Say, I have a command command which prints huge number of lines to stdout:
line1
line2
.....
lineN
I want to save the output to disk, but not as a single file, but as a sequence of files each having 1000 lines of stdout:
file0001.txt:
-------------
line1
....
line1000
file0002.txt:
-------------
line1001
....
line2000
etc
I've tried to google the answer, but every time google points me to tee command, which is useless in this situation. Probably, I'm entering wrong queries.
Once you are done saving the file, you could always split the file into file pieces or multiple files based on the number of lines.
split -l 1000 output_file
or even better just try
command | split -l 1000 -
This will split the output stream into files with each 1000 lines (default is 1000 lines without -l option).
The below command will give you additional flexibility to put or enforce a prefix to the filename that will be generated when the output is generated and splitted to store into the file.
command | split -l 1000 - small-
split [arguments...] [input e.g. "-" for stdin] [output_prefix], for example: tar -c somedir | split --byes 100MB --numeric-suffixes --suffix-length=3 - somedir.tar.part- would output a bunch of 100MB files named somedir.tar.part-000, 001, 002 ans so on.
--bytes?
You can use a bash script lines.bash
#!/bin/bash
a=0
while IFS='' read -r line
do
printf -v filename "%04d.txt" "$((a++/1000))"
echo "$line" >> $filename
done
and use it as:
cat long_file.txt | bash lines.bash
The only problem I noticed is with * sign in long_file.txt (somebody could correct it).
IFS to empty string to avoid word splitting on read. Use -r to disable backslash escaping on read. Remove -e to avoid backslash escaping on echo. Use quoting to avoid word splitting on echo. Use -v in bash since 4.0 to avoid starting a sub-process. Use post-incrementing as your current code will put in the first file only 999 lines. a=0; while IFS='' read -r line; do printf -v filename "%04d.txt" $((a++/1000)); echo "$line" >> "$filename"; done
printf does not have -v switch. (bash 4.2.10). At least it's not in manpage of printf
man printf documents /usr/bin/printf, that could never in life set an environment variable. See help printf for the printf shell built-in's documentation.
++/ part yet.