sed command to extract from xml

I'm using my mac terminal to do a script, it basically does:

wget http://p2.edms-pr.ccomrcdn.com/player/player_dispatcher.html?section=radio&action=listen_live

This file returns an XML which I can save to txt or XML, I'm saving it as "url.xml"

<PlayerContent>
<ListenLiveInitialize>
<StreamInfo>
<stream id="4694" primary_location="rtmp://cp58082.live.edgefcs.net/live/COR_5103_OR@s5137?auth=daEaIcRcbb.afahbOdwbWdjdYcEdYaOaDdc-bn7nM7-4q-PN0X1_3nqDHom4EBvmEuwr&aifp=1234&CHANNELID=4694&CPROG=_&MARKET=PREMIERE&REQUESTOR=EDMS-PR&SERVER_NAME=p2.edms-pr.ccomrcdn.com&SITE_ID=13293&STATION_ID=EDMS-PR&MNM=_&TYPEOFPLAY=0" backup_location=""/>
</StreamInfo>
<JustPlayed/>

I want to used SED to return the AUTH code inside "primary_location". So basically I want to store

daEaIcRcbb.afahbOdwbWdjdYcEdYaOaDdc-bn7nM7-4q-PN0X1_3nqDHom4EBvmEuwr

o开发者_运维技巧n a variable.

I found this online but it doesn't seem to be working.

sed -n 's/.*\(auth=......................................... ...........................\).*/\1/p' url.xml

Try

sed -n 's|^<stream.*auth\=\(.*\)\&ai.*|\1|p' url.xml

which reads the file and matches the line up to the = before the auth code, stores everything from there up to the & in &ai as \1 which is then substituted for the whole pattern space.

You have a stray space () in the middle of your .s!

This is neater and will output auth= with the value (it looks like it's a string of alphanumerics with hyphens and underscores):

% grep -o 'auth=[[:alnum:]_-]\+' url.xml

You could even use it like so:

% eval $(grep -o 'auth=[[:alnum:]_-]\+' url.xml) 
% echo ${auth}
daEaIcRcbb.afahbOdwbWdjdYcEdYaOaDdc-bn7nM7-4q-PN0X1_3nqDHom4EBvmEuwr

Works on OSX.