Substitute pattern within a file with the content of other file

Question

I have a text file (devel.xml).

I added the word REPLACETHIS to it in order to replace this string with the content within a different file (temp.txt).

The closest thing I have is this:

sed -i -e "/REPLACETHIS/r temp.TXT" -e "s///" devel.txt;

This inserts the content after the string, and then deletes the string afterwards.

Is this the best way to do it?

Answer 1

What you've done is to remove SUBSTITUTETHIS wherever it appears in the file (but not the rest of the line where it appears) and insert the content of temp.TXT below that line. If SUBSTITUTETHIS appears multiple times on a line, only the first occurrence is removed, and only one copy of temp.TXT is added.

If you want to replace the whole line when SUBSTITUTETHIS appears, use the d command. Since you need to run both r and d when there's a match, put them in a braced group.

sed -e '/SUBSTITUTETHIS/ {' -e 'r temp.TXT' -e 'd' -e '}' -i devel.txt

Some sed implementations let you use semicolons to separate commands and omit separators altogether around braces, but you still need a newline to terminate the argument to the r command:

sed -e '/SUBSTITUTETHIS/ {r temp.TXT
                          d}' -i devel.txt

If you want to replace SUBSTITUTETHIS by the content of the file, but retain what comes before and after it on the line, it's more complicated. The simplest method is to include the content of the file in the sed command; note that you'll have to properly quote its contents.

sed -e "s/SUBSTITUTETHIS/$(<temp.TXT sed -e 's/[\&/]/\\&/g' -e 's/$/\\n/' | tr -d '\n')/g" -i devel.txt

Or use Perl. This is short but runs cat once for each substitution:

perl -pe 's/SUBSTITUTETHIS/`cat temp.TXT`/ge' -i devel.txt

To read the file once when the script starts, and avoid depending on a shell command:

perl -MFile::Slurp -pe 'BEGIN {$r = read_file("temp.TXT"); chomp($r)}
                        s/SUBSTITUTETHIS/$r/ge' -i devel.txt

(presented on two lines for readability but you can omit the line break). If the file name is variable, to avoid quoting issues, pass it to the script via an environment variable:

replacement_file=temp.TXT perl -MFile::Slurp -pe 'BEGIN {$r = read_file($replacement_file); chomp($r)}
                        s/SUBSTITUTETHIS/$r/ge' -i devel.txt

Answer 2

sed 's/<\!--insert \(.*\.html\) content here-->/&/;s//\ncat "\1"/e'

or this more concise and (I suspect) more efficient refinement from @KlaxSmashing:

sed 's/<\!--insert \(.*\.html\) content here-->/\ncat "\1"/e'

This is a one-liner templating system I made for inserting html partials into another html file. It lets you specify a filename within a pattern. That pattern is then replaced with the content from the filename specified in the pattern.

For example, the line <!--insert somehtmlfile.html content here--> gets replaced with the content of somehtmlfile.html. It appears to work fine without any special handling for characters like & or \.

$ cat first.html
first line of first.html
second line of first.html

$ cat second.html
this is second.html
\ slashes /
and ampersand &
and ! exclamation point
end content from second.html

$ cat file\ name\ space.html
the file containing this content has spaces in the name
and it still works with no escaping.

$ cat input
<!--insert first.html content here-->
input line 1
<!--insert first.html content here-->
input line 2
<!--insert second.html content here-->
input line 3
<!--insert file name space.html content here-->

$ sed 's/<\!--insert \(.*\.html\) content here-->/&/;s//\ncat "\1"/e' input
first line of first.html
second line of first.html
input line 1
first line of first.html
second line of first.html
input line 2
this is second.html
\ slashes /
and ampersand &
and ! exclamation point
end content from second.html
input line 3
the file containing this content has spaces in the name
and it still works with no escaping.

$

Answer 3

Using Raku (formerly known as Perl_6)

~$ raku -pe 'BEGIN my $donor_file = "donor_file.txt".IO.slurp;   \
             s:g/REPLACETHIS/{$donor_file}/;'  recipient_file.txt

This Raku answer uses the familiar s/// substitution idiom, and -pe autoprinting (sed-like) flags are used.

In the first statement the donor_file.txt is slurped (read all-at-once) into the $donor_file variable. In the second statement the :g adverb tells s/// to perform a :global substitution. On the RHS of the substitution operator (replacement half), enclosing the $donor_file variable in {…} curly braces triggers interpolation, inserting the desired text.

https://raku.org