print characters between a specific pattern containing <>

Question

i am asking for your help please.

I reviewed a lot of sources and did some testing with awk and sed, but i can't get it to work. Below is a snippet of a config file which i could get an output via grep, but not the way i need it.

> file.txt
> 
> "<property>name="DBName"><value>ABC</value>name="DBName"><value>DEF</value></property>
> 
> cat file.xml | grep -o -P '.name="DBName"><value>.{0,20}'
> name="DBName"><value>ABC</value>
> name="DBName"><value>DEF</value></propert

The desired output is:

ABC
DEF

Thanks for any help.

Andi

Answer 1

If (huge, enormous "if") your file really only has the very simple case where you want the exact string <value> followed by a few non-< characters and then </value>, so your problem can be formuated as "fetch me the string of simple, non-newline characters found between each occurrence of <value> and the first < after it", then you can do (using GNU grep):

grep -oP '<value>\K[^<]+' file

Of course, this will fail on anything even slightly different. If you have multi-line values, for example, or if the value tag can have something like <value foo=bar> or on any number of other, perfectly valid XML cases. The Right Way© is to use an XM: parser instead. You might want to check out xmllint or XMLStarlet among others.

Answer 2

If all of your input looks exactly like the one line of sample input you posted:

$ cat file
"<property>name="DBName"><value>ABC</value>name="DBName"><value>DEF</value></property>

then using any awk:

$ awk -F'[<>]+' '{for (i=5; i<=NF; i+=4) print $i}' file
ABC
DEF

but, like any other solution that doesn't use an XML parser, that is fragile.

Answer 3

Using XMLStarlet:

XML file taken from the answer.

<config>
 <property name="DBName"><value>ABC</value></property>
 <property name="DBName" year="2023"><value>DEF</value></property>
 <property name="SystemName"><value>s70</value></property>
</config>

$ xmlstarlet select -t -v '//value' --nl ex.xml
ABC
DEF
s70

$ xmlstarlet select -t -m '//property[@name="DBName"]' -v 'value' --nl ex.xml
ABC
DEF

---

Using awk:

$ awk -v pat1="<value>" -v pat2="</value>" '
   {
       while (match($0, pat1)){ 
           $0=substr($0,RSTART+RLENGTH);
           if (match($0, pat2)) print substr($0,1,RSTART-1)
       }
   }
'

The awk and pcregrep solutions work for this question but may fail on many occasions.

$ pcregrep -o1 '<value>(.*?)</value>'

Answer 4

Thank a lot, this worked for me. Here to sum it up more sample code and the final solution for mr:

host123:~ # grep -oP '.{0,50}name="DBName"><value>.{0,100}' file.xml
ystemName"><value>XYZ</value></property><property name="DBName"><value>ABC-SERVER1</value></property><property name="SystemHome"><value>host123</value></property><prope 
ystemName"><value>XYZ</value></property><property name="DBName"><value>DEF</value></property><property name="SystemHome"><value>host123</value></property><property name=


host123:~ # grep -oP 'name="DBName"><value>\K[^<]+' file.xml 
ABC-SERVER1 
DEF

Answer 5

If you have a well formed XML input such us

<config>
 <property name="DBName"><value>ABC</value></property>
 <property name="DBName" year="2023"><value>DEF</value></property>
 <property name="SystemName"><value>s70</value></property>
</config>

You can use a XML-aware tool, and use Xpath (or similar) to select the parts to extract. Example with xidel:

## 1) get all values:
$ xidel -e "//value" ex.xml 
ABC
DEF
s70
## 2) get the values inside "property" with attribute "name" "DBname"
$ xidel -e "//property[@name='DBname']/value" ex.xml 
ABC
DEF