Version 0.4, December 04, 2010
Albert Graef <Dr.Graef@t-online.de>
XML, the Extensible Markup Language, facilitates the exchange of complex structured data between applications and systems. XSLT allows you to transform XML documents to other XML-based formats such as HTML. Together, XML and XSLT let you create dynamic web content with ease. Both XML and XSLT are open standards by the W3C consortium (http://www.w3.org).
Pure’s XML interface is based on the libxml2 and libxslt libraries from the GNOME project. If you have a Linux system then you most likely have these libraries, otherwise you can get them from http://xmlsoft.org. For Windows users, the required dlls are available from the GnuWin32 project (http://gnuwin32.sourceforge.net) and are already included in the Pure MSI package.
Copyright (c) 2009 by Albert Graef <Dr.Graef@t-online.de>.
pure-xml is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.
pure-xml is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public License along with this program. If not, see <http://www.gnu.org/licenses/>.
Get the latest source from http://pure-lang.googlecode.com/files/pure-xml-0.4.tar.gz.
Run make and then sudo make install to compile and install this module. This requires libxml2, libxslt and Pure.
Use the following declaration to make the operations of this module available in your programs:
using xml;
The module defines two namespaces xml and xslt for the XML and the XSLT operations, respectively. For convenience, you can open these in your program as follows:
using namespace xml, xslt;
A number of complete examples illustrating the use of this module can be found in the examples directory in the source distribution.
This module represents XML documents using pointers to the xmlDoc and xmlNode structures provided by the libxml2 library. Similarly, stylesheets are simply pointers to the xmlStylesheet structure from libxslt (cf. Transformations). This makes it possible to use these objects directly with the operations of the libxml2 and libsxslt libraries (via Pure’s C interface) if necessary. Note, however, that these are all “cooked” pointers which take care of freeing themselves automatically when they are no longer needed, therefore you shouldn’t free them manually.
You can also check for these types of pointers using the following predicates:
checks whether x is an XML document pointer.
checks whether x is a pointer to a node in an XML document.
checks whether x is an XSLT stylesheet pointer.
An XML document is a rooted tree which can be created, traversed and manipulated using the operations of this module. There are different types of nodes in the tree, each carrying their own type of data. In Pure land, the node data is described using the following “node info” constructors.
An XML element with given (possibly qualified) name tag, a (possibly empty) list of namespace declarations ns and a (possibly empty) list of attributes attrs. Namespace declarations normally take the form of a pair of strings (prefix,href), where prefix is the prefix associated with the namespace and href the corresponding URI (the name of the namespace), but they can also be just a string href if the namespace prefix is missing. Attributes are encoded as key=>value pairs, where key is the attribute name and value the associated value; both key and value are strings.
A convenience function which denotes a combination of an element node with a text child. This is only used when creating a new node, and indicates that a text node child is to be added to the node automatically.
An attribute node. These only occur as results of the select and attrs functions, and cannot be inserted directly into a document.
A text node with the given content (a string).
A comment.
An entity reference (&name;).
Processing instructions. name is the application name, content the text of the processing instructions.
Besides the node types described above, there are some additional node types used in the document type definition (DTD), which can be extracted from a document using the int_subset and ext_subset functions. These are for inspection purposes only; it is not possible to change the DTD of a document in-place. (However, you can create a new document and attach a DTD to it, using the new_doc function.)
DTDs are represented using this special type of node, where name is the name of the root element, and extid is a pair consisting of the external identifier and the URI of the DTD (or just the URI if there is no external identifier). The xml::doctype node has as its children zero or more of the following kinds of DTD declaration nodes (these are just straightforward abstract syntax for the !ELEMENT, !ATTLIST and !ENTITY declarations inside a DTD declaration; see the XML specification for details).
Element declarations: Here, name is the element tag and content the definition of the element structure, see element content below. XML supports various kinds of element types, please refer to document type definition in the XML specification for details.
An undefined element. This is in libxml2, but not in the XML specification, you shouldn’t see this in normal operation.
An element without any content.
An element with unrestricted content.
A “mixed” element which can contain character data, optionally interspersed with child elements, as given in the content specification.
A standard element consisting only of child elements, as given in the content specification.
Attribute declarations: These are used to declare the attributes of an element. elem_name is the name of an element which describes the attribute type, name is the name of the attribute itself, and default specifies the default value of the attribute, see attribute defaults below. XML supports a bunch of different attribute types, please refer to document type definition in the XML specification for details.
Entity declarations: These are used for internal and external entity declarations. name is the entity name and content its definition. External entities also have an extid (external identifier/URI pair) identifying the entity.
The element content type (content argument of the element declaration nodes) is a kind of regular expression formed with tags (specified as strings) and the following constructors:
Attribute defaults (the default argument of attribute declaration nodes) are represented using the following constructor symbols:
a required attribute, i.e., the user must specify this
an implied attribute, i.e., the user does not have to specify this
an attribute with the given default value val
an attribute with the given fixed value val
This module provides all operations necessary to create, inspect and manipulate XML documents residing either in memory or on disk. Operations for formatting XML documents using XSLT stylesheets are also available.
The following functions allow you to create new XML documents, load them from or save them to a file or a string, and provide general information about a document.
This function creates an XML document. It returns a pointer to the new document. version is a string denoting the XML version (or "" to indicate the default). info is the node info of the root node (which should denote an element node). dtd denotes the document type which can be () to denote an empty DTD, a string (the URI/filename of the DTD), or a pair (pubid,sysid) where pubid denotes the public identifier of the DTD and sysid its system identifier (URI).
Note that only simple kinds of documents with an internal DTD can be created this way. Use the load_file or load_string function below to create a skeleton document if a more elaborate prolog is required.
Load an XML document from a file name or a string s. flags denotes the parser flags, a bitwise disjunction of any of the following constants, or 0 for the default:
The return value is the document pointer. These operations may also fail if there is a fatal error parsing the document.
Save an XML document doc to a file or a string. When saving to a file, encoding denotes the desired encoding (or "" for the default), compression the desired level of zlib compression (0 means none, 9 is maximum, -1 indicates the default). Note that with xml::save_string, the result is always encoded as UTF-8.
Retrieve general information about a document. Returns a tuple (version,encoding,url,compression,standalone), where version is the XML version of the document, encoding the external encoding (if any), url the name/location of the document (if any), compression the level of zlib compression, and standalone is a flag indicating whether the document contains any external markup declarations “which affect the information passed from the XML processor to the application”, see the section on the standalone document declaration in the XML spec for details. (Apparently, in libxml2 standalone is either a truth value or one of the special values -1, indicating that there’s no XML declaration in the prolog, or -2, indicating that there’s an XML declaration but no standalone attribute.)
Retrieve the internal and external DTD subset of a document. Returns a doctype node (fails if there’s no corresponding DTD).
Example
Read the sample.xml document distributed with the sources (ignoring blank nodes) and retrieve the document info:
> using xml;
> let sample = xml::load_file "sample.xml" xml::NOBLANKS;
> xml::doc_info sample;
"1.0","","sample.xml",0,-2
These operations are used to traverse the document tree, i.e., the nodes of the document. They take either a document pointer doc or a node pointer node as argument, and yield a corresponding node pointer (or a document pointer, in the case of xml::doc). The node pointers can then be used with the Node Information and Node Manipulation operations described below.
the root node of doc
the document node belongs to
the parent of node
All these operations fail if the corresponding target node does not exist, or if the type of the given node is not supported by this implementation.
There are also two convenience functions to retrieve the children and attribute nodes of a node:
returns the list of all child nodes of node
returns the list of all attribute nodes of node
Moreover, given a node pointer node, node!i can be used to retrieve the ith child of node.
Example
Peek at the root node of the sample document and its children:
> let r = xml::root sample; r;
#<pointer 0xe15e10>
> xml::node_info r;
xml::element "story" [] []
> #xml::children r;
5
> xml::node_info (r!0);
xml::cdata "<greeting>Hello, world!</greeting>"
These operations retrieve information about the nodes of an XML document.
Retrieve nodes using an XPath specification. Given an XPath (a string) xpath, this operation returns the list of all matching nodes in the given document doc. You can also specify a node as the first argument, in which case the document of the given node is searched and paths are interpreted relative to the given node (rather than the root node of the document).
Moreover, instead of just an XPath you can also specify a pair (xpath,ns) consisting of an XPath xpath and a list ns of prefix=>uri string pairs which describe the namespaces to be recognized in the XPath expression. This is necessary to select nodes by qualified tag or attribute names. Note that only the namespace URIs must match up with those used in the queried document; the corresponding namespace prefixes can be chosen freely, so you can use whatever prefixes are convenient to formulate the XPath query. However, for each namespace prefix used in the XPath expression (not the document!), there must be a corresponding binding in the ns list. Otherwise the underlying libxml2 function will complain about an undefined namespace prefix and xml::select will fail.
Retrieve the node data from node. Returns a node info value, as described in Data Structure above. Fails if the node does not belong to one of the supported node types.
Checks whether a node is a blank node (empty or whitespace only) and thus possibly ignorable.
Returns the base URI of the given node.
Returns the path of a node in the document, in the format accepted by select.
Returns the text carried by the node, if any (after entity substitution).
In addition, you can retrieve and change attributes of element nodes with the following operations:
Retrieves the value of the attribute with the given name (after entity substitution).
Sets or unsets an attribute value.
Examples
Set and unset a node attribute:
> xml::set_node_attr r "foo" "4711";
()
> xml::node_info r;
xml::element "story" [] ["foo"=>"4711"]
> xml::node_attr r "foo";
"4711"
> xml::unset_node_attr r "foo";
()
> xml::node_info r;
xml::element "story" [] []
The select function is very powerful, and probably the single most important operation of this module if you want to extract information from an existing XML document without traversing the entire structure. Here is a very simple example of its use:
> [xml::node_content n, xml::node_path n | n = xml::select sample "//author"];
[("John Fleck","/story/storyinfo/author")]
Note that if the XPath expression contains qualified names, the corresponding namespace prefixes and their URIs must be given in the second argument along with the XPath, as follows:
xml::select doc ("//foo:bar", ["foo"=>"http://www.foo.org"]);
These operations enable you to manipulate the document structure by adding a new node to the document tree (specified through its node info), and by removing (unlinking) existing nodes from the tree.
Add the new node specified by info in place of the given node node.
Add the new node as the first or last child of node, respectively.
Add the new node as the next or previous sibling of node, respectively.
The operations above all return a pointer to the new XML node object.
Deletes an existing node from the document tree. Returns ().
Examples
Replace the first child of the root node in the sample document:
> xml::node_info (r!0);
xml::cdata "<greeting>Hello, world!</greeting>"
> xml::replace (r!0) (xml::text "bla bla");
#<pointer 0xd40d80>
> xml::node_info (r!0);
xml::text "bla bla"
Delete that node:
> xml::unlink (r!0);
()
> xml::node_info (r!0);
xml::comment "This is a sample document for testing the xml interface."
The following operations provide basic XSLT support. As already mentioned, stylesheets are represented as pointers to the xsltStylesheet structure provided by libxslt. Note that, in difference to XML document pointers, this is an opaque type, i.e., there is no direct means to inspect and manipulate parsed stylesheets in memory using the operations of this module. However, a stylesheet is just a special kind of XML document and thus can be manipulated after reading the stylesheet as an ordinary XML document. The load_stylesheet function then allows you to convert the document pointer to an XSLT Stylesheet object.
Applying a stylesheet to an XML document generally involves the following steps:
Here is a brief summary of the XSLT operations. Please check the XSLT documentation for details of the transformation process.
Load a stylesheet. x can be either an XML document pointer, or a string denoting the desired .xsl file.
Apply the stylesheet style to the given document doc with the given parameters params. The third argument is a (possibly empty) list of key=>value string pairs which allows you to pass additional parameters to the stylesheet.
Save the transformation result doc obtained with the stylesheet style to a file or a string. This works pretty much like save_file or save_string, but also keeps track of some output-related information contained in the stylesheet.
Example
Load the recipes.xml document and the recipes.xsl stylesheet distributed with the sources:
> let recipes = xml::load_file "recipes.xml" xml::DTDVALID;
> let style = xslt::load_stylesheet "recipes.xsl";
Apply the stylesheet to the document and save the result in a html file:
> let res = xslt::apply_stylesheet style recipes [];
> xslt::save_result_file "recipes.html" res style 0;
()
That’s all. You can now have a look at recipes.html in your favourite web browser.