Batch write truncated filename to IPTC caption-abstract

[grole]

Hello,

I've already learned a command to batch write the file name of a jpg to the caption (or caption-abstract as it appears to be called) field:

exiftool "-iptc:caption-abstract<filename" FILE

What I now need is a way to only write part of the filename.

I have a number of photos with 5 digit filenames such as "12345.jpg". Some of these files have an additional caption in the filename, such as "12345-caption text.jpg".

What I want is, of course, only the "caption text" and not the rest of the filename. So what I need is a way to truncate the first 5 digits, or everything before and including the "-", plus the extension. Those files without text would have nothing written, or just blank.

Thanks in advance!

[StarGeek]

Try this
exiftool -if "$filename=~/-/" "-Caption-Abstract<${filename;s/^\d{5}-//;s/(.*)\.[^.]+$/$1/}" FILE

The if checks to see if there is a dash in the filename.  If there is, that means the file has text to be written to Caption-Abstract.  Then, it will take the filename, remove the five digits and the dash from the beginning, remove everything from the last dot to the end, and copy the rest to Caption-Abstract.

[grole]

That's great, thanks! Just what I'm looking for.

Didn't quite work yet though. I get the following warning:

Warning: [minor] Tag '1' not defined - f:temp/02465-Workshop about 1986.jpg
Warning: No writable tags set from f:temp/02465-Workshop about 1986.jpg

Command as entered:

exiftool -if "$filename=~/-/" "-Caption-Abstract<${filename;s/^\d{5}-//;s/(.*)\.[^.]+$/$1/}" f:temp -overwrite_original

[StarGeek]

Ah, sorry, my mistake.  The $/ near the end is interpreted as a new line.  Usually I remember that.  Two ways to fix it.  Either double the dollar sign or put it in parenthesis.  Either of these should work.

exiftool -if "$filename=~/-/" "-Caption-Abstract<${filename;s/^\d{5}-//;s/(.*)\.[^.]+$$/$1/}" f:temp -overwrite_original
exiftool -if "$filename=~/-/" "-Caption-Abstract<${filename;s/^\d{5}-//;s/(.*)\.[^.]+($)/$1/}" f:temp -overwrite_original

[grole]

Hmmm no dice . Still the same warning. I tried both $$ and ($), same result.

I'd troubleshoot myself but I'm not familiar wtih this syntax at all.

Thanks!

[StarGeek]

What version of exiftool? exiftool -ver to find out. 
I'm assuming this is on windows?

I just double checked and my original command and the command with the parenthesis worked properly (double $ didn't).

Code Select Expand

C:\>exiftool -if "$filename=~/-/" "-Caption-Abstract<${filename;s/^\d{5}-//;s/(.*)\.[^.]+$/$1/}" "X:\!temp\12345-Test-Copy.jpg"
    1 image files updated

C:\>exiftool -caption-abstract "X:!temp\12345-Test-Copy.jpg"
Caption-Abstract                : Test-Copy

The syntax is exiftool advanced formatting, which in this case is Perl regex substitution.

[Phil Harvey]

To explain the $/ confusion:

Expressions which allow tag names with a preceding dollar sign (eg. arguments of -if and -p, and the STR of -TAG<STR) interpret $/ as a newline.  The advanced formatting expression doesn't allow tag names, so no pre-processing is done by ExifTool and $/ isn't special (ie. the expression is standard Perl).

- Phil

[grole]

It worked! I was using 8.74 on windows 7. I upgraded to 10.37 and it still wouldn't work, said the file xxx could not be read. I then moved them onto a local disk, as I was working on a USB HDD, and then it worked.

That will help me out for now, but I may need to modify the command for other situations. If I've understood it correctly, I could probably change the trigger of the IF condition by simply changing the first hyphen? For example if it was a blank instead. In that case I'd want to also remove the first x digits, let's say 7 this time, plus the blank. Would this work, or what is the syntax for a blank?

-if "$filename=~/' '/" "-Caption-Abstract<${filename;s/^\d{7}' '//;s/(.*)\.[^.]+($)/$1/}"

Thanks again!

[StarGeek]

Close.  No need for the single quotes.  A space is just a space.

-if "$filename=~/ /" "-Caption-Abstract<${filename;s/^\d{7} //;s/(.*)\.[^.]+($)/$1/}"

To break it down, there are two substitutions, separated by the semi-colon.  Perl regex substitutions take the pattern following the first slash and replace it by what's after the second slash (though there are other characters you can use beside slashes)
s/^\d{7} //
^ - Matches the start of the string
\d - Any Digit character (0-9)
{7} - The previous character, exactly 7 times.  In this case, 7 of any digit.
Then a single space.  Not as important for a filename, but there is also \s, which is any whitespace character, Space, Tab, New Line, Carriage Return, and Form Feed

The second substitution is more complex
s/(.*)\.[^.]+($)/$1/
The parens indicate we're going to capture the match that is between them. 
.*  The dot indicates any single character (except a New Line, but there's a mod for that).  The Asterisk means the previous character is matched 0 or more times, as many times as possible.  This pattern is sorta like a simple asterisk wildcard in Windows, but much more greedy.  For example, if you had a string "ABCDABCDA" and used A.*A as the pattern, the .* would match "BCDABCD", not "BCD".
\. Escapes the dot so it's matched literally
[^.]+ Brackets indicate a group of characters.  The caret inside the brackets indicate a negative, we don't want to match the characters in the brackets.  So the brackets in this case will match any non-dot character.  The + is similar to the asterisk, except it will match 1 or more characters.  This is what is stripping the extension off
The dollar sign matches the end of the string, opposite of the Caret.
The $1 in the second half takes the value that was captured between the parens in the first half as the replacement.

Regex is a complex subject.  The tutorial site I recommend is Regular-Expressions.info and you can test patterns out at Regex101.com.

[grole]

Thanks for the explanation. It confirms that I'd be way over my head in Pearl! However I'll hang onto this for future reference.

Thanks, and Happy Holidays!