Plasma GitLab Archive
Projects Blog Knowledge


Editorial note: The PXP tree API has some complexity. The definition in Pxp_document is hard to read without some background information. This introduction is supposed to provide this information.

Also note that there is also a stream representation of XML. See Intro_events for more about this.

The structure of document trees

In a document parsed with the default parser settings every node represents either an element or a character data section. There are two classes implementing the two aspects of nodes: Pxp_document.element_impl, and Pxp_document.data_impl. There are configurations which allow more node types to be created, in particular processing instruction nodes, comment nodes, and super root nodes, but these are discussed later. Note that you can always add these extra node types yourself to the node tree no matter what the parser configuration specifies.

The following figure shows an example how a tree is constructed from element and data nodes. The circular areas represent element nodes whereas the ovals denote data nodes. Only elements may have subnodes; data nodes are always leaves of the tree. The subnodes of an element can be either element or data nodes; in both cases the O'Caml objects storing the nodes have the class type Pxp_document.node.

A tree with element nodes, data nodes, and attributes

Attributes (the clouds in the picture) do not appear as nodes of the tree, and one must use special access methods to get them.

You would get such a tree by parsing with

let config = Pxp_types.default_config let source = Pxp_types.from_string "An orangeCherries" let spec = Pxp_tree_parser.default_spec let doc = Pxp_tree_parser.parse_document_entity config source spec let root = doc#root

The config record sets a lot of parsing options. A number of these options are explained below. The source argument says from where the parsed text comes. For the mysterious spec parameter see below.

The parser returns doc, which is a Pxp_document.document. You have to call its root method to get the root of the tree. Note that there are other parsing functions that return directly nodes; these are intended for parsing XML fragments, however. For the usual closed XML documents use a function that returns a document.

The root is, as other nodes of the tree, an object instance of the Pxp_document.node class type.

What about other things that can occur in XML text? As mentioned, by default only elements and data nodes appear in the tree, but it is possible to enable more node types by setting appropriate Pxp_types.config options:

  • Comments are ignored by default. By setting the config option enable_comment_nodes, however, comments are added to the tree. There is a special node type for comments.
  • Processing instructions (denoted by <? ... ?> parentheses) are not ignored, but normally no nodes are created for them. The instructions are only gathered up, and attached to the surrounding node, so one can check for their presence but not for their exact location. By setting the config option enable_pinstr_nodes, however, processing instructions are added to the tree as normal nodes. There is also a special node type for them.
  • Usually, the topmost element is the root of the tree. There is, however, the difficulty that the XML syntax allows one to surround the topmost element by comments and processing instructions. For an exact representation of this, it is possible to put an artificial root node at the top of the tree, so that the topmost element is one of the children, and the other surrounding material appears as the other children. This mode is enabled by setting enable_super_root_node. The node is called super root node, and is also a special type of node.
  • It is possible to also get attributes and even namespaces as node objects, but they are never put into the regular tree. To get these very special nodes, one has to use special access methods.
  • CDATA sections (like <![CDATA[some text]]>) are simply added to the surrounding data node, so they do not appear as nodes of their own.
  • Entity references (like &amp;) are automatically resolved, and the resolution is added to the surrounding node
The parser collapses as much data material into one data node as possible such that there are normally never two adjacent data nodes. This invariant is enforced even if data material is included by entity references or CDATA sections, or if a data sequence is interrupted by comments. So

 a &amp; b <!-- comment --> c <![CDATA[<> d]]> 

is represented by only one data node, for instance (for the default case where no comment nodes are created). Of course, you can create document trees manually which break this invariant; it is only the way the parser forms the tree.

All types of nodes are represented by the same Ocaml objects of class type Pxp_document.node. The method Pxp_document.node.node_type returns a hint which type of node the object is. See the type Pxp_document.node_type for details how the mentioned node types are reflected by this method. For instance, for elements this method returns T_element n where n is the name of the element.

Note that this means formally that all access methods are implemented for all node types. For example, you can get the attributes of data nodes by calling the Pxp_document.node.attributes method, although this does not make sense. This problem is resolved on a case-by-case basis by either returning an "empty value" or by raising appropriate exceptions (e.g. Pxp_types.Method_not_applicable). For the chosen typing it is not possible to define slimmer class types that better fit the various node types.

Attributes are usually represented as pairs string * att_value of names and values. Here, Pxp_types.att_value is a conventional variant type. There are lots of access methods for attributes, see below. It is optionally possible to wrap the attributes as nodes (method Pxp_document.node.attributes_as_nodes), but even in this case the attributes are outside the regular document tree.

Normally, the processing instructions are also not included into the document tree. They are considered as an extra property of the element to which they are attached, and can be retrieved by the Pxp_document.node.pinstr method of the element node. If this way of handling processing instructions is not exact enough, the parser can optionally create processing instruction nodes that are regular members of the document tree.

Access methods

An overview over some relevant access methods:

Mutation methods

Trees are mutable, and nodes are mutable. Note that the tree is not automatically (re-)validated when it is changed. You have to explicitly call validation methods, or the Pxp_document.validate function for the whole tree.

Links between nodes

The node tree has links in both directions: Every node has a link to its parent (if any), and it has links to the subnodes (see the following picture). Obviously, this doubly-linked structure simplifies the navigation in the tree; but has also some consequences for the possible operations on trees.

Nodes are doubly linked trees

(Definitions: Pxp_document.node.parent, Pxp_document.node.sub_nodes.)

Because every node must have at most one parent node, operations are illegal if they violate this condition. The following figure shows on the left side that node y is added to x as new subnode which is allowed because y does not have a parent yet. The right side of the picture illustrates what would happen if y had a parent node; this is illegal because y would have two parents after the operation.

A node can only be added if it is a root

(Definition: Pxp_document.node.append_node.)

The remove operation simply removes the links between two nodes. In the following picture the node x is deleted from the list of subnodes of y. After that, x becomes the root of the subtree starting at this node.

A removed node becomes the root of the subtree

(Definition: Pxp_document.node.remove.)

It is also possible to make a clone of a subtree; illustrated in the next picture. In this case, the clone is a copy of the original subtree except that it is no longer a subnode. Because cloning never keeps the connection to the parent, the clones are called orphaned.

The clone of a subtree

(Definition: Pxp_document.node.orphaned_clone.)

Optional features of document trees

As already pointed out, the parser does only create element and data nodes by default. The configuration of the parser can be controlled by the Pxp_types.config record. There are a number of optional features that change the way the document trees are constructed by the parser:

Note that the parser configuration only controls the parser. If you create trees of your own, you can simply add all the additional node types to the tree without needing to enable these features.

  • When enable_super_root_node is set, the extra super root node is generated at the top of the tree. This node has type T_super_root.
  • The option enable_comment_nodes lets the parser add comment nodes when it parses comments. These nodes have type T_comment.
  • The option enable_pinstr_nodes changes the way processing instructions are added to the document. Instead of attaching such instructions to their containing elements as additional properties, this mode forces the parser to create real nodes of type T_pinstr for them.
  • The option drop_ignorable_whitespace (enabled by default) can be turned off. It controls whether the parser skips over so-called ignorable whitespace. The XML standard allows that elements contain whitespace characters even if they are declared in the DTD not to contain character data. Because of this, the parser considers such whitespace as ignorable detail of the XML instance, and drops the characters silently. You can change this by setting drop_ignorable_whitespace to false. In this case, every character of the XML instance will be accepted by the parser and will be added to a data node of the document tree.
  • By default, the parser creates elements with an annotation about the location in the XML source file. You can query this location by calling the method position. As this requires a lot of memory, it is possible to turn this off by setting store_element_positions to false.

There are a number of further configuration options; however, these options do not change the structure of the document tree.

Optional features of nodes

The following features exist per node, and are simply invoked by using the methods dealing with them.

  • Attribute nodes: These are useful if you want to have data structures that contain attributes together with other types of nodes. The method attributes_as_nodes returns the attributes wrapped into node objects. Note that these nodes are read-only.
  • Validation: The document nodes contain the routines validating the document body. Of course, the validation checks depend on what is stored in the DTD object. (There is always a DTD object - even in well-formedness mode, only that it is mostly empty then, and validation is a no-op.)

    The DTD object contains the declarations of elements, attribute lists, entities, and notations. Furthermore, the DTD knows whether the document is flagged as "standalone". As a PXP extension to classic XML processing, the DTD may specify a mixed mode between "validating mode" and "well-formedness mode". It is possible to allow non-declared elements in the document, but to check declared elements against their declaration at the same time. Moreover, there is a similar feature for attribute lists; you can allow non-declared attributes and check declared attributes. (Well, the whole truth is that the parser always works in this mix mode, and that the "validating mode" and the "well-formedness mode" are only the extremes of the mix mode.)

Creating nodes and trees

Often, the parser creates the trees, but on occasion it is useful to create trees manually. We explain here only the basic mechanism. There is a nice camlp4 syntax extension called pxp-pp (XXX: LINK) allowing for a much better notation in programs.

The most basic way of creating new nodes are the create_element, create_data, and create_other methods of nodes. It is not recommended to use them directly, however, as they are very primitive.

In the Pxp_document module there are a number of functions creating individual nodes (without children), the node constructors:

There are no functions to create attribute and namespace nodes - these are always created automatically by their containing nodes, so the user does not need to do anything for creating them.

The node constructors must be equipped with all required data to create the requested type of node. This includes the data that would have been available in the textual XML representation, and some of the meta data passed to the parsers, and meta data implicitly generated by the parsers. For an element, this is at minimum:

  • The name of the element (e.g. the "foo" in <foo>)
  • The attributes of the element
  • The DTD object to use (a rudimentary DTD object is even required if only well-formedness checks will be applied but no validation)
  • The specification which classes are instantiated to create the nodes
For the latter two, see below. Optionally one can provide:

  • The position of the element in the XML text
  • Whether the attribute list is to be validated at creation time
  • Whether name pools are to be used for the attributes
Regarding validation, the default is to validate local data such as attributes, but to omit any checks of the position the node has in the tree. The tree is still a singleton, and consists only of one node after creation, so non-local checks do not make sense.

After some nodes have been created, they can be combined to more complex trees by mutation methods (e.g. Pxp_document.node.append_node).

As mentioned, a node must always be connected with a DTD object, even if no validation checks will be done. It is possible to create DTD objects that do not impose restrictions on the document:

  let dtd = Pxp_dtd_parser.create_empty_dtd config
  dtd # allow_arbitrary

Even such a DTD object can contain entity definitions, and can demand a certain way of dealing with namespaces. Also, the character encoding of the nodes is taken from the DTD. See Pxp_dtd.dtd for DTD methods, and Pxp_dtd_parser for convenient ways to create DTD objects. Note that all nodes of a tree must be connected to the same DTD object.

PXP is not restricted to using built-in classes for nodes. When the parser is invoked and a tree is built, it is looked up in a so-called document model specification how the new objects have to be created (type Pxp_document.spec. Basically, it is a list of sample objects to use (which are called exemplars), and these objects are cloned when a node is actually created.

When calling the node constructors directly (bypassing the parser), the document model specification has also to be passed to them as argument. It is used in the same way as the parser uses it.

For getting the built-in classes without any modification, just use Pxp_tree_parser.default_spec. For the variant with enabled namespaces, prefer Pxp_tree_parser.default_namespace_spec.

Extended nodes

Every node in a tree has a so-called extension. By default, the extension is practically empty and only present for formal uniformity. However, one can also define custom extension classes, and effectively add new methods to the node classes.

Node extensions are explained in detail here: Intro_extensions


As an option, PXP processes namespace declarations in XML text. See this separate introduction for details: Intro_namespaces.

Details of the mapping from XML text to the tree representation

If an element declaration does not allow the element to contain character data, the following rules apply.

If the element must be empty, i.e. it is declared with the keyword EMPTY, the element instance must be effectively empty (it must not even contain whitespace characters). The parser guarantees that a declared EMPTY element never contains a data node, even if the data node represents the empty string.

If the element declaration only permits other elements to occur within that element but not character data, it is still possible to insert whitespace characters between the subelements. The parser ignores these characters, too, and does not create data nodes for them.

Example. Consider the following element types:

<!ELEMENT x ( #PCDATA | z )* >
<!ELEMENT y ( z )* >

Only x may contain character data, the keyword #PCDATA indicates this. The other types are character-free.

The XML term

<x><z/> <z/></x>

will be internally represented by an element node for x with three subnodes: the first z element, a data node containing the space character, and the second z element. In contrast to this, the term

<y><z/> <z/></y>

is represented by an element node for y with only two subnodes, the two z elements. There is no data node for the space character because spaces are ignored in the character-free element y.

Parser option: By setting the parser option drop_ignorable_whitespace to false, the behaviour of the parser is changed such that even ignorable whitespace characters are represented by data nodes.

The representation of character data

The XML specification allows all Unicode characters in XML texts. This parser can be configured such that UTF-8 is used to represent the characters internally; however, the default character encoding is ISO-8859-1. (Currently, no other encodings are possible for the internal string representation; the type Pxp_types.rep_encoding enumerates the possible encodings. Principally, the parser could use any encoding that is ASCII-compatible, but there are currently only lexical analyzers for UTF-8 and ISO-8859-1. It is currently impossible to use UTF-16 or UCS-4 as internal encodings (or other multibyte encodings which are not ASCII-compatible) unless major parts of the parser are rewritten - unlikely...)

The internal encoding may be different from the external encoding (specified in the XML declaration <?xml ... encoding="..."?>); in this case the strings are automatically converted to the internal encoding.

If the internal encoding is ISO-8859-1, it is possible that there are characters that cannot be represented. In this case, the parser ignores such characters and prints a warning (to the collect_warning object that must be passed when the parser is called).

The XML specification allows lines to be separated by single LF characters, by CR LF character sequences, or by single CR characters. Internally, these separators are always converted to single LF characters.

The parser guarantees that there are never two adjacent data nodes; if necessary, data material that would otherwise be represented by several nodes is collapsed into one node. Note that you can still create node trees with adjacent data nodes; however, the parser does not return such trees.

Note that CDATA sections are not represented specially; such sections are added to the current data material that is being collected for the next data node.

The representation of entities within documents

Entities are not represented within documents! If the parser finds an entity reference in the document content, the reference is immediately expanded, and the parser reads the expansion text instead of the reference.

The representation of attributes

As attribute values are composed of Unicode characters, too, the same problems with the character encoding arise as for character material. Attribute values are converted to the internal encoding, too; and if there are characters that cannot be represented, these are dropped, and a warning is printed.

Attribute values are normalized before they are returned by methods like attribute. First, any remaining entity references are expanded; if necessary, expansion is performed recursively. Second, newline characters (any of LF, CR LF, or CR characters) are converted to single space characters. Note that especially the latter action is prescribed by the XML standard (but &#10; is not converted such that it is still possible to include line feeds into attributes).

The representation of processing instructions

Processing instructions are parsed to some extent: The first word of the PI is called the target, and it is stored separated from the rest of the PI:

<?target rest?>

The exact location where a PI occurs is not represented (by default). The parser attaches the PI to the object that represents the embracing construct (an element, a DTD, or the whole document); that means you can find out which PIs occur in a certain element, in the DTD, or in the whole document, but you cannot lookup the exact position within the construct.

Parser option: If you require the exact location of PIs, it is possible to create regular nodes for them instead of attaching them to the surrounding node as property. This mode is controlled by the option enable_pinstr_nodes. The additional nodes have the node type T_pinstr target, and are created from special exemplars contained in the spec (see Pxp_document.spec).

The representation of comments

Normally, comments are not represented; they are dropped by default.

Parser option: However, if you require comment in the document tree, it is possible to create T_comment nodes for them. This mode can be specified by the option enable_comment_nodes. Comment nodes are created from special exemplars contained in the spec (see Pxp_document.spec). You can access the contents of comments through the method comment.

The attributes xml:lang and xml:space

These attributes are not supported specially; they are handled like any other attribute.

Note that the utility function Pxp_document.strip_whitespace respects xml:space

This web site is published by Informatikbüro Gerd Stolpmann
Powered by Caml