Shell Programming and Scripting

Hi,

I'm trying to extract information from one file to update another one and am a bit stuck.

the first file is made up of tags e.g.

<item>a@b.com</item>
jksdhfjkdsh sldkjfds l klsjdf
<item> c@d.com </item>

what i'd like to do is extract the email addresses between these tags, sometimes they have spaces between the email address and the tag as well.

I'm able to grep for the lines but havent been able to extract the string I need.

I then would like to take the email addresses found and update a line in another file which is:

To: x@x.com

I'd like this line to have the email addresses from the other file e.g.

To: a@b.com, c@d.com

Thanks for all your help.

Hi,

Once you grepped the right line, you can use this :
... | sed -e "s/^<item> *\([^ ]*\) *<\/item>/\1/"
It will simply select everything between item tags, without spaces. So, it does not check if the mail adress is right.

After, to update your second file, if the "To: " is unique, you can do :
sed -i -e "s/^\(To:.*\)$/\1, $newAdress/" myFile.txt

Hopping it will help you ^^
Bye

Thanks for the tip, it didnt quite work though. I tested it as follows:

echo "<item> testst <item>" | sed -e "s/^<item> *\([^]*\) *<\/item>/\1/"
sed: -e expression #1, char 33: Unmatched [ or [^

---------- Post updated at 04:25 PM ---------- Previous update was at 04:24 PM ----------

I corrected the item tag but still same error:

echo "<item> testst </item>" | sed -e "s/^<item> *\([^]*\) *<\/item>/\1/"
sed: -e expression #1, char 33: Unmatched [ or [^

---------- Post updated at 04:27 PM ---------- Previous update was at 04:25 PM ----------

Hi,

Sorry it did work, I didnt realise at first there is a space in here [^ ].

Can you explain what the code is doing please?

I'll try the second part now

---------- Post updated at 04:37 PM ---------- Previous update was at 04:27 PM ----------

Hi,

I'm trying to do:

cat file | sed -e "s/^<item> *\([^]*\) *<\/item>/\1/"

but it just seems to output all the file rather than just extracting the strings I need.

Thanks for your help again.

newb1000, this may be a simpler way to do it, and follow:

Code:

echo "<item> a@b.com   </item>" | sed -e 's/<item>//' -e 's/<\/item>//' -e 's/ //'

What is happening here is, you search for pattern that you are looking for, for example, <item>, and replace it with nothing - the part between the second and the third /.

- GP

Yes, the space is useful ^^

So, let's explain what happens : sed -e "s/^<item> *\([^ ]*\) *<\/item>/\1/"

"s/blabla1/blabla2/" means we'll replace blabla1 by blabla2.
"^<item> *" means we want a line, which starts with "<empty>" and spaces (or not)
" *<\/item>" means we want a line, which finishes by spaces (or not) and </item>. The backslash is used to not interpret the following slash.
\(blabla\) means "blabla" is useful and we need to save it for after.
[^ ]* means we are looking for everything except spaces.
"\1", in the second part is calling the first important thing saved in the first part -> \(blabla\)

So, we are selecting all the line surrounded by <item>...</item> and only keep what is inside, except space.

Thanks, that does make sense, basically you're removing everything apart from what I'm searching for.

I tried this with:

cat file | grep "<item>" | sed <and your command>

which leaves me with the email addresses I want.

How is it best to then save these results and apply them to the other file?

---------- Post updated at 04:54 PM ---------- Previous update was at 04:51 PM ----------

Thanks man, that makes sense as well now.

There is one bit, I'm still a bit confused with though:

[^ ]*

doesnt that mean look to the start of the line again for one or more spaces?

---------- Post updated at 05:01 PM ---------- Previous update was at 04:54 PM ----------

Hi LeNouveau,

There is something interesting, if I run:

cat file | grep "<item>" | sed <your command> nothing happens

but if i do echo "<item>.......</item>" | sed <your command> then I can see the string gets extracted.

If I try the first method with g.pi's sed command, it works as expected.
Do you know what's happening differently with yours?

Thanks

Usually, when you put something between [], that means you're looking for one of the things inside. But if you insert "^" at the begining, that means you're looking something that is not inside.
It inverts rule matching.

My function is more restrictives than g.pi's.
For example, if there is a space before the "<item>", because of my "^<item>", it will not match anymore and it become useless.

My first idea was :
... | grep -o "[a-zA-Z0-9.-]@[a-zA-Z0-9.-]"
but if there is a not expected character, it does not work anymore.

So, I think the simplest way is g.pi's ^^'