The pipeline therefore relies on a default, basic set of named paragraph and character styles which it uses to detect semantic elements in the source document. These names can be configured if desired by editing pipeline module properties. The set of supported style names will be discussed later.

The pipeline implementation supports two DocBook root elements: book and article. Since a book has more structuring layers before reaching the recursive nesting of sections, the pipeline's mapping of styles to the respective container elements may need to be adjusted with respect to your actual documents. The hierarchical mapping relies on an ordered set of named paragraph styles for the respective (required) title elements in the DocBook DTD. The pipeline infers the level of a title (and as a result the required boxing of container elements) from these named paragraph styles.

For each of the two root elements, book and article, there is a separate default set of named paragraph styles. You can override the effective set of styles (resulting from the chosen root element) with your own one if needed.

DocBook supports the use of both the CALS and the HTML table model. In the CALS table model, the pipeline does not support nested tables. If your further workflow requires CALS tables and you can be sure that your source documents do not contain nested tables, then using the CALS table model should be safe.

In any other case, it is highly recommended to use the HTML table model for conversion so that table structures are properly expressed in DocBook and that you do not lose any information, as any nested table when outputting to CALS will be dropped from the output completely.

The names of paragraph styles in the source document will be mirrored in the role attribute of the resulting para element.

Many elements in DocBook (tables, images) can be either of the informal variant, i.e. they do not have a name or caption. Translation of these elements is as expected and will create informaltable or inlinemediaobject elements.

When you want a table or image to have a caption or title, you must use the named paragraph style "caption" for this in your source document and you must ensure that it immediately precedes or follows the table or image it belongs to. To resolve ambiguous situations when mixing the position (before, after) within a single document, you can specify a bias parameter that tells the pipeline if it should prefer captions before or after elements when unclear. However, it is recommended that you decide for one placement option (either before or after the element) throughout the whole document to avoid false caption/element combinations.

Both, paragraph and character styles are supported, with special handling where appropriate. As a general rule, the name of the style is preserved in the role attribute in the destination element.

Consecutive paragraphs with paragraph style names contained in the following pipeline's parameters are handled specially by grouping them into an element of the respective name:

Character styles with names listed in the pipeline's parameter dbInlineClasses create an inline element of the respective name in the result.

You can modify the default setting (=style names) for all of the above variables by editing the Custom pipeline variables parameter in the PVAR: set semantic style names module.

The manual inline formattings italic, bold and underline are converted to emphasis elements with no role attribute, a role attribute of bold and a role attribute of underline, respectively.

^{Superscripted} and _subscripted text is automatically converted to superscript and subscript elements based on its layout traits.

Lists are converted in a best-effort manner. The pipeline supports both normal and compact lists:

Pagebreaks and column breaks in the original document are marked using XML comments:

The pipeline supports both the HTML and CALS^[1] table model with full support for horizontal and vertical spans. Colors and other layout traits are not preserved.

Unless you can be sure that the source document does not contain nested tables, use if the HTML table model is preferred.

upCast RT comes with full Unicode support, including special font encodings like Zapf Dingbats, Symbol and (as far as possible) Wingdings. Encodings for custom fonts can easily be added. Here are some examples:

❤ (Zapf Dingbats hearts), © (copyright symbol), … (ellipsis).

RTF has the concept of automatically generated content at the time of publishing (i.e.: printing) a document. Generated content can be output either by using the phrase element with a role attribute of GEN-contenttype :

or by using a uc-gentext Processing Instruction:

Generation can be controlled by setting the respective pipeline parameter.

Examples of generated content are the current date (), the current time () or the source document's filename ().

A second reference to an earlier footnote^[1] is handled specially and creates a footnoteref element.

Cross references to earlier sections (see Configuration parameters) are automatically generated using the xref element.

Links to interesting targets (like Lists) within the same document create link elements.

Hyperlinks to interesting external destinations create ulink elements.

RTF has no concept of titles for specific elements. Instead, they are formatted as standard paragraphs in the immediate vicinity of the object that should get a title and use a special paragraph style name, caption (or its localized equivalent):

Fig. upCast Product Logo

The DocBook pipeline tries to make a good guess for the matching title of an image or table it finds. You can set a bias as to whether the caption should be searched first before an element or after it by setting the corresponding pipeline parameter.

This document has been annotated with index terms using Word's tools for marking index entries. Index entries are converted to indexterm elements up to tertiary level.

Floating textboxes are converted to sidebar elements.

Word annotations or comments are converted to remark elements. If author information is available, it is includedChristian RothCRShould be revised to include more detail on authorname markup..

Generating a table of contents and its placement is not supported and left to the final publishing engine. Same holds true for other lists (LoT, LoF, …).

Objects currently are not marked for inclusion in a list of objects. To include them, you need to edit the corresponding templates in the processing sheet.

The default mapping for the book root element looks like this:

The default mapping for the article root element looks like this:

Using the Style to level mapping parameter, you can define a custom overlay. This parameter is a parameter of type list and can have up to 9 elements. In the graphical UI, each line corresponds to a list element. Each list element consists again of a list of style names which map onto the respective DocBook element's title child (and implicitly creating the container nesting in that way).

As an example, have a look at the following style to level mapping overlay:

As you see, for the list elements 1 and 3, two paragraph style names have been defined each. For syntax, enclose the style names in double quotes and separate them by whitespace or commas.

When using the article DocBook root element, combining this with the default settings, this result in the following, final style to level mapping table:

As you see, for each level where there is no overlay defined (empty line), the default settings are used, where for the levels where there are non-empty definitions, those replace the default settings. When you do not want any of the default settings to apply, you must define non-empty entries for all 9 level mapping slots.