Execute rm with string on file and delete line - bash

Question

I have a file (log.txt) with multiples lines.
Uploaded 1Y3JxCDpjsId_f8C7YAGAjvHHk-y-QVQM at 1.9 MB/s, total 3.9 MB
Uploaded 14v58hwKP457ZF32rwIaUFH216yrp9fAB at 317.3 KB/s, total 2.1 MB
Each line in log.txt represents a file that needs to be deleted.
I want to delete the file and then delete the respective line.
Example:
rm 1Y3JxCDpjsId_f8C7YAGAjvHHk-y-QVQM
and after deleting the file that the log.txt contains, delete the line, leaving only the others.
Uploaded 14v58hwKP457ZF32rwIaUFH216yrp9fAB at 317.3 KB/s, total 2.1 MB

Answer 1

Try this:
#!/bin/bash
logfile="logfile.txt"
logfilecopy=$( mktemp )
cp "$logfile" "$logfilecopy"
while IFS= read -r line
do
filename=$( echo "$line" | sed 's/Uploaded \(.*\) at .*/\1/' )
if [[ -f "$filename" ]]
then
tempfile=$( mktemp )
rm -f "$filename" && grep -v "$line" "$logfile" >"$tempfile" && mv "$tempfile" "$logfile"
fi
done < "$logfilecopy"
# Cleanup
rm -f "$logfilecopy"
It does:
keep a copy of the original log file.
read each line of this copy using while and read.
for each line, extract the filename. Note, done with sed since a filename could contain spaces. Therefore cut would not work as required.
If the file exists, delete it, remove the line from the log file and store it in a temporary file, move the temporary file into the log file.
that last step is done with && between commands to ensure that the last command is done before continuing. If the rm fails, the log entry must not be deleted.
finally delete the original log file copy.
you can add echo statements and-or -x to $!/bin/bash to debug if required.

Answer 2

The following code reads log.txt line by line, captures the filename with a bash ERE and tries to delete that file. When the regex or the deletion fails it outputs the original line.
#!/bin/bash
tmpfile=$( mktemp ) || exit 1
while IFS='' read -r line
do
[[ $line =~ ^Uploaded\ (.*)\ at ]] &&
rm -- "${BASH_REMATCH[1]}" ||
echo "$line"
done < log.txt > "$tmpfile" &&
mv "$tmpfile" log.txt
remark: the while loop final result is true unless there's a problem reading log.txt or generating "$tmpfile", so chaining the mv with && makes it so that you won't overwrite the original logfile abusively.

Answer 3

Another approach using bash4+ and GNU tools.
#!/usr/bin/env bash
##: Save the file names in an array named files using mapfile aka readarray.
##: Process Substitution and With GNU grep that supports the -P flag.
mapfile -t files < <(grep -Po '(?<=Uploaded ).*(?= at)' log.txt)
##: loop through the files ("${files[#]}") and check if it is existing (-e).
##: If it does, save them in an array named existing_file.
##: Add an additional test if need be, see "help test".
for f in "${files[#]}"; do
[[ -e $f ]] && existing_file+=("$f")
done
##: Format the array existing_file into a syntax that is accepted
##: by GNU sed, e.g. "/file1|file2|file3|file4/d" and save it
##: in a variable named to_delete.
to_delete=$(IFS='|'; printf '%s' "/${existing_file[*]}/d")
##: delete/remove the existing files.
##: Not sure if ARG_MAX will come up.
echo rm -v -- "${existing_file[#]}"
##: Remove the deleted files (lines that contains the file name)
##: from log.txt using GNU sed.
echo sed -E -i "$to_delete" log.txt
Remove all the echo if you're satisfied with the output.
This not exactly what you asked for and it is not perfect but it just might be what you need.