Delete string of text of output using SED

I have went to https://sed.js.org/ and https://regex101.com and finally https://tecadmin.net/delete-a-line-containing-specific-string-using-sed/. I am unable to delete the front end and the trailing / in a directory string. This would be set to a variable in a bash. Expecting dir=$(sed ...)

Folder: /config/folder/donwloads/Video/Epoc/

I would like to only have the exact test of "Epoc" as the output. I am unable to figure this out using this above resources.

I have tested the string below using both the sed.js.org website and my bash prompt.

echo "/config/folder/downloads/Epoc/" | sed 's/.*[downloads]//'

but let with "c/" as the output.

I have tried to at least remove the trailing / at end .

echo "/config/folder/downloads/Video/Epoc/" | sed 's/.*[Video]//' | sed '/\//d'

Then I get nothing for output.

Am I using the wrong command to do this or just using the command in the incorrect way?

Output of Epoc only with no trailing or fronted "/"

What goes wrong with your regex:

1. s/.*[downloads]// : this performs a greedy search for anything before any of the letters "d","o","w","n","l","a" and "s". It does not search for the word "download", this would be: s/.*downloads//. Anything between square brackets means a set of characters, but represents a single character. Hence the result is c/ because of the letter o in "Epoc".
2. The same applies to the string s/.*[Video]//
3. The string /\//d means delete every line that has a / in it.

This might work for you (GNU sed):

sed -n 's/.*[/]\([^/]\+\)[/]\?$//p' file

Use greed .* to search for the last or last but one /, then capture anything which is not a / until the last optional / before the end of the line.

Alternative:

sed -nE 's/.*\/([^/]+)\/?$/p' file