + "}"
Actor2$ = Actor2$ + "," + "{" + DQ$ + "@type" + ":" + DQ$ + "Person" + DQ$
Actor2$ = Actor2$ + "," + DQ$ + "url" + DQ$ + ":" + DQ$ + "/name/nm0000705/" + DQ$
Actor2$ = Actor2$ + "," + DQ$ + "name" + DQ$ + ":" + DQ$ + "Robin Wright" + DQ$
Actor2$ = Actor2$ + "}" + "," + "{" + DQ$ + "@type" + DQ$ + ":" + DQ$
Actor2$ = Actor2$ + DQ$ + "Person" + DQ$ + "," + DQ$ + "url" + DQ$
Actor2$ = Actor2$ + ":" + DQ$ + "/name/nm1882152/" + DQ$ + ","
Actor2$ = Actor2$ + DQ$ + "name" + DQ$ + ":" + DQ$ + "Xavier Samuel" + DQ$

'"@type":"Person","url":"/name/nm0915208/","name":"Naomi Watts"},{"@type":"Person","url":"/name/nm0000705/","name":"Robin Wright"},{"@type":"Person","url":"/name/nm1882152/","name":"Xavier Samuel"
PRINT Actor2$

Actor2$ = REMCHAR$(Actor2$, DQ$)

PRINT Actor2$

Ubnd = FN.GetAllFields(Actor2$, "NAME:", 0)
PRINT Ubnd
FOR I = 0 TO Ubnd
 PRINT I, Ary$(I)
NEXT I
END

'-----------------------------------------------------------------
'-----------------------------------------------------------------

FUNCTION FN.GetAllFields(TxtIn$, Dlm$, MatchCase)

DlmLen  = 0
NumFlds = 0
AryCntr = 0
NumChrs = 0
EndPosn = 0
Posn    = 0
I       = 0

TmpIn$  = ""

DIM FldPos(1000)
DIM Ary$(-1)

TmpIn$ = TxtIn$
IF MatchCase = 0 THEN
 Dlm$ = UPPER$(Dlm$)
 TmpIn$ = UPPER$(TmpIn$)
END IF

DlmLen = LEN(Dlm$)
trace 2
I       = 1
NumFlds = -1
 [GET.FIELDS]
I = INSTR(TmpIn$, Dlm$, I)

IF I = 0 THEN GOTO [END.FIELDS]

IF I THEN
 NumFlds = NumFlds + 1
 FldPos(NumFlds) = I
 I = I + 1
 GOTO [GET.FIELDS]
END IF

 [END.FIELDS]
IF NumFlds < 0 THEN EXIT FUNCTION

REDIM Ary$(NumFlds + 1)

Ary$(0) = LEFT$(TxtIn$, FldPos(0) - 1)
PRINT "ary0 = ";Ary$(0), FldPos(0)
FOR I = 0 TO NumFlds - 1
 Posn = FldPos(I) + DlmLen
 EndPosn = FldPos(I + 1)
 NumChrs = EndPosn - Posn
 AryCntr = AryCntr + 1
 Ary$(AryCntr) = MID$(TxtIn$, Posn, NumChrs)
NEXT I
AryCntr = AryCntr + 1
Posn = FldPos(I) + DlmLen
Ary$(AryCntr) = MID$(TxtIn$, Posn)

FN.GetAllFields = NumFlds + 1
END FUNCTION
'

[Walt Decker]

Here are a couple of string functions that may be of use to someone:

'
FUNCTION FN.GetField$(Source$, Dlm$, FieldNum, MatchCase)

NumFlds = 0
FldStrt = 0
NumChrs = 0
FldEnd  = 0
DlmLen  = 0
I       = 0

DIM FldPos(1000)

TmpSrc$ = ""

IF Source$ = "" THEN
 EXIT FUNCTION
END IF

TmpSrc$ = Source$

IF MatchCase = 0 THEN
 Dlm$ = UPPER$(Dlm$)
 TmpSrc$ = UPPER$(TmpSrc$)
END IF

DlmLen = LEN(Dlm$)

IF FieldNum > 0 THEN FieldNum = FieldNum - 1

NumFlds = -1
I = 1
 [FIND.FIELDS]
I = INSTR(TmpSrc$, Dlm$, I)

IF I = 0 THEN GOTO [END.FIELDS]

NumFlds = NumFlds + 1
FldPos(NumFlds) = I
I = I + 1
GOTO [FIND.FIELDS]

 [END.FIELDS]
IF NumFlds < 0 THEN
 REDIM FldPos(-1)
 FN.GetField$ = Source$
 EXIT FUNCTION
END IF

IF FieldNum > NumFlds + 1 THEN
 REDIM FldPos(-1)
 EXIT FUNCTION
END IF

IF FieldNum = 0 THEN
 FN.GetField$ = LEFT$(Source$, FldPos(0) - 1)
 REDIM FldPos(-1)
 EXIT FUNCTION
END IF

IF FieldNum = NumFlds + 1 THEN
 FldStrt = FldPos(NumFlds) + DlmLen
 FN.GetField$ = MID$(Source$, FldStrt)
 REDIM FldPos(-1)
 EXIT FUNCTION
END IF

FldStrt = FldPos(FieldNum - 1) + DlmLen
FldEnd = FldPos(FieldNum)
NumChrs = FldEnd - FldStrt

REDIM FldPos(-1)
FN.GetField$ = MID$(Source$, FldStrt, NumChrs)

END FUNCTION

'----------------------------------------------------------------
'----------------------------------------------------------------

FUNCTION FN.GetFieldCount(Source$, Dlm$, MatchCase)

FldCnt = 0
I      = 0

IF Source$ = "" THEN EXIT FUNCTION

IF MatchCase = 0 THEN
 Source$ = UPPER$(Source$)
 Dlm$ = UPPER$(Dlm$)
END IF

I = 1

 [START.COUNT]
I = INSTR(Source$, Dlm$, I)

IF I = 0 THEN GOTO [END.COUNT]

FldCnt = FldCnt + 1
GOTO [START.COUNT]

 [END.COUNT]

FN.GetFieldCount = FldCnt + 1
END FUNCTION

'

[dkl]

WOW, so much response Thank you. I'm sorry I should have mentioned that the data was in JSON. Stripping the code to what I wanted would be a lot easier if I had been using Python, but I can't actually code in Python (!), although have seen many examples of web scraping using it.
Rod's example was similar to what I was looking and I think he may have helped me out before, but I couldn't find the example.

I actually managed to come up with a solution, with code which I added to the code in my first post

actor$ = actor$ +actor4$
pos = pos + len(match$)'or +1 - both work
actor2$ = mid$(actor2$,pos,len(actor2$)):pos = 1'<- this shortens the string and keeps 'pos' at position 1, where Rod moves 'pos' along/thru the array, which was what I was initially trying to do
actor$ = replstr$(actor$,chr$(34)," ")
print "ACTOR$-> ";actor$

It effectively shortens the array$ (actors2$) after every find of 'name'

It's difficult to emulate here as the initial array$ (actor2$) string was plucked out of a long piece of JSON code. Trying to make a new array$ from it requires adding all the extra code - as WALT did.
However, many thanks for all your help. Now I have other ways of looking at the solution next time