=head1	NAME

ODF::lpOD::Element - Common features available with any ODF element


This manual page describes the C<odf_element> class.

Note that C<odf_element> is the super-class of various specialized objects,
so the present manual page introduces only the I<basic> methods that apply
to I<every> C<odf_element>. The C<odf_element> class is not always explicitly
used in a typical application; it's generally used through more specialized,
user-oriented specialized objects.

C<odf_element> is an alias for C<ODF::lpOD::Element> package.

Every XML element (loaded from an existing ODF document or created by any
lpOD-based application) is a C<odf_element>. This class is the base class
for any document element.

An element may be explicitly created using the C<odf_create_element> class
constructor (or a the constructor of any derivative of C<odf_element> (such
as C<odf_paragraph>, C<odf_table>, etc), then inserted somewhere is an document.
However, in most cases, elements are either retrieved in the existing structure
or implicitly created ad put in place as children of existing elements through
various element-based C<set_xxx> methods (where "xxx" depends on the kind of
newly created data).

Among the C<odf_element> methods, we distinguish I<element> methods from
I<context> methods, while both kinds of methods belong to C<odf_element>
objects. An element method gets or sets one or more properties of the calling
element, while a context method uses the calling element as its operating
context and may produce effects regarding other elements located somewhere in
the hierarchy of the calling element (generally below, and sometimes above).
As examples, C<set_attribute> is an element method (it changes an attribute of
the current element), while C<get_element> (in its element-based version, that
is not the same as its part-based one) retrieves an element somewhere below
the current one.

=head1  Constructor and retrieval tools

=head3  odf_create_element(data)

Creates an odf_element from an argument which may be an arbitrary tag, a
well-formed XML string, a file handle, or a web link. Beware that it's a
generic, low level constructor, allowing the user to create arbitrary
elements. Its explicit use should be exceptional.

The following instruction creates a regular XML element whose tag is "foo":

        $e = odf_create_element('foo');

If the given argument is a string whose first character is "<", it's parsed
as an XML element definition. As an example, the instruction below creates
a "foo" element with a "bar" attribute whose value is "baz" while the text
of the element is "xyz":

        $e = odf_create_element('<foo bar="baz">xyz</foo>');

If the given string contains a colon (such as "http://xy.z"), then it's
regarded as the URL of an XML resource available through the web. If the
resource is available, then it's parsed as XML.

If the given string doesn't start with C<'http:'> but ends with C<'.xml'>, then
it's regarded as the path to an XML file in the local filesystem, and the
content of this file is parsed as the XML definition of the new element.

If the argument is a reference (and not a string), it's regarded as a text file
handle; if it's really a file handle, the content of this file is loaded and
parsed as XML.

The use of file handles or HTTP links allows the applications to easily import
ODF element definitions from remote locations and/or to reuse element
definitions stored in application-specific XML files or databases.

Note that the provided XML content must be UTF8-encoded as soon as it contains
one or more characters that don't belong to the 7-bit ASCII character set.

The new element is not attached to a document; it's free for later use.

C<odf_create_element()> is an alias for one of the following instructions,
which are equivalent:


The same principle applies for every subclass of C<ODF::lpOD::Element>; as
a consequence, whatever C<"xxx"> in the name of a particular lpOD element
constructor, the two following instructions are equivalent:

        $e = odf_create_xxx(@args);
        $e = odf_xxx->create(@args);

=head3  get_element(tag [options])

This method returns the first element (if any) matching the given XML tag.
It's the most generic context-based retrieval method.

The given tag may be replaced by a regular expression, so the search space will
include all the elements whose tags match the expression.

For example, the following instruction, assuming C<$context> is a previously
retrieved element, returns the first element in C<$context> that is either a
paragraph or a heading (knowing that the corresponding tags are C<text:p> and

        my $text_element = $context->get_element(qr'text:(p|h)');

A lot of more specialized and more user-friendly element retrieval methods are
provided by lpOD, so the direct use of C<get_element()> (that implies some XML
awareness) may and should be avoided in most situations. This method is just
intended to allow arbitrary search operations that are not explicitly covered
by lpOD.

The allowed options are:



C<position>: The sequential zero-based position of the element among the
set of elements matching the given tag; negative positions are counted
backward from the end.


C<attribute>: The name of an attribute used as a selection criterion; if this
option is set, the C<value> option is required.


C<value>: the value of the selection attribute.


C<content>: a search string (or a regexp) restricting the search space to the
elements with matching content.


The example below (that combines all the options) returns the 4th level 1
heading before the end of the current context:

                attribute       => 'outline level',
                value           => 1,
                position        => -4

I<Caution: the C<get_element> method of C<odf_part> is not the same as the
C<get_element> method of C<odf_element>.>

Note that such an instruction should not appear in a real application, knowing
that C<get_paragraph()>, introduced in L<ODF::lpOD::TextElement>, does the job
in a more user-friendly way.

=head3  get_elements(tag)

Returns the full list of elements matching the given tag, whose tags match
the given regexp.

The C<attribute> and C<value> options are allowed in order to restrict the

The next example returns the list of paragraphs whose style is "Standard":

        my @std_paragraphs = $context->get_elements(
                attribute       => 'style name',
                value           => 'Standard'

=head1  Context import

The C<import_children()> method allows the user to directly append elements
coming from an external context. As an example, the following instruction
appends appends all the children of a given C<$source> element in the context
of the calling C<$destination> element:


In this example, both C<$destination> and C<$source> are elements. The source
context may belong to the same document as the destination context, or not, so
this method may be used for content replication across documents. Knowing that
the imported elements are clones of the original ones, the source context is
unchanged. The imported elements are appended at the end of the calling context
in the same order as in the source context.

An optional filter may be specified as the 2nd argument, in order to import a
particular kind of the source context only. So, the instruction below imports
only the I<paragraphs> of the source context (remember that C<'text:p'> is the
ODF tag for paragraphs):

        $destination->import_children($source, 'text:p');

The C<substitute_children()> method, that takes the same arguments as
C<import_children()>, removes all the children of the calling element (if any),
then imports the children of a source element. The following sequence replaces
all the content of a given document body by the content of another document
body (see L<ODF::lpOD::Document> for details about C<get_body()>):


=head1  Top level contexts

As introduced in L<ODF::lpOD::Document>, the C<odf_part> handlers provide
methods that automatically return high level elements that may be the preferred
contexts in most cases. The most common one is the I<root> element; its context
is the whole document part. The C<body> element, that is sometimes the same as
the C<root> one, is a bit more restricted in the document C<content> part (it
includes only the content objects, and excludes other objects such as style
definitions). Both the C<root> and the C<body> may be got using the part-based
C<get_root> and C<get_body> methods.

The following sequence, starting from the creation of a document instance,
selects a part, then the root element of the part, than selects the list of
table styles defined in the part:

        my $doc = odf_document->get("/home/jmg/report.odt");
        my $content = $doc->get_part(CONTENT);
        my $context = $content->get_root;
        my @table_styles = $context->get_element_list(
                attribute       => 'family',
                value           => 'table'

Note that the sequence above is shown in order to illustrate a principle but
that it should not be needed in a real application, knowing that lpOD provides
more user-friendly style retrieval tools.

=head1  Child element creation methods

The methods described in this section allows the user to insert elements
(previously existing or not) as children of the calling element.

=head3  insert_element(element [options])

Insert the given odf_element at a given location, that is defined according
to a C<position> parameter. This parameter specifies the position of the element
to be inserted (provided as the 1st argument) relatively to the calling element.
The possible values are:



C<FIRST_CHILD>: the odf_element will be the first child (default).


C<LAST_CHILD>: the odf_element will be the last child.


C<NEXT_SIBLING>: the odf_element will be inserted just after.


C<PREV_SIBLING>: the odf_element will be inserted just before.


C<WITHIN>: the odf_element will be inserted as a child within the text content;
if C<position> is C<WITHIN>, then the C<offset> parameter is required.


C<offset>: specifies the position in the text of the context element where the
new child element must be inserted (the position is zero-based).


C<before>: the value of this option, if set, must be another child
C<odf_element> of the calling one; the new element will be inserted as the
previous sibling of this child element.


C<after>: like C<before>, but the new element will be inserted I<after> the
value of this option.


C<PARENT>: if this option is set, the calling element is moved and becomes the
I<last child> of the element given as C<PARENT>. So the two instructions below
are equivalent:

        $e2->insert_element($e1, position => PARENT);
        $e1->insert_element($e2, position => LAST_CHILD);

Note that, with the C<PARENT> option, if the first argument is an element that
already belongs to a tree, it's not really inserted: the calling element is
just inserted inside it.


The WITHIN option splits the text content of the container in two parts
and inserts the elements between them, at a given offset. So if position is
WITHIN, the offset optional parameter is used.
By default, if no offset argument is provided, or if the calling element
doesn't contain any text, WITHIN produces the same result as FIRST_CHILD.
The offset argument must be an integer; it specifies the position of the
inserted child element within the text content of the calling element.
A zero offset means that the element must be inserted before the 1st
character. A negative offset value means that the insert position must be
counted down from the end of the text, knowing that -1 is the position just
before the last character. Of course, if the insertion must be done after
the end of the text, the simplest way is to select LAST_CHILD instead of

If C<before> or C<after> is provided, the other options are ignored. Of course,
C<before> and C<after> are mutually exclusive.

The following example inserts a previously existing element between the 4th and
the 5th characters of the text of the calling element:

                position        => WITHIN,
                offset          => 4

The next example inserts a new empty paragraph before the last paragraph of the
calling context:

        my $last_p = $context->get_paragraph(position => -1);
        my $new_p = odf_paragraph->create();
                $new_p, before => $last_p

The sequence below (that uses features described in L<ODF::lpOD::TextElement>),
creates a protected section whose name is "Conclusion" and inserts it as the
container of the last paragraph of the document:

        $paragraph = $context->get_paragraph(position => -1);
                odf_section->create("Conclusion", protected => TRUE),
                position => PARENT

=head3  insert_element(tag)

Like the first version of C<insert_element>, but the argument is an arbitrary,
application defined XML tag (i.e. technically a text string instead of a
C<odf_element> instance); in such a case a new element is created then inserted
according to the same rules and options, with the exception of the C<PARENT>
option, that is supported only if the first argument is an existing element
and not the tag of a element to be created.

As an example, the next instruction creates and inserts a 'foo:bar' XML element
at the end of the content of the calling element:

        $context->insert_element('foo:bar', position => LAST_CHILD);

=head3  append_element(element/tag)

Like C<insert_element>, but without options; appends the element as the I<last
child> of the calling element. So these tow lines are equivalent:

        $context->insert_element($elt, position => LAST_CHILD);

=head1  Element methods

The methods introduced in this section are accessors that get or set the
own properties of the calling element. However, in some cases they may have
indirect consequence on other elements.

Note that C<odf_element> is a subclass of C<XML::Twig::Elt>. So, beyond the
methods specifically described below, programmers familiar with L<XML::Twig>
are allowed to directly (but cautiously) call any C<XML::Twig::Elt> method
from any element, if the methods described below don't meet all their needs.

=head3  append

Appends the calling element to another element provided as argument. The calling
element becomes the last child of the given one.

Note that the two following instructions produce the same result:


=head3  clear

Erases the text of an element and all its children. Beware that this method
is overridden by some specialized element classes.

=head3  clone

Returns a copy of the calling element, with all its attributes, its text, and
its children elements. Allows the user to copy a high-level structured element
(like a section or a table) as well as a single paragraph. The copy is a free
element, that may be inserted somewhere in the same document as the prototype,
or in another document.

=head3  delete

Removes the calling element with all its descendants.

=head3  del_attribute(name)

Deletes the attribute whose name is given in argument. Nothing is done if
the attribute doesn't exist. The argument may be the exact XML name of the
attribute, or an "approximative" name according to the same logic as
C<get_attribute> below.

=head3  export

Same as C<serialize()> introduced later.

=head3  get_attribute(name)

Returns the string value of the attribute having this name. The argument may
be the exact XML name of the attribute. However, if a name without name space
prefix is provided, the prefix is automatically supposed to be the same as the
prefix of the context element. In addition, any white space or underscore
character in the given name is interpreted as a "-". As a consequence, some
attributes may be designated without care of the exact XML syntax. As an
example, assuming C<$p> is a paragraph, the two instructions below are
equivalent, knowing that the name space prefix of a paragraph is C<'text'>:

        $style = $p->get_attribute('text:style-name');
        $style = $p->get_attribute('style name');

The attribute values are returned in a character set that depends on the
global configuration. See L<ODF::lpOD::Common> for details about the character
set handling.

=head3  get_attributes

Returns all the attributes of the calling element as a hash ref where keys are
the full XML names of the attributes.

=head3  get_document

Returns the C<odf_document> object that owns the element, or C<undef> if the
calling element doesn't belong to a document.

=head3  get_document_part

If the element belongs to a document, returns the C<odf_part> where it's
registered, or returns C<undef> if the element is not in a document.

=head3  get_document_type

Returns the type of the C<odf_document> object that owns the element, or
C<undef> if the calling element doesn't belong to a document.

=head3  get_parent

Returns the I<parent> of the calling element, if it's included in a hierarchy.
Returns C<undef> if the calling element is free or if it's at the top level of
a hierarchy. May be used to check if the element is free or not. This method
is an alias for C<XML::Twig::Elt::parent> (see L<XML::Twig> for details).

=head3  get_root

Returns the top level element of the document part that contains the calling

=head3  get_style

Returns the name of the style used by the calling element (this accessor makes
sense for objects that may be displayed according to a layout). Returns
C<undef> if no style is used.

Note: if your style names contain non-ASCII characters and if your preferred
output character set is not C<utf8>, see L<ODF::lpOD::Common> for details
about character sets handling.

=head3  get_tag

Returns the XML tag of the element with its name space prefix.

=head3  get_text(recursive => FALSE)

Returns the text contents of the element as a string. By default this method is
not recursive (i.e. it just returns the own text of the element, not the text
belonging to children and descendant elements). However, if the optional
C<recursive> parameter is provided and set to C<TRUE>, then the method returns
the concatenated contents of all the descendants of the given element.

In a default configuration, the character set of the output is C<utf8>. If
that is not convenient for you, see the character set handling section in

=head3  get_url

Returns the URL if the element owns a hyperlink property, or C<undef> otherwise.

=head3  insert

Insert the calling element at a given location, depending on various options.
The first argument must be any C<odf_element>, that will be a position
reference. The other arguments may be the same as C<insert_element>.
See C<insert_element> for details, knowing that the two following instructions
are equivalent:

        $e1->insert($e2 [, options]);
        $e2->insert_element($e1 [, options]);

=head3  serialize

Returns an XML export of the calling element, allowing the lpOD applications to
store and/or transmit particular pieces of documents, and not only full
documents. The C<pretty> or C<indent> option is allowed, like with the
C<serialize> method of C<odf_part> objects, described in

Note that this XML export is not affected by the content encoding/decoding
mechanism that works for user content, so it's character doesn't depend on
the custom text output character set possibly selected through the
C<set_output_charset()> method introduced in L<ODF::lpOD::Common>.

=head3  set_attribute(attribute => value)

Sets or changes an attribute of the calling element. The attribute is created
if it didn't exist. If the provided value is C<undef>, the attribute is
deleted (if it didn't exist, nothing is done). The attribute name may be
specified according to the same rules as with C<get_attribute>.

About the character set of the input values, the same rules as with any text
input apply; see the character set handling section in L<ODF::lpOD::Common>.

=head3  set_attributes(attr_hash_ref)

Sets several attributes at a time. The attributes to change or create must be
passed as a hash ref (like the hash ref returned by C<get_attributes>). The
attribute names may be provided in simplified form like with C<set_attribute>.

=head3  set_comment(text)

Intended for debugging purposes, this method puts a XML comment I<before> the
calling element. This comment produces a "<!--xyz-->" tag, where "xyz" is the
given text, in the XML output if the document is later serialized. Beware that
such comments are not always preserved if the document is changed by an office
application software.

=head3  set_child(tag, text, attributes)

Synonym of C<set_first_child()> (see below).

=head3  set_first_child(tag, text, attributes)

Makes sure that the calling element contains at least one element with the
given XML tag. If there is no compliant child, a new element is created with
the given tag and inserted as the first child of the calling element. If one
or more compliant child exist, the first one is selected and its text content
(if any) is deleted.

The second argument (optional) is a string that becomes the new text content
of the created or selected child. The remainder of the argument list, if any,
is a hash specifying attribute/value pairs for this element.

The return value is the selected or created element.

=head3  set_last_child(tag, text, attributes)

Same as C<set_first_child()> but, in case of creation, the new element is
inserted as the I<last> child. If compliant children already exist, the
result is the same as C<set_first_child()>.

=head3  set_parent(tag, text, attributes)

Makes sure that the current element is a child of an element whose tag is
specified by the first argument. If the calling element is free or if its
immediate parent has not the given tag, a new element with the given tag is
inserted as the same place of the calling element, that becomes the first
child of the new element.

The return value is the new element.

The other arguments are the same as with C<set_first_child()>.

=head3  set_style(style_name)

Changes or sets the style name of the calling object. Caution: a lot of ODF
elements should not have any style, so don't use this accessor unless you know
that the calling object needs a style.

Note: if your style names contain non-ASCII characters and if your preferred
input character set is not C<utf8>, see L<ODF::lpOD::Common> for details
about character sets handling.

=head3  set_tag(new_tag)

Changes the XML tag of the calling element. Not for usual business; it's a low
level technical feature.

=head3  set_text(text_string)

Sets the text content of the calling element, replacing any previous content.

        my $paragraph = $context->get_element('text:p', position => 15);
        $paragraph->set_text("The new content");

The character set of the provided string must comply to the currently active
input character set (default is C<utf8>). See the character set handling
section in L<ODF::lpOD::Common> if you get troubles about text encoding.

If C<set_text> is called with an empty string, its effect is the same as

=head3  set_url(url)

Sets a URL attribute (C<xlink:href>) with the argument.

=head1  Custom element classes

The C<ODF::lpOD::Element> package provides a C<associate_tag> class method
allowing developers to create custom subclasses and associate them to
particular elements. The code example below defines C<CustomParagraph> as a
subclass of C<odf_pragraph> (introduced in L<ODF::lpOD::TextElement>) and
specifies that every C<text:p> XML element must be mapped to this new
class instead of C<ODF::lpOD::Paragraph>:

        package CustomParagraph;
        use ODF::lpOD;
        use base 'ODF::lpOD::Paragraph';

        sub custom_method {


This extensibility mechanism must be used very cautiously if the specified tag
is already associated with a lpOD class, knowing that a wrongly overridden
method could produce destructive side effects.


Developer/Maintainer: Jean-Marie Gouarne L<http://jean.marie.gouarne.online.fr>
Contact: jmgdoc@cpan.org

Copyright (c) 2010 Ars Aperta, Itaapy, Pierlis, Talend.
Copyright (c) 2011 Jean-Marie Gouarne.

This work was sponsored by the Agence Nationale de la Recherche

License: GPL v3, Apache v2.0 (see LICENSE).