Regex question

[Danp2]

I could use some help with a Regex pattern. I am attempting to extract some names and addresses (US format) from some free form text. Examples:

Jane Doe
PO Box 1234
Pensacola, FL 32524


Jane Doe
123 Sunset Blvd
Apt 456
Pensacola, FL 32524


Jane Doe
555 Any Street
Pensacola, FL 32524-1234

Second address line is optional. There are instances where the city or state may be missing. I came up with the following pattern through research / trial & error, but it isn't working exactly as I desire:

(.+)(?>\r\n)(.+)(?>\r\n)(.+)(?>\r\n)?(.+)?,\h+(\w+)? ([-\d]+)

When the second address line is missing, the value for city ends up in the wrong place. Suggestions on how I can fix / improve the pattern?

TIA, Dan

 

[Melba23]

Danp2,

First rule when asking for SRE help - give an example of what you want as the result.

Each of those addresses has a second line, so the "when second line is missing" comment is not a great deal of help. What exactly do you want to see returned from the RegEx?

M23

[Danp2]

Hey Melba,

Sorry for being vague. When I said second line, I'm actually meant the optional address line (Apt 456 in the second example). I'm trying to extract the following elements: name, address1, address2, city, state, and zip.

Dan

[Danp2]

This appears to mostly work, although the 3rd group is capturing the CRLF.

(.+)(?>\R)(.+)(?>\R)((?:.+)(?>\R))?(.+)?,\h+(\w+)?\h([-\d]+)

[Melba23]

Danp2,

I would not bother with an SRE for this - just split the address and read the required lines:

#include <StringConstants.au3>

Global $aAddress[3] = ["Jane Doe" & @CRLF & "PO Box 1234" & @CRLF & "Pensacola, FL 32524", _
                "Jane Doe" & @CRLF & "123 Sunset Blvd" & @CRLF & "Apt 456" & @CRLF & "Pensacola, FL 32524", _
                "Jane Doe" & @CRLF & "555 Any Street" & @CRLF & "Pensacola, FL 32524-1234"]

For $i = 0 To 2
    $aSplit = StringSplit($aAddress[$i], @CRLF, $STR_ENTIRESPLIT)
    ; Combine the first 2 and last lines
    $sExtract = $aSplit[1] & @CRLF & $aSplit[2] & @CRLF & $aSplit[$aSplit[0]]
    ConsoleWrite($sExtract & @CRLF & @CRLF)
Next

Now wait for a real SRE guru to show you how it should be done!

M23

[Danp2]

I mentioned the free form text, but failed to add that the text contains other details unrelated to the address I am attempting to capture. It's essentially a series of form letters where the data has already been merged, and I'm trying to capture the separate portions of the mailing address.

[jchd]

Having fought with international addresses for a long time, I wouldn't be surprised if even addresses in "US format" would prove very hard to deal with in general. Sure one can come up with a fairly good pattern but I strongly suspect that you'll get exceptions that would ruin your efforts. Here a more pedestrian approach is best, AFAICT.

[Melba23]

Danp2,

Second rule (applicable to all requests for help) - give all the necessary data when posing the question so people do not waste their time coding solutions which are obviously not going to work.

M23

[Danp2]

I tried to be detailed in my post, but obviously I left out some critical details. My bad.

[mikell]

A possible way could be to get rid of all the 'unwanted details', it depends a lot on the shape of the original text

Otherwise considered jchd's comment this looks like an impossible task

[Melba23]

Danp2,

Following on from mikell's comment: is there any way you can distinguish the address lines from the rest of the text? Are they the only lines surrounded by blank lines? Are they in a specific place in the document? Do they have an indentation? Anything of that sort would help extract those particular lines for further analysis.

M23

[jdelaney]

If you use option 4, in stringregexp, you can then fix the ubound of the array of arrays as the city, state, zip, and depending on the size of the array of arrays, you can infer if an address2 exists.

[Danp2]

Yes, they are surrounded by blank lines. They are in the same general vicinity for each letter, but the positioning may vary slightly at times.

I believe that I can make it work using the SRE I posted earlier. Thanks for the help!

[BrewManNH]

The best thing I ever learned about RegEx was knowing when not to use it. This is one of those places where a RegEx isn't going to help all that much, and may in fact take far longer to develop than just a FileReadToArray and parsing each line.