How to Replace Multiple Lines using Files on Termux

Question

Assume I have many *.txt files on directory texts with the below contents.

Lorem ipsum dolor sit amet, consectetuer adipiscing elit.
Aliquam tincidunt mauris eu risus.
Vestibulum auctor dapibus neque.

And I want to replace them with the following contents recursively.

Vestibulum commodo felis quis tortor.
Ut aliquam sollicitudin leo.
Cras iaculis ultricies nulla.
Donec quis dui at dolor tempor interdum.

As this is a quite large replacement. Typing each one of them can be time consuming.

Hence, I think it would be better if there is an option like this.

Copy and Paste the original texts into file original.txt and the required replacements into another file update.txt.

And then execute a command to find all the *.txt files in the directory texts that consist of the content in original.txt and replace them with the contents of update.txt.

Similar to simple replacements like:

find texts -name "*.txt" -exec sed -i 's/original/update/g' {} \;

I think this way there will be no mistakes as manual typing and less time will be consumed.

But I don't know what command I should use to achieve this? Is this possible.

However, first of all I must be able to verify the availability and number of occurrence of the original text.

Similar to Simple Checks like:

cd texts
grep -r --color=always "original" | wc -l

Thanks.

Answer 1

I'd use perl instead of sed (or awk):

find texts/ -name '*.txt' \
  -exec perl -0777 -p -i.bak -e '
    BEGIN {
      $search = q{Lorem ipsum dolor sit amet, consectetuer adipiscing elit.
Aliquam tincidunt mauris eu risus.
Vestibulum auctor dapibus neque};
  $replace = q{Vestibulum commodo felis quis tortor.

Ut aliquam sollicitudin leo.
Cras iaculis ultricies nulla.
Donec quis dui at dolor tempor interdum.};
    };
s/\Q$search\E/$replace/mg' {} +

• -0777 tells perl to "slurp" in the entire file at once and process it as one long string

• -p makes perl behave similarly to sed (and the counterpart -n option makes it work like sed -n).

• -i.bak does an "in-place" edit of the file, saving the original with a .bak extension. Again, similar to sed -i.

  If you don't want the backup copies, use just -i instead of -i.bak.

• \Q in a perl regex tells perl to treat the following pattern (until it sees a \E) as literal string even if it contains regex special characters.

  From man perlre:

  \Q quote (disable) pattern metacharacters until \E

  \E end either case modification or quoted section

• q{} uses the perl q quoting operator that works exactly the same as single-quotes. It's particularly useful in a one-liner where the perl script is already in single-quotes (which can't be backslash-escaped because escape codes are ignored inside single quotes). See man perlop and search for "Quote and Quote-like Operators". See also perldoc -f q (and compare with perldoc -f qq, the double-quote operator).

BTW, I recommend testing just the perl portion of this on a single file and examine the output to make sure it's going to do what I want (i.e. without find and especially without -i.bak).

Answer 2

Here's one way, done in Debian 11:

• bash v.5.1.4(1)
• tr (GNU Coreutils) v.8.32
• sed (GNU sed) v.4.7

Written this way this script

1. assumes every file in question has .txt extension
2. must be run in the same directory where the .txt -files are
3. displays the name and contents of every file while processing
4. creates new files instead of replacing in place to ensure original files are still available in case of failure. Also this way the newly created files can be deleted with a single command without affecting the originals, if for example the replacement text needs to be changed.

Directory contents:

pg1@TREX:~/foo$ ls
repla.sh  text1.txt  text2.txt

Text file contents:

pg1@TREX:~/foo$ for i in {1..2}; do cat text$i.txt; echo; done
This is the beginning
of the first text.
These lines
will be replaced.
This is the ending
of the first text.
This is the beginning
of the second text.
These lines
will be replaced.
This is the ending
of the second text.

The replacement text:

New lines
that will replace
old lines.

repla.sh:

#!/bin/sh
for i in *.txt; do
    echo $i
    cat text.txt | tr '\n' '\r' | sed -e 's/These lines\rwill be replaced./New lines\rthat will replace\rold lines./g' | tr '\r' '\n' | tee $i.new
done

Result:

pg1@TREX:~/foo$ ./repla.sh 
text1.txt
This is the beginning
of the first text.
New lines
that will replace
old lines.
This is the ending
of the first text.
text2.txt
This is the beginning
of the second text.
New lines
that will replace
old lines.
This is the ending
of the second text.
pg1@TREX:~/foo$ ls
repla.sh  text1.txt  text1.txt.new  text2.txt  text2.txt.new

Script needs to be adjusted as needed, for example to accommodate the situation if there are also text files with different extensions (or without one) to modify or if the files are in different directory/directories. Commenting out echo $i prevents displaying the file names while processing, replacing | tee with > redirection prevents the file contents from showing. In-place replacement requires creating new files, the originals can be replaced simply by adding mv to the script.

Please note Ed Morton's comments below.

There are other solutions, quite probably more elegant; and more suitable for longer paragraphs. For example in this Stack Overflow question the old and replacement paragraphs are in variables.