19.5. Modules de traitement XML
*******************************

Les interfaces de Python de traitement de XML sont regroupées dans le
paquet "xml".

Avertissement: The XML modules are not secure against erroneous or
  maliciously constructed data.  If you need to parse untrusted or
  unauthenticated data see Vulnérabilités XML.

Il est important de noter que les modules dans le paquet "xml"
nécessitent qu’au moins un analyseur compatible SAX soit disponible.
L’analyseur Expat est inclus dans Python, ainsi le module
"xml.parsers.expat" est toujours disponible.

La documentation des *bindings* des interfaces DOM et SAX  se trouve
dans "xml.dom" et "xml.sax".

Les sous-modules de traitement XML sont :

* "xml.etree.ElementTree": l’API ElementTree, un processeur simple
  et léger

* "xml.dom": la définition de l’API DOM

* "xml.dom.minidom": une implémentation minimale de DOM

* "xml.dom.pulldom": gestion de la construction partiel des arbres
  DOM

* "xml.sax": classes de bases SAX2 base et fonctions utilitaires

* "xml.parsers.expat": le *binding* de l’analyseur Expat


19.6. Vulnérabilités XML
************************

The XML processing modules are not secure against maliciously
constructed data. An attacker can abuse vulnerabilities for e.g.
denial of service attacks, to access local files, to generate network
connections to other machines, or to or circumvent firewalls. The
attacks on XML abuse unfamiliar features like inline DTD (document
type definition) with entities.

The following table gives an overview of the known attacks and if the
various modules are vulnerable to them.

+---------------------------+----------------+-----------------+----------------+----------------+----------------+
| type                      | sax            | etree           | minidom        | pulldom        | xmlrpc         |
+===========================+================+=================+================+================+================+
| *billion laughs*          | **Vulnérable** | **Vulnérable**  | **Vulnérable** | **Vulnérable** | **Vulnérable** |
+---------------------------+----------------+-----------------+----------------+----------------+----------------+
| *quadratic blowup*        | **Vulnérable** | **Vulnérable**  | **Vulnérable** | **Vulnérable** | **Vulnérable** |
+---------------------------+----------------+-----------------+----------------+----------------+----------------+
| *external entity          | **Vulnérable** | Sûr    (1)      | Sûr    (2)     | **Vulnérable** | Sûr    (3)     |
| expansion*                |                |                 |                |                |                |
+---------------------------+----------------+-----------------+----------------+----------------+----------------+
| Récupération de DTD       | **Vulnérable** | Sûr             | Sûr            | **Vulnérable** | Sûr            |
+---------------------------+----------------+-----------------+----------------+----------------+----------------+
| *decompression bomb*      | Sûr            | Sûr             | Sûr            | Sûr            | **Vulnérable** |
+---------------------------+----------------+-----------------+----------------+----------------+----------------+

1. "xml.etree.ElementTree" doesn’t expand external entities and
   raises a ParserError when an entity occurs.

2. "xml.dom.minidom" n’étend pas les entités externe et renvoie
   simplement le verbatim de l’entité non étendu.

3. "xmlrpclib" n’étend pas les entités externes et les omet.

*billion laughs* / *exponential entity expansion*
   The Billion Laughs attack – also known as exponential entity
   expansion – uses multiple levels of nested entities. Each entity
   refers to another entity several times, the final entity definition
   contains a small string. Eventually the small string is expanded to
   several gigabytes. The exponential expansion consumes lots of CPU
   time, too.

*quadratic blowup entity expansion*
   A quadratic blowup attack is similar to a Billion Laughs attack; it
   abuses entity expansion, too. Instead of nested entities it repeats
   one large entity with a couple of thousand chars over and over
   again. The attack isn’t as efficient as the exponential case but it
   avoids triggering countermeasures of parsers against heavily nested
   entities.

*external entity expansion*
   Entity declarations can contain more than just text for
   replacement. They can also point to external resources by public
   identifiers or system identifiers. System identifiers are standard
   URIs or can refer to local files. The XML parser retrieves the
   resource with e.g. HTTP or FTP requests and embeds the content into
   the XML document.

Récupération de DTD
   Certaines bibliothèques XML comme "xml.dom.pulldom" de Python
   récupère les documents de définitions de types (DTD) depuis des
   emplacements distants ou locaux. La fonctionnalité a des
   implications similaires que le problème d’extension d’entités
   externes.

*decompression bomb*
   The issue of decompression bombs (aka ZIP bomb) apply to all XML
   libraries that can parse compressed XML stream like gzipped HTTP
   streams or LZMA-ed files. For an attacker it can reduce the amount
   of transmitted data by three magnitudes or more.

The documentation of defusedxml on PyPI has further information about
all known attack vectors with examples and references.


19.6.1. defused packages
========================

These external packages are recommended for any code that parses
untrusted XML data.

defusedxml is a pure Python package with modified subclasses of all
stdlib XML parsers that prevent any potentially malicious operation.
The package also ships with example exploits and extended
documentation on more XML exploits like xpath injection.

defusedexpat provides a modified libexpat and patched replacement
"pyexpat" extension module with countermeasures against entity
expansion DoS attacks. Defusedexpat still allows a sane and
configurable amount of entity expansions. The modifications will be
merged into future releases of Python.

The workarounds and modifications are not included in patch releases
as they break backward compatibility. After all inline DTD and entity
expansion are well-defined XML features.
