OpenOffice::OODoc::File(3) I/O operations with OpenDocument files


The explicit use of this module is generally required only in programs which need to do some raw data import or export operations in OpenDocument files. This module can be used as a special wrapper of Archve::Zip.

However, a look at the constructor and the save() methods is recommended, in order to get a better knowledge of the file interface which is used by the document-oriented modules (i.e. OODoc::XPath and its derivatives).

If the program is only concerned with a single XML element from the file, it is unnecessary to create an OODoc::File explicitly. Just build an OODoc::XPath object with a filename as parameter. The XPath will create a ``private'' File itself and use it to access the data invisibly.

Please note that OODoc::File is able to handle all standard zip archives and not just files. Filenames do not have to have a ``.sx?'' extension. It allows access by other modules, if required, to XML/OpenOffice data in compressed archives which are not necessarily in format.


Constructor : OpenOffice::OODoc::File->new(filename|handle)

        Short Forms (equivalent):
                odfFile(), odfContainer(), or odfPackage()
        Returns an instance of OODoc::File if the argument corresponds to a
        valid and accessible zip archive. The argument may be either a
        file path/name or an open (in binmode) and seekable IO::File.
        The file or file handle used here is used in order to initialize the
        ODF Connector for subsequent processing through the other submodules
        of the OpenOffice::OODoc API. It becomes the default target file if
        the data are updated and the changes committed using save().
        If the OODoc::File object is loaded for update, it's strongly
        recommended to avoid the use of a single IO::File object to read and
        write. However, if the source is a file name, the same file name may
        be safely used as the source and the target (of course and as usual,
        any sensitive document must be backed up before processing).
        If the "create" option is set (see below), a file path/name may be
        provided, but a file handle must be avoided.
        See the explanations about the save() method below.
            my $container1 = odfContainer("doc1.odt");
            my $fh = new IO::File "< doc2.odt";
            binmode $fh;           
            my $container2 = odfContainer($fh);
        An optional argument can be passed in hash format (key => value),
        after the filename. Like this:
                work_dir        => "path"
        which designates the path to the XML working files also generated
        during a save for this object (each OpenOffice::OODoc::File object
        can have its own working directory); without this option, the
        working directory is set according to the content of the class
        variable $OpenOffice::OODoc::File::WORKING_DIRECTORY.
        Note: no content checking is carried out. The archive can be opened
        whether or not it is an document.
        It's possible to create an OpenOffice::OODoc::File object without
        providing an existing file. To do so, there is a
        special option:
                create          => "class"
        where "class" is the document class according to the
        terminology, so it is one of the following values: "text",
        spreadsheet", "presentation", "drawing". These values are the same
        as the legal parameters of contentClass() in OpenOffice::OODoc::XPath.
        If "create" is provided, the first argument must not be a file handle,
        knowing that the target (if the data is later saved) can't be an
        open file, and that there is by definition no source file. Note that,
        for a creation, the value of the file argument is ignored so it may be
        undef or an empty string; however a valid path/name may be provided in
        order to provide a default target for a subsequent save().
        For a very advanced use, it's possible to combine "create" with an
        additional option
                template_path   => "path"
        to generate the new file from special, user-provided ODF templates
        instead of those included in the installation. If this option is
        not provided, the general template path (possibly changed with
        the templatePath() function) is used. The template path must
        indicate a directory which contains a set of ODF files whose
        names are "template.???", where "???" represents the conventional
        ODF and OOo suffixes, knowing that the supported suffixes are
        presently odt, ods, odp, odg, sxw, sxc, sxi, sxd. Of course, some
        applications don't need all the templates; for example, the
        "" files should be omitted if the user don't want to
        use the legacy 1.0 format for new documents.
        When the "create" option is used, it's possible to provide an
        "opendocument" option in order to override the installation-level
        default file format for new documents. If this option is set to
        "1", "on" or "true", the new document will comply to the OASIS
        OpenDocument format; if it's set to "0", "off" or "false", the new
        document will be created according to the 1.0 format.
        The "opendocument" option is ignored without the "create" one (this
        tool is not a format converter for existing documents). Remember that
        the default format is set to OpenDocument in the CPAN distribution
        and the 1.0 format is deprecated.
        The returned object of new(), if successful, is a valid File object,
        whose methods listed below become available.
        If unsuccessful (generally due to non-existent file or invalid zip
        archive or even a corrupt zip archive), the constructor returns a
        null value (undef), and an error message is sent to the standard


        Returns the decompressed content of the requested member, if
        contained in the archive and corresponds to an XML element of the
        currently active file instance. The <member>
        parameter must therefore correspond to one of the members of the
        file (see Introduction). If the application uses any of the words
        "content", "meta", "styles" or "settings", in upper or lower case,
        the .xml extension is automatically added but any other names are
        accepted without change if they are indeed existing members of the
        The following statements are equivalent:
            my $content = $archive->extract('META');
            my $content = $archive->extract('meta.xml');
            my $content = $archive->extract('meta);
        After the above calls, the variable $content contains the XML
        document which represents the metadata of an file.
        This content can be used, for example, to instance a Meta object.
        Note: in most "normal" cases, this method does not have to be called
        explicitly as it is called silently by each occurrence of XPath
        (therefore by Text and Meta which are derivatives of it), but only
        if XPath is constructed referencing an OODoc::File object as a
        parameter (see OODoc::XPath). An extract call is only useful when
        exporting the XML or handling it outside of OODoc::XPath.
        On error (e.g. unknown archive member), a null value is returned and
        an error message is produced.


        Connects a File object to an XPath object given as an argument. This
        connection has two output products:
            - immediately calls the extract method using the corresponding
            "specialist" component of the XPath object (metadata if
            OODoc::Meta, content if OODoc::Text, etc).
            - stores the link for later updates to all file
            members which may have been modified by XPath objects (in case a
            save is called, see below).
        Note: This method is used by OODoc::XPath to connect as "clients" to
        OODoc::File objects. It does not have to be called directly by
        highest-level programs which only use OODoc::XPath objects.


        Orders the deletion of any file member.
        deletes the physical content of an image loaded in the file.
        It is entirely up to the application to ensure that such a deletion
        does not compromise the integrity of the file as no dependency
        checking is carried out here. In the above example, the delete
        operation could be particularly justified if the "image" member
        which referenced this content had been (or was going to be)
        otherwise removed, or if it had been replaced by an external
        This method can be used to remove any XML or non-XML member. It can
        be combined with raw_import() in order to effect a raw replacement
        of content without interpretation. Caution: this method should not
        be used for an XML member (content, style, meta, etc.) which is
        currently "active" (i.e. linked to an active OODoc::XPath instance),
        unless the member has been loaded as "read only" (search in the
        OpenOffice::OODoc::XPath for the "read_only" option).
        Note: calls to this method only prepare the deletion, which is
        actually carried out by the save() method if it occurs before the
        end of the program. If save() is called with a filename which is
        different from the source filename, the source file remains
        unchanged and the deleted member is simply not transferred to the
        target file.

raw_export(member [, destination])

        Decompresses and exports the physical content of a given member (XML
        or non-XML) of an archive. If the second argument is used, it passes
        the destination filename (perhaps with access path). If not, the
        file is exported using its internal archive name. Examples:
        exports the "styles.xml" member into a file of the same name in the
        current directory.
            $archive->raw_export("styles.xml", "/tmp/my_style.xml");
        exports the same XML member to a given path.
        raw_export executes immediately (and is not deferred like
        If successful, the returned value is the filename of the exported

raw_import(member [, source])

        Creates or replaces the indicated member by importing an external
        source file. If the second argument is omitted, the source file is
        taken to have the same access path as the internal member.
            $arch1->raw_export("styles.xml", "/tmp/styles.xml");
            $arch2->raw_import("styles.xml", "/tmp/styles.xml");
        or, in more compact form:
        The above sequence requests the import of the member "styles.xml"
        from an archive called $arch1 into $arch2 (a direct means of using
        the styles and page layout of one document as a template for
        The imported files can be any type and have any content. This "raw"
        method treats an file as any other zip archive. It
        notably allows the import of non-XML members (images, sounds,
        programs, etc) which the application deals with (and which can be
        ignored by the office application).
        Caution: the import is only completed when a save() method is called
        by the importing object. It can only succeed if the source file is
        available at that very moment. A raw_import method can be called
        before the imported file is available (no check of availability is
        made). An error will be caused if the file is absent at the time of
        the save() call. If several raw_import statements are run against the
        same filename, there will actually be a corresponding number of
        copies of the file in its final state which are imported at the
        moment of the save() call, even if it had perhaps been modified in the
        meantime (probably not a very useful outcome).


        Commits all the changes made in the members of the archive and makes
        the resulting content persistent.
        In addition, the external, non-XML resources (i.e. image files) that
        have previously been targeted by import methods, if any, are
        physically imported when save() is called (and not before). If these
        resources are not available at this time, they are ignored and a
        warning is issued for each missing file.
        The target may be a file name (with optional path) or an IO::File
        object (that must be open, in binmode and writable).
        Without argument, save(), attempts to write the on the file or handle
        that was specified as the source file at the creation time, if any.
        However, if the OODoc::File was initialized with a IO::File, the use of
        the same IO::File for writing the output is presently not recommended;
        in some situations, bi-directional I/Os against an application-provided
        open IO::File may result in unpredictable results with the current
        implementation. Of course, if the OODoc::File object has been
        initialized with the "create" option and with an undef value or an
        empty string as file name, save() requires an explicit target (a valid
        file/path or an open and writable IO::File).
        Please note that OODoc::File does not check the content, and the save()
        method can be used to force through any data which may produce a
        file not compliant with the ODF packaging specification.
        The filename argument is optional. If it is omitted, the source file
        previously supplied by the constructor call (if any) is updated.
        Without a filename argument, save() if the source document was got
        through a IO::Handle.
        Even though the life of an OODoc::File object does not necessarily
        end with a save, it is recommended that you avoid repeated
        alternation between save and extract (the object's behaviour in this
        situation has not been tested). Normally it is preferable to call a
        save once and for all at the end of a series of updates.
        Only a call to OODoc::File's save() method saves content, metadate and
        presentation changes made by other OODoc components to the file, including raw imports of external data
        (raw_import). However, the XML members currently associated with
        "read only" OODoc::XPath objects are not changed in the file.
        No file is created or modified before this method is
        called, with the exception of external files created by raw_export.
        Nevertheless File's save can be called automatically and silently by
        an OODoc::XPath object but only where it has been called as a
        parameter explicitly for this purpose (see the chapter on
        All XPath objects which are "connected" to a File object by link
        must be present at the time of the save call. If one of these
        objects has meanwhile been deleted, the consequences are
        unpredictable and, in any case, any document updates it could have
        made are lost.


        Class function (not to be used as a method).
        Can be replaced by odfTemplatePath(), see OODoc man page.
        Accessor to get/set the path for a user-defined set of ODF templates,
        to be used in case of new document creation. This path is empty by
        default. Without an explicit template path, the default XML templates
        provided with the OpenOffice::OODoc distribution are automatically
        The template path must designate a directory containing 4 regular
        ODF files, each one corresponding to a supported ODF document class,
        i.e. "template.odt", "template.ods", "template.odp", "template.odg".
        However, the user don't need to provide templates that will not be
        Note that it's always possible to override the default template path
        when a new document is created, thanks to the 'template_path' option.


        No class variables are exported.
        The class variable $OpenOffice::OODoc::File::WORKING_DIRECTORY
        indicates the directory to be used for temporary files (used but the
        save() method) when no object-specific path is provided through the
        'work_dir' option. By default, the working directory is the current
        directory ('.').
        The $OpenOffice::OODoc::File::TEMPLATE_PATH variable, empty by
        default, can contain an alternative path for document generation
        template files; it can be set with the templatePath() function.
        The $OpenOffice::OODoc::File::$DEFAULT_OFFICE_FORMAT variable,
        whose default is 2, controls the default format for newly created
        files (when the format is not explicitly selected by the application).
        Allowed values are "1" for 1.0 and "2" for
        OASIS OpenDocument. In a regular installation, this variable is
        automatically set according to the <File-DEFAULT_OFFICE_FORMAT>
        element of the config.xml file (see INSTALL).
        Instance hash variables are:
            'linked'            => list of connected OODoc::XPath instances
            'members'           => list of file member (*.xml and others)
            'raw_members'       => list of import files
            'temporary_files'   => created temporary files.
        Where $f is a given instance of OODoc::File, the table
        is a list of OODoc::XPath objects which were connected to $f by the
        link method and
        is a list of members found in the archive when new is called.
        These variables can be read at any time even though they were
        normally designed to be used internally by OODoc::File. Unless you
        are just finding out exactly what they do, it is dangerous to modify
        them. Applications do not normally need to access them.


Developer/Maintainer: Jean-Marie Gouarne <>

Contact: [email protected]

Copyright 2004-2008 by Genicorp, S.A. <>

Initial English version of the reference manual by Graeme A. Hunter ([email protected]).

License: GNU Lesser General Public License v2.1