I have an XML file that looks like this:

<?xml version="1.0" encoding="utf-8"?>
<!DOCTYPE rrd SYSTEM ".dtd">
<!-- Round Robin Database Dump -->
<rrd>
    <version>0003</version>
    <step>10</step> <!-- Seconds -->
    <lastupdate>1743206258</lastupdate> <!-- 2025-03-28 16:57:38 PDT -->

    <ds>
        <name> uptime_remote </name>
        <type> GAUGE </type>
        <minimal_heartbeat>600</minimal_heartbeat>
        <min>2.0000000000e+02</min>
        <max>2.0000000000e+02</max>

        <!-- PDP Status -->
        <last_ds>15508</last_ds>
        <value>2.0000000000e+02</value>
        <unknown_sec> 0 </unknown_sec>
    </ds>

    <!-- Round Robin Archives -->
    <rra>
        <cf>AVERAGE</cf>
        <pdp_per_row>1</pdp_per_row> <!-- 10 seconds -->

        <params>
        <xff>5.0000000000e-01</xff>
        </params>
        <cdp_prep>
            <ds>
            <primary_value>2.0000000000e+02</primary_value>
            <secondary_value>2.0000000000e+02</secondary_value>
            <value>2.0000000000e+02</value>
            <unknown_datapoints>0</unknown_datapoints>
            </ds>
        </cdp_prep>
        <database>
            <!-- 2025-03-27 16:57:40 PDT / 1743119860 --> <row><v>2.0000000000e+02</v></row>
            <!-- 2025-03-27 16:57:50 PDT / 1743119870 --> <row><v>2.0000000000e+02</v></row>
            <!-- 2025-03-27 16:58:00 PDT / 1743119880 --> <row><v>2.0000000000e+02</v></row>
            <!-- 2025-03-27 16:58:10 PDT / 1743119890 --> <row><v>2.0000000000e+02</v></row>
            <!-- 2025-03-27 16:58:20 PDT / 1743119900 --> <row><v>2.0000000000e+02</v></row>
            <!-- 2025-03-27 16:58:30 PDT / 1743119910 --> <row><v>2.0000000000e+02</v></row>
      </database>
    </rra>
</rrd>

I would like to be able to edit (replace) the value of each database entry, but I am first learning to manipulate this file with xmlstarlet.

To that end, first I want to extract the value of each / any row on the command line using xmlstarlet, but the only "key" here appears to be the timestamp in the comment. I can't figure out (and the manual doesn't mention this) how to match on comments. For example, this query just dumps every single value out:

$ xmlstarlet sel -t --value-of "rrd/rra/database/row/v" test_modified.xml

My desired search query should be the timestamp (e.g "2025-03-27 16:57:40 PDT") and the return value should be the entry:

2.0000000000e+02

Is there way to use xmlstarlet to match on the comment part of the tag & return the value? If no, what tool can I use (for bash scripting) this?

I have an XML file that looks like this:

<?xml version="1.0" encoding="utf-8"?>
<!DOCTYPE rrd SYSTEM "https://oss.oetiker.ch/rrdtool/rrdtool.dtd">
<!-- Round Robin Database Dump -->
<rrd>
    <version>0003</version>
    <step>10</step> <!-- Seconds -->
    <lastupdate>1743206258</lastupdate> <!-- 2025-03-28 16:57:38 PDT -->

    <ds>
        <name> uptime_remote </name>
        <type> GAUGE </type>
        <minimal_heartbeat>600</minimal_heartbeat>
        <min>2.0000000000e+02</min>
        <max>2.0000000000e+02</max>

        <!-- PDP Status -->
        <last_ds>15508</last_ds>
        <value>2.0000000000e+02</value>
        <unknown_sec> 0 </unknown_sec>
    </ds>

    <!-- Round Robin Archives -->
    <rra>
        <cf>AVERAGE</cf>
        <pdp_per_row>1</pdp_per_row> <!-- 10 seconds -->

        <params>
        <xff>5.0000000000e-01</xff>
        </params>
        <cdp_prep>
            <ds>
            <primary_value>2.0000000000e+02</primary_value>
            <secondary_value>2.0000000000e+02</secondary_value>
            <value>2.0000000000e+02</value>
            <unknown_datapoints>0</unknown_datapoints>
            </ds>
        </cdp_prep>
        <database>
            <!-- 2025-03-27 16:57:40 PDT / 1743119860 --> <row><v>2.0000000000e+02</v></row>
            <!-- 2025-03-27 16:57:50 PDT / 1743119870 --> <row><v>2.0000000000e+02</v></row>
            <!-- 2025-03-27 16:58:00 PDT / 1743119880 --> <row><v>2.0000000000e+02</v></row>
            <!-- 2025-03-27 16:58:10 PDT / 1743119890 --> <row><v>2.0000000000e+02</v></row>
            <!-- 2025-03-27 16:58:20 PDT / 1743119900 --> <row><v>2.0000000000e+02</v></row>
            <!-- 2025-03-27 16:58:30 PDT / 1743119910 --> <row><v>2.0000000000e+02</v></row>
      </database>
    </rra>
</rrd>

I would like to be able to edit (replace) the value of each database entry, but I am first learning to manipulate this file with xmlstarlet.

To that end, first I want to extract the value of each / any row on the command line using xmlstarlet, but the only "key" here appears to be the timestamp in the comment. I can't figure out (and the manual doesn't mention this) how to match on comments. For example, this query just dumps every single value out:

$ xmlstarlet sel -t --value-of "rrd/rra/database/row/v" test_modified.xml

My desired search query should be the timestamp (e.g "2025-03-27 16:57:40 PDT") and the return value should be the entry:

2.0000000000e+02

Is there way to use xmlstarlet to match on the comment part of the tag & return the value? If no, what tool can I use (for bash scripting) this?

• xml
• xpath
• command-line
• comments
• xmlstarlet

• While asking a question, you need to provide a minimal reproducible example: Please edit your original question and provide the following: (1) Well-formed XML file sample with all relevant namespaces. (2) What you need to do, i.e. logic, and your code attempt trying to implement it. (3) Desired output based on the sample data in #1 above. – Yitzhak Khabinsky Commented Apr 1 at 18:50
• "I would like to be able to edit (replace) the value of each database entry" It is better to use XSLT for your task via xsltproc. Check it out here: stackoverflow/questions/79489144/… – Yitzhak Khabinsky Commented Apr 1 at 19:10
• Edited to add the requested information! Hope it's clearer now. Thanks for the suggestion. – flyingfishfinger Commented Apr 1 at 19:17

4 Answers 4

An XPath expression like //database/comment()[contains(., '2025-03-27 16:57:40 PDT')]/following-sibling::row[1]/v should do to select the row/v element directly following a comment that contains that particular timestamp (e.g. 2025-03-27 16:57:40 PDT).

I know nothing about XMLStarlet. I read it supports XPath.

The XPath expression to select the v element that immediately follows the comment with the given timestamp would be:

/rrd/rra/database/comment()[contains(., '2025-03-27 16:57:40 PDT')]/following-sibling::row[1]/v

HTH.

I think xmlstarlet doesn’t support comments. But if Perl is an alternative, LibXML can handle comments like a node:

#!/usr/bin/perl
use strict;
use warnings;
use XML::LibXML;

# Load the XML file
my $file = "test_modified.xml";
my $parser = XML::LibXML->new();
my $doc = $parser->parse_file($file);

# Define the target timestamp
my $target_timestamp = "2025-03-27 16:57:40 PDT";

# Find all comments in the document
foreach my $comment ($doc->findnodes('//comment()')) {
    if ($comment->textContent =~ /\Q$target_timestamp\E/) {
        # The comment is right before the desired <v> value
        my $next_node = $comment->nextNonBlankSibling;
        if ($next_node && $next_node->nodeName eq 'row') {
            my ($value_node) = $next_node->findnodes('.//v');
            if ($value_node) {
                print "Value for timestamp $target_timestamp: " . $value_node->textContent . "\n";
            }
        }
    }
}

Thanks for the above guidance using the XPath string, this helped solve the original question. The following replaces, in place, the original value on the line containing "2025-03-27 16:57:40 PDT" with "5" (as an example)

xmlstarlet ed -L --update "//database/comment()[contains(., '2025-03-27 16:57:40 PDT')]/following-sibling::row[1]/v" --value "5" test_modified.xml