Finding a pattern in a stream?

[Terry Olsen]

Is there a good way to find a pattern of bytes/chars in a stream? I've got
a serial port connected to a tcp port. I need to be able to catch a unique
character string in the stream so that I can perform certain functions. For
example, I have a telnet client connected to an Apple II through the serial
port. The user at the telnet terminal is using the BBS running on the Apple
II just like the good ole days of dialup BBS's. I need to be able to catch
a "command string" sent out from the Apple II (like the "+++ATH") that tells
my app to disconnect the telnet client (in lieu of the modem hanging up).
Is there a good way to do this?

 

[Ken Tucker [MVP]]

Hi,

Regular expression are the best way to find patterns.
http://msdn.microsoft.com/library/d...s/cpguide/html/cpconCOMRegularExpressions.asp

Ken
-----------------------------
Is there a good way to find a pattern of bytes/chars in a stream? I've got
a serial port connected to a tcp port. I need to be able to catch a unique
character string in the stream so that I can perform certain functions. For
example, I have a telnet client connected to an Apple II through the serial
port. The user at the telnet terminal is using the BBS running on the Apple
II just like the good ole days of dialup BBS's. I need to be able to catch
a "command string" sent out from the Apple II (like the "+++ATH") that tells
my app to disconnect the telnet client (in lieu of the modem hanging up).
Is there a good way to do this?

 

[Terry Olsen]

Just doing a quick read of the page it looks as though Regex functions
similarly to the InStr() function. Is Regex actually faster and more
efficient than using InStr?

Also, how would I find my pattern if it is broken up into two "reads"? For
example, I wait for a DataArrival event, read from the serial port and pass
the data out the TCP port. Say the string i'm looking for is "+++ATH". So
I read the serial port and receive "<other data>+++A" and then on the next
read of the serial port I get "TH<more data>". Or maybe it could
conceivably be broken up into 3 or more reads. Is this a situation where I
just have to concatenate the current read with the previous read and then
look for the pattern? Or is there a better way?

Thanks for the help.
Terry

 

[Jay B. Harlow [MVP - Outlook]]

Terry,
You would need to check each read to see if it contains the characters you
are looking for. Be careful, depending on how you defined the "Read" your
characters may be spread across two reads.


InStr allows you to find a fixed set of characters a "word".

RegEx allows you to find a pattern (a variable set of characters), the Like
operator is a simplified form of RegEx.

For example: The "^\d+$" RegEx pattern says to find the beginning of the
line/string "^" followed by one or more "+" digits "\d", followed by the end
of the line/string "$"


A tutorial & reference on using regular expressions:
http://www.regular-expressions.info/

The MSDN's documentation on regular expressions:
http://msdn.microsoft.com/library/d...l/cpconRegularExpressionsLanguageElements.asp


Hope this helps
Jay

 

[Terry Olsen]

Yeah, that was my question in the post. What if my characters ARE spread
across two reads? Do I need to concatenate the previous read with the
current read and check for the pattern? Or is there another way?

 

[Jay B. Harlow [MVP - Outlook]]

Terry,
You could concatenate the two reads together, or match as much as the
pattern in the first read, remember how much you did match, then match the
rest in the second read.

I would recommend which ever you understand & are the most comfortable with.
Concatenating the two reads obviously will be easier... Where as depending
on what else you are doing (you already have a "parser") the second might be
easy enough to have the parser just handle...

Hope this helps
Jay