Skip to main content

large output in XML using unicode and namespaces

Project description

loxun is a Python module to write large output in XML using Unicode and namespaces. Of course you can also use it for small XML output with plain 8 bit strings and no namespaces.

loxun’s features are:

  • small memory foot print: the document is created on the fly by writing to an output stream, no need to keep all of it in memory.

  • easy to use namespaces: simply add a namespace and refer to it using the standard namespace:tag syntax.

  • mix unicode and io.BytesIO: pass both unicode or plain 8 bit strings to any of the methods. Internally loxun converts them to unicode, so once a parameter got accepted by the API you can rely on it not causing any messy UnicodeError trouble.

  • automatic escaping: no need to manually handle special characters such as < or & when writing text and attribute values.

  • robustness: while you write the document, sanity checks are performed on everything you do. Many silly mistakes immediately result in an XmlError, for example missing end elements or references to undeclared namespaces.

  • open source: distributed under the GNU Lesser General Public License 3 or later.

Here is a very basic example. First you have to create an output stream. In many cases this would be a file, but for the sake of simplicity we use a io.BytesIO here:

>>> from __future__ import unicode_literals
>>> import io
>>> out = io.BytesIO()

Then you can create an XmlWriter to write to this output:

>>> xml = XmlWriter(out)

Now write the content:

>>> xml.addNamespace("xhtml", "http://www.w3.org/1999/xhtml")
>>> xml.startTag("xhtml:html")
>>> xml.startTag("xhtml:body")
>>> xml.text("Hello world!")
>>> xml.tag("xhtml:img", {"src": "smile.png", "alt": ":-)"})
>>> xml.endTag()
>>> xml.endTag()
>>> xml.close()

And the result is:

>>> print out.getvalue().rstrip("\r\n")
<?xml version="1.0" encoding="utf-8"?>
<xhtml:html xmlns:xhtml="http://www.w3.org/1999/xhtml">
  <xhtml:body>
    Hello world!
    <xhtml:img alt=":-)" src="smile.png" />
  </xhtml:body>
</xhtml:html>

Writing a simple document

The following example creates a very simple XHTML document.

To make it simple, the output goes to a BytesIO, but you could also use a binary file that has been created using io.open(filename, "wb").

>>> from __future__ import unicode_literals
>>> import io
>>> out = io.BytesIO()

First create an XmlWriter to write the XML code to the specified output:

>>> xml = XmlWriter(out)

This automatically adds the XML prolog:

>>> print out.getvalue().rstrip("\r\n")
<?xml version="1.0" encoding="utf-8"?>

Next add the <html> start tag:

>>> xml.startTag("html")

Now comes the <body>. To pass attributes, specify them in a dictionary. So in order to add:

<body id="top">

use:

>>> xml.startTag("body", {"id": "top"})

Let’ add a little text so there is something to look at:

>>> xml.text("Hello world!")

Wrap it up: close all elements and the document.

>>> xml.endTag()
>>> xml.endTag()
>>> xml.close()

And this is what we get:

>>> print out.getvalue().rstrip("\r\n")
<?xml version="1.0" encoding="utf-8"?>
<html>
  <body id="top">
    Hello world!
  </body>
</html>

Specifying attributes

First create a writer:

>>> import io
>>> out = io.BytesIO()
>>> xml = XmlWriter(out)

Now write the content:

>>> xml.tag("img", {"src": "smile.png", "alt": ":-)"})

Attribute values do not have to be strings, other types will be converted to Unicode using Python’s unicode() function:

>>> xml.tag("img", {"src": "wink.png", "alt": ";-)", "width": 32, "height": 24})

And the result is:

>>> print out.getvalue().rstrip("\r\n")
<?xml version="1.0" encoding="utf-8"?>
<img alt=":-)" src="smile.png" />
<img alt=";-)" height="24" src="wink.png" width="32" />

Using namespaces

Now the same thing but with a namespace. First create the prolog and header like above:

>>> out = io.BytesIO()
>>> xml = XmlWriter(out)

Next add the namespace:

>>> xml.addNamespace("xhtml", "http://www.w3.org/1999/xhtml")

Now elements can use qualified tag names using a colon (:) to separate namespace and tag name:

>>> xml.startTag("xhtml:html")
>>> xml.startTag("xhtml:body")
>>> xml.text("Hello world!")
>>> xml.endTag()
>>> xml.endTag()
>>> xml.close()

As a result, tag names are now prefixed with “xhtml:”:

>>> print out.getvalue().rstrip("\r\n")
<?xml version="1.0" encoding="utf-8"?>
<xhtml:html xmlns:xhtml="http://www.w3.org/1999/xhtml">
  <xhtml:body>
    Hello world!
  </xhtml:body>
</xhtml:html>

Working with non ASCII characters

Sometimes you want to use characters outside the ASCII range, for example German Umlauts, the Euro symbol or Japanese Kanji. The easiest and performance wise best way is to use Unicode strings. For example:

>>> import io
>>> out = io.BytesIO()
>>> xml = XmlWriter(out, prolog=False)
>>> xml.text(u"The price is \u20ac 100") # Unicode of Euro symbol
>>> out.getvalue().rstrip("\r\n")
'The price is \xe2\x82\xac 100'

Notice the “u” before the string passed to XmlWriter.text(), it declares the string to be a unicode string that can hold any character, even those that are beyond the 8 bit range.

Also notice that in the output the Euro symbol looks very different from the input. This is because the output encoding is UTF-8 (the default), which has the advantage of keeping all ASCII characters the same and turning any characters with a code of 128 or more into a sequence of 8 bit bytes that can easily fit into an output stream to a binary file or io.BytesIO.

If you have to stick to classic 8 bit string parameters, loxun attempts to convert them to unicode. By default it assumes ASCII encoding, which does not work out as soon as you use a character outside the ASCII range:

>>> import io
>>> out = io.BytesIO()
>>> xml = XmlWriter(out, prolog=False)
>>> xml.text("The price is \xa4 100") # ISO-8859-15 code of Euro symbol
Traceback (most recent call last):
    ...
UnicodeDecodeError: 'ascii' codec can't decode byte 0xa4 in position 13: ordinal not in range(128)

In this case you have to tell the writer the encoding you use by specifying the the sourceEncoding:

>>> import io
>>> out = io.BytesIO()
>>> xml = XmlWriter(out, prolog=False, sourceEncoding="iso-8859-15")

Now everything works out again:

>>> xml.text("The price is \xa4 100") # ISO-8859-15 code of Euro symbol
>>> out.getvalue().rstrip("\r\n")
'The price is \xe2\x82\xac 100'

Of course in practice you will not mess around with hex codes to pass your texts. Instead you just specify the source encoding using the mechanisms described in PEP 263, Defining Python Source Code Encodings.

Pretty printing and indentation

By default, loxun starts a new line for each startTag and indents the content with two spaces. You can change the spaces to any number of spaces and tabs you like:

>>> out = io.BytesIO()
>>> xml = XmlWriter(out, indent="    ") # <-- Indent with 4 spaces.
>>> xml.startTag("html")
>>> xml.startTag("body")
>>> xml.text("Hello world!")
>>> xml.endTag()
>>> xml.endTag()
>>> xml.close()
>>> print out.getvalue().rstrip("\r\n")
<?xml version="1.0" encoding="utf-8"?>
<html>
    <body>
        Hello world!
    </body>
</html>

You can disable pretty printing all together using pretty=False, resulting in an output of a single large line:

>>> out = io.BytesIO()
>>> xml = XmlWriter(out, pretty=False) # <-- Disable pretty printing.
>>> xml.startTag("html")
>>> xml.startTag("body")
>>> xml.text("Hello world!")
>>> xml.endTag()
>>> xml.endTag()
>>> xml.close()
>>> print out.getvalue().rstrip("\r\n")
<?xml version="1.0" encoding="utf-8"?><html><body>Hello world!</body></html>

Changing the XML prolog

When you create a writer, it automatically write an XML prolog processing instruction to the output. This is what the default prolog looks like:

>>> import io
>>> out = io.BytesIO()
>>> xml = XmlWriter(out)
>>> print out.getvalue().rstrip("\r\n")
<?xml version="1.0" encoding="utf-8"?>

You can change the version or encoding:

>>> out = io.BytesIO()
>>> xml = XmlWriter(out, encoding=u"ascii", version=u"1.1")
>>> print out.getvalue().rstrip("\r\n")
<?xml version="1.1" encoding="ascii"?>

To completely omit the prolog, set the parameter prolog=False:

>>> out = io.BytesIO()
>>> xml = XmlWriter(out, prolog=False)
>>> out.getvalue()
''

Adding other content

Apart from text and tags, XML provides a few more things you can add to documents. Here’s an example that shows how to do it with loxun.

First, create a writer:

>>> import io
>>> out = io.BytesIO()
>>> xml = XmlWriter(out)

Let’s add a document type definition:

>>> xml.raw("<!DOCTYPE html PUBLIC \"-//W3C//DTD XHTML 1.0 Strict//EN\" SYSTEM \"http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd\">")
>>> xml.newline()

Notice that loxun uses the generic XmlWriter.raw() for that, which allows to add any content without validation or escaping. You can do all sorts of nasty things with raw() that will result in invalid XML, but this is one of its reasonable uses.

Next, let’s add a comment:

>>> xml.comment("Show case some rarely used XML constructs")

Here is a processing instruction:

>>> xml.processingInstruction("xml-stylesheet", "href=\"default.css\" type=\"text/css\"")

And finally a CDATA section:

>>> xml.cdata(">> this will not be parsed <<")

And the result is:

>>> print out.getvalue().rstrip("\r\n")
<?xml version="1.0" encoding="utf-8"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" SYSTEM "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
<!-- Show case some rarely used XML constructs -->
<?xml-stylesheet href="default.css" type="text/css"?>
<![CDATA[>> this will not be parsed <<]]>

Optimization

Loxun automatically optimized pairs of empty start/end tags. For example:

>>> out = io.BytesIO()
>>> xml = XmlWriter(out)
>>> xml.startTag("customers")
>>> xml.startTag("person", {"id": "12345", "name": "Doe, John"})
>>> xml.endTag("person") # without optimization, this would add </person>.
>>> xml.endTag()
>>> xml.close()
>>> print out.getvalue().rstrip("\r\n")
<?xml version="1.0" encoding="utf-8"?>
<customers>
  <person id="12345" name="Doe, John" />
</customers>

Despite the explicit startTag("person") and matching endtag(), the output only contains a simple <person ... /> tag.

Contributing

If you want to help improve loxun, you can access the source code at <http://github.com/roskakori/loxun>.

Future

Currently loxun does what it was built for.

There are is no real plans to improve it in the near future, but here is a list of features that might be added at some point:

  • Add validation of tag and attribute names to ensure that all characters used are allowed. For instance, currently loxun does not complain about a tag named “a#b*c$d_”.

  • Raise an XmlError when namespaces are added with attributes instead of XmlWriter.addNamespace().

  • Logging support to simplify debugging of the calling code. Probably XmlWriter would get a property logger which is a standard logging.Logger. By default it could log original exceptions that loxun turns into an XmlError and namespaces opened and closed. Changing it to logging.DEBUG would log each tag and XML construct written, including additional information about the internal tag stack. That way you could dynamically increase or decrease logging output.

  • Rethink pretty printing. Instead of a global property that can only be set when initializing an XmlWriter, it could be a optional parameter for XmlWriter.startTag() where it could be turned on and off as needed. And the property could be named literal instead of pretty (with an inverse logic).

  • Add a DomWriter that creates a xml.dom.minidom.Document.

Some features other XML libraries support but I never saw any real use for:

  • Specify attribute order for tags.

Version history

Version 2.0, 2014-07-28

  • Added support for Python 3.2+ while retaining the option to run with Python 2.6+ (issue #5; thanks go to Stefan Schwarzer who offered his guidance during a “Python 2 to 3” sprint at EuroPython 2014).

  • Dropped support for Python 2.5, keep using loxun 1.3 if you are stuck with with this version.

Version 1.3, 2012-01-01

  • Added endTags() to close several or all open tags (issue #3, contributed by Anton Kolechkin).

  • Added ChainXmlWriter which is similar to XmlWriter and allows to chain methods for more concise source code (issue #3, contributed by Anton Kolechkin).

Version 1.2, 2011-03-12

  • Fixed AttributeError when XmlWriter(..., encoding=...) was set.

Version 1.1, 08-Jan-2011

  • Fixed AssertionError when pretty was set to False (issue #1; fixed by David Cramer).

Version 1.0, 11-Oct-2010

Version 0.8, 11-Jul-2010

  • Added possibility to pass attributes to XmlWriter.startTag() and XmlWriter.tag() with values that have other types than str or unicode. When written to XML, the value is converted using Python’s built-in unicode() function.

  • Added a couple of files missing from the distribution, most important the test suite.

Version 0.7, 03-Jul-2010

  • Added optimization of matching start and end tag without any content in between. For example, x.startTag("some"); x.endTag() results in <some /> instead of <some></some>.

  • Fixed handling of unknown name spaces. They now raise an XmlError instead

    of ValueError.

Version 0.6, 03-Jun-2010

  • Added option indent to specify the indentation text each new line starts with.

  • Added option newline to specify how lines written should end.

  • Fixed that XmlWriter.tag() did not remove namespaces declared immediately before it.

  • Cleaned up documentation.

Version 0.5, 25-May-2010

  • Fixed typo in namespace attribute name.

  • Fixed adding of namespaces before calls to XmlWriter.tag() which resulted in an XmlError.

Version 0.4, 21-May-2010

  • Added option sourceEncoding to simplify processing of classic strings. The manual section “Working with non ASCII characters” explains how to use it.

Version 0.3, 17-May-2010

  • Added scoped namespaces which are removed automatically by XmlWriter.endTag().

  • Changed text() to normalize newlines and white space if pretty printing is enabled.

  • Moved writing of XML prolog to the constructor and removed XmlWriter.prolog(). To omit the prolog, specify prolog=False when creating the XmlWriter. If you later want to write the prolog yourself, use XmlWriter.processingInstruction().

  • Renamed *Element() to *Tag because they really only write tags, not whole elements.

Version 0.2, 16-May-2010

  • Added XmlWriter.comment(), XmlWriter.cdata() and XmlWriter.processingInstruction() to write these specific XML constructs.

  • Added indentation and automatic newline to text if pretty printing is enabled.

  • Removed newline from prolog in case pretty printing is disabled.

  • Fixed missing “?” in prolog.

Version 0.1, 15-May-2010

  • Initial release.

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

loxun-2.0.zip (27.8 kB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page