Announcement

**Michael Mattias** · 9 Jan 2009, 11:15 AM

Try using "," for param three to EXPLICITLY define the parse character instead of relying on defaults.

No promises but worth a shot.

MCM

**keith shelton** · 9 Jan 2009, 01:24 PM

Looks like it is working correctly to me. Here is a quote from the help file on PARSE$

.

"If delim$ is not specified or is null (zero-length), standard comma-delimited (optionally quoted) fields are presumed. In this case only, the following parsing rules apply. If a standard field is enclosed in optional quotes, they are removed. If any characters appear between a quoted field and the next comma delimiter, they are discarded. If no leading quote is found, any leading or trailing blank spaces are trimmed before the field is returned."

KS

**John Gleason** · 9 Jan 2009, 02:52 PM

Looks like it is working correctly to me.

Agreed, so one solution is below:

Code:

#COMPILE EXE
DEFLNG A-Z

FUNCTION PBMAIN () AS LONG
   'input string below
   '123,9" PAN,2.99
   x$ = "123,9"" PAN,2.99"
   PRINT x$
   PRINT PARSECOUNT(x$) 'returns 3
   PRINT PARSE$(x$,1) 'returns 123
   PRINT PARSE$(x$,2) 'returns 9" PAN
   PRINT PARSE$(x$,3) 'returns 2.99
   WAITKEY$

END FUNCTION

**Donnie Hughes** · 9 Jan 2009, 03:05 PM

Originally posted by keith shelton View Post

"If delim$ is not specified or is null (zero-length), standard comma-delimited (optionally quoted) fields are presumed. In this case only, the following parsing rules apply. If a standard field is enclosed in optional quotes, they are removed. If any characters appear between a quoted field and the next comma delimiter, they are discarded. If no leading quote is found, any leading or trailing blank spaces are trimmed before the field is returned."

the string is: 123,"9" PAN",2.99
the second field should be 9" PAN or a 9 inch pan. This is standard notation for a standard comma-delimited field parsing. Excel and others parser parse this string correctly. The question is if this is by design to ignore part of a field or an oversite.
Thanks,
Donnie

**Donnie Hughes** · 9 Jan 2009, 10:24 PM

Originally posted by John Gleason View Post

Agreed, so one solution is below:

Code:

#COMPILE EXE
DEFLNG A-Z

FUNCTION PBMAIN () AS LONG
   'input string below
   '123,9" PAN,2.99
   x$ = "123,9"" PAN,2.99"
   PRINT x$
   PRINT PARSECOUNT(x$) 'returns 3
   PRINT PARSE$(x$,1) 'returns 123
   PRINT PARSE$(x$,2) 'returns 9" PAN
   PRINT PARSE$(x$,3) 'returns 2.99
   WAITKEY$

END FUNCTION

Thanks for the suggestion but I can find solutions to work arounds the issue. I would prefer to not have to rewrite the parse function for this one anomaly. I wrote one back in the QBX days but only recently found this issue with PB. One of the solutions however couldn't be to alter the source data. We have a scripting process that reads data from many external data sources. In my opinion the function should handle Standard comma-delimited fields.
Actually in doing some more research the standard definition for this case is defined as:

Fields with embedded double-quote characters must be enclosed within double-quote characters, and the embedded double-quote characters must be represented by a pair of double-quote characters.
1997,Ford,E350,"Super ""luxurious"" truck"

Of which I would have no problems telling our customer they are not adhearing to standard encoding rules.
Thanks,
Donnie

**John Gleason** · 10 Jan 2009, 09:24 AM

It's not too much of a workaround to keep the data as is. Try specifying the delimiter like MCM pointed out above, then additionally trim the double quotes.

added: I was still changing the input string, so to avoid any change, replace any $DQ pairs with a single $DQ first.

Code:

#COMPILE EXE
DEFLNG A-Z

FUNCTION PBMAIN () AS LONG
   'input string below
   x$ = "1997,Ford,E350,""Super """"luxurious"""" truck""
   PRINT x$
   REPLACE CHR$($DQ,$DQ) WITH $DQ IN x$  'added
   PRINT PARSECOUNT(x$)
   PRINT TRIM$(PARSE$(x$,",",1), $DQ)
   PRINT TRIM$(PARSE$(x$,",",2), $DQ)
   PRINT TRIM$(PARSE$(x$,",",3), $DQ)
   PRINT TRIM$(PARSE$(x$,",",4), $DQ)
   WAITKEY$

END FUNCTION

**Donnie Hughes** · 11 Jan 2009, 11:24 AM

Originally posted by John Gleason View Post

It's not too much of a workaround to keep the data as is. Try specifying the delimiter like MCM pointed out above, then additionally trim the double quotes.

added: I was still changing the input string, so to avoid any change, replace any $DQ pairs with a single $DQ first.

Code:

#COMPILE EXE
DEFLNG A-Z

FUNCTION PBMAIN () AS LONG
   'input string below
   x$ = "1997,Ford,E350,""Super """"luxurious"""" truck""
   PRINT x$
   REPLACE CHR$($DQ,$DQ) WITH $DQ IN x$  'added
   PRINT PARSECOUNT(x$)
   PRINT TRIM$(PARSE$(x$,",",1), $DQ)
   PRINT TRIM$(PARSE$(x$,",",2), $DQ)
   PRINT TRIM$(PARSE$(x$,",",3), $DQ)
   PRINT TRIM$(PARSE$(x$,",",4), $DQ)
   WAITKEY$

END FUNCTION

This then would fail on strings like 123,"Hughes,Donnie",7.25.

I know one relatively easy work around would be REPLACE CHR$($DQ,$DQ) WITH CHR$(255) IN x$ or some other non ascii character then replace the result back to a single $DQ.

The point is Powerbasic parse function is not handing imbedded $DQ character based on the standard definition of a CSV. Excel and Microsoft parsers do but PB does not.

Thanks,
Donnie

**Michael Mattias** · 11 Jan 2009, 11:46 AM

The point is Powerbasic parse function is not handing imbedded $DQ character based on the standard definition of a CSV

A "standard for CSV?"

I've probably only seen five or six "standards" for CSVs. ( This does not include all the new 'standards' developed by imaginative users.)

e.g.

>This then would fail on strings like 123,"Hughes,Donnie",7.25.

So what would be "correct?"

Four fields:
123
"Hughes
Donnie"
7.25
Four Fields:
123
Hughes
Donnie
7.25
Three fields:
123,
Hughes,Donnie
7.25
Three fields:
123
"Hughes,Donnie"
7.25

Go back to square one: CSV is not a good choice for data containing quotes and/or commas. Change now while you still can.

MCM

**John Gleason** · 11 Jan 2009, 12:45 PM

Here's what Wikipedia has to say about the quandary:

No general standard specification for CSV exists. Variations between CSV implementations in different programs are quite common and can lead to interoperation difficulties. For Internet communication of CSV files, an Informational IETF document (RFC 4180 from October 2005) describes the format for the "text/csv" MIME type registered with the IANA. Another relevant specification is provided by Fielded Text which also covers the CSV format.

Many informal documents exist that describe the CSV format. How To: The Comma Separated Value (CSV) File Format provides an overview of the CSV format in the most widely used applications and explains how it can best be used and supported.

But, since it looks like you can't change your data Donnie, your double substitution idea should work if "Hughes and Donnie" are the desired output as two fields. If it's a one field Hughes,Donnie sort of answer, then... um... there's more work to do.

**Aslan Babakhanov** · 11 Jan 2009, 01:49 PM

This works fine:

Code:

#COMPILE EXE
DEFLNG A-Z

FUNCTION PBMAIN () AS LONG
   'input string below
   '123,"9'' PAN",2.99
   x$ = "123," & $DQ & "9'' PAN" & $DQ & ",2.99"
   
   ? x$
   ? STR$(PARSECOUNT(x$)) 'returns 3
   ? PARSE$(x$,1) 'returns 123
   ? PARSE$(x$,2) 'returns  9'' PAN
   ? PARSE$(x$,3) 'returns 2.99
END FUNCTION

I suppose, that INCH must be identified as dbl Short REMARK, but not as dbl. quote...

**Donnie Hughes** · 11 Jan 2009, 06:29 PM

RFC 4180 defines a Common Format and MIME Type for Comma-Separated Values (CSV) Files

To quote from this document:

If double-quotes are used to enclose fields, then a double-quote
appearing inside a field must be escaped by preceding it with
another double quote. For example:

"aaa","b""bb","ccc"

Likewise, this format adhears to the way excel saves data when saved as a csv. The above is three fields with the second field containing b"bb. Powerbasic parse would only return the b portion of this field and ignore "bb portion of this field.

I have an issue with data being ignored...

Donnie

**John Gleason** · 11 Jan 2009, 08:11 PM

Voila!

Ok, blast away and see if you can break this:

Code:

#COMPILE EXE
#DIM ALL

FUNCTION PBMAIN () AS LONG
   LOCAL x AS STRING, ii AS LONG
   'input string below
   x$ =  "123,""Hughes,Donnie"",7.25,,"""""""",""aaa"",""b""""bb"",""ccc"",""1997"",Ford,""""E350"""",""Super """"luxurious"""" truck""
   PRINT x$
   REPLACE CHR$($DQ,$DQ) WITH $BEL IN x$       'forces PARSE to include whole field by removing $DQ's temporarily
   PRINT PARSECOUNT(x$): PRINT

   DIM yArray(1 TO PARSECOUNT(x$)) AS STRING
   PARSE x, yArray()

   FOR ii = 1 TO PARSECOUNT(x$)
       REPLACE ANY $BEL WITH $DQ IN yArray(ii) 'with single chr replacements, use the ANY keyword for speed.
       PRINT yArray(ii)
   NEXT
   WAITKEY$

END FUNCTION

(I used $BEL here instead of chr$(255). Any entirely unused chr in the data is fine.)

**Donnie Hughes** · 11 Jan 2009, 09:13 PM

Originally posted by John Gleason View Post

Voila!

Ok, blast away and see if you can break this:

Code:

#COMPILE EXE
#DIM ALL

FUNCTION PBMAIN () AS LONG
   LOCAL x AS STRING, ii AS LONG
   'input string below
   x$ =  "123,""Hughes,Donnie"",7.25,,"""""""",""aaa"",""b""""bb"",""ccc"",""1997"",Ford,""""E350"""",""Super """"luxurious"""" truck""
   PRINT x$
   REPLACE CHR$($DQ,$DQ) WITH $BEL IN x$       'forces PARSE to include whole field by removing $DQ's temporarily
   PRINT PARSECOUNT(x$): PRINT

   DIM yArray(1 TO PARSECOUNT(x$)) AS STRING
   PARSE x, yArray()

   FOR ii = 1 TO PARSECOUNT(x$)
       REPLACE ANY $BEL WITH $DQ IN yArray(ii) 'with single chr replacements, use the ANY keyword for speed.
       PRINT yArray(ii)
   NEXT
   WAITKEY$

END FUNCTION

(I used $BEL here instead of chr$(255). Any entirely unused chr in the data is fine.)

Thanks for your effort and suggestion but actually an empty string may look like like ,"", instead of ,"""",. In this case the parse would return " as a result.

It took a little more than my initial easy solution and I solved it with this.

Code:

#COMPILE EXE
DEFLNG A-Z

FUNCTION ParseDQReplace(SPtr AS DWORD,strLen&)
   LOCAL lsptr AS BYTE PTR
   lsPtr = sPtr
   qstr& = 0
   FOR w& = 0 TO strLen&-2
      IF @lsptr[w&] = 34 THEN
         IF w& = 0 OR @lsptr[w&-1] = 44 THEN
            qstr& = -1
         ELSEIF qstr& THEN
            IF @lsptr[w&+1] = 44 THEN
               qstr& = 0
            ELSE
               @lsPtr[w&] = 255
            END IF
         END IF
      END IF
   NEXT
END FUNCTION

FUNCTION PBMAIN () AS LONG
   '123,"Hughes,Donnie",7.25,,"","aaa","b""bb","ccc","1997",Ford,""E350"","Super ""luxurious"" truck"
   x$ = "123,""Hughes,Donnie"",7.25,,"""",""aaa"",""b""""bb"",""ccc"",""1997"",Ford,""""""E350"""""",""Super """"luxurious"""" truck""
   ParseDQReplace(STRPTR(x$),LEN(x$))
   DIM yArray(1 TO PARSECOUNT(x$)) AS STRING
   PARSE x$, yArray()

   FOR ii = 1 TO PARSECOUNT(x$)
       REPLACE CHR$(255,255) WITH $DQ IN yArray(ii) 'with single chr replacements, use the ANY keyword for speed.
       PRINT yArray(ii)
   NEXT
   WAITKEY$
END FUNCTION

Fast and functional.
Thanks,
Donnie

**Vidar Hanto** · 11 Jan 2009, 10:17 PM

My understanding is that PARSE$ works correctly, and the data is used incorrectly.

Code:

#COMPILE EXE
DEFLNG A-Z

FUNCTION PBMAIN () AS LONG
   'input string below
   '123,"9" PAN",2.99
   x$ = "123,""9"" PAN"",2.99" ' < -----Incorrect
   x$ = "123,9"" PAN,2.99"     ' < -----Correct
   PRINT PARSECOUNT(x$) 'returns 3
   PRINT PARSE$(x$,1) 'returns 123
   PRINT PARSE$(x$,2) ' did return 9 but now returns 9" PAN
   PRINT PARSE$(x$,3) 'returns 2.99
   WAITKEY$


END FUNCTION

The thing is that X$, being defined as "123,""9"" PAN"",2.99" is internally stored as:

Code:

123,"9" PAN",2.99

so will "123,9"" PAN,2.99" be stored as

Code:

123,9" PAN,2.99

before PARSE$ gets to work with it. This format PARSE$ can process as "expected". PARSE$ does not claim to be a CSV parser (on the contrary), and PB do reduce the DQs correctly when storing the string. I agree with Keith when he says PARSE$ do work as documented.

If you need to preprocess your strings prior to feeding them to PARSE$ as a result of this, it seems that there are a few CSV parsers available in the source forum. You may even get all you need from such a parser, and be able to do without PARSE$ entirely. Here's one CSV parser: http://www.powerbasic.com/support/pb...V+string+parse

ViH

PS: I see you've got yourself a working solution while I was typing

, but my comment on PARSE$ still applies.

**John Gleason** · 12 Jan 2009, 09:50 AM

Donnie said: an empty string may look like like ,"", instead of ,"""",. In this case the parse would return " as a result.

Is it possible that an empty string can look like like """"? If so, your algo returns ". Also I noted that for ""E350"" your algo adds a leading space and returns no quotes, that is, | E350| but I was thinking it should be |"E350"|.

**Donnie Hughes** · 12 Jan 2009, 03:25 PM

Originally posted by John Gleason View Post

Is it possible that an empty string can look like like """"? If so, your algo returns ". Also I noted that for ""E350"" your algo adds a leading space and returns no quotes, that is, | E350| but I was thinking it should be |"E350"|.

Thanks for looking at it and the comments you made. Actually the string of ,"""", is by standard definition a single double quote.

You are correct that the ""E350"" is not a standard definition and returns | E350 | since it should be """E350""". What is actually happening is the inside single occurance of the double quote was replace with a single char 255 and it does not display with a print statement which apprears like a space. Because a supplier that is providing data may send me a string like this in the production program I decide to to treat that as
replace chr$(255,255) with $dq in x$
replace chr$(255) with $dq in x$

This way it will will not discard any data, adhear to the standard definition and allow a little forgiveness to the standard without abandoning it.
Thanks,
Donnie

**Michael Mattias** · 13 Jan 2009, 10:39 AM

[CSV Files] ..by standard definition...is not a standard definition .... adhear[sic] to the standard definition ... to the standard ...

You, sir, must see even empty glasses as half-full.

MCM

**Donnie Hughes** · 13 Jan 2009, 11:23 PM

Not sure why the attack... I may have overused a term that was used in the PB help document or somehow personally offended you... Sorry if I did.

I work for a fairly large software company ($220m) and one of the things we provide is a scripting tool that converts data from many sources from about 8 format types. We don't define the format the source provider sends to our customers, we just provide the conversions and tranformation mapping scripts. Once converted to XML, we have a standard import process to load data into our databases. The transformation tool was written with PB.

One of the formats the conversion tool supports is CSV. Our customers call our help desk if they are trying to load a description of 9" Pan for an item they're selling through their scanning system and it only scans as 9.

One thing we've learned years ago was to abandon the phrase "works as designed". This tends to tick off our customers. We challenge ourselves to make sure the design is correct. I didn't post this thread in an effort to query a solution but rather to challenge the design. It doesn't matter to me any more since I've worked around the problem.

Thanks for listening...

Announcement

Parse$ error

Parse$ error

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment