Pythonic XML parsing/reading
Project description
dunder_xml_reader
Description/Overview
The dunder_xml_reader
package provides classes and functions that aim to make navigating
and extracting data from an XML document easier and more "Pythonic". All you have to do is
pass a string containing valid XML to the parse_xml()
function and it will return a Python
object that exposes properties as dictionary keys and child-nodes as attributes. The
following sample code illustrates. Consider the following XML:
<cXML payloadID="1233444-2001@premier.workchairs.com" xml:lang="en-CA" timestamp="2000-10-12T18:41:29-08:00">
<Header>
<To>
<Credential domain="AribaNetworkUserId" type="marketplace">
<Identity>bigadmin@marketplace.org</Identity>
</Credential>
<Credential domain="AribaNetworkUserId">
<Identity>admin@acme.com</Identity>
</Credential>
</To>
</Header>
</cXML>
It can be easily nagivated with the following Python code:
>>> with open('order_request.xml') as infile:
... xml = dunder_xml_reader.parse_xml(infile.read())
>>>
>>> print(xml['payloadID'])
1233444-2001@premier.workchairs.com
>>>
>>> print(xml.Header.To.Credential[0].Identity.text())
bigadmin@marketplace.org
>>>
As you can see, the property payloadID
is accessed as if it were a dictionary key of the xml
object. Likewise, the Header
child node of the cXML document is accessed as if it were an
attribute of the xml
object.
The Python object(s) returned by parse_xml()
use Python dunder methods to fulfill
functionality provided by other Python built-ins:
>>> # Python's "in" operator
>>>
>>> 'payloadID' in xml
True
>>> 'someOtherThing' in xml
False
>>>
>>> # Python's "hasattr" function
>>>
>>> hasattr(xml, 'Header')
True
>> hasattr(xml, 'Footer')
False
>>>
>>> # Python's "len" function
>>>
>>> len(xml.Header.To.Credential)
2
>>>
The Python objects returns by parse_xml()
have the following methods:
tag()
- Returns the XML element tag (i.e. "<cXML>" returns 'cXML')text()
- Returns the text between the tag and it's closer (i.e. "<p>blah</p>" returns 'blah')
Nodes that have siblings with the same element tag (i.e. Credential
in the above example) are
represented as a XmlNodeList
, a subclass of Python's built-in list
. The XmlNodeList
class
has a number of extra convenience methods:
first()
- Returns the first item in the list, or if the list is empty, a default value.last()
- Returns the last item in the list, or if the list is empty, a default value.filter()
- Filters the list items given a lambda expression.filter_prop()
- Filter's the list items looking for items that have a property with a given value.filter_text()
- Filter's the list items looking for items that have text with the given value.map()
- Transforms each item in the list to another given a lambda expression.map_attr()
- Transforms each item in the list to the value of the item's given attribute.map_prop()
- Transforms each item in the list to the value of the item's given property.map_text()
- Transforms each item in the list to the value of the item'stext()
.join_text()
- Returns string created by concatenating thetext()
values from each item.join_prop()
- Returns a string craeted by concatenating the given property value from each item.
Installation
The easiest way to install dunder_xml_reader is to use pip
:
$ pip install dunder-xml-reader
Some Examples
The following are some examples of how you can use the Python objects returns by parse_xml()
:
Getting property values by dereferencing as a dictionary
>>> xml = parse_xml(raw_xml_text)
>>> payload_id = xml['payloadID']
>>> print(payload_id)
1233444-2001@premier.workchairs.com
>>>
You can even check to see if a property exists using Python's built-in in
operator
>>> xml = parse_xml(raw_xml_text)
>>> 'payloadID' in xml
True
>>> 'someOtherWeirdThing` in xml
False
Getting siblings by dereferencing as a list
>>> xml = parse_xml(raw_xml_text)
>>> second_credential = xml.Header.To.Credential[1]
>>> print(second_credential.Identity.text())
admin@acme.com
>>>
A nice side effect of being able to dereference an CxmlNode as a list is that you can for-loop over CxmlNodes:
>>> xml = parse_xml(raw_xml_text)
>>> for cred in xml.Header.To.Credential
... print(cred['domain'], cred.Identity.text())
...
AribaNetworkUserId bigadmin@marketplace.org
AribaNetworkUserId admin@acme.com
>>>
You can also use the built-in len()
function with CxmlNodes:
>>> xml = parse_xml(raw_xml_text)
>>> print(len(xml.Header.To.Credential))
2
>>>
Looking up the original element name with the .tag()
method
>>> request = parse_xml(raw_xml_text).Request.ConfirmationRequest
>>> print(request.tag())
ConfirmationRequest
>>>
Getting the inner-text of the element with the .text()
method
>>> xml = parse_xml(raw_xml_text)
>>> node = xml.Header.From.Credential.Identity
>>> print(node.text())
942888711
>>>
Getting the first of a list of siblings with the .first()
method
>>> xml = parse_xml(raw_xml_text)
>>> node = xml.Header.To.Credential.first()
>>> print node['type']
marketplace
>>>
Getting the last of a list of siblings with the .last()
method
>>> xml = parse_xml(raw_xml_text)
>>> print(xml.Header.To.Credential.last().Identity.text())
admin@acme.com
>>>
Filtering nodes with the .filter()
method
>>> extrinsics = parse_xml(raw_xml_text).Request.OrderRequest.OrderRequestHeader.Extrinsic
>>> user_nodes = extrinsics.filter(lambda n: n['name'].startswith('User'))
>>> print(len(user_nodes), len(extrinsics))
5 12
>>>
Simpler filtering nodes with the .filter_prop()
method
>>> extrinsics = parse_xml(raw_xml_text).Request.OrderRequest.OrderRequestHeader.Extrinsic
>>> user_nodes = extrinsics.filter_prop('name', 'User', startswith_predicate)
>>> print(len(user_nodes), len(extrinsics))
5 12
>>>
Mapping nodes to something else with a lambda expression
>>> line_item = parse_xml(raw_xml_text).Request.OrderRequest.ItemOut[0]
>>> taxes = line_item.tax.map(lambda n: float(n.TaxDetail.TaxAmount.text()))
>>> print(taxes)
[0.805063025, 0.0, 3.703289915]
>>>
Simpler mapping with the .map_attr()
method
>>> from_header = parse_xml(raw_xml_text).Header.From
>>> identity_nodes = from_header.Credential.map_attr('Identity')
>>> print(identity_nodes[1].text())
test_customer
>>>
Simpler mapping with the .map_prop()
method
>>> from_header = parse_xml(raw_xml_text).Header.From
>>> credential_domains = from_header.Credential.map_prop('domain')
>>> print(credential_domains)
['DUNS', 'CompanyName', 'InteropKey']
>>>
Simpler mapping with the .map_text()
method
>>> bill_to = parse_xml(raw_xml_text).Request.OrderRequest.OrderRequestHeader.BillTo
>>> street = bill_to.Address.PostalAddress.Street.map_text()
>>> print(street)
['1242 West Main Street', 'Suite 6a']
>>>
String joining with the .join_text()
method
>>> address = parse_xml(raw_xml).Request.OrderRequest.OrderRequestHeader.BillTo.Address
>>> print(address.PostalAddress.Street.join_text("\n"))
1231 West Main Street
Suite 6a
>>>
String joining with the .join_prop()
method
>>> xml = parse_xml(raw_xml_text)
>>> print(xml.Header.From.Credential.join_prop('domain'))
DUNS, CompanyName, InteropKey
>>>
XML with namespaces
The dunder_xml_reader
package can handle XML with varying defined namespaces. It does
this by optionally replacing namespaces in tags with an alias prefix. Take the following XML
for example:
<?xml version="1.0" encoding="UTF-8"?>
<soapenv:Envelope xmlns:soapenv="http://schemas.xmlsoap.org/soap/envelope/">
<soapenv:Body>
<ReceiveClaimResponse xmlns="http://insurance.com/webservices/ReceiveClaim.job">
<ClaimResponse>
<status>S</status>
<errorDetail/>
<claimNumber>LC0000001742</claimNumber>
</ClaimResponse>
</ReceiveClaimResponse>
</soapenv:Body>
</soapenv:Envelope>
Using the namespaces
argument to parse_xml()
, you can have the attributes prefixed with
an alias:
>>> soap = parse_xml(raw_soap_text, namespaces={
... 'http://schemas.xmlsoap.org/soap/envelope/': 'soap',
... 'http://insurance.com/webservices/ReceiveClaim.job': 'claim'
... }
>>> soap.soap_Body.claim_ReceiveClaimResponse.claim_ClaimResponse.claim_status.text()
S
>>>
If you don't want any prefixes and you just want to strip out the namespace references you
can either pass a set
of namespace strings to parse_xml()
:
>>> soap = parse_xml(raw_soap_text, namespaces={
... 'http://schemas.xmlsoap.org/soap/envelope/',
... 'http://insurance.com/webservices/ReceiveClaim.job'
... }
>>> soap.Body.ReceiveClaimResponse.ClaimResponse.status.text()
S
>>>
Bonus: SafeReference
SafeReference is a wrapper that you can wrap a xmlNode instance in (or any other Python object graph for that matter) that will prevent errors like:
TypeError: 'NoneType' object is not subscriptable
AttributeError: 'NoneType' object has no attribute 'blah'
A quick example. Say you want to issue the following line of code:
>>> x = xml.Request.OrderRequest.OrderRequestHeader.BillTo.Address.PostalAddress.City.text()
But, the XML that was parsed in order to create the xml
object doesn't have a BillTo
element. You'd expect to get the following exception:
AttributeError: 'OrderRequestHeader' object has no attribute 'BillTo'
Now you could test your way through the graph heirchy (aka look-before-you-leap):
>>> if hasattr(xml, 'Request') and \
... hasattr(xml.Request, 'OrderRequest') and \
... hasattr(xml.Request.OrderRequest, 'OrderRequestHeader') and \
... hasattr(xml.Request.OrderRequest.OrderRequestHeader, 'BillTo'):
... x = xml.Request.OrderRequest.OrderRequestHeader.BillTo.Address.PostalAddress.City.text()
... else:
... x = ''
Or, you could just wrap your assignment in a try/except (aka ask-for-forgiveness):
>>> try:
... x = xml.Request.OrderRequest.OrderRequestHeader.BillTo.Address.PostalAddress.City.text()
... except AttributeError:
... x = ''
But in both cases you're writing a lot of "check to see if it's really there"-type code. The goal of SafeReference is to prevent that code from needing to be written in the first place. If your object-graph is wrapped in a SafeReference instance, it will check at each level for the requested attribute to be there or missing. If it's there, it will return the requested attribute. However, if it's not there, it will return a default-value. Here's an example:
>>> xml = parse_xml(raw_xml_text_thats_missing_a_BillTo_element)
>>> safe_xml = safe_reference(xml, 'n/a')
>>> x = safe_xml.Request.OrderRequest.OrderRequestHeader.BillTo.Address.PostalAddress.City.text()
>>> print(x)
n/a
>>>
This result of n/a
will be the result no matter what item is missing (Request
, OrderRequest
,
OrderRequestHeader
, BillTo
, etc.) along the way. If any of the objects in that long
dereferencing are missing, the default will be returned. However, if nothing is missing, the
actual graph value will be returned.
>>> xml = parse_xml(raw_xml_text_with_nothing_missing)
>>> safe_xml = safe_reference(xml, 'n/a')
>>> x = safe_xml.Request.OrderRequest.OrderRequestHeader.BillTo.Address.PostalAddress.City.text()
>>> print(x)
New York
>>>
Bonus 2: extract_namespaces()
If you have a chunk of XML or a SOAP message that conains numerous namespace references, figuring
our what the namespaces are (so you can pass them as an argument to the parse_xml()
), you can use
the extract_namespaces()
function:
>>> from dunder_xml_reader import extract_namespaces
>>> xml = """
... <section xmlns="http://www.ibm.com/events" xmlns:bk="urn:loc.gov:books" xmlns:pi="urn:personalInformation"
... xmlns:isbn='urn:ISBN:0-395-36341-6'>
... <title>Book-Signing Event</title>
... <signing>
... <bk:author pi:title="Mr" pi:name="Jim Ross"/>
... <book bk:title="Writing COBOL for Fun and Profit" isbn:number="0426070806"/>
... <comment xmlns=''>What a great issue!</comment>
... </signing>
... </section>
... """
...
>>> ns = extract_namespaces(xml)
>>> print(ns)
{(None, 'http://www.ibm.com/events'), ('bk', 'urn:loc.gov:books'), ('pi', 'urn:personalInformation'), ('isbn', 'urn:ISBN:0-395-36341-6') }
>>>
Project details
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
File details
Details for the file dunder_xml_reader-0.1.1.tar.gz
.
File metadata
- Download URL: dunder_xml_reader-0.1.1.tar.gz
- Upload date:
- Size: 12.2 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: poetry/1.7.1 CPython/3.8.18 Linux/6.2.0-1018-azure
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 688fd2d240443505e3be46f523abe1f6ad5d179227eed35a972782df1cc6fb02 |
|
MD5 | 0bc22d2de8fe558acf8a72ddb81e5ef5 |
|
BLAKE2b-256 | 055be21bb8c35599cc4784fdbd31a8ed43edd5e47d3844da1907972060b56e65 |
File details
Details for the file dunder_xml_reader-0.1.1-py3-none-any.whl
.
File metadata
- Download URL: dunder_xml_reader-0.1.1-py3-none-any.whl
- Upload date:
- Size: 10.0 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: poetry/1.7.1 CPython/3.8.18 Linux/6.2.0-1018-azure
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | f38d738de4991e4c12924f4a19465ba3a9f11df74b397c6fa4c86afc21b0b8b5 |
|
MD5 | c6af272f457ff8c448f9b65abe3ece70 |
|
BLAKE2b-256 | 01dffb8f3f8d4d7f17d9bf98d67ad6c0398353774f8a6db1746a8d5f4d488fc3 |