Shell Programming and Scripting

Hi, I have the following command that parses an xml file to read a node <port>'s value. Hoever the output comes with spaces.

My requirement is to trim the spaces around the value and assign to a variable.

Code:

 
sed -n 's|<port>\(.*\)</port>|\1|p' ../cfg.xml

How do I go about it?

Here is how you assign output to a variable.

Code:

var=$(sed -n 's|<port>\(.*\)</port>|\1|p' ../cfg.xml)

If you post the input and how you like the output would be, it would be more easy to give you a correct guide.

Using awk to remove leading and trailing spaces:

Code:

awk -F'[<>]' '/<port>/{ gsub(/^[ \t]+|[ \t]+$/,x,$3); print $3 } ' cfg.xml

Just add a space before and after your parenthesis:

Code:

sed -n 's|<port> \(.*\) </port>|\1|p' file
8900

:

Thank you variable assignment worked.

cfg.xml has XML as below....

Code:

 
<server>
........
........
<port>7001</port>
........
........
</server>

Code:

 
var=$(sed -n 's|<port>\(.*\)</port>|\1|p' ../cfg.xml)

O/P:" 7001" (quotes included here for showing space. not part of output)

Expected O/p:"7001"

---------- Post updated at 03:24 PM ---------- Previous update was at 03:17 PM ----------

How do I use awk in conjunction with sed?

My true requirement is to retrieve the value of an xml node that may or may not contain spaces.

the sed code, in my case returns value with spaces even though the XML does not contain spaces.

Code:

sed -n 's|.*<port> *\(.*[^ ]\) *</port>.*|\1|p' ../cfg.xm

Narrative: If you find a line with '<PORT>', optional spaces, some not-space-value, optional spaces and '</port>', turn the entire line into the not-space-value and print it. The * is greedy, so the ' * after <port> will stop just before the first not space value, and we must stop the captured \(...\) value before we scoop up any spaces, so it must end in a non-space; the .* before that is ensured to start with a not space by the greedy prior ' *' -- left greedy beats right greedy. Not that I ensure anything on the line before and after is picked up and not laid back down. We could still be fooled by '<port> 1 2 </port>', but we are assuming that the data is 'well behaved' with one port on any line. If it was really well behaved, there would be no spaces in there. Maybe they were before or after it all, not inside. Never give a white space your trust, it could be spaces, backspaces, tabs, carriage returns, form feeds, iso8859-1 nonbreaking spaces (space + 128), and other control characters that leave no glyph. In ascii, there is [!-~] that you can see, and then there is the rest.

Thanks for the code and narrative.

I tried the code but nothing was returned. No error. No output.


---------- Post updated at 03:31 PM ---------- Previous update was at 03:28 PM ----------

Tried with a single space after the <port> and it worked. Spaces trimmed and output assigned. Thanks much!

Code:

 
sed -n 's|.*<port>  *\(.*[^ ]\) *</port>.*|\1|p' ../cfg.xm