xmlser is a python library that makes generating XML less painful.

Generating XML from python objects can and will never be as easy as generating JSON or YAML without heavy restriction on the resulting XML. Instead, this library uses specialized format strings allowing terse expression of the desired transformation from a known python object structure.

To run the examples below, start off with the following definition:

>>> import xmlser
>>> def ser(fmt, obj=None):
...     print xmlser.serialize(fmt, obj)

We start off with a plain document containing only two tags:

>>> ser('<doc<item>>')
<doc><item></item></doc>

As you can see, the closing tag is dropped by using the greater-than (">") character as a terminator. So how do we add text elements?

>>> ser('<doc<item&text>>')
<doc><item>text</item></doc>

Non-ASCII letters can also be used:

>>> ser(u'<cities<city&東京><city&Zürich>>')
<cities><city>東京</city><city>Zürich</city></cities>
>>> # The first city should be Tokyo, someone tell me if I'm wrong

Strings with special characters used by xmlser must be quoted:

>>> ser(r'<doc<"item.name"&"*=<>?.&~\"\\· ·\t·\n·">>')
<doc><item.name>*=&lt;&gt;?.&amp;~"\· ·        ·
·</item.name></doc>
>>> # Output may differ due to tab

Finally, attributes:

>>> ser('<doc<item=id"1">>')
<doc><item id="1"></item></doc>

So far, so good; but not very interesting.

There is always a "current" object. It can be displayed using a question mark ("?") as follows:

>>> ser('<doc<item&?>>', "text & more text")
<doc><item>text &amp; more text</item></doc>

This lookup can be used in all sorts of places as long as the result is legal XML:

>>> ser('<doc<item=?"1"><item=id?><?>>', "name")
<doc><item name="1"></item><item id="name"></item><name></name></doc>
>>> ser('<doc<item=?"1">>', "a&b")
ValueError: XML attribute name contains invalid characters

Attribute, key and index lookups can be performed using the dot (".") operator:

>>> ser('<doc<item&.content>>', {"content": "Ich habe eine Nase!"})
<doc><item>Ich habe eine Nase!</item></doc>
>>> ser('<doc<item&.0>>', ["Chuchichästli"])
<doc><item>Chuchichästli</item></doc>

They can easily be chained:

>>> ser('<doc<item&.content.0>>', {"content": ["Kilimanjaro"]})
<doc><item>Kilimanjaro</item></doc>

When repeatingly accessing sub-objects of an object, a group with a lookup can shorten things:

>>> ser('<doc<item&.content.0><item&.content.1>>', {"content": ["Vinson Massif", "Puncak Jaya"]})
<doc><item>Vinson Massif</item><item>Puncak Jaya</item></doc>
>>> ser('<doc{.content<item&.0><item&.1>>}>', {"content": ["Vinson Massif", "Puncak Jaya"]})
<doc><item>Vinson Massif</item><item>Puncak Jaya</item></doc>

Groups will come in handy again for conditionals.

Tag repetition can be performed over any iterable. If the iterable has an "items" method, it is called and the result is used for iteration. This allows iteration of mappings as lists of two-element tuples.

Within the repeated tags, the current element from the repetition is the current object for lookups:

>>> ser('<longest_rivers<river*?&?>>', [u"النيل", u"Amazonas", u"长/長江"])
<longest_rivers><river>ﺎﻠﻨﻴﻟ</river><river>Amazonas</river><river>长/長江</river></longest_rivers>
>>> ser('<person<.0*?&.1>>', {"fullname": "Thomas Anderson", "title": "Mr.", "nick": "Neo"})
<person><nick>Neo</nick><fullname>Thomas Anderson</fullname><title>Mr.</title></person>

A notable quirk about repetition is that despite the source iterable coming after the tag name, lookups in the tag name use the iterated items.

Conditionals start with a tilde ("~") character, followed by the condition, the true case and optionally another tilde and the false case:

>>> ser('<doc~??<item&?>>', 'hello')
<doc><item>hello</item></doc>
>>> ser('<doc~??<item&?>>', '')
<doc></doc>

Operators are infix, i.e. preceded by the first parameter. Four conditional operators are available:

Else clauses are made by adding a another tilde after the true-case, followed by the false-case:

>>> ser('<doc<item~??&?~&world>>', 'hello')
<doc><item>hello</item></doc>
>>> ser('<doc<item~??&?~&world>>', '')
<doc><item>world</item></doc>

Thus, else-if chains can be made simply by beginning another conditional within the else clause:

>>> ser('<doc<part*4~?=0&Hello~~?=1", "~~?=2World~"!">>')
<doc><part>Hello</part><part>, </part><part>World</part><part>!</part></doc>

Finally, grouping with { and } allow for complex expressions within conditionals:

>>> ser('<person<.0*?~.0=name{.1<.0*?&.1><fullname&.givenname&" "&.surname>}~~.0=groups<group*.1&?>~&.1>>',
        { 'name': {'givenname': 'John', 'surname': 'Smith'},
          'groups': ['employees', 'generic'],
          'birthday': '1970-01-01',
        })
<person><birthday>1970-01-01</birthday><name><givenname>John</givenname><surname>Smith</surname><fullname>John Smith</fullname></name><groups><group>employees</group><group>generic</group></groups></person>

TODO