====== Sed ======
{{htmlmetatags>metatag-robots=(follow, index) 
metatag-keywords=(bsd,editor,editors,gnu,linux,regex,regexp,regular expression,sed,unix) 
metatag-description=(Text editing tools, real life examples, *BSD, GNU/linux, linux)
}}

{{tag>editing editor regex text}}

[[.:|{{mdi>arrow-left-circle}}]]

===== GNU/sed =====
==== Case transformation ====
GNU/sed has got the functionality to transform upper-case letters to lower-case letters and vice-versa. Such a transformation can be applied to single letters, or from a start point to an end point.
<code bash>
echo bee | \
sed 's/\([a-z]\)/\u\1/'
  
Bee
</code>

In the example above you see that the first letter matched by the character class '[a-z]' was 'b' and got translated to its upper-case counterpart 'B' by **\u**. The translation from upper-case to lower-case works the same way but instead with **\l**.

The translation of the first letter of every word to upper-case could be done with
<code bash>
echo 'hi this is the sed way of case transformation' | \
sed 's/\b\([a-z]\)/\u\1/g'
  
Hi This Is The Sed Way Of Case Transformation
</code>
Another way is to define a start point and an end point. To start a continuous transformation to upper-case you have to use **\U** instead of **\u** as start point. It ends at the **\E** or at the **\L** wich would start a transformation to lower-case from that point.
<code bash>
echo 'hi this is the sed way of case transformation' | \
sed 's/^\([^t]\+\)\(t[^s]\+s\)\(.*\)/\1\U\2\E\3/'
  
hi THIS is the sed way of case transformation
</code>

==== Substitution ====
=== N-th occurence ===

To substitute the second occurence of 'e' with '3'
<code bash>
$ echo 'hey there' | sed 's/e/3/2'
hey th3re
</code>

Substitute every 'e' with '3' but first
<code bash>
$ echo 'hey there' | sed 's/e/3/2g'
hey th3r3
</code>
To substitute the third 'e' with '3'
<code bash>
$ echo 'hey there' | sed 's/e/3/3'
hey ther3
</code>
==== Snippets ====
=== Extract text from .docx documents ===
<code bash>
unzip -p file.docx word/document.xml | \
sed -e 's/<\/w:p>/\n/g; s/<[^>]\{1,\}>//g; s/[^[:print:]\n]\{1,\}//g'
</code>

 [[https://www.commandlinefu.com/commands/view/4311/extract-plain-text-from-ms-word-docx-files|CommandlineFU]]

=== Join every line in file to single line ===

<code>
sed ':a; N; s/\n/ /; ta' <file>
</code>

  - '':a'' puts a label.
  - ''N'' appends the next line into pattern space (which contains the current line)
  - ''s/\n/ /'' substitues the newline character with a space character
  - ''ta'' jumps to label ''a'' if it's previous command ends successfully

https://askubuntu.com/questions/164056/how-do-i-combine-all-lines-in-a-text-file-into-a-single-line/996766#996766