Replace spaces in a matched substring with underscores

Question

I'm new to the forum so please forgive any syntactical errors in my question.

I'm trying to replace spaces with underscores in a matched substring only. I figured sed would be the best editor for this but I cannot find the proper code to do this.

Sample line from file1 below:

Some text before pattern to match href="./Dynamic Directory name - Junk_files/irrelevant stuff after match">

Would like to change to this:

Some text before pattern to match href="./Dynamic_Directory_name_-_Junk_files/irrelevant stuff after match">

I thought I was close with this cat file1 |sed '/\.\/.*. Junk_files/ { s/ /_/g; }' but all it did was replace all spaces on the matched line with underscores.

Any help with this would be greatly appreciate. Thanks

Answer 1

Try this, it finds the first pair of slashes and removes all spaces between!

 awk -F'/' '{for(i=2;i<=NF;i++)if(i==2)gsub(" ","_",$i);}1' OFS="/"

Example

file='href="./Dynamic Directory name - Junk_files/irrelevant stuff after match">' 
echo $file | awk -F'/' '{for(i=2;i<=NF;i++)if(i==2)gsub(" ","_",$i);}1' OFS="/"
# Output: 
href="./Dynamic_Directory_name_-_Junk_files/irrelevant stuff after match">

Answer 2

Through python,

$ echo 'href="./Dynamic Directory name - Junk_files/irrelevant stuff after match"' |
> python -c "import re;
> import sys;
> print re.sub(r'(?<=\./).*?(?=/)', lambda m: m.group().replace(' ', '_'), sys.stdin.read())
> "
href="./Dynamic_Directory_name_-_Junk_files/irrelevant stuff after match"

Through perl,

$ echo 'href="./Dynamic Directory name - Junk_files/irrelevant stuff' | perl -pe '
> s/\s(?=(?:(?!\.\/).)*?\/)/_/g
> '
href="./Dynamic_Directory_name_-_Junk_files/irrelevant stuff

Answer 3

It's better to use an XML parser.

If you insist on using sed; assuming the pattern stays consistent:

sed -r 's#^([^/]+/[^ ]+) ([^ ]+) ([^ ]+) - ([^ ]+/)#\1_\2_\3_-_\4#' file.txt

This will replace all spaces between two forward slashes (/) with underscores (_). As the input contains /, i have used # as the pattern separator for sed.

Example:

% sed -r 's#^([^/]+/[^ ]+) ([^ ]+) ([^ ]+) - ([^ ]+/)#\1_\2_\3_-_\4#' <<<'Some text before pattern to match href="./Dynamic Directory name - Junk_files/irrelevant stuff after match">'
Some text before pattern to match href="./Dynamic_Directory_name_-_Junk_files/irrelevant stuff after match">

Answer 4

That's HTML, and unless you have a very well defined a simple enough subset of HTML in your file, parsing HTML using regular expressions is a pretty bad idea.

This Perl one-liner works for replacing that substring in that specific context:

printf 'Some text before pattern to match href="./Dynamic Directory name - Junk_files/irrelevant stuff after match">\n' | perl -ne 'if(/(.*?")(.*\/)(.*)/){$x = $1; $y = $2; $z = $3; $y =~ s/ /_/g; print("$x$y$z")}'

Meaning: it will just replace spaces with underscores in the first " and / delimited substring found. But that's about it. If you're parsing a complex document, don't use it. You could make the pattern more strict (for example you could use /href=(.*?")(.*\/)(.*)/ and print("href=$x$y$z")), but that could still fail upon any occurrence of /href=(.*?")(.*\/)(.*)/.

Unless you're parsing a very well defined and simple enough subset of HTML in your file and you're sure something like that won't fail, just use an HTML parser.