django-import-data 0.1.4

A Django command line tool for importing HTML, XML and JSON data to models via XSLT mapping

Django import data is a command-line tool for importing XML and HTML data to django models via XSLT mapping.

Source code is located here - https://github.com/lev-veshnyakov/django-import-data.

Basic features

Django import data can take any XML or HTML source file or URL as an input and save entities from it to the django models without need to modify an existing code.

It also supports saving of a related data in form one-to-many and many-to-many.

Dependencies

It uses lxml library for all xml manipulations.

Installation

First you need to install dependencies for lxml library:

sudo apt-get install libxml2-dev libxslt-dev python-dev

Then install django-import-data using pip:

pip install django-import-data

If you want the latest version you can install it from Github:

pip install git+https://github.com/lev-veshnyakov/django-import-data

Add import_data to INSTALLED_APPS:

INSTALLED_APPS = [
    ...
    'import_data',
]

Usage

Django import data is a management command-line tool, that can be used from the code as well.

Too see the list of console commands type:

python manage.py help

In the output you will find data_import section like below:

[data_import]
    process_xslt
    validate_xml

To get help for the particular command type:

python manage.py help process_xslt

python manage.py help validate_xml

To call console commands from your code use django.core.management.call_command:

from django.core.management import call_command

call_command('process_xslt', 'http://stackoverflow.com/', 'transform.xslt', '--save')

How it works

In a few words it takes a source in either XML or HTML, then takes provided by you XSLT file, transforms the source into the specific XML representation, and then saves the data from this XML to the database using models.

The point is, that you don’t need to write procedural code for saving data. You only need to write XSLT files, which is actually XML. One file for one source. By the source I mean a range of XML or HTML files in the same format. For example all google search result pages have one schema. That means that you can write only one XSLT transformation file to import all search pages data.

The difficult moment is that you have to be familiar with XSLT and Xpath.

XSLT and XPath

XSLT is a language for transforming XML documents into XHTML documents or to other XML documents.

XSLT uses XPath to find information in an XML document. XPath is used to navigate through elements and attributes in XML documents.

If you are not familiar with that I reccomend you to read a short tutorial on www.w3school.com.

Moreover, you have to know what an XML Schema is and a particular schema language RELAX NG.

XML Schema and RELAX NG

Django import data uses RELAX NG to validate resuls of transformations. That means if you write XSLT file wrong, it wouldn’t be accepted.

But you dont have to write RELAX NG schema yoursef, it’s already included in the module.

Resulting XML

After XSLT transformation and schema validation the resulting XML file should be like following:

<?xml version="1.0" encoding="UTF-8"?>
<mapping>
    <model model="app.Author">
        <item key="1">
            <field name="name">Andrew Tanenbaum</field>
        </item>
        <item key="2">
            <field name="name">Donald Knuth</field>
        </item>
    </model>
    <model model="app.Book">
        <item key="1">
            <field name="name">Computer Networks</field>
            <field name="ISBN">0130661023</field>
            <fk model="app.Author" key="1"/>
        </item>
        <item key="2">
            <field name="name">The Art of Computer Programming</field>
            <field name="ISBN">0321751043</field>
            <m2mk model="app.Author" key="2"/>
        </item>
    </model>
</mapping>

This XML can be automatically saved to the models.

It contains the root element <mapping/>. Into it are nested <model/> elements. Each model element represents a particular django model. You must provide model="" attributes, in which specify a related model. Path to the model is in following format: application_name.ModelName, the same format like manage.py dumpdata uses.

Model elements don’t have to be unique. If you specify several model elements with the same model attribute, they will be merged together. This concerns to item elements as well.

Model elements contain <item/> elements, representing particular records in the database. They have only one required attribute name="", which sets the name of a related model field.

Foreign keys

Django import data supports import of related entities in the form one-to-many and many-to-many. To save such entities your models should have appropriate foreign keys.

In a resulting XML you can use <fk/> and <m2m/> elements (see above). They have model="" and key="" attributes, pointing to the related <item/> elements.

Setting key attribute

The key="" attribute of <item/> elements must be unique by each unique record. It has not to be the same as a primary key value in the database. It even will not be stored (if you want to store a primary key value, use <field/> element).

Therefore, the value of the key="" attribute not obliged to be integer. You can use any sring. Often it’s convenient to use an URL as the key.

You can even omit filling that attribute if you don’t have related items.

But one case is special. That’s when you don’t have any unique attributes in the source. In that case you can use generate-id(..) XPath function. It will generate unique IDs for every separate XML node in the source.

Examples

Save data to one model

In this simple example we will parse the main page of stackoverflow.com and save titles of recent questions to this model:

from django.db import models

class Question(models.Model):
    title = models.CharField(max_length=255)

First we need to write an XSLT file:

<?xml version="1.0" encoding="UTF-8"?>
<mapping xsl:version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
    <model model="test_app.Question">
        <xsl:for-each select="//a[@class='question-hyperlink']">
            <item key="">
                <field name="title">
                    <xsl:value-of select="."/>
                </field>
            </item>
        </xsl:for-each>
    </model>
</mapping>

Name it transform.xslt and perform the following command:

python manage.py process_xslt http://stackoverflow.com/questions transform.xslt --validate

The output will be like this (but longer):

<?xml version="1.0" encoding="utf-8"?>
<mapping>
  <model model="xml_json_import.Article">
    <item key="">
      <field name="title">customizing soap response attribute format</field>
    </item>
    <item key="">
      <field name="title">Second fragment loaded but not visible on screen</field>
    </item>
    <item key="">
      <field name="title">django-oscar :first time use "python manage.py migrate" gets error</field>
    </item>
    <item key="">
      <field name="title">JTable fireTableDataChanged() method doesn't refresh table</field>
    </item>
    <item key="">
      <field name="title">why the dynamic nodes dont respond to click in jstree?</field>
    </item>
    <item key="">
      <field name="title">Connecting kdb+ to R</field>
    </item>
  </model>
</mapping>

Parameter --validate adds to output Document is valid.

To save the result add the parameter --save to the command above.

Save data to related models

Use the same source and add two other models with foreign keys:

class Question(models.Model):
    title = models.CharField(max_length=255)
    user = models.ForeignKey('User', null=True)
    tags = models.ManyToManyField('Tag')

class Tag(models.Model):
    title = models.CharField(max_length=255)

class User(models.Model):
    title = models.CharField(max_length=255)

The XSLT file will be like following:

<?xml version="1.0" encoding="UTF-8"?>
<mapping xsl:version="1.0" xmlns:xsl="http://www.w3.org/1999/XSL/Transform">
    <model model="test_app.Question">
        <xsl:for-each select="//div[@class='question-summary narrow']">
            <item key="">
                <field name="title">
                    <xsl:value-of select=".//a[@class='question-hyperlink']"/>
                </field>
                <fk model="test_app.User">
                    <xsl:attribute name="key">
                        <xsl:value-of select="generate-id(.//div[@class='started']/a[2])"/>
                    </xsl:attribute>
                </fk>
                <xsl:for-each select=".//a[@class='post-tag']">
                    <m2mk model="test_app.Tag">
                        <xsl:attribute name="key">
                            <xsl:value-of select="generate-id(.)"/>
                        </xsl:attribute>
                    </m2mk>
                </xsl:for-each>
            </item>
        </xsl:for-each>
    </model>

    <model model="test_app.Tag">
        <xsl:for-each select="//a[@class='post-tag']">
            <item>
                <xsl:attribute name="key">
                    <xsl:value-of select="generate-id(.)"/>
                </xsl:attribute>
                <field name="title">
                    <xsl:value-of select="."/>
                </field>
            </item>
        </xsl:for-each>
    </model>

    <model model="test_app.User">
        <xsl:for-each select="//div[@class='started']/a[2]">
            <item>
                <xsl:attribute name="key">
                    <xsl:value-of select="generate-id(.)"/>
                </xsl:attribute>
                <field name="title">
                    <xsl:value-of select="."/>
                </field>
            </item>
        </xsl:for-each>
    </model>
</mapping>

Take notice how calculated attributes are set and how to use generate-id function. In tis example we use both types of relationship: one-to-many and many-to-many. This means that one question can have several tags, but only one related user (which has changed the question last).

The output will be like following (but essential longer):

<?xml version="1.0" encoding="utf-8"?>
<mapping>
  <model model="test_app.Question">
    <item key="">
      <field name="title">C TCP Server doesn't send data before closing</field>
      <fk model="test_app.User" key="idp1407668180"/>
      <m2mk model="test_app.Tag" key="idp1407657924"/>
      <m2mk model="test_app.Tag" key="idp1407659948"/>
      <m2mk model="test_app.Tag" key="idp1407660732"/>
      <m2mk model="test_app.Tag" key="idp1407661540"/>
    </item>
    <item key="">
      <field name="title">Ninject Factory Extension</field>
      <fk model="test_app.User" key="idp1407665492"/>
      <m2mk model="test_app.Tag" key="idp1407676788"/>
      <m2mk model="test_app.Tag" key="idp1407674900"/>
      <m2mk model="test_app.Tag" key="idp1407678572"/>
      <m2mk model="test_app.Tag" key="idp1407678508"/>
      <m2mk model="test_app.Tag" key="idp1407652988"/>
    </item>
  <model model="test_app.Tag">
    <item key="idp1407657924">
      <field name="title">c</field>
    </item>
    <item key="idp1407659948">
      <field name="title">linux</field>
    </item>
    <item key="idp1407660732">
      <field name="title">sockets</field>
    </item>
    <item key="idp1407661540">
      <field name="title">tcp</field>
    </item>
    <item key="idp1407676788">
      <field name="title">c#</field>
    </item>
    <item key="idp1407674900">
      <field name="title">dependency-injection</field>
    </item>
    <item key="idp1407678572">
      <field name="title">ninject</field>
    </item>
    <item key="idp1407678508">
      <field name="title">ninject.web.mvc</field>
    </item>
    <item key="idp1407652988">
      <field name="title">ninject-extensions</field>
    </item>
  </model>
  <model model="test_app.User">
    <item key="idp1407668180">
      <field name="title">user3809727</field>
    </item>
    <item key="idp1407665492">
      <field name="title">user2119597</field>
    </item>
  </model>
</mapping>