Tricky Regexp Problem, How to specify a slightly varying pattern

Hello all,

I am working with some files that have weird/dumb naming conventions and within the file's name, I need to extract some info so some other stuff I'm doing for a school database project.

The name of a typical file looks something like this:

G001E94A13g1_NM_MAE1317.xls

Where the combination of letters and numbers allow for users to know at a glance the student's name, major, and department among other things.

I want to be able to extract only G001E94A13 from the file name as the g1 is typically associated with something else that isn't necessary to know in the project's scope.

I am hoping to apply this regexp to an entire catalog of these files where the naming convention varies slightly. For example, some students will have two last names and other info that changes this "at a glance" file name. For reference it could look something like:

JG002C96E15X_TX_EE1317.xls

And all I would want is JG002C96E15. The big change is the extra last name that gives the initials JG and the big X after 15. The commonality is that the naming conventions don't change drastically enough to where I can always say that the first portion of the file name that I want will always end with the last two digits of the year the student started the program (15 in this one and 13 in the last one).

I am hoping there is a way to tell the regexp to grab the letters and numbers starting at the beginning of the file name, stopping at the first _, while not grabbing _ or the first set of numbers/characters to the left of _ but keeping the next number that comes after the first character left of the _.

Sounds convoluted but hopefully this explains what I'm trying to do. I'm thinking it might have to be two regexps or something.

What I have tried is:

fileName(1) = G001E94A13g1_NM_MAE1317.xls;
splitStr = regexp(fileName(1),'\_','split');
[a b] = regexpr(splitStr(1),'^([a-Az-0-9]+)_','match')

And I get:

a = G001E94A13g1

But I want:

a = G001E94A13

Thank you in advance and please let me know if more clarification is needed.

[OCDER]

Can you simplify the issue by providing more examples like you had at the way end? What are the EXACT patterns you are looking for, exceptions, etc. It's easier for us just to see a bunch of examples (instead of an essay) that we can check the results, simply like:

G001E94A13g1_sdfsdf.xls    ==> G001E94A13
JG002C96E15X_TX_EE1317.xls ==> JG002C96E15

What are the patterns that are okay? Start by filling in the brackets. EXAMPLE

"[1-2 letters][numbers][letters][numbers][IGNORE 1-2 char]_[IGNORE THE REST AFTER UNDERSCORE].xls"

[Walter Roberson]

Is it 1 or more uppercase alphabetic letters, followed by upper case hex, ending in the first non-hex, with you wanting to stop just before that non-hex ?

[L]

@Walter I believe the answer to your question is yes.

@OCDER [1-2 letters][numbers][letter][numbers][letter][numbers][IGNORE 1-3 char]_[IGNORE]

The only variances would be that sometimes people forgot to keep things all caps so something like:

SJ003h97I16jg1_NM_CIE1317.xls ==> SJ003h97I16

To be honest, from what I've seen, sifting through the files, this seems to be what it is like. If I see anything other than what has been described than I'll post it. I appreciate both of your help!

Answer 1

[Guillaume]

If I understood correctly, the pattern you want to match is:

• one or more uppercase character at the start: ^[A-Z]+
• followed by three digits: \d{3}
• followed by a single uppercase character: [A-Z]
• followed by two digits: \d{2}
• followed by another single uppercase character: [A-Z]
• followed by another two digits: \d{2}

So:

match = regexp(filename, '^[A-Z]+\d{3}[A-Z]\d{2}[A-Z]\d{2}', 'match', 'once')

[L]

Hey @Guillaume, thanks for your post. I will run this regexp on some of the varying files and let you know how it goes. If it works like I think it will then I'll mark this answer. Thanks again!

[OCDER]

This one seems to work for the examples given, but since sometimes you have lowercases, you could use regexpi instead.

 match = regexpi(filename, '[A-Z]+\d{3}[A-Z]\d{2}[A-Z]\d{2}', 'match', 'once')

[Guillaume]

Yes, regexpi or replace the [A-Z] by [A-Za-z]

[L]

Ok so I have seen one minor hiccup in the pattern. The last set of numbers can either be two or three digits. I tried to modify the above like so:

match = regexpi(filename, '[A-Z]+\d{3}[A-Z]\d{2}[A-Z]\d{2-3}', 'match', 'once')

But that didn't work. Other than this minor thing, that should be it in terms of what can throw the regexp off.

An example:

Na004r92I013nr01_TX_IEE1317.xls ==> Na004r92I013

[OCDER]

match = regexpi(filename, '[A-Z]+\d{3}[A-Z]\d{2}[A-Z]\d{2,3}', 'match', 'once')

[L]

Thank you both! @OCDER you deserve a medal too haha.

[OCDER]

You're welcome! @Guillaume did all the heavy lifting! I only fixed a comma and added an "i" after regexp. :)

[L]

I found another nuance (I swear this should be the last one!);

N004TR92I013nr01_TX_IEE1317.xls ==> N004TR92I013

An extra character in the mix. I've fiddled around with [A-Z,A-Z] but that didn't work.

Additionally, is there any in depth documentation on to increase me knowledge of regexp? How did you guys learn to throw those combinations of characters together?

[OCDER]

match = regexp(filename, '^[A-Z]+\d{3}[A-Z]+\d{2}[A-Z]\d{2}', 'match', 'once')

The "+" is for 1 or more, so [A-Z]+ is 1 or more alphabet from A-Z.

For any help of matlab function, type

 > help regexp
 > doc regexp

Or do an web search for it.