Edit value under specific tag using awk

Question

I'm trying to understand AWK command. The problem that I'm trying to solve is as follow: I have XML file with structure:

<root>
...
<elem_name name='type1'>
...
<prop name='a' type="xxx" value="000"/>
<prop name='b' type="xxx" value="000"/>
<prop name='c' type="xxx" value="000"/>
...
</elem_name>
<elem_name name='type2'>
....
<prop name='a' type="xxx" value="000"/>
<prop name='b' type="xxx" value="000"/>
<prop name='c' type="xxx" value="000"/>
...
</elem_name>
...
</root>

And I have to edit value of prop 'b' under 'type1' root. As I mentioned I would like to do it with awk or sed. I'm aware of better tools to do it.

For now, I created following command, but it is not working properly.

gawk '/elem_name name="type1"/ {for(i=1; i<=4; i++) {getline;found=index($0,"a");if(found != 0){sub("value=", "test", $0)}; print > "test.xml"}} {print > "test.xml"    }' original_file.xml

Firstly, I the value is not changed after script execution, the 'name="type1" is removed from output file.

[Don't Parse XML/HTML With Regex.](https://stackoverflow.com/a/1732454/3776858) I suggest to use an XML/HTML parser (xmlstarlet, xmllint ...). — Cyrus, Sep 25 '21 at 14:57
No, I think that I can use Perl but only like oneliner to execute it from Bash script. — Guinea, Sep 25 '21 at 14:57
@Cyrus I can use xmlint, but I have no idea how to edit xml file using that tool. — Guinea, Sep 25 '21 at 14:58
@Guinea The python you have in your OS should be able to run the answer I have posted. — balderman, Sep 25 '21 at 15:01
Regarding [I can use Perl but only like oneliner to execute it from Bash script.](https://stackoverflow.com/questions/69327243/edit-value-under-specific-tag-using-awk#comment122534412_69327243) - why do you care how many lines anything is if it's inside a bash script? Please [edit] your question to show the expected output so we can help you. — Ed Morton, Sep 26 '21 at 15:41

balderman · Answer 1 · 2021-09-25T15:58:32.060

2

I have to edit value of prop 'b' under 'type1' root

Since I assume you have python in the OS you are running I share the below solution.

import xml.etree.ElementTree as ET

FILE_NAME = 'myxml.xml'

root = ET.parse(FILE_NAME)
ele = root.find(".//elem_name[@name='type1']")
b = ele.find(".//prop[@name='b']")
b.attrib['value'] = 'new_value_goes_here'
root.write(FILE_NAME)

edited Sep 25 '21 at 15:58

answered Sep 25 '21 at 14:59

balderman

22,927
7
34
52

Thank you, I appreciate your answer, but as I mentioned I want to avoid other tools. Main reason is that I would like to encapsulate everything in one bash file. – Guinea Sep 25 '21 at 15:10
your bash file can be a 1 liner like `python3 myscirpt.py myxml.xml` :-) If you have specific questions about the py script- I will be happy to answer. – balderman Sep 25 '21 at 15:14
Yea, but the 'myscript.py' has to be stored somewhere, that's not the solution for me ;) – Guinea Sep 25 '21 at 15:17
It can be part of the bash which will redirect it to a file (`myscript.py`) and call it later - isnt it? – balderman Sep 25 '21 at 15:18
Yeah, but how to save it to original file? Instead of getting to stdout? ET.write(original_file_name)? – Guinea Sep 25 '21 at 15:52
OK - so now your question is: "how should the python script save myxml.xml back to disk after modification" - is that correct? – balderman Sep 25 '21 at 15:53
Well, instead of parsing from string, as you have done in the example ET.fromstring(xml), i will use ET.parse(path). And after the change operation, I want to see the chagnes in the same file, that I passed as a path above. – Guinea Sep 25 '21 at 15:57
@Guinea Code was modified in order to support your needs. – balderman Sep 25 '21 at 15:58

Cyrus · Answer 2 · 2021-09-25T17:03:30.000

1

Update attribute with xmlstarlet:

xmlstarlet edit --omit-decl \
  --update '//root/elem_name[@name="type1"]/prop[@name="b"]/@value' \
  --value "test" file.xml

Output with removed ...:

<root>
  <elem_name name="type1">
    <prop name="a" type="xxx" value="000"/>
    <prop name="b" type="xxx" value="test"/>
    <prop name="c" type="xxx" value="000"/>
  </elem_name>
  <elem_name name="type2">
    <prop name="a" type="xxx" value="000"/>
    <prop name="b" type="xxx" value="000"/>
    <prop name="c" type="xxx" value="000"/>
  </elem_name>
</root>

See: xmlstarlet edit

edited Sep 25 '21 at 17:03

answered Sep 25 '21 at 15:33

Cyrus

84,225
14
89
153

Question. Does `xmlstarlet` comes out of the box on every linux distro? – balderman Sep 25 '21 at 16:00
Unfortunately no, that solution is probably ok, but not for me :). – Guinea Sep 25 '21 at 16:16

LMC · Answer 3 · 2021-09-25T17:18:32.687

It can be done with xmllint in one-liner taking advantage of its --shell option as

 (echo 'cd //elem_name[@name="type1"]/prop[@name="b"]/@value' ; echo "set some other xxxxx value"; echo "save test-e.xml" ; echo "bye") | xmllint --shell test.xml

The commands to xmllint are intuitive, "shell" alike

Change directory to the desired node expressed as an XPath expression:
cd //elem_name[@name="type1"]/prop[@name="b"]/@value

Set element value (no quotes):
set some other xxxxx value

Save xml doc to a new file:
save test-e.xml

Result:

<?xml version="1.0"?>
<root>
  <elem_name name="type1">
    <prop name="a" type="xxx" value="000"/>
    <prop name="b" type="xxx" value="some other xxxxx value"/>
    <prop name="c" type="xxx" value="000"/>
  </elem_name>
  <elem_name name="type2">
    <prop name="a" type="xxx" value="000"/>
    <prop name="b" type="xxx" value="000"/>
    <prop name="c" type="xxx" value="000"/>
  </elem_name>
</root>

The whole node content could be changed also

xmllint --shell test.xml

Commands at prompt

/ > cd //elem_name[@name="type1"]
elem_name > set <new>cdvfbg</new>
elem_name > cat
<elem_name name="type1">
  <new>cdvfbg</new>
</elem_name>
elem_name > save test-e.xml

To get an interactive prompt and play around any xml/html file do:

xmllint --shell test.xml

Same commands above could be executed there. Try help to get a list of possible commands.

Ed Morton · Answer 4 · 2021-09-26T16:02:25.543

It sounds like this is what you're trying to do, assuming your input is always formatted as shown in your example, using any awk in any shell on every Unix box:

awk -v elem='type1' -v prop='b' -v val='test' '
    $1 == "</elem_name>" {
        inElem = 0
    }
    inElem {
        if ( ($1 == "<prop") && ($2 == ("name=\047" prop "\047")) ) {
            match($0,/value="[^"]*"/)
            $0 = substr($0,1,RSTART+6) val substr($0,RSTART+RLENGTH-1)
        }
    }
    $1 == "<elem_name" {
        inElem = ( $2 == ("name=\047" elem "\047>") )
    }
    { print }
' file
<root>
...
<elem_name name='type1'>
...
<prop name='a' type="xxx" value="000"/>
<prop name='b' type="xxx" value="test"/>
<prop name='c' type="xxx" value="000"/>
...
</elem_name>
<elem_name name='type2'>
....
<prop name='a' type="xxx" value="000"/>
<prop name='b' type="xxx" value="000"/>
<prop name='c' type="xxx" value="000"/>
...
</elem_name>
...
</root>

Edit value under specific tag using awk

4 Answers4

Linked