Help refining command

[philbond87]

I am constructing a command that uses a regex to take parts of a filename to form a string that I will insert into an xmp tag.
The filename will have a four digit string, followed by an arbitrary number of characters, an underscore, two or more uppercase letters, an underscore, one or more specific letters ('a', 'b', or 'c') then four digits. (eg. 1234_some-text_XYZ_ab1234)
What I want it to do is construct a string from:

• the first four numbers
• the two or more capital letters
• the last four numbers (removing the letters that may appear before those four digits).

The resulting string (for the example above) should be:
1234XYZ1234

Here is the command that I have thus far:

Code Select Expand

'-transmissionreference<${filename;m/(\d{4}).*_([A-Z]{2,})_?([abc]*\d{4})/;$_=$1.$2.($3=~ s/a|b|c//gr)}'

It almost works...but if my string has an 'a', 'b' or 'c' in the last group (before the final 4 digits) it removes the a, b or c and everything before that, leaving only the last four digits.

The message I get from the ExifTool command is:

Code Select Expand

Warning: Use of uninitialized value $2 in concatenation (.) or string for 'filename'


I would really appreciate any help in correctly constructing this command.
Thank you!

[Phil Harvey]

Ah.  Your regex to remove abc from $3 resets all of the $1,$2,$3 variables.

Maybe try this:

Code Select Expand

'-transmissionreference<${filename;m/(\d{4}).*_([A-Z]{2,})_?[abc]*(\d{4})/;$_=$1.$2.$3}'

...but are you sure you want the "?" after the second "_" ?

- Phil

[philbond87]

Thanks Phil. That does work.

However the reason I was trying to extract the characters from the last group is because of one additional requirement (that I failed to add) –
In addition to 'a', 'b' and/or 'c' there may also be a 'z'... which I do want to keep.

So:
1234_some-additional-characters_XYZ_abz1234

Should yield:
1234_XYZ_z1234

[Phil Harvey]

OK, so this:

Code Select Expand

'-transmissionreference<${filename;m/(\d{4}).*_([A-Z]{2,})_?[abc]*(z?\d{4})/;$_=$1.$2.$3}'

- Phil

[philbond87]

Thanks Phil.

Adding

Code Select Expand

z? to the last group didn't work but adding

Code Select Expand

[z?] sort of did.
However if the 'z' character doesn't appear at the end of the other optional letters it doesn't seem to be found.

[Phil Harvey]

If the z doesn't come last in the letters then my suggestion won't work.

I guess I don't know enough about the possibilities to give you a complete answer.

- Phil

[philbond87]

I really appreciate the help you've provided. It has been tremendous help!

I apologize for not providing enough clarification at the outset, though. The letters that prefix the last 4 digits – including the 'z' – may appear in any order. I just want to keep the 'z' and remove the rest.

I will continue experimenting.
Thanks, Phil

[Luuk2005]

Greetings Philbond87. Its hard to understand, because Im not knowing if you like to keep underscores?  Also the file-extensions are unknown, so I just make regex for any 2-4chars in the extension? Im also thinking maybe you like to add -if '$Filename =~ /(\d{4})_[^_]+_([A-Z]{2,})_(.*\d{4})\.[^.]{2,4}$$/' if there can be filenames in other formats?

Please to test on command line, then to modify like you really need it, because it keeps not only z, but everything besides [abc]

Code Select Expand

exiftool -fast4 -p '$Filename     ${Filename; s/^(\d{4})_[^_]+_([A-Z]{2,})_(.*\d{4})\.[^.]{2,4}$/\1\2\3/; $_=$1.$2.$3=~s/[abc]//gr}' . 
If preferring the m// method ...

Code Select Expand

exiftool -fast4 -p '$Filename     ${Filename; m/^(\d{4})_[^_]+_([A-Z]{2,})_(.*\d{4})\.[^.]{2,4}$/; $_=$1.$2.$3=~s/[abc]//gr}' .

Edit:
If there can only be [abcz] chars between Last_ and Last4Digits ??  Improve match by changing .* into [abcz]*
If there can be any chars between 1st4Digits and 1st_ ??                 Improve match by changing ^(\d{4})_ into ^(\d{4})[^_]*_

[philbond87]

Hello Luuk2005,

Thank you for looking at this. I really appreciate the assistance.

Correct, a file name will always have:

• four digits, followed by an underscore
• an arbitrary number of characters, ending with an underscore
• two or more uppercase letters
• an underscore followed by 'a' and/or 'b' and/or 'c' and/or 'z', in any order (I want to remove all of those letters, if they appear, except for a 'z')

So a sample file name might look like this:

Code Select Expand

1234_any-arbitrary_underscores_and-dashes_XY_azbc5678
And with that example I would like the resulting string to be:

Code Select Expand

1234XYz5678

The current expression I am using is:

Code Select Expand

'-transmissionreference<${filename;s/^(\d{4})[^_]*_([A-Z]{2,})_([abcz]*\d{4}) $/\1\2\3/; $_=$1.$2.$3=~s/[abc]//gr}'

I'm getting the following message:

Code Select Expand

Warning: Use of uninitialized value $3 in concatenation (.) or string for 'filename'

[StarGeek]

Try this
'-transmissionreference<${filename;m/^(\d{4})_.*_([A-Z]{2,})_([abcz]*\d{4})/;$_=$1.$2.($3=~tr/abc//rd)}'

Example

Code Select Expand

C:\>exiftool -p "${filename;m/^(\d{4})_.*_([A-Z]{2,})_([abcz]*\d{4})/;$_=$1.$2.($3=~tr/abc//rd)}"   Y:\!temp\ccccc\1234_any-arbitrary_underscores_and-dashes_XY_azbc5678.jpg
1234XYz5678

[philbond87]

StarGeek,

Thank you – that seems to have done it.

Was the issue that

Code Select Expand

tr/// returns the altered string and

Code Select Expand

s/// simply returns the number of substitutions, if the characters are found?

Very much obliged.

[StarGeek]

Was the issue that

Code Select Expand

tr/// returns the altered string and

Code Select Expand

s/// simply returns the number of substitutions, if the characters are found?

There were other problems in the whole thing, but the problem there was that you can't do a substitution (or use the tr translates operator) on a capture group variable.  An example in straight Perl

Code Select Expand

C:\>perl -e "$1=10;print $1"
Modification of a read-only value attempted at -e line 1.

For tr/abc//rd, the d option means it will delete characters and the r option will return a new string and leave the original untouched. 

[Luuk2005]

Greetings Philbond87. Sorry Im not replying earlier (we have baby kittens!)  Both my regex assume names to be like first examples.
Im not realized that many underscores can be inside of "arbitrary number of characters, ending with an underscore".
So both my regex must change [^_]+ into .+ like StarGeek already fixes for you.

After that, Im thinking its ok to use either $3=~tr/abc//rd or $3=~s/[abc]//gr  ?
Im on Windows so maybe its different, but this often a big problem Im having when either tr///r or s///r ...

With ${Tag; s/whatever/$1$2$3$4/; $_=$1.$2.($3=~s///r).$4;   The $4 never conducts because /r-modifier resets groups like Phil says.
So instead Im just do this instead:     $_=$1.$2.$4($3=~s///r);   with another s/// to move $3 later.
Maybe there is better ways, but Im just learning and not discovered any yet.

So both my $Tags could be like ....
'${Filename; s/^(\d{4})_.+_([A-Z]{2,})_(.*\d{4})\.[^.]{2,4}$/\1\2\3/; $_=$1.$2.($3=~s/[abc]//gr)}'
'${Filename; m/^(\d{4})_.+_([A-Z]{2,})_(.*\d{4})\.[^.]{2,4}$/; $_=$1.$2.($3=~s/[abc]//gr)}'

The \.[^.]{2,4}$ makes sure that names end with _Uppers_4Digits, but not _Uppers_4DigitsAnyText.
I wish I have Linux to test ($3=~s///r) instead of $3=~s///r, so far Im only needing () to do the math.
Sorry if there was any confusion.