Board index » Visual Studio » Newbie Needs Design Help

Visual Studio116
I work with text files that contain messages passed between POS systems and

financial institutions. The messages are text strings that vary in length,

up to 400 characters, with each string being composed of variable length

fields. Here's an example of one of the message strings:



EFQC000855050413180103ACME004200031403 05041300000000SWTH

780000000000 01010356001234567890123456=14121104485478004008PB09901

05044123 Main Street Home Town RI40008

41020091004663 42020840901000325542

47024002480010990001000000000810100000000000830260101000000009900200606663299



The messages themselves are requests sent from the POS terminal to the

financial institution, and replies from the institution back to the POS

terminal. Reply messages are matched to their corresponding request message

when certain fields within the messages (message number, transaction

sequence id, etc.) are identical between the two messages.



I'm trying to write a vb app that will scan the text file, parse the

messages and match request messages to reply messages. I've got a working

program that performs all but the matching part. I'm stumped on how/where to

temporarily store each request message (in its entirety) while the program

is searching for the matching reply message (the reply message string could

be located hundreds of lines after the request message string in the text

file). I'd like to store each complete message so that I can perform

comparisons on the individual fields to report on the status of the

transaction.



I've thought of writing each request message to its own individual file and

naming the file using some value I can use to match the request message to

its reply message, delete the file after the match is made, and use the

remaining undeleted files as an indicator of the number of unmatched

messages. The only drawback to this is that there could be thousands of

messages in the original text file. I guess this same drawback would apply

to using an array as well.



Any suggestions would be appreciated.



Thanks!



Dave


-

"DJ" <ddjames@tampabay.rr.com>wrote



If I read your message correctly, you have a large text file where each line is

a recorded transaction (either request, or reply) and you want to match up

replies with their requests using a few fields to identify the request/reply.



Can it be safe to assume that the request will always be ahead of the reply?

(Going on that assumtion....)



Use the file pointer to identify and locate the different lines in the file. Build

a fixed length identifier that will match the replies with the requests (using

the message number, transaction ID, etc). Finally, use 2 collections to sort

the file into replies, and requests. As you read each line in the file, first note

where the file pointer points to, then read the line, then determine if it is a

request, or reply. Build the transaction ID and add the file pointer value

to either the Requests collection, or the Replies collection, using the transaction

ID as its key. If that line was a reply, check for a matching key in the Requests

collection. If it is found there, you have enough info to make a status report on

that transaction.



In pseudo code, it might look similar to:



Open file

Do While Not EOF

Get File Pointer (Seek)

Input line (Line Input)

Decide Transaction type

If Request type Then

Build ID

Add to request collection

If Reply type Then

Build ID

Check for matching request

If Request found Then

Add status to status report

(Remove Request ???)





For a short example (minus any file activity) paste the following

into a new form:



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

Option Explicit

Private ReQ As Collection

Private ReP As Collection



Private Sub Form_Load()

Dim msg As String, ID As String



Set ReQ = New Collection ' Requests

Set ReP = New Collection ' Replies



' Reading file adds to collections

ReQ.Add 1, BuildID("9", "1000")

ReQ.Add 101, BuildID("10", "1001")

ReP.Add 201, BuildID("9", "1002")

ReQ.Add 301, BuildID("10", "1014")

ReQ.Add 401, BuildID("11", "1016")

ReP.Add 501, BuildID("10", "1001")



' Test current reply (10, 1001)

ID = BuildID("10", "1001")

If IsPresent(ID) Then

msg = "STATUS:" & vbCrLf & vbCrLf

msg = msg & "The Request that begins at file position "

msg = msg & CStr(ReQ(ID)) & vbCrLf

msg = msg & "is followed by a reply that begins at file position "

msg = msg & CStr(ReP(ID)) & vbCrLf

MsgBox msg, vbInformation, "Match Found"

End If



End Sub





Function BuildID(MsgNum As String, TransID As String) As String

BuildID = Space$(24)

Mid(BuildID, 1) = MsgNum

Mid(BuildID, 12) = TransID

End Function



Function IsPresent(ID As String) As Boolean

On Error Resume Next

IsPresent = (ReQ(ID)>0)

Err.Clear

End Function





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



Note that keeping the functions separate helps in re-use. Also, once

you know the positions in the file where the data is at, you can, at your

next convenience, go back and re-read those lines if needed. (Read

Seek Function and Seek Statement in VB Help) Leaving the data in the

file means you won't be using up computer memory as much as storing

each line would require. Chances are, the file pointer is all you need....



If there is only one reply per request, then you could remove any matches

found (from both collections) so that at the end, you'd know how many

were not matched in the two collections, and if needed, where they are

in the file.



HTH

LFS













-

Larry,



Thanks for the great suggestion. I've utilized collections in Word VBA but

had not even considered their use here. I really appreciate the informative

pseudo code and demo examples!



The replies do not always follow the requests (due to a preprocessor

application that takes the raw data and formats it somewhat) so the Seek tip

will definitely come in handy!!



Thanks again!



~ Dave



"Larry Serflaten" <serflaten@usinternet.com>wrote in message



-