How to loop through a series of files with Bash - bash

I am trying to loop through a series of files and modify them. The files follow a pattern but I can't use pattern because I don't need all the files that match a pattern but just those between a certain sequence of numbers.
Example:
for files in D70_3113.NEF...D70_3330.NEF;do exiftool -GPS...; done

If you want to loop through the list of numbers, you can use a brace expansion:
for files in D70_{3113..3330}.NEF; do exiftool -GPS...; done

It depends on what you can expect from your naming scheme. I can't tell if your files can range from
D70_3113.NEF to D79_9999.NEF
or
D70_3113.NEF to D70_3999.NEF
or what have you. Assuming the latter, you could do:
for files in D70_3[0-9][0-9][0-9].NEF; do exiftool -GPS...; done
...just let the shell's pattern matching do the job for you.
Caveat: If you have too many files, the "for" command line may be too long. In that case you'd need to do find and pipe its output into a "while" loop. But today's command lines can run quite long... over 100,000 characters. See Bash command line and input limit

Related

Randomly shuffling lines in multiple text files but keeping them as separate files using a command or bash script

I have several text files in a directory. All of them unrelated. Words doesn't repeat in each file. Each line has 1 to 3 words in it such as:
apple
potato soup
vitamin D
banana
guinea pig
life is good
I know how to randomize each file:
sort -R file.txt > file-modified.txt
That's great but I want to do this in over 500+ files in a directory and it would take me ages. There must be something better.
I would like to do something like:
sort -R *.txt -o KEEP-SAME-NAME-AS-ORIGINAL-FILE-ADD-SUFFIX-TO-ALL.txt
Maybe this is possible with an script that go through each file in the directory until finished.
Very important is every file should only randomize the words within itself and do not mix with the other files.
Thank you.
Something like this one-liner:
for file in !(*-modified).txt; do shuf "$file" > "${file%.txt}-modified.txt"; done
Just loop over the files and shuffle each one in turn.
The !(*-modified).txt pattern uses bash's extended pattern matching to not match .txt files that already have -modified at the end of the name so you don't shuffle a pre-existing already shuffled output file and end up with file-modified-modified.txt. Might require a shopt -s extglob first, though that's usually turned on already in an interactive shell session.

Iterating through pairs of files with glob

I'm having a difficult time trying to iterate through a long set of files that I need to pair up to run through some process. I'd like to generate a bit of a batch file, pairing each set of matching files one per line. I've done this kind of thing before when it's a simple replacement (e.g. file1 = something.txt, file2 = something.csv). But in this case, the end of the file string is a random UUID, and I can't figure out how to get bash to properly expand the glob the second file.
Given a directory of files like this:
banana_pre-proc_b101a65a-31c7-5e4f-b433-bac4fb1efc1f.txt
banana_proc_a75b3a3e-7140-1cb6-2ad1-c10f7db6743f.txt
cherry_pre-proc_f5d0716f-c205-b0b4-5c63-d33755767de4.txt
cherry_proc_025ff6d5-534d-0020-5446-5da3ed04adc6.txt
kiwi_pre-proc_26075f3b-e3a2-fc1a-a741-615cacfc1a7e.txt
kiwi_proc_be1760f6-413d-edc0-1efc-a134b1b6bfbb.txt
peach_pre-proc_ecafbb30-3df0-6014-61ee-11d1d5745b53.txt
peach_proc_bb3ea3fc-671e-e024-6e61-06a2bc147363.txt
pear_pre-proc_c2db376f-f351-7141-114e-a2ebc3cfc410.txt
pear_proc_ccb2f16a-27cd-c70d-7aac-ce72c3af6575.txt
How can I get a file that looks like:
banana_pre-proc_b101a65a-31c7-5e4f-b433-bac4fb1efc1f.txt banana_proc_a75b3a3e-7140-1cb6-2ad1-c10f7db6743f.txt
cherry_pre-proc_f5d0716f-c205-b0b4-5c63-d33755767de4.txt cherry_proc_025ff6d5-534d-0020-5446-5da3ed04adc6.txt
kiwi_pre-proc_26075f3b-e3a2-fc1a-a741-615cacfc1a7e.txt kiwi_proc_be1760f6-413d-edc0-1efc-a134b1b6bfbb.txt
peach_pre-proc_ecafbb30-3df0-6014-61ee-11d1d5745b53.txt peach_proc_bb3ea3fc-671e-e024-6e61-06a2bc147363.txt
pear_pre-proc_c2db376f-f351-7141-114e-a2ebc3cfc410.txt pear_proc_ccb2f16a-27cd-c70d-7aac-ce72c3af6575.txt
I thought I could do something like
for f in *pre-proc_*txt; do echo "$f" "${f/-pre-proc_/-proc_}"; done
But that doesn't deal with the UUID at the end of the file. I've tried a few other iterations of this strategy too, but none get any closer. What is the trick to doing this? Obviously for a few files like this, I can just manually do it. But, the actual set of files I need to process is quite long and apart from just pulling them all into a text doc and then using some Vim macro or something, I'm a bit baffled as to how to get Bash to expand the glob like I'm intending.
This seems to work:
for preproc in *_pre-proc*; do
base=${preproc%_pre-proc*}
proc=${base}_proc*
echo $preproc $proc
done
We get a base name by stripping of the _pre_proc<uuid> part, and
then use the base name to find the matching _proc file.
This I think should be sufficient:
printf "%s %s\n" *[-_]proc_*.txt
Glob expansions are sorted and the pairs of files share the same prefix.

For loop in shell script - colons and hash marks?

I am trying to make heads or tails of a shell script. Could someone please explain this line?
$FILEDIR is a directory containing files. F is a marker in an array of files that is returned from this command:
files=$( find $FILEDIR -type f | grep -v .rpmsave\$ | grep -v .swp\$ )
The confusing line is within a for loop.
for f in $files; do
target=${f:${#FILEDIR}}
<<do some more stuff>>
done
I've never seen the colon, and the hash before in a shell script for loop. I haven't been able to find any documentation on them... could someone try and enlighten me? I'd appreciate it.
There are no arrays involved here. POSIX sh doesn't have arrays (assuming you're not using another shell based upon the tags).
The colon indicates a Bash/Ksh substring expansion. These are also not POSIX. The # prefix expands to the number of characters in the parameter. I imagine they intended to chop off the directory part and assign it to target.
To explain the rest of that: first find is run and hilariously piped into two greps which do what could have been done with find alone (except breaking on possible filenames containing newlines), and the output saved into files. This is also something that can't really be done correctly if restricted only to POSIX tools, but there are better ways.
Next, files is expanded unquoted and mutalated by the shell in more ridiculous ways for the for loop to iterate over the meaningless results. If the rest of the script is this bad, probably throw it out and start over. There's no way that will do what's expected.
The colon can be as a substring. So:
A=abcdefg
echo ${A:4}
will print the output:
efg
I'm not sure why they would use a file directory as the 2nd parameter though...
If you are having problems understanding the for loop section, try http://www.dreamsyssoft.com/unix-shell-scripting/loop-tutorial.php

Create a new sequence of files from an existing sequence, along with numbering

I know this question has been asked, but I can't find more than one solution, and it does not work for me. Essentially, I'm looking for a bash script that will take a file list that looks like this:
image1.jpg
image2.jpg
image3.jpg
And then make a copy of each one, but number it sequentially backwards. So, the sequence would have three new files created, being:
image4.jpg
image5.jpg
image6.jpg
And yet, image4.jpg would have been an untouched copy of image3.jpg, and image5.jpg an untouched copy of image2.jpg, and so on. I have already tried the solution outlined in this stackoverflow question with no luck. I am admittedly not very far down the bash scripting path, and if I take the chunk of code in the first listed answer and make a script, I always get "2: Syntax error: "(" unexpected" over and over. I've tried changing the syntax with the ( around a bit, but no success ever. So, either I am doing something wrong or there's a better script around.
Sorry for not posting this earlier, but the code I'm using is:
image=( image*.jpg )
MAX=${#image[*]}
for i in ${image[*]}
do
num=${i:5:3} # grab the digits
compliment=$(printf '%03d' $(echo $MAX-$num | bc))
ln $i copy_of_image$compliment.jpg
done
And I'm taking this code and pasting it into a file with nano, and adding !#/bin/bash as the first line, then chmod +x script and executing in bash via sh script. Of course, in my test runs, I'm using files appropriately titled image1.jpg - but I was also wondering about a way to apply this script to a directory of jpegs, not necessarily titled image(integer).jpg - in my file keeping structure, most of these are a single word, followed by a number, then .jpg, and it would be nice to not have to rewrite the script for each use.
Perhaps something like this. It will work well for something like script image*.jpg where the wildcard matches a set of files which match a regular pattern with monotonously increasing numbers of the same length, and less ideally with a less regular subset of the files in the current directory. It simply assumes that the last file's digit index plus one through the total number of file names is the range of digits to loop over.
#!/bin/sh
# Extract number from final file name
eval lastidx=\$$#
tmp=${lastidx#*[!0-9][0-9]}
lastidx=${lastidx#${lastidx%[0-9]$tmp}}
tmp=${lastidx%[0-9][!0-9]*}
lastidx=${lastidx%${lastidx#$tmp[0-9]}}
num=$(expr $lastidx + $#)
width=${#lastidx}
for f; do
pref=${f%%[0-9]*}
suff=${f##*[0-9]}
# Maybe show a warning if pref, suff, or width changed since the previous file
printf "cp '$f' '$pref%0${width}i$suff'\\n" $num
num=$(expr $num - 1)
done |
sh
This is sh-compatible; the expr stuff and the substring extraction up front is ugly but Bourne-compatible. If you are fine with the built-in arithmetic and string manipulation constructs of Bash, converting to that form should be trivial.
(To be explicit, ${var%foo} returns the value of $var with foo trimmed off the end, and ${var#foo} does similar trimming from the beginning of the value. Regular shell wildcard matching operators are available in the expression for what to trim. ${#var} returns the length of the value of $var.)
Maybe your real test data runs from 001 to 300, but here you have image1 2 3, and therefore you extract one, not three digits from the filename. num=${i:5:1}
Integer arithmetic can be done in the bash without calling bc
${#image[#]} is more robust than ${#image[*]}, but shouldn't be a difference here.
I didn't consult a dictionary, but isn't compliment something for your girl friend? The opposite is complement, isn't it? :)
the other command made links - to make copies, call cp.
Code:
#!/bin/bash
image=( image*.jpg )
MAX=${#image[#]}
for i in ${image[#]}
do
num=${i:5:1}
complement=$((2*$MAX-$num+1))
cp $i image$complement.jpg
done
Most important: If it is bash, call it with bash. Best: do a shebang (as you did), make it executable and call it by ./name . Calling it with sh name will force the wrong interpreter. If you don't make it executable, call it bash name.

How to rename files keeping a variable part of the original file name

I'm trying to make a script that will go into a directory and run my own application with each file matching a regular expression, specifically Test[0-9]*.txt.
My input filenames look like this TestXX.txt. Now, I could just use cut and chop off the Test and .txt, but how would I do this if XX wasn't predefined to be two digits? What would I do if I had Test1.txt, ..., Test10.txt? In other words, How would I get the [0-9]* part?
Just so you know, I want to be able to make a OutputXX.txt :)
EDIT:
I have files with filename Test[0-9]*.txt and I want to manipulate the string into Output[0-9]*.txt
Would something like this help?
#!/bin/bash
for f in Test*.txt ;
do
process < $f > ${f/Test/Output}
done
Bash Shell Parameter Expansion
A good tutorial on regexes in bash is here. Summarizing, you need something like:
if [[$filenamein =~ "^Test([0-9]*).txt$"]]; then
filenameout = "Output${BASH_REMATCH[1]}.txt"
and so on. The key is that, when you perform the =~" regex-match, the "sub-matches" to parentheses-enclosed groups in the RE are set in the entries of arrayBASH_REMATCH(the[0]entry is the whole match,1` the first parentheses-enclosed group, etc).
You need to use rounded brackets around the part you want to keep.
i.e. "Test([0-9]*).txt"
The syntax for replacing these bracketed groups varies between programs, but you'll probably find you can use \1 , something like this:
s/Test(0-9*).txt/Output\1.txt/
If you're using a unix shell, then 'sed' might be your best bet for performing the transformation.
http://www.grymoire.com/Unix/Sed.html#uh-4
Hope that helps
for file in Test[0-9]*.txt;
do
num=${file//[^0-9]/}
process $file > "Output${num}.txt"
done

Resources