Command line XPath expressions generator. Works on XML/HTML files and includes attributes as well.
A wrapper around xmllint tool that provides convenience options to inspect documents listing namespaces and xpaths expressions.
By default, shows xpaths expressions starting at the root element but it can start at a predefined element passed from command line as an xpath expression.
ⓘ Check pyxml2xpath python implementation here!
- Basic usage
- Using found XPaths
- XPaths on HTML
- Absolute paths and namespaces
- Default namespaces
- XPath expressions at a given element
- Using found XPaths
- TL;DR
- Generate an XML from an XSD and show its XPaths
- Performance
- Script help
- Known issues
- Generate man page
- Rationale
Get all element xpaths from an XML file. To show attribute XPaths also, add -a (absolute XPaths)
xml2xpath.sh -x soap.xml
Result:
xml2xpath: find XPath expressions on soap.xml
-x ; XML file: soap.xml
Namespaces:
soap1=http://schemas.xmlsoap.org/soap/envelope/xxxxxxxxx
soap=http://schemas.xmlsoap.org/soap/envelope/
defaultns=http://example.com
Found 4 XPath expressions (unique, use -r to override):
/soap:Envelope
/soap:Envelope/soap:Body
/soap:Envelope/soap:Body/defaultns:incident
/soap:Envelope/soap:Body/defaultns:incident/defaultns:Company
Found relative or absolute XPaths expressions can be tested on browser development tools.
Inspector or Elements tabs
Browsers might expect elements on xpath that do not exists on html source.
The script found:
/html/body/center[2]/table/tr[2]/td[1]
But browser will accept (tbody must be added)
/html/body/center[2]/table/tbody/tr[2]/td[1]
Relative expressions are also valid. All below could point to the same td element
//center[2]/table[@id='t1']//tr[2]/td[1]
//center[2]/*//tr[2]/td[1]
//center[2]/descendant::*/td[1]
//center[2]/descendant::*/td[position()=1]
Console
Firefox and Chrome offer $x built-in object to search elements using absolute or relative xpath expressions.
Hovering over the found elements on console might highlight the element on the page.
Clicking on the elements will show the element on Inspector tab
$x("/html/body/center[2]/table/tbody/tr")
Result:
(4) [tr, tr, tr, tr]
Or for a single element:
$x("//center[2]/table/tbody/tr[2]")
Result (expanded):
(1) […]
0: <tr>
length: 1
<prototype>: Array []
Generic numeric XPaths to Xpaths with Element names
Given this xpaths generated with -a -g it says there are 4 entry elements where the first has the absolute position 8 for any element.
/ > whereis //defaultns:entry
/*/*[8]
/*/*[9]
/*/*[10]
/*/*[11]
The first one /*/*[8] would be equivalent to //defaultns:entry[1], the eight element and the first entry element for that xpath expression.
Find all attributes under a qualified path with this shell command
(echo "setrootns"; echo "cat /defaultns:feed/defaultns:entry/defaultns:link/@*") | xmllint --shell wiki.xml
Result showing rel, type and href attributes:
/ > setrootns
/ > cat /defaultns:feed/defaultns:entry/defaultns:link/@*
-------
rel="alternate"
-------
type="text/html"
-------
href="https://en.wikipedia.org/w/index.php?title=Title%202"
-------
rel="alternate"
-------
type="text/html"
-------
href="https://en.wikipedia.org/w/index.php?title=some_title"
/ >
Find @rel attribute under a specific path with this shell command
(echo "setrootns"; echo "cat /*/*[8]/*[3]/@rel") | xmllint --shell wiki.xml
Result of specific @rel attribute:
/ > setrootns
/ > cat /*/*[8]/*[3]/@rel
-------
rel="alternate"
Adding p allows to pass a namespace prefix to search for (experimental).
xml2xpath.sh -a -s '//table[2]/thead/tr' -l resources/test.html
Result:
Namespaces:
xml http://www.w3.org/XML/1998/namespace
Found Xpath (absolute):
/html/body/table[2]/thead/tr
/html/body/table[2]/thead/tr/@class
/html/body/table[2]/thead/tr/th[1]
/html/body/table[2]/thead/tr/th[2]
/html/body/table[2]/thead/tr/th[3]
/html/body/table[2]/thead/tr/th[4]
Given an XML with namespaces, the following command will show absolute paths in numeric format
xml2xpath.sh -a -n -x wiki.xml
xmllint does not show element names in full paths
/*
/*/*[8]
/*/*[9]
/*/*[8]/*[6]
/*/*[9]/*[6]
/*/*[8]/*[3]/@rel
/*/*[8]/*[3]/@type
/*/*[8]/*[3]/@href
Adding -g would show the qualified xpath used to generate absolute paths
/ > whereis /defaultns:feed/defaultns:entry
/*/*[8]
/*/*[9]
/ > whereis /defaultns:feed/defaultns:entry/defaultns:author
/*/*[8]/*[6]
/*/*[9]/*[6]
/ > whereis /defaultns:feed/defaultns:entry/defaultns:author/defaultns:name/@*
/ > whereis /defaultns:feed/defaultns:entry/defaultns:id/@*
/ > whereis /defaultns:feed/defaultns:entry/defaultns:link/@*
/*/*[8]/*[3]/@rel
/*/*[8]/*[3]/@type
/*/*[8]/*[3]/@href
/*/*[9]/*[3]/@rel
/*/*[9]/*[3]/@type
/*/*[9]/*[3]/@href
This XPath
/defaultns:feed/defaultns:entry/defaultns:author
Could generate
/*/*[8]/*[6]
/*/*[9]/*[6]
This start root element has a default namespace, i.e. without a prefix
<feed xmlns="http://www.w3.org/2005/Atom" xml:lang="en">
Passing -n will map that namespace to a defaultns prefix by default.
To change that, pass -o option with the desired definition in the form <prefix>=<uri>.
xml2xpath.sh -n -o 'ns1=http://www.w3.org/2005/Atom' -s '//ns1:entry[descendant::ns1:name[.="author2"]]' -x tests/resources/wiki.xml
Try this command:
xml2xpath.sh -a -g -n -o 'ns1=http://www.w3.org/2005/Atom' -s '//ns1:entry[descendant::ns1:name[.="author2"]]' -t -x tests/resources/wiki.xml
Passing -s option to show xpath expressions starting at an specific element or elements.
xml2xpath.sh -n -s '//defaultns:entry/defaultns:author' -x wiki.xml
Result:
xml2xpath: find XPath expressions on wiki.xml
-n ; default ns prefix: defaultns
-s ; Start tree at: '//defaultns:entry/defaultns:author' (du_path)
-x ; XML file: wiki.xml
Namespaces:
defaultns=http://www.w3.org/2005/Atom
Found 6 XPath expressions (unique, use -r to override):
//defaultns:entry/defaultns:author
//defaultns:entry/defaultns:author/defaultns:name
You see, in this world there's two kinds of people, my friend:
Those with loaded guns and those who dig.
You dig.(The Good, The Bad and The Ugly)
... those with TL;DR, and those who read.
You read.
😏 😆
Start by passing the xpath expression of a known element in the tree and get the xpath expression of the elements under it. Try these commands over already provided samples:
./xml2xpath.sh -a -s "//table[@id[.='t1']]" -l tests/resources/test.html
./xml2xpath.sh -o 'defns=urn:hl7-org:v3' -s '//defns:addr' -x tests/resources/HL7.xml
Smoke tests for the script to quickly verify that changes did not break any functionality. Tests results show the tested command which can be tried on console to witness how those options work.
Run all at once with test-all.sh or one at a time
cd tests
./test-all.sh
./test-html-base.sh
Result:
*** HTML tests ***
TC01 : Basic test (-l)
cmd : ../xml2xpath.sh -l resources/test.html
PASSED
...
...
Print more details by running as
dbg=1 ./test-all.sh
The command used for testing is shown in details. Run it to see it in action (adding missing quotes if needed :-p )
../xml2xpath.sh -a -s "//table[@id[.='t1'] and descendant::tr[@class='headerRow']]" -l resources/test.html
To log a trace of test commands to a file run
dbg=1; trace="$HOME/tmp/all-tests.log" ./test-all.sh
If an XSD file is provided and xmlbeans package is installed, try to create an XML instance and print the XPath from it.
Taking an XSD example from w3schools the script will print
xml2xpath.sh -a -f shiporder -d shiporder.xsd
Result:
Creating XML instance starting at element shiporder from shiporder.xsd
Found Xpath (absolute):
/shiporder
/shiporder/@orderid
/shiporder/item
/shiporder/item/note
/shiporder/item/price
/shiporder/item/quantity
/shiporder/item/title
/shiporder/orderperson
/shiporder/shipto
/shiporder/shipto/address
/shiporder/shipto/city
/shiporder/shipto/country
/shiporder/shipto/name
Relative paths starting at an element are also possible
xml2xpath.sh -s '//shipto' -f shiporder -d shiporder.xsd
Result:
Creating XML instance starting at element shiporder from shiporder.xsd
Found XPath:
//shipto
//shipto/name
//shipto/address
//shipto/city
//shipto/country
If -t option is passed it will print XML elements tree also
shiporder
orderperson
shipto
name
address
city
country
item
title
note
quantity
price
It is a wrapper around xmllint interactive shell that automates the inspection of XML/HTML files, looking for XPath expressions taking into account namespaces.
The goal is to get as much information as possible without writing a parser. Code is focused on getting information and presenting the results as usefully as possible.
Main commands sent to interactive shell are:
Find namespaces
echo "ls /some/xpath/namespace::*" | xmllint --shell file.xml
/ > ls /*/namespace::*[local-name()!='xml']
n 1 default -> http://example.com/ns1
n 1 soap -> http://schemas.xmlsoap.org/soap/envelope/
...
echo "xpath /some/xpath" | xmllint --shell file.xml
/ > xpath /*//*
Object is a Node Set :
Set contains 3 nodes:
1 ELEMENT soap:Body
2 ELEMENT incident
default namespace href=http://example.com/ns2
...Get elements tree
Key command is du since it offers a tree representation with indented elements so "flattening" that tree provides simple XPath
expressions like /feed/entry/title.
echo "du /" | xmllint --shell resources/wiki.xml
# result
/ > du /
/
feed
id
title
link
subtitle
entry
id
title
Get absolute xpath expressions
XPath expressions found by du command are used to find absolute XPath expressions. Namespaces must be set for this command to success.
XPath for attributes could look like /feed/entry/title/@*.
(echo "setrootns"; echo "whereis /defaultns:feed/defaultns:entry/defaultns:title") | xmllint --shell resources/wiki.xml
# result
/ > setrootns
/ > whereis /defaultns:feed/defaultns:entry/defaultns:title
/*/*[8]/*[2]
/*/*[9]/*[2]
/*/*[10]/*[2]
Parsing big documents might take a long time. As an example, this 1M elements sample document took almost 2 hours to just find 17 different expressions. As many large xml documents, that sample has the same elements repeated many times.
This wrapper looks for XPath expressions, not content so inspecting just one of those many elements would be enough.
That can be done in 2 ways, specifying the xpath of the first known element or extracting that first element to another file as we'll see next.
Occasionally an due to external factors like computer performance, antivirus, etc., the script could fail with a timeout error
Timeout reading from file descriptor 142 (parse ns stage1) . Current timeout: 5 secs. Try extending the timeout with: XML_XPATH_RTOUT=[int or float > 5] xml2xpath.sh ...
Read timeout can be extended by setting/using XML_XPATH_RTOUT environmental variable
export XML_XPATH_RTOUT=15
./xml2xpath.sh -x tests/resources/HL7.xml
Or
XML_XPATH_RTOUT=15 ./xml2xpath.sh -x tests/resources/HL7.xml
time xml2xpath.sh -a -g -s '/table/T[1]//*' -x big/lineitem.xml
This simple command will extract the first entry element to wiki-1.xml file. Parsing that file would be more efficient.
(echo "setrootns"; echo "cd /defaultns:feed/defaultns:entry[1]"; echo "write wiki-1.xml"; echo "bye") | xmllint --shell --format wiki-big.xml
Print XPath present on xml or (if possible) xsd files. Based on xmllint utility, try to build all possible XPaths from an XML instance. The latter could be constructed from a provided XSD file.
Usage: xml2xpath.sh [-h -v] [-d file -f <tag name>] [-a -g -t -s <xpath>] [-n -p <ns prefix> -o <prefix>=URI -x <file>] [-l <file>]
xml2xpath.sh [-h -v] [XSD OPTIONS] [COMMON XML/HTML OPTIONS] [XML OPTIONS] [HMTL OPTIONS]
Options:
Basic:
-h print this help message.
-v print version
XML/HTML Common Options:
-a Show absolute Xpaths. Use -g too to add details. -s is used to filter but absolute paths are shown.
-g Print xmllint command for debugging or clarity. Implies -a.
-r Print repeated xpaths when -a is used. For debugging only.
-s Start printing XPath at an absolute or relative xpath, e.g.: /shiporder/shipto, //shipto.
Must contain namespace prefix if needed. e.g.: //defaultns:entry, //xs:element
-t print XML element tree as provided by xmllint 'du' shell command.
HTML options:
-l Use HTML parser
XML options:
-n Set namespaces found on root element. Default namespace prefix is 'defaultns' but may be overriden with -o option.
-o Override the default namespace definition by passing <prefix>=URI, e.g.: -o 'defns=urn:hl7-org:v3'
-p Namespace prefix to use. No need to pass -n if used. EXPERIMENTAL.
-x xml file, will take precedence over -d option.
XSD options:
-d xsd file path.
-D Same as -d but saves created xml instance to <xsd file path>.xml
-f name of the root element to build xml from xsd.
Examples:
xml2xpath.sh -t -x test.xml Print all xpaths and elements tree
xml2xpath.sh -s '//shipto' -x test.xml Print xpaths starting at //shipto element
xml2xpath.sh -a -f shiporder -d tests/resources/shiporder.xsd Print xpaths from generated xml
xml2xpath.sh -a -n -g -x wiki.xml Use namespaces, show absolute paths and xmllint shell messages
xml2xpath.sh -o 'defns=urn:hl7-org:v3' -s '//defns:addr' -x HL7.xml Add a namespace definition and use it in a relative expression
xml2xpath.sh -a -n -l test.html Html file with absolute paths option
Reporting bugs:
https://github.com/mluis7/xml2xpath/issues
Use this command with help2man utility
help2man --locale=en_US --no-info --help-option='-h' --version-option='-v' xml2xpath.sh -o man/xml2xpath.sh.1
To test the generated man page use:
MANPATH="./man" man man/xml2xpath.sh.1
- Multiple default namespaces in document:
-oand/or-smay give incorrect results.