Command-line tools that can be called from shell scripts include:

• 4xpath - command-line wrapper around Python's 4Suite package

• XMLStarlet

• xpath - command-line wrapper around Perl's XPath library

  sudo apt-get install libxml-xpath-perl
  

• Xidel - Works with URLs as well as files. Also works with JSON

I also use xmllint and xsltproc with little XSL transform scripts to do XML processing from the command line or in shell scripts.

I am not aware of any pure shell XML parsing tool. So you will most likely need a tool written in an other language.

My XML::Twig Perl module comes with such a tool: xml_grep, where you would probably write what you want as xml_grep -t '/html/head/title' xhtmlfile.xhtml > titleOfXHTMLPage.txt (the -t option gives you the result as text instead of xml)

Well, you can use xpath utility. I guess perl's XML::Xpath contains it.

While there are quite a few ready-made console utilities that might do what you want, it will probably take less time to write a couple of lines of code in a general-purpose programming language such as Python which you can easily extend and adapt to your needs.

Here is a python script which uses lxml for parsing — it takes the name of a file or a URL as the first parameter, an XPath expression as the second parameter, and prints the strings/nodes matching the given expression.

Example 1

#!/usr/bin/env python
import sys
from lxml import etree

tree = etree.parse(sys.argv[1])
xpath_expression = sys.argv[2]

#  a hack allowing to access the
#  default namespace (if defined) via the 'p:' prefix    
#  E.g. given a default namespaces such as 'xmlns="http://maven.apache.org/POM/4.0.0"'
#  an XPath of '//p:module' will return all the 'module' nodes
ns = tree.getroot().nsmap
if ns.keys() and None in ns:
    ns['p'] = ns.pop(None)
#   end of hack    

for e in tree.xpath(xpath_expression, namespaces=ns):
    if isinstance(e, str):
        print(e)
    else:
        print(e.text and e.text.strip() or etree.tostring(e, pretty_print=True))

lxml can be installed with pip install lxml. On ubuntu you can use sudo apt install python-lxml.

Usage

python xpath.py myfile.xml "//mynode"

lxml also accepts a URL as input:

python xpath.py http://www.feedforall.com/sample.xml "//link"

Note: If your XML has a default namespace with no prefix (e.g. xmlns=http://abc...) then you have to use the p prefix (provided by the 'hack') in your expressions, e.g. //p:module to get the modules from a pom.xml file. In case the p prefix is already mapped in your XML, then you'll need to modify the script to use another prefix.

Example 2

A one-off script which serves the narrow purpose of extracting module names from an apache maven file. Note how the node name (module) is prefixed with the default namespace {http://maven.apache.org/POM/4.0.0}:

pom.xml:

<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
    <modules>
        <module>cherries</module>
        <module>bananas</module>
        <module>pears</module>
    </modules>
</project>

module_extractor.py:

from lxml import etree
for _, e in etree.iterparse(open("pom.xml"), tag="{http://maven.apache.org/POM/4.0.0}module"):
    print(e.text)

While it seems like "never parse XML, JSON... from bash without a proper tool" is sound advice, I disagree. If this is side job, it is waistfull to look for the proper tool, then learn it... Awk can do it in minutes. My programs have to work on all above mentioned and more kinds of data. Hell, I do not want to test 30 tools to parse 5-7-10 different formats I need if I can awk the problem in minutes. I do not care about XML, JSON or whatever! I need a single solution for all of them.

As an example: my SmartHome program runs our homes. While doing it, it reads plethora of data in too many different formats I can not control. I never use dedicated, proper tools since I do not want to spend more than minutes on reading the data I need. With FS and RS adjustments, this awk solution works perfectly for any textual format. But, it may not be the proper answer when your primary task is to work primarily with loads of data in that format!

The problem of parsing XML from bash I faced yesterday. Here is how I do it for any hierarchical data format. As a bonus - I assign data directly to the variables in a bash script.

To make thins easier to read, I will present solution in stages. From the OP test data, I created a file: test.xml

Parsing said XML in bash and extracting the data in 90 chars:

awk 'BEGIN { FS="<|>"; RS="\n" }; /host|username|password|dbname/ { print $2, $4 }' test.xml

I normally use more readable version since it is easier to modify in real life as I often need to test differently:

awk 'BEGIN { FS="<|>"; RS="\n" }; { if ($0 ~ /host|username|password|dbname/) print $2,$4}' test.xml

I do not care how is the format called. I seek only the simplest solution. In this particular case, I can see from the data that newline is the record separator (RS) and <> delimit fields (FS). In my original case, I had complicated indexing of 6 values within two records, relating them, find when the data exists plus fields (records) may or may not exist. It took 4 lines of awk to solve the problem perfectly. So, adapt idea to each need before using it!

Second part simply looks it there is wanted string in a line (RS) and if so, prints out needed fields (FS). The above took me about 30 seconds to copy and adapt from the last command I used this way (4 times longer). And that is it! Done in 90 chars.

But, I always need to get the data neatly into variables in my script. I first test the constructs like so:

awk 'BEGIN { FS="<|>"; RS="\n" }; { if ($0 ~ /host|username|password|dbname/) print $2"=\""$4"\"" }' test.xml

In some cases I use printf instead of print. When I see everything looks well, I simply finish assigning values to variables. I know many think "eval" is "evil", no need to comment :) Trick works perfectly on all four of my networks for years. But keep learning if you do not understand why this may be bad practice! Including bash variable assignments and ample spacing, my solution needs 120 chars to do everything.

eval $( awk 'BEGIN { FS="<|>"; RS="\n" }; { if ($0 ~ /host|username|password|dbname/) print $2"=\""$4"\"" }' test.xml ); echo "host: $host, username: $username, password: $password dbname: $dbname"