write keywords from a db to the Keywords tag in file

[flammekueche]

Hi,
I extract keywords/filename from a sql db and want to add them to the Keywords tag in file.
File can already contains keywords or not.
- I want the file to be update only if a keyword is add
- print the filename of the files changed

I created a shell with 10000+ command lines for each keyword/file like this:

> exiftool -p '$filename: $keywords' -execute -overwrite_original -keywords+="MYKEYWORD" -common_args -q -if 'not $keywords or $keywords!~/MYKEYWORD/'   FILENAME"

This works but is'nt very efficient.

I tried with
> exiftool -stay_open True -@ args.cmd &

and stream some commands, but couldnt make it.

Thanks for help

[Boyd]

I'm sure somebody will have a good answer.

When I read your post I did wonder though is it worth querying your database so that instead of getting a list of 10,000+ images and the keywords for each and every individual image, you get a list of images and which keyword(s) you need to add to those images?

This might be more efficient. If there's 1000 images that need a keyword such as "sunset", 500 that need the keyword "car" and 120 that need both "sunset" and "car" there might be less processing?

[flammekueche]

Oki how do you get those new lists ?

By running a first exiftool on the list of files to create the list of keywords/files ?
And then running exiftool to update them ?
With 300 keywords, there is potentialy 90000 two keywords lists, and with 3 keywords ... 27000000 lists much more than thr number of files.

Im sure there is a simpler direct solution with stream @, I just can make it work.

[Boyd]

As I understand it you extract your keywords and filenames from an SQL database. If there are 300 keywords and 10,00 files you should only need to write 300 lines instead of 10,000. You would specify one keyword and several files on one command line. You might even be able to use a wildcard - eg file42** if you know there are blocks of files that all have the same keyword applying.

Keyword             Add to files
001 sunset    >     file1233, file4345, file3533, file3560, file3355, file8944 etc
002 evening   >     file1233, file1293, file6445 etc
003 river     >     file1233, file1293, file6445 etc
-----
297 snow      >     file1733, file223, file6445 etc
298 horse     >     file0450, file4776, file5654 etc
299 etc etc

[flammekueche]

Yes, but to create the list without duplicate keywords, I need to check if the keyword already exist with exiftool.
Im doing that with:
     -if 'not $keywords or $keywords!~/MYKEYWORD/'
So just call to exiftool is done instead of two.

I can optimize the way you said
But the @stream is by far the best way.
filename list can be very long with long path and sometimes 1000 files for one keyword

BTW I have not 10.00 but 30.000+ images.

[Phil Harvey]

- I want the file to be update only if a keyword is add
- print the filename of the files changed

I created a shell with 10000+ command lines for each keyword/file like this:

> exiftool -p '$filename: $keywords' -execute -overwrite_original -keywords+="MYKEYWORD" -common_args -q -if 'not $keywords or $keywords!~/MYKEYWORD/'   FILENAME"

I assume MYKEYWORD is different for every file?

Your first command to print the filename is slowing down your processing by a factor of 2.  Instead, if you add the -v0 option and capture the output and search for lines beginning with "========" you will have a list of all files that changed.  So your command will look like this:

exiftool -@ my.args -common_args -overwrite_original -q -v0 > file_list.txt

with the my.args file:

Code Select Expand


-if
not $keywords or $keywords !~ /KEYWORD1/
-keywords+=KEYWORD1
FILENAME1
-execute
-if
not $keywords or $keywords !~ /KEYWORD2/
-keywords+=KEYWORD2
FILENAME2
-execute
...

- Phil

[flammekueche]

Wow exactly what I was looking for 

Thank you Phil