Edit
After a long process of roaming the web, re-runs and troubleshoot the script with this wonderful community, the script is functional and does what it’s intended to do. The script itself is probably even further improvable in terms of efficiency/logic, but I lack the necessary skills/knowledge to do so, feel free to copy, edit or even propose a more efficient way of doing the same thing.
I’m greatly thankful to @AernaLingus@hexbear.net, @GenderNeutralBro@lemmy.sdf.org, @hydroptic@sopuli.xyz and Phil Harvey (exiftool) for their help, time and all the great idea’s (and spoon-feeding me with simple and comprehensive examples ! )
How to use
Prerequisites:
parallelpackage installed on your distribution
Copy/past the below script in a file and make it executable. Change the start_range/end_range to your needs and install the parallel package depending on your OS and run the following command:
time find /path/to/your/image/directory/ -type f | parallel ./script-name.sh
This will order only the pictures from your specified time range into the following structure YEAR/MONTH in your current directory from 5 different time tag/timestamps (DateTimeOriginal, CreateDate, FileModifyDate, ModifyDate, DateAcquired).
You may want to swap ModifyDate and FileModifyDate in the script, because ModifyDate is more accurate in a sense that FileModifyDate is easily changeable (as soon as you make some modification to the pictures, this will change to your current date). I needed that order for my specific use case.
From:
'-directory<$DateAcquired/' '-directory<$ModifyDate/' '-directory<$FileModifyDate/' '-directory<$CreateDate/' '-directory<$DateTimeOriginal/'
To:
'-directory<$DateAcquired/' '-directory<$FileModifyDate/' '-directory<$ModifyDate/' '-directory<$CreateDate/' '-directory<$DateTimeOriginal/'
As per exfitool’s documentation:
ExifTool evaluates the command-line arguments left to right, and latter assignments to the same tag override earlier ones.
#!/bin/bash
if [ $# -eq 0 ]; then
echo "Usage: $0 <filename>"
exit 1
fi
# Concatenate all arguments into one string for the filename, so calling "./script.sh /path/with spaces.jpg" should work without quoting
filename="$*"
start_range=20170101
end_range=20201230
FIRST_DATE=$(exiftool -m -d '%Y%m%d' -T -DateTimeOriginal -CreateDate -FileModifyDate -DateAcquired -ModifyDate "$filename" | tr -d '-' | awk '{print $1}')
if [[ "$FIRST_DATE" != '' ]] && [[ "$FIRST_DATE" -gt $start_range ]] && [[ "$FIRST_DATE" -lt $end_range ]]; then
exiftool -api QuickTimeUTC -d %Y/%B '-directory<$DateAcquired/' '-directory<$ModifyDate/' '-directory<$FileModifyDate/' '-directory<$CreateDate/' '-directory<$DateTimeOriginal/' '-FileName=%f%-c.%e' "$filename"
else
echo "Not in the specified time range"
fi
Hi everyone !
Please no bash-shaming, I did my outmost best to somehow put everything together and make it somehow work without any prior bash programming knowledge. It took me a lot of effort and time.
While I’m pretty happy with the result, I find the execution time very slow: 16min for 2288 files.
On a big folder with approximately 50,062 files, this would take over 6 hours !!!
If someone could have a look and give me some easy to understand hints, I would greatly appreciate it.
What Am I trying to achieve ?
Create a bash script that use exiftool to stripe the date from images in a readable format (20240101) and compare it with an end_range to order only images from that specific date range (ex: 2020-01-01 -> 2020-12-30).
Also, some images lost some EXIF data, so I have to loop through specific time fields:
- DateTimeOriginal
- CreateDate
- FileModifyDate
- DateAcquired
The script in question
#!/bin/bash
shopt -s globstar
folder_name=/home/user/Pictures
start_range=20170101
end_range=20180130
for filename in $folder_name/**/*; do
if [[ $(/usr/bin/vendor_perl/exiftool -m -d '%Y%m%d' -T -DateTimeOriginal "$filename") =~ ^[0-9]+$ ]]; then
DateTimeOriginal=$(/usr/bin/vendor_perl/exiftool -d '%Y%m%d' -T -DateTimeOriginal "$filename")
if [ "$DateTimeOriginal" -gt $start_range ] && [ "$DateTimeOriginal" -lt $end_range ]; then
/usr/bin/vendor_perl/exiftool -api QuickTimeUTC -r -d %Y/%B '-directory<$DateTimeOriginal/' '-FileName=%f%-c.%e' "$filename"
echo "Found a value"
echo "Okay its $(tput setab 22)DateTimeOriginal$(tput sgr0)"
fi
elif [[ $(/usr/bin/vendor_perl/exiftool -m -d '%Y%m%d' -T -CreateDate "$filename") =~ ^[0-9]+$ ]]; then
CreateDate=$(/usr/bin/vendor_perl/exiftool -d '%Y%m%d' -T -CreateDate "$filename")
if [ "$CreateDate" -gt $start_range ] && [ "$CreateDate" -lt $end_range ]; then
/usr/bin/vendor_perl/exiftool -api QuickTimeUTC -r -d %Y/%B '-directory<$CreateDate/' '-FileName=%f%-c.%e' "$filename"
echo "Found a value"
echo "Okay its $(tput setab 27)CreateDate$(tput sgr0)"
fi
elif [[ $(/usr/bin/vendor_perl/exiftool -m -d '%Y%m%d' -T -FileModifyDate "$filename") =~ ^[0-9]+$ ]]; then
FileModifyDate=$(/usr/bin/vendor_perl/exiftool -d '%Y%m%d' -T -FileModifyDate "$filename")
if [ "$FileModifyDate" -gt $start_range ] && [ "$FileModifyDate" -lt $end_range ]; then
/usr/bin/vendor_perl/exiftool -api QuickTimeUTC -r -d %Y/%B '-directory<$FileModifyDate/' '-FileName=%f%-c.%e' "$filename"
echo "Found a value"
echo "Okay its $(tput setab 202)FileModifyDate$(tput sgr0)"
fi
elif [[ $(/usr/bin/vendor_perl/exiftool -m -d '%Y%m%d' -T -DateAcquired "$filename") =~ ^[0-9]+$ ]]; then
DateAcquired=$(/usr/bin/vendor_perl/exiftool -d '%Y%m%d' -T -DateAcquired "$filename")
if [ "$DateAcquired" -gt $start_range ] && [ "$DateAcquired" -lt $end_range ]; then
/usr/bin/vendor_perl/exiftool -api QuickTimeUTC -r -d %Y/%B '-directory<$DateAcquired/' '-FileName=%f%-c.%e' "$filename"
echo "Found a value"
echo "Okay its $(tput setab 172)DateAcquired(tput sgr0)"
fi
elif [[ $(/usr/bin/vendor_perl/exiftool -m -d '%Y%m%d' -T -ModifyDate "$filename") =~ ^[0-9]+$ ]]; then
ModifyDate=$(/usr/bin/vendor_perl/exiftool -d '%Y%m%d' -T -ModifyDate "$filename")
if [ "$ModifyDate" -gt $start_range ] && [ "$ModifyDate" -lt $end_range ]; then
/usr/bin/vendor_perl/exiftool -api QuickTimeUTC -r -d %Y/%B '-directory<$ModifyDate/' '-FileName=%f%-c.%e' "$filename"
echo "Found a value"
echo "Okay its $(tput setab 135)ModifyDate(tput sgr0)"
fi
else
echo "No EXIF field found"
done
Things I have tried
- Reducing the number of
ifcalls
But it didn’t much improve the execution time (maybe a few ms?). The syntax looks way less readable but what I did, was to add a lot of or ( || ) in the syntax to reduce to a single if call. It’s not finished, I just gave it a test drive with 2 EXIF fields (DateTimeOriginal and CreateDate) to see if it could somehow improve time. But meeeh :/.
#!/bin/bash
shopt -s globstar
folder_name=/home/user/Pictures
start_range=20170101
end_range=20201230
for filename in $folder_name/**/*; do
if [[ $(/usr/bin/vendor_perl/exiftool -m -d '%Y%m%d' -T -DateTimeOriginal "$filename") =~ ^[0-9]+$ ]] || [[ $(/usr/bin/vendor_perl/exiftool -m -d '%Y%m%d' -T -CreateDate "$filename") =~ ^[0-9]+$ ]]; then
DateTimeOriginal=$(/usr/bin/vendor_perl/exiftool -d '%Y%m%d' -T -DateTimeOriginal "$filename")
CreateDate=$(/usr/bin/vendor_perl/exiftool -d '%Y%m%d' -T -CreateDate "$filename")
if [ "$DateTimeOriginal" -gt $start_range ] && [ "$DateTimeOriginal" -lt $end_range ] || [ "$CreateDate" -gt $start_range ] && [ "$CreateDate" -lt $end_range ]; then
/usr/bin/vendor_perl/exiftool -api QuickTimeUTC -r -d %Y/%B '-directory<$DateTimeOriginal/' '-directory<$CreateDate/' '-FileName=%f%-c.%e' "$filename"
echo "Found a value"
echo "Okay its $(tput setab 22)DateTimeOriginal$(tput sgr0)"
else
echo "FINISH YOUR SYNTAX !!"
fi
fi
done
- Playing around with find
To recursively find my image files in all my folders I first tried the find function, but that gave me a lot of headaches… When my image file name had some spaces in it, it just broke the image path strangely… And all answers I found on the web were gibberish, and I couldn’t make it work in my script properly… Lost over 4 yours only on that specific issue !
To overcome the hurdle someone suggest to use shopt -s globstar with for filename in $folder_name/**/* and this works perfectly. But I have no idea If this could be the culprit of slow execution time?
- Changing all
[ ]into[[ ]]
That also didn’t do the trick.
How to Improve the processing time ?
I have no Idea if it’s related to my script or the exiftool call that makes the script so slow. This isn’t that much of a complicated script, I mean, it’s a comparison between 2 integers not a hashing of complex numbers.
I hope someone could guide me in the right direction :)
Thanks !
I have not tested this, but I have a couple ideas off the top of my head.
#1 - Retrieve all fields with a single exiftool command. e.g. ALL_DATES=$(exiftool -m -d '%Y%m%d' -T -DateTimeOriginal -CreateDate -FileModifyDate -DateAcquired -ModifyDate "$filename")
Then retrieve individual fields from $ALL_DATES with something like awk. e.g. echo $ALL_DATES | awk '{print $1}' will return the first field (DateTimeOriginal), and changing that to ‘{print $2}’ will return the second field (CreateDate).
#2 - Perhaps process multiple files with a single exiftool call. e.g. exiftool -m -d '%Y%m%d' -T -DateTimeOriginal -CreateDate -FileModifyDate -DateAcquired -ModifyDate ~/Pictures/*. You might compare whether running just this exiftool query once vs running it in a loop takes a significantly different amount of time. If not, it’s probably simpler to use one call per file.
Edit: I doubt the either find or globbing will use a significant amount of time, however, the issues you have with find and spaces in file names can be worked around by using find’s -print0 option. This prints out file paths separated by NUL bytes (i.e. ASCII value 0). You can then loop through them without needing to guess when whitespace is part of the path vs a delimiter. A common way of dealing with this is to pipe the output of find into xargs like so: find ~/Pictures -type f -print0 | xargs -0 -L 1 echo 'File path: '. That will execute echo 'File path: ' <file> for every file in your Pictures folder. It’s a little more complicated. You can also use a for loop like so:
find ~/Pictures -type f -print0 | while IFS= read -r -d '' file_path; do
echo "Processing: $file_path"
done
Note that when you pass a blank string with read -d '', it reads to a NUL char, as documented here: https://www.gnu.org/software/bash/manual/bash.html#index-read . I’m not 100% sure if this is true in older versions of Bash or other similar shells.
Hey ha :) Sorry to pin you, I just wanted to give you a little update !
I was maybe a bit to hasty to cheer up. With the last updated script, It actually behaved strangely, some if statements were skipped, looped strangely through my folder and it wrongly ordered many files with the wrong Field. Also it raised many strange errors I couldn’t make any sense of while looking at the script…
HOWEVER !!
I changed how my script loops through my files according to your find suggestions:
#!/bin/bash
start_range=20160101
end_range=20221212
folder_name=/home/user/Pictures
find ~/Pictures -type f -print0 | while IFS= read -r -d '' file_path; do
image_path=$file_path
ALL_DATES=$(/usr/bin/vendor_perl/exiftool -m -d '%Y%m%d' -T -DateTimeOriginal -CreateDate -FileModifyDate -DateAcquired -ModifyDate "$file_path")
for images in $file_path; do
if [[ $(echo $ALL_DATES | awk '{print $1}') =~ ^[0-9]+$ ]] && [[ $(echo $ALL_DATES | awk '{print $1}') -gt $start_range ]] && [[ $(echo $ALL_DATES | awk '{print $1}') -lt $end_range ]]; then
/usr/bin/vendor_perl/exiftool -api QuickTimeUTC -d %Y/%B '-directory<$DateTimeOriginal/' '-FileName=%f%-c.%e' "$images"
echo "Okay its $(tput setab 22)DateTimeOriginal$(tput sgr0)"
While it doesn’t give me the “BIG improvement” I dreamed yesterday, it does give a non-neglieable improvement ! Also there are no errors anymore and the script loops perfectly through all files… So it’s a win :D.
real 11m5,139s
user 9m47,445s
sys 1m16,224s
To further improve the script, I probably need to find a way to make only a single exiftool call, as you suggested, but that’s probably out of my league… Also I already lost enough amount of time to somehow put this together without any coding skills and just wrongly make use of a tool I have no clue about !!!
Thank you very much for your help :))))
Glad it’s working! Couple more quick ideas:
Since you’re looping through the results of find, $file_path will be a single path name, so you don’t need to loop over it with for images in $file_path; anymore.
I think you’re checking each field of the results in its own if statement, e.g. if [[ $(echo $ALL_DATES | awk '{print $1}')... then if [[ $(echo $ALL_DATES | awk '{print $2}')... etc. While I don’t think this is hurting performance significantly, it would make your code easier to read and maintain if you first found the correct date, and then did only one comparison operation on it.
For example, exiftool -m -d '%Y%m%d' -T -DateTimeOriginal -CreateDate -FileModifyDate -DateAcquired -ModifyDate "$file_path" returns five columns, which contain either a date or “-”, and it looks like you’re using the first column that contains a valid date. You can try something like this to grab the first date more easily, then just use that from then on:
FIRST_DATE=$(exiftool -m -d '%Y%m%d' -T -DateTimeOriginal -CreateDate -FileModifyDate -DateAcquired -ModifyDate "$file_path" | tr -d '-' | awk '{print $1}')
tr -d '-' will delete all occurrences of ‘-’. That means the result will only contain whitespace and valid dates, so awk '{print $1}' will print the first valid date. Then you can simply have one if statement:
if [[ "$FIRST_DATE" != '' ]] && [[ "$FIRST_DATE" -gt $start_range ]] && [[ "$FIRST_DATE" -lt $end_range ]]; then
Hope this helps!
Update
I found something interesting ! It seems that the tag FileModifyDate is not being processed in the script! After removing all time tags except FileModifyDate the file is not even processed, it directly goes to the else statement ! I’m still digging :D
Hey again !
Thank you very much for sharing your knowledge and ELI !
I gave it a try and while I understand what it does (thanks to your concise and easy to understand examples) and how it should be implemented exiftool seems to behave very strangely (maybe a bug, but I guess skill issue).
FIRST_DATE=$(exiftool -m -d '%Y%m%d' -T -DateTimeOriginal -CreateDate -FileModifyDate -DateAcquired -ModifyDate "$file_path" | tr -d '-' | awk '{print $1}')
if [[ "$FIRST_DATE" != '' ]] && [[ "$FIRST_DATE" -gt $start_range ]] && [[ "$FIRST_DATE" -lt $end_range ]]; then
/usr/bin/vendor_perl/exiftool -api QuickTimeUTC -d %Y/%B '-directory<$DateTimeOriginal/' '-directory<$CreateDate/' '-directory<$FileModifyDate/' '-directory<$DateAcquired/' '-directory<$ModifyDate/' '-FileName=%f%-c.%e' "$file_path"
As per the exiftool documentation (source example 12)
-directory<$DateTimeOriginal/' '-directory<$CreateDate/' '-directory<$FileModifyDate/' '-directory<$DateAcquired/' '-directory<$ModifyDate/'
In this command the Directory tag is set from multiple other date/time tags. ExifTool evaluates the command-line arguments left to right, and latter assignments to the same tag override earlier ones, so the Directory for each image is ultimately set by the rightmost copy argument that is valid for that image.
However, it sometimes skip the DateTimeOriginal field and takes FileModifyDate instead even if the first one is present. My guess is that exiftool needs more time to correctly process the file, but it’s only a guess ! Because with the for loop and all elif calls it works without any issues.
Thanks again for your insightful help :)
Side note: I gave a test run with only one time field to see If there is any time gain with calling only the first valid date, while it seems ~2ms slower per file I think it would really make the difference without all the elif calls on the long run !!
Thanks again !!
I would assume that exiftool is where most time is spent. Spitballing here and I’m just ever so slightly drunk (shut up I know it’s Thursday) so don’t take me too seriously, but maybe you could parallelize this somehow?
Something like modifying that script so that it takes a list of files as arguments (escaping might become a problem though because fucking bash) and then munges them, and then either in the shell or in another script you use eg GNU parallel to run that script on yourdir/**/*?
edit: you’d want to check if it’s easier to use cli arguments or stdin in a script being run with parallel, been a while since I used it. Could be either one is fine
Hey :)
Thanks for your input, but I have absolutely no idea what you’re talking about XD. Not because you were slightly drunk (cheers BTW a bit late… But I also had a couple yesterday after bagging my head with bash) but I’m really not familiar with bash or programming logic etc…
Just changing how my script loops through the variables, I went from 16min to 11min not bad at all :) There’s probably quite some room to even further reduce the processing time… But hey that’s the best I can do right now and I’m tired running the script over and over again to “troubleshoot” !
Ah sorry lol, I happily assumed you’re just a bit unfamiliar with bash and not programming in general – you’re definitely doing a great job if that’s the case, shell scripting isn’t exactly easy in a lot of ways like you discovered with how spaces in file names can bork things sometimes. But yeah, 11min isn’t terrible at all and if you don’t have to run that script very often there’s not much point in spending time optimizing it.
Just for future reference in case you end up scripting again, what I was thinking of was instead of processing every filename in $folder_name/**/* one at a time, you could do it in parallel with some creative rewriting, meaning you’d possibly be able to take advantage of having a multicore CPU – but whether it’ll actually be faster depends on a lot of things so it’d basically just have to be tried and timed.
Something along these lines, totally untested, the script can be further optimized and might not even run as-is, but should give you the idea. Starting off with the script, I modified it to just take a single file name as an argument:
#!/bin/bash
# exif_date_filter.sh
if [ $# -eq 0 ]; then
echo "Usage: $0 <filename>"
exit 1
fi
# Concatenate all arguments into one string for the filename, so calling "./script.sh /path/with spaces.jpg" should work without quoting
filename="$*"
start_range=20170101
end_range=20201230
DateTimeOriginal=$(/usr/bin/vendor_perl/exiftool -m -d '%Y%m%d' -T -DateTimeOriginal "$filename")
CreateDate=$(/usr/bin/vendor_perl/exiftool -m -d '%Y%m%d' -T -CreateDate "$filename")
if [[ "$DateTimeOriginal" =~ ^[0-9]+$ ]] || [[ "$CreateDate" =~ ^[0-9]+$ ]]; then
if ([ "$DateTimeOriginal" -gt $start_range ] && [ "$DateTimeOriginal" -lt $end_range ]) ||
([ "$CreateDate" -gt $start_range ] && [ "$CreateDate" -lt $end_range ]); then
/usr/bin/vendor_perl/exiftool -api QuickTimeUTC -d %Y/%B '-directory<$DateTimeOriginal/' '-directory<$CreateDate/' '-FileName=%f%-c.%e' "$filename"
echo "Found a value"
echo "Okay its $(tput setab 22)DateTimeOriginal$(tput sgr0)"
else
echo "FINISH YOUR SYNTAX !!"
fi
fi
So it’s called as ./exif_date_filter.sh /some/image.jpg
Then, assuming you have GNU parallel installed (should be easy to find on just about every distro), you could do… uh, I think just eg. find /home/user/Pictures -type f | parallel ./exif_date_filter.sh to automagically run as many “copies” of that script at the same time as you have CPU cores
Hello :)
Thanks again for your input and spoon feeding me ! I took some time to understand but when I got it I felt excited. Because I’m unable to further improve to script itself using my hardware is such a great idea !!! I would never though of that !!
And after a single test drive on a sample, it was BLAZING FAST (3min on a 2200 sample thanks CPU cores :D)! Thank you very much for sharing your knowledge !! You brought my hope back.
Thanks 😘 👍 🙏
Edit:
I rewrote the script and combined your solution with the script
#!/bin/bash
if [ $# -eq 0 ]; then
echo "Usage: $0 <filename>"
exit 1
fi
# Concatenate all arguments into one string for the filename, so calling "./script.sh /path/with spaces.jpg" should work without quoting
filename="$*"
start_range=20170101
end_range=20201230
FIRST_DATE=$(/usr/bin/vendor_perl/exiftool -m -d '%Y%m%d' -T -DateTimeOriginal -CreateDate -FileModifyDate -DateAcquired -ModifyDate "$filename" | tr -d '-' | awk '{print $1}')
#echo $FIRST_DATE
if [[ "$FIRST_DATE" != '' ]] && [[ "$FIRST_DATE" -gt $start_range ]] && [[ "$FIRST_DATE" -lt $end_range ]]; then
/usr/bin/vendor_perl/exiftool -api QuickTimeUTC -d %Y/%B '-directory<$DateAcquired/' '-directory<$ModifyDate/' '-directory<$FileModifyDate/' '-directory<$CreateDate/' '-directory<$DateTimeOriginal/' '-FileName=%f%-c.%e' "$filename"
else
echo "Error"
fi
running time find /home/user/Pictures/ -type f | parallel ./exif-test.bash took 1min21s for 2200 files without any errors and everything is perfectly organized !!! Thanks a lot !!!
