progress 4gl OpenEdge parse key/value-pair string aka Query-String - query-string

Question

I cant find anything on how i should parse a string of key/value paris AKA query-string like this one:
FieldType="String"&FieldFormat="^[a-z0-9!#$%&'*+/=?^_`{|}~-]+$"
field separators might be contained in the value like the above example
this is not to be used as a web-request paramlist.
i found this: running a loop on a comma delimited list of items progress 4GL
but entry() does not care if the data is in a qoutation.
= EDIT =
So i found a not so ideal solution that i hope nobody needs to mimic
DO jj=1 TO NUM-ENTRIES(curr,"&"):
DEFINE VARIABLE pos AS INTEGER NO-UNDO.
ASSIGN
k = entry( 1, ENTRY(jj,curr,"&"), "=")
v = entry( 2, ENTRY(jj,curr,"&"), "=")
pos = INDEX( curr, k + "=" ).
/* Check if this is a qouted value*/
IF NUM-ENTRIES( SUBSTRING( curr, pos, ABS( INDEX(curr, "&", pos) - pos) ) ,'"') > 1 THEN
ASSIGN v = ENTRY( 2, SUBSTRING( curr, pos) , '"').
end.
The IF-statment is what nightmares are made of!

Answer 1

Building from Tom's and TheMadDBA's answers.
Assumption: The first & will be the one we want to split on.
define variable cQryString as character no-undo.
define variable iSplitIndex as integer no-undo.
define variable cType as character no-undo format "x(30)" label " Type".
define variable cFormat as character no-undo format "x(30)" label "Format".
assign
cQryString = 'FieldType=String&FieldFormat="^[a-z0-9!#$%&~'*+/=?^_`~{|}~~-]+$"'
iSplitIndex = index(cQryString, "&")
cType = substring(cQryString, 1, iSplitIndex - 1)
cFormat = substring(cQryString, iSplitIndex + 1, length(cQryString))
cType = substring(cType, index(cType, "=") + 1, length(cType))
cFormat = substring(cFormat, index(cFormat, "=") + 1, length(cFormat))
.
assign cType = entry(2, cType, '"') when substring(cType, 1, 1) = '"'.
assign cFormat = entry(2, cFormat, '"') when substring(cFormat, 1, 1) = '"'.
display
cType skip
cFormat
with side-labels.

Answer 2

define variable qryString as character no-undo.
define variable sep1 as character no-undo.
define variable sep2 as character no-undo.
define variable trimlist as character no-undo.
define variable sep1pos as integer no-undo.
define variable sep2pos as integer no-undo.
define variable part1 as character no-undo format "x(60)".
define variable part2 as character no-undo format "x(60)".
define variable name1 as character no-undo format "x(60)".
define variable name2 as character no-undo format "x(60)".
define variable valu1 as character no-undo format "x(60)".
define variable valu2 as character no-undo format "x(60)".
qryString = 'FieldType="String"&FieldFormat="^[a-z0-9!#$%&~'*+/=?^_`~{|}~~-]+$"'.
sep1 = '&'.
sep2 = '='.
trimlist = '"' + "'".
sep1pos = index( qryString, sep1 ).
part1 = substring( qryString, 1, sep1pos - 1 ).
part2 = substring( qryString, sep1pos + 1 ).
sep2pos = index( part1, sep2 ).
name1 = trim( substring( part1, 1, sep2pos - 1 ), trimlist ).
valu1 = trim( substring( part1, sep2pos + 1 ), trimlist ).
sep2pos = index( part2, sep2 ).
name2 = trim( substring( part2, 1, sep2pos - 1 ), trimlist ).
valu2 = trim( substring( part2, sep2pos + 1 ), trimlist ).
display
part1 skip
part2 skip
name1 skip
valu1 skip
name2 skip
valu2 skip
with
side-labels
.
(I have escaped special characters with a "~" inside the quoted string in order to include it in the program rather than get it from whatever input source you have. In real life qryString probably isn't embedded in the program.)

Answer 3

If you know the list of keys you can use the INDEX function to find the starting positions for all of key names and use SUBSTRING to pick apart the string.
display INDEX(<yourvar>,"&FieldFormat").
You could also use the extra option with NUM-ENTRIES and ENTRY to supply a delimiter to use ('"') as long as that will not show up inside of other quotes.

Answer 4

I think unless you want to do a brute force dissection, the only progress function which takes delimiters and quotes into account is IMPORT.
I think you'd have to mess about with streams, but have a look at this:
http://knowledgebase.progress.com/articles/Article/P112126
which discusses importing CSVs with commas in the fields.
I've tried using it, and while this test had issues with stream time-out, it kind of works:
def var test1 as char initial 'FldTp="String"&FldFmt="^[az0-9!#$%&+-/=^]+$"'.
def var kvp as char extent 10 format "x(50)".
def stream test.
input-output stream test through 'cat -u' unbuffered.
put stream test test1 format "X(100)" skip(1).
import stream test delimiter "=" kvp .
input-output stream test close.
display kvp with 1 column.
The interesting thing is that to get the "desired" result, you need to break on the "=" not the "&" since the quotes follow the "=". However, if you had some quoted and some non-quoted values this wouldn't work.
The output form this was:
┌────────────────────────────────────────────────────────────────────┐
│ kvp[1]: FldTp │
│ kvp[2]: String │
│ kvp[3]: &FldFmt │
│ kvp[4]: ^[a-z0-9!#$%&+-/=^]+$ │
│ kvp[5]: │
│ kvp[6]: │
│ kvp[7]: │
│ kvp[8]: │
│ kvp[9]: │
│ kvp[10]: │
└────────────────────────────────────────────────────────────────────┘
so it breaks the values correctly and you just need to remove the "&" from field +