UNIX for Dummies Questions & Answers

I need help with this:
Can any one tell me what does these below mean:

1. "\(.\).*") != '/'

2. sed 's+^\./++;s+/.*++'

3. sed "s+${f}/+ +


Thanks in advance

It does not conform with a known regular expression.
I don't know what it means.
The above 'sed' has two replacement commands:
a) s+^\./++
Removes './' from the beginning of each line/string.

b) s+/.*++
Removes anything that starts with '/' from each line/string.
You are missing a double quote (") at the end -- it should be:
sed "s+${f}/+ +"
It replaces the content of variable 'f' followed by one '/' with one space.

I think Shell did a nice job explaining stuff, but since I went to all the trouble of writing this post... before he beat me to it ... I will post anyway

1 - The first one makes no sense, cause it has no context... it looks to me like it was part of an if statement or something like that, in which case it was probably more like this:

Code:

if m/\(.\).*/ != '/'

in which case it still makes no sense, cause you are asking whether "(oneCharacter)anyNumberOfCharacters" = "/"
Which quite obviously it never would

2 - Sed is short for "String Editor", and allows you to manipulate strings, so the 's+++' that you see is really just the same as 's~~~' or 's///' and means:

"Search for the thing in between the first two delimiters (+ for you) and replace it with what is between the second two delimiters... but only do it the first time you find it, and never again"

next regEx knowledge for you is that ^ represents the absolute beginning of a string and $ the absolute end... also, since sed is technically running everything in between the ticks (') as a script, you can make multiple search/replace calls to the same info, and they are run sequentially, each applying changes to the string that was passed to them...

SO... the second one is looking for a period followed by a forward slash, and removing it (replacing with nothing)... then it passes that new string into the second half which searches for a forward slash followed by any number of single characters (absolutely anything basically) and removes the stuff...

so basically a use for the second one would be to get a list of directory names from a list... the only problem is that you never give any input to sed, so it will never really do anything... so try going to a shell and doing this:

Code:

ls -R | sed 's+^\./++;s+/.*++'

Not that this does anything useful, but you can see what it does compared to the ls -R command without the pipe to sed...

3 - Again, this is just a search/replace, this time replacing with a space... but the difference is that this is looking for a shell variable named 'f', and it is going to put the content of 'f' in that search... so if 'f' is 'Fred' then we are going to search for 'Fred' and replace it with ' '

Actually it's "stream(s) editor"

Sorry folks;
the first one that you mentioned it didn't mean any thing, it was part of this:

DZERO=${0}
PROG=$(basename ${DZERO})
while [ -h ${DZERO} ]; do
DZERO=$(ls -l ${DZERO} | awk '{print $NF}')
done
PDIR=$(dirname ${DZERO})
if [ ${PDIR} = '.' ]; then
PDIR=${PWD}
elif [ $(expr ${PDIR} : "\(.\).*") != '/'[/B] ]; then
PDIR=${PWD}/${PDIR}
fi

For the third one, it's a part of this:

for f in ${files}
do
echo " ${f}"
find ${f} -name '*.*' -print | sed "s+${f}/+ +"
rm -ri ${f}
n=$((${n} + 1))
done

From what I can gather, the third one looks like you want a verbose recursive removal script...only I am not so sure about the 'find ${f}' part...

Here is the deal (I think)...

The ${files} is weird, cause if you are actually looping through file names, then the 'find ${f}' line isnt going to find much unless your directories are named the same as your files, and even then, it is only going to match within the current directory and if it does that, then the sed is pointless, cause it is intended to strip paths...

So, if the '${files}' actually was intended to be a "loop through directory names", then it makes a little more sense. In this case, each 'find ${f}' line would produce a list of all files in all subdirectories under the directory specified by '${f}' which comes from the list '${files}'. As each line comes out, the first occurrence of that directory name would be printed as its own line and removed from all subsequent echo's...

So basically this script prints out everything under a directory that you "theoretically" want to delete, and asks you whether or not you want to delete every file therein... oh, and it keeps a count... not that that seems to matter, since you do nothing with it.

I had a friend once that aliased a script like this to his "rm -rf" command so that if he stupidly ran an "rm -rf /" he would get some 3 million lines of echoed files, and realize that he was dumb...

Hope this helps.