TL;DR
How to filter an ls/find output using grep
with an array as a pattern?
Background story:
I have a pipeline which I have to rerun for datasets which run into an error.
Which datasets are run into an error is saved in a tab separated file.
I want to delete the files where the pipeline has run into an error.
To do so I extracted the dataset names from another file containing the finished dataset and saved them in a bash array {ds1 ds2 ...} but now I am stuck because I cannot figure out how to exclude the datasets in the array from my deletion step.
This is the folder structure (X=1-30):
datasets/dsX/results/dsX.tsv
Not excluding the finished datasets, meaning deleting the folders of the failed and the finished datasets works like a charm
#1. move content to a trash folder
ls /datasets/*/results/*|xargs -I '{}' mv '{}' ./trash/
#2. delete the empty folders
find /datasets/*/. -type d -empty -delete
But since I want to exclude the finished datasets I thought it would be clever to save them in an array:
#find finished datasets by extracting the dataset names from a tab separated log file
mapfile -t -s 1 finished < <(awk '{print $2}' $path/$log_pf)
echo ${finished[#]}
which works as expected but now I am stuck in filtering the ls output using that array:
*pseudocode
#trying to ignore the dataset in the array - not working
ls -I${finished[#]} -d /datasets/*/
#trying to reverse grep for the finished datasets - not working
ls /datasets/*/ | grep -v {finished}
What do you think about my current ideas?
Is this possible using bash only? I guess in python I could do that easily
but for training purposes, I want to do it in bash.
grep can get the patterns from a file using the -f option. Note that file names containing newlines will cause problems.
If you need to process the input somehow, you can use process substitution:
grep -f <(process the input...)
I must admit I'm confused about what you're doing but if you're just trying to produce a list of files excluding those stored in column 2 of some other file and your file/directory names can't contain spaces then that'd be:
find /datasets -type f | awk 'NR==FNR{a[$2]; next} !($0 in a)' "$path/$log_pf" -
If that's not all you need then please edit your question to clarify your requirements and add concise testable sample input and expected output.
Related
I have a very long list of files stored in a text file (missing-files.txt) that I want to locate on my drive. These files are scattered in different folders in my drive. I want to get whatever closest available that can be found.
missing-files.txt
wp-content/uploads/2019/07/apple.jpg
wp-content/uploads/2019/08/apricots.jpg
wp-content/uploads/2019/10/avocado.jpg
wp-content/uploads/2020/04/banana.jpg
wp-content/uploads/2020/07/blackberries.jpg
wp-content/uploads/2020/08/blackcurrant.jpg
wp-content/uploads/2021/06/blueberries.jpg
wp-content/uploads/2021/01/breadfruit.jpg
wp-content/uploads/2021/02/cantaloupe.jpg
wp-content/uploads/2021/03/carambola.jpg
....
Here's my working bash code:
while read p;
do
file="${p##*/}"
/usr/local/bin/fd "${file}" | /usr/local/bin/rg "${p}" | /usr/bin/head -n 1 >> collected-results.txt
done <missing-files.txt
What's happening in my bash code:
I iterate from my list of files
I use FD (https://github.com/sharkdp/fd) command to locate those files in my drive
I then piped it to RIPGREP (https://github.com/BurntSushi/ripgrep) to filter the results and find the closest match. The match I'm looking for should match the same file and folder structure. I only limit it to one result.
Then finally stored it on another text file where I can later then evaluate the lists for next step
Where I need help:
Is this the most effecient way to do this? I have over 2,000 files that I need to locate. I'm open to other solution, this is something I just divised.
For some reason my coded broke, It stopped returning results to "collected-results.txt". My guess is that it broke somewhere in the second pipe right after the FD command. I haven't setup any condition in case it encounters an error or it can't find the file so it's hard for me to determine.
Additional Information:
I'm using Mac, and running on Catalina
Clearly this is not my area of expertise
"Missing" sounds like they do not exist where expected.
What makes you think they would be somewhere else?
If they are, I'd put the filenames in a list.txt file with enough minimal pattern to pick them out of the output of find.
$: cat list.txt
/apple.jpg$
/apricots.jpg$
/avocado.jpg$
/banana.jpg$
/blackberries.jpg$
/blackcurrant.jpg$
/blueberries.jpg$
/breadfruit.jpg$
/cantaloupe.jpg$
/carambola.jpg$
Then search the whole machine, which is gonna take a bit...
$: find / | grep -f list.txt
/tmp/apricots.jpg
/tmp/blackberries.jpg
/tmp/breadfruit.jpg
/tmp/carambola.jpg
Or if you want those longer partial paths,
$: find / | grep -f missing-files.txt
That should show you the actual paths to wherever those files exist IF they do exist on the system.
From the way I understand it, you want to find all files what could match the directory structure:
path/to/file
So it should return something like "/full/path/to/file" and "/another/full/path/to/file"
Using a simple find command you can get a list of all files that match this criteria.
Using find you can search your hard disk in a single go with something of the form:
$ find -regex pattern
The idea is now to build pattern, which we can do from the file missing_files.txt. The pattern should look something like .*/\(file1\|file2\|...\|filen\). So we can use the following awk to do so:
$ sed ':a;N;$!ba;s/\n/\|/g' missing_files.txt
So now we can do exactly what you did, but a bit quicker, in the following way:
pattern="$(sed ':a;N;$!ba;s/\n/\|/g' missing_files.txt)"
pattern=".*/\($pattern\)"
find -regex "$pattern" > file_list.txt
In order to find the files, you can now do something like:
grep -F -f missing_files file_list.txt
This will return all the matching cases. If you just want the first case, i.e.
awk '(NR==FNR){a[$0]++;next}{for(i in a) if (!(i in b)) if ($0 ~ i) {print; b[i]}}' missing_files file_list.txt
Is this the most effecient way to do this?
I/O is mostly usually the biggest bottleneck. You are running some software fd to find the files for one file one at a time. Instead, run it to find all files at once - do single I/O for all files. In shell you would do:
find . -type f '(' -name "first name" -o -name "other name" -o .... ')'
How can I iterate from a list of source files and locate those files on my disk drive?
Use -path to match the full path. First build the arguments then call find.
findargs=()
# Read bashfaq/001
while IFS= read -r patt; do
# I think */ should match anything in front.
findargs+=(-o -path "*/$patt")
done < <(
# TODO: escape glob better, not tested
# see https://pubs.opengroup.org/onlinepubs/009604499/utilities/xcu_chap02.html#tag_02_13
sed 's/[?*[]/\\&/g' missing-files.txt
)
# remove leading -o
unset findargs[0]
find / -type f '(' "${findargs[#]}" ')'
Topics to research: var=() - bash arrays, < <(...) shell redirection with process substitution and when to use it (bashfaq/024), glob (and see man 7 glob) and man find.
I'm trying to create a bash script based on a input file (list.txt). The input File contains a list of files with absolute path. The output should be a bash script (move.sh) which moves the files to another location, preserve the folder structure, but changing the target folder name slightly before.
the Input list.txt File example looks like this :
/In/Folder_1/SomeFoldername1/somefilename_x.mp3
/In/Folder_2/SomeFoldername2/somefilename_y.mp3
/In/Folder_3/SomeFoldername3/somefilename_z.mp3
The output file (move.sh) should looks like this after creation :
mv "/In/Folder_1/SomeFoldername1/somefilename_x.mp3" /gain/Folder_1/
mv "/In/Folder_2/SomeFoldername2/somefilename_y.mp3" /gain/Folder_2/
mv "/In/Folder_3/SomeFoldername3/somefilename_z.mp3" /gain/Folder_3/
The folder structure should be preserved, more or less.
after executing the created bash script (move.sh), the result should looks like this :
/gain/Folder_1/somefilename_x.mp3
/gain/Folder_2/somefilename_y.mp3
/gain/Folder_3/somefilename_z.mp3
What I've done so far.
1. create a list of files with absolute path
find /In/ -iname "*.mp3" -type f > /home/maars/mp3/list.txt
2. create the move.sh script
cp -a /home/maars/mp3/list.txt /home/maars/mp3/move.sh
# read the list and split the absolute path into fields
while IFS= read -r line;do
fields=($(printf "%s" "$line"|cut -d'/' --output-delimiter=' ' -f1-))
done < /home/maars/mp3/move.sh
# add the target path based on variables at the end of the line
sed -i -E "s|\.mp3|\.mp3"\"" /gain/"${fields[1]}"/|g" /home/maars/mp3/move.sh
sed -i "s|/In/|mv "\""/In/|g" /home/maars/mp3/move.sh
The script just use the value of ${fields[1]}, which is Folder_1 and put this in all lines at the end. Instead of Folder_2 and Folder_3.
The current result looks like
mv "/In/Folder_1/SomeFoldername1/somefilename_x.mp3" /gain/Folder_1/
mv "/In/Folder_2/SomeFoldername2/somefilename_y.mp3" /gain/Folder_1/
mv "/In/Folder_3/SomeFoldername3/somefilename_z.mp3" /gain/Folder_1/
rsync is not an option since I need the full control of files to be moved.
What could I do better to solve this issue ?
EDIT : #Socowi helped me a lot by pointing me in the right direction. After I did a deep dive into the World of Regex, I could solve my Issues. Thank you very much
The script just use the value of ${fields[1]}, which is Folder_1 and put this in all lines at the end. Instead of Folder_2 and Folder_3.
You iterate over all lines and update fields for every line. After you finished the loop, fields retains its value (from the last line). You would have to move the sed commands into your loop and make sure that only the current line is replaced by sed. However, there's a better way – see down below.
What could I do better
There are a lot of things you could improve, for instance
Creating the array fields with mapfile -d/ fields instead of printf+cut+($()). That way, you also wouldn't have problems with spaces in paths.
Use sed only once instead of creating the array fields and using multiple sed commands. You can replace step 2 with this small script:
cp -a /home/maars/mp3/list.txt /home/maars/mp3/move.sh
sed -i -E 's|^/[^/]*/([^/]*).*$|mv "&" "/gain/\1"|' /home/maars/mp3/move.sh
However, the best optimization would be to drop that three step approach and use only one script to find and move the files:
find /In/ -iname "*.mp3" -type f -exec rename -n 's|^/.*?/(.*?)/.*/(.*)$|/gain/$1/$2|' {} +
The -n option will print what will be renamed without actually renaming anything . Remove the -n when you are happy with the result. Here is the output:
rename(/In/Folder_1/SomeFoldername1/somefilename_x.mp3, /gain/Folder_1/somefilename_x.mp3)
rename(/In/Folder_2/SomeFoldername2/somefilename_y.mp3, /gain/Folder_2/somefilename_y.mp3)
rename(/In/Folder_3/SomeFoldername3/somefilename_z.mp3, /gain/Folder_3/somefilename_z.mp3)
It's not builtin to bash, but the mmv command is nice for this kind of mv where you need to use wildcards in paths. Something like the following should work:
mmv "in/*/*/*" "#1/#3"
Note that this won't create the directories for you - but in your example above it looks like these already exist?
i'm new in Linux, and i've been trying to run a script which processes all the files in a folder using ImageMagick's convert (I rather do this task in Shell than using mogrify, because as far as I know it doesn't save different files). The files have to be processed in 'last modified' order, so I used this code:
for file in `ls -1tr {*.jpg,*.png}`; do
# imagemagick processes with the filename...
done
This code breaks for files with spaces, and according to this answer using ls is wrong for these purposes.
I also tried this solution from this response, but apparently I got it totally wrong (It raised an 'ambiguous redirect' error) and I decided I needed help.
while read LINE; do
...
done `(ls -1tr {*.png,*.jpg}`
So how do I get an ordered list of filenames for a loop? (It doesn't necessarily have to be a FOR...IN loop, it can be a WHILE, or anything.)
try this :
for file in `ls -1tr {*.jpg,*.png} | awk '{print $9}'`; do
# imagemagick processes with the filename...
done
ls -lrth give's 9 coulmns in output, out of which you need only 9th column(file names), you can get that using awk
If you have space seperated filenames, modify awk print to print all data after 9th column
I have a folder filled with ~300 files. They are named in this form username#mail.com.pdf. I need about 40 of them, and I have a list of usernames (saved in a file called names.txt). Each username is one line in the file. I need about 40 of these files, and would like to copy over the files I need into a new folder that has only the ones I need.
Where the file names.txt has as its first line the username only:
(eg, eternalmothra), the PDF file I want to copy over is named eternalmothra#mail.com.pdf.
while read p; do
ls | grep $p > file_names.txt
done <names.txt
This seems like it should read from the list, and for each line turns username into username#mail.com.pdf. Unfortunately, it seems like only the last one is saved to file_names.txt.
The second part of this is to copy all the files over:
while read p; do
mv $p foldername
done <file_names.txt
(I haven't tried that second part yet because the first part isn't working).
I'm doing all this with Cygwin, by the way.
1) What is wrong with the first script that it won't copy everything over?
2) If I get that to work, will the second script correctly copy them over? (Actually, I think it's preferable if they just get copied, not moved over).
Edit:
I would like to add that I figured out how to read lines from a txt file from here: Looping through content of a file in bash
Solution from comment: Your problem is just, that echo a > b is overwriting file, while echo a >> b is appending to file, so replace
ls | grep $p > file_names.txt
with
ls | grep $p >> file_names.txt
There might be more efficient solutions if the task runs everyday, but for a one-shot of 300 files your script is good.
Assuming you don't have file names with newlines in them (in which case your original approach would not have a chance of working anyway), try this.
printf '%s\n' * | grep -f names.txt | xargs cp -t foldername
The printf is necessary to work around the various issues with ls; passing the list of all the file names to grep in one go produces a list of all the matches, one per line; and passing that to xargs cp performs the copying. (To move instead of copy, use mv instead of cp, obviously; both support the -t option so as to make it convenient to run them under xargs.) The function of xargs is to convert standard input into arguments to the program you run as the argument to xargs.
Basically I need to find a few hundred *.DB3 files on three or four of our network shares and execute a SQLite3 script against all of them, extracting the outputs to a new DB3 file or even CSV file; but my shell scripting is very rusty and I would appreciate any help.
Using either Windows CLI or something that can be executed in Cygwin, how would I build a script to search up to three named shares (or mounted drive letters) for all files with a specific file extension (*.DB3) store these full paths and filenames in an array or file and then remove any duplicated file names.
For all files in this list, I then need to run a predefined SQLite3.exe script (example below) against all of these files and output this data into a new DB3 file or even a CSV file.
SELECT Product.Type AS 'Product', Product.Identifier AS SKU;
The basic footwork is to create a suitably formatted list of file names. The following will create a file where the filename is in the first column, the full path in the second, tab delimited.
find /share/one /path/to/share/two \
/export/third/share -type f -name '*.DB3' -printf '%f\t%p\n'
Pass that to sort | awk and create a simple script to print only unique names. Does the order of the duplicates matter? Perhaps you would like to check which duplicate is older, instead of blindly keeping only the first, or the last? But here is a simple Awk script which does the former.
awk -F '\t' '!a[$1]++ { print $2 }'
To connect the pieces, run find | sort | awk | xargs sqlite3 'your commands'
If you need something a bit more involved, you can read the list of non-duplicate files in a loop:
find /share/one /path/to/share/two \
/export/third/share -type f -name '*.DB3' -printf '%f\t%p\n' |
sort |
awk -F '\t' '!a[$1]++ { print $2 }' |
while read file; do
sqlite3 "complex things with $file; more things; commit; done; etc"
done >output