<?xml version="1.0" encoding="utf-8"?>

<!-- 
    ConvertMorph rules for the 19th Century Fiction collection (NCF). 
    
    These rules define the translation of MorphAdorner XML output files for NCF into WordHoard 
    XML input files for NCF.
    
    We've included quite a few comments in this file about the format of the file in general and 
    what all the many rules do in general, in addition to comments that are specific to the NCF 
    collection. These comments do not add up to complete formal specifications, but they should 
    give you a good idea of what everything means and how it works.
    
    The general idea is to make ConvertMorph fully parameterized and rule-driven, so that
    it can be used to process morphadorned XML files from many different sources and in many
    different formats. Indeed, in an ideal world, these files would not even need to be in
    TEI format or a TEI subset or "simple" format. They could be in almost any format at all,
    as long as it is well-formed XML that has been processed by MorphAdorner. We're still short
    of fully reaching this goal, but we've made progress.
    
    Even within the realm of "TEISimple", there are many variations in the way collections are
    encoded. For an example which is quite different from NCF, see the ConvertMorph rules file
    for the Stein collection.
-->

<ConvertMorphRules>

    <!-- 
        The WordHoard corpus tag for this collection is "ncf".
        
        This element is required. Every WordHaord corpus must have a unique corpus tag, and
        there must be a definition for the specified corpus in the WordHoard "corpora.xml" file.
    -->

    <corpusTag>ncf</corpusTag>
    
    <!--
        Title page rules.
        This element is optional.
        
        This rule contains text to be generated for the WordHoard title pages of each work. 
        There may be any number of <respStmt> elements and at most one <publicationStmt> element, 
        in the format described in the WordHoard manual.
    -->
    
    <titlePageRules>
        <publicationStmt>
            <p>This version of the text is an experimental derivative of
            its version in the Chadwyck-Healey archive of
            Nineteenth-Century Fiction (hereafter NCF) and will be
            improved or replaced in the course of the MONK Project.</p>
            <p>The copyright to this text is owned by ProQuest, and this
            version may be accessed only by members who are affiliate
            with institutions that have purchased the NCF source files
            or by members directly affiliated with the MONK Project, who
            have been granted special permission by ProQuest to use the
            source files for purposes that bear directly on the
            development of the MONK Project.</p> <p>As an experimental
            derivative, this version differs from its source file in the
            following ways:</p> <p>1. The text file has been transformed
            from its original encoding in a Chadwyck-Healey SGML dtd
            into an XML file that parses under a TEI P5 dtd that uses a
            sharply reduced number of elements. This dtd, provisionally
            called teisimple.dtd, employs some minor extensions and
            relaxations of the TEI content model to accommodate a
            variety of text archive, including the Text Creation
            Partnership (TCP), Early American Fiction (EAF), the Wright
            Archive of American Fiction 1851-75), and Documenting the
            American South. Teisimple will evolve with the final release
            of TEI P5, scheduled for October 2007.</p> <p>2. While
            letters are explicitly tagged as such in other fiction
            collections, they are not identified in NCF, unless they
            occur in epistolary novels that consist only of letters. In
            order to enhance the comparability of NCF texts with texts
            in other fiction archives, letters in NCF have been
            explicitly tagged by Martin Mueller.  Nested narratives and
            other inserted documents have also been tagged explicitly.
            In the WordHoard environment, this tagging is not yet
            available for querying.</p> <p>3. The use of the apostrophe
            as an elision marker at the beginning or end of a word is a
            very common feature in poetic, regional or colloquial
            speech, but in digital environments its consistent
            disambiguation from opening or closing single quotations
            marks is a non-trivial task, with considerable consequences
            for tokenization and POS tagging. For various reasons, the
            current version does not distinguish between the apostrophe
            and the single quote as  accurately as it should and
            includes some unnecessary errors in tokenization or POS
            tagging. We know where to look for and correct these errors
            but will be grateful for users to point them out anyhow by
            using the error reporting feature of WordHoard.</p>
        </publicationStmt>
    </titlePageRules>
    
    <!-- We have no file rules for NCF. For an example of file rules, see the Stein rules. -->

    <!-- 
        Header rules for extracting WordHoard bibliographic information for each work. 
        
        These rules use regular epxression pattern matching. For details, see
        Sun's javadoc for the class java.util.regex.Pattern.
        
        NCF uses some rather complicated and sometimes inconsistent conventions for encoding
        header information, and we often need to use rather complex patterns to get the
        information we need for WordHoard. The examples given below along with the patterns
        illustrate all the different cases.
        
        These rules extract the following bibliographic information about each work for 
        WordHoard:
        
        workTag = Work tag. Each work in a collection must have a unique tag. For NCF we
        use the numbers as assigned by the C-H encoders.
        
        title = Work title.
        
        pubDateStart = Publication year, if known, or first year in a range of publication 
        years.
        
        pubDateEnd = Last year in a range of publication years.
                
        Note that for both works and work parts, WordHoard has the notion of
        "full" and "short" titles. ConvertMorph currently only has one notion of
        "title", and sets both WordHoard titles to the same value.
    -->

    <headerRules>
    
        <headerRule>
        
            <!-- Extracts the WordHoard work tag from the <idno> element. -->
                            
            <path>TEI/teiHeader/fileDesc/publicationStmt/idno</path>
            
            <pattern>
                <!--
                    Example:
                    In ANCF0101.xml, the <idno> is "NCF0101". The WordHoard work tag for this 
                    work is "0101", with the "NCF" prefix stripped off.
                    
                    Note that the WordHoard "full work tag" for a work is always 
                    "corpusTag-workTag", in this example "ncf-0101". Thus, from the point
                    of view of WordHoard, encoding "NCF" in the work tag itself would be
                    redundant, and we don't do it.
                    
                    In the pattern below, parentheses are used to "group" everything after
                    the "NCF" prefix, and "$1" is used to extract the value of this group
                    as the WordHard work tag.
                -->
                <match>NCF(.*)</match>
                <extract item="workTag">$1</extract>
            </pattern>
            
        </headerRule>
        
        <headerRule>
        
            <!-- 
                Extracts the WordHoard work title from the <title> element. 
                
                The titles encoded in NCF are verbose and include author names and
                other junk. It would be nice to have "cleaner" titles, but this would
                most likely need to be done by hand by a human being. It does not seem that
                regular expressions are adequate for this task. So for now we just use the
                work titles as is, without making any attempt to clean them up.
                
                Note that for NCF, ConvertMorph often generates very long titles for both
                works and their parts. The WordHoard client is ill-prepared to deal
                with long titles, which present human interface problems in quite a few places.
                To solve these problems, the WordHoard ingest program BuildWorks currently 
                truncates all titles to 50 characters.
                
                Whenever pattern matching is used to examine a string extracted from an
                XML source file, ConvertMorph replaces any runs of line feeds and carriage 
                returns in the string by a space before trying to match patterns against the
                string. Believe it or not, such garbage actually appears in at least one NCF
                work title string!
            -->
            
            <path>TEI/teiHeader/fileDesc/titleStmt/title</path>
            
            <pattern>
                <!-- 
                    Note that .* matches the whole string, and $0 extracts the whole string.
                    This is the simplest possible kind of pattern matching and header value
                    extraction.
                -->
                <match>.*</match>
                <extract item="title">$0</extract>
            </pattern>
            
        </headerRule>
        
        <headerRule>
        
            <!-- 
                Extracts WordHoard publication dates from the <date> element.
                
                WordHoard supports both simple publication dates (yyyy) and publication date
                ranges (yyyy-yyyy). NCF specifies this information in a variety of different
                formats, and we need quite a few different patterns to catch all the
                variations.
            -->
            
            <path>TEI/teiHeader/fileDesc/sourceDesc/biblFull/publicationStmt/date</path>
            
            <pattern>
                <!--
                    Four digit pub date, optionally enclosed in square brackets.
                    Examples:
                    ANCF0101.xml: 1839
                    ANCF22505.xml: [1850]
                    
                    The square brackets presumably encode some kind of meaningful information
                    in NCF, but we just ignore them. In the second example above, the
                    WordHoard publication date is simply 1850, and whatever information is
                    represented by the square brackets is lost.
                -->
                <match>\[?(\d\d\d\d)\]?</match>
                <extract item="pubDateStart">$1</extract>
            </pattern>
            
            <pattern>
                <!--
                    Two four digit dates separated by dash or en-dash.
                    Example:
                    ANCF22503.xml: 1840-1841
                    
                    Yes, NCF sometimes uses dash, and sometimes en-dash. It's hard to even
                    see the difference in the source text, but it's there and must be dealt
                    with in our patterns.
                -->
                <match>(\d\d\d\d)[--](\d\d\d\d)</match>
                <extract item="pubDateStart">$1</extract>
                <extract item="pubDateEnd">$2</extract>
            </pattern>
            
            <pattern>
                <!--
                    Four digit date, dash or en-dash, then two digits.
                    Example:
                    ANCF1602.xml: 1794-97
                    In this example, the WordHoard pub date range is 1794-1797.
                -->
                <match>(\d\d)(\d\d)[--](\d\d)</match>
                <extract item="pubDateStart">$1$2</extract>
                <extract item="pubDateEnd">$1$3</extract>
            </pattern>
            
            <pattern>
                <!--
                    Four digit date, dash or en-dash, then one digit.
                    Example:
                    ANCF3701.xml: 1826-7
                    In this example, the WordHoard pub date range is 1826-1827.
                -->
                <match>(\d\d\d)(\d)[--](\d)</match>
                <extract item="pubDateStart">$1$2</extract>
                <extract item="pubDateEnd">$1$3</extract>
            </pattern>
            
        </headerRule>
        
    </headerRules>
    
    <!-- 
        Rules for extracting WordHoard author information for each work. 
        
        A work can have more than one author. For each author, we extract the following 
        information for WordHoard:
        
        authorName = Author name.
        
        authorBirthYear = Author birth year, if known.
        
        authorDeathYear = Author death year, if known.
        
        authorEarliestWorkYear = Author earliest work year, if known. (Not used for NCF.)
        
        authorLatestWorkYear = Author latest work year, if known, (Not used for NCF.)
        
        ConvertMorph gathers together all the author information and updates the WordHoard
        authors definition file "authors.xml". Any new authors are added to the file. 
        If any author is encounterd which is already in the authors.xml file with 
        conflicting attribute values, an error message is issued.
    -->
    
    <authorRules>
    
        <!-- 
            The path to author elements. There may be more than one instance of this path
            for multiple authors, although for NCF this is not the case.
        -->
    
        <path>TEI/teiHeader/fileDesc/titleStmt/author</path>
        
        <headerRule sep=" / ">
        
            <!--
                Extracts WordHoard author information from the <author> element.
                
                NCF encodes author names, birth dates, and death dates in a variety of 
                different formats.
                
                The "sep" attribute is used to specify the separator for multiple values 
                encoded in a single element. In this rule, sep=" / " is specified for NCF. 
                The string is split using the separator, and each part is processed using 
                the patterns specified. For example, in ANCF25901.xml there are two 
                authors specified as:
                
                     Somerville, E. OE. (Edith OEnone), 1858-1949 / Ross, Martin, 1862-1915
                    
                In this example, the two parts are split out of the string and matched against 
                the patterns separately:
                
                    Somerville, E. OE. (Edith OEnone), 1858-1949
                    Ross, Martin, 1862-1915
                    
                This results in two WordHoard "author" elements being generated, one for
                the author "Somerville, E. OE. (Edith OEnone)" with birth and death dates
                1858 and 1949, and one for the author "Ross, Martin" with birth and death dates
                1862 and 1915.
                
                Patterns are matched in order. When a patten matches, the specified extraction
                rules are applied, and the remaining patterns are skipped. In this example, 
                the third and last pattern is used only if the first two patterns fail. The order
                of the patterns is therefore important. For this rule, if we put the last pattern
                first in the list, the rule would not work as intended!
            -->
            
            <!--
                The path to the element for this rule. This path is relative to the base
                path given above to author elements. For NCF, the author information is
                encoded directly in the author element rather than in child elements, so this
                relative path is empty.
            -->
            
            <path></path>
            
            <pattern>
                <!-- 
                    Both birth and death dates specified.
                    Examples:
                    ANCF24501.xml: Linton, E. Lynn (Elizabeth Lynn) 1822-1898
                    ANCF1501.xml: Hays, Mary, 1759/60-1843
                     
                    Either a dash (-) or an en-dash (-) can be used to separate the dates.
                    
                    In the second example above, the WordHoard birth and death dates are set
                    to 1759 and 1843 respectively. The "/60" following "1759" is ignored.
                -->
                <match>(.*?),? *(\d+)(/\d+)?[--](\d+)(/\d+)?</match>
                <extract item="authorName">$1</extract>
                <extract item="authorBirthYear">$2</extract>
                <extract item="authorDeathYear">$4</extract>
            </pattern>
            
            <pattern>
                <!-- 
                    Only birth date specified.
                    Example: 
                    ANCF0901.xml: Dacre, Charlotte, b. 1782 
                -->
                <match>(.*?),? *b. *(\d+)(/\d+)?</match>
                <extract item="authorName">$1</extract>
                <extract item="authorBirthYear">$2</extract>
            </pattern>
            
            <pattern>
                <!-- 
                    No dates specified.
                    Example: 
                    ANCF1101.xml: Fenwick, E. (Eliza) 
                -->
                <match>.*</match>
                <extract item="authorName">$0</extract>
            </pattern>
            
        </headerRule>
    
    </authorRules>
    
    <!-- 
        The path to the text for the work is TEI/text. 
        This element is required.
    -->
    
    <textPath>TEI/text</textPath>
    
    <!-- 
        Text element rules. 
        This element is required.
        
        For each element in the subtree rooted at the text path, these rules specify how to 
        process the element. The rules specify both when to create work parts and how to 
        format the text within the work parts in WordHoard.
        
        Each rule has the following possible attributes. Only the "name" attribute is 
        required.
        
        name = the name of the element.
        
        parBreak = true to force a paragraph break before and after the element. Default = false. 
        In WordHoard, paragraphs are separated by blank lines.
        
        lineBreak = true to force a line break before and after the element. Default = false.
        
        Note that parBreak implies lineBreak.
        
        lineStyle = left, center, or right. Default = no change in current line style.
        
        indent = indentation in pixels. Default = no change in current line indentation.
        Indentation is cumulative and is only used with the left justification line style.
        
        wordStyles = list of word styles separated by commas. Default = no change in current
        word styles. The word styles may be any styles supported by WordHoard: bold, italic,
        extended, underline, overline, superscript, subscript, monospaced, and plain. Word
        styles are cumulative, except for "plain", which is used to remove any current other
        word styles and revert to plain unstyled text.
        
        ignoreChildren = true to ignore any children of this element. Default = false.
        
        createPart = never, sometimes, or always. Default = never. "always" means to always
        create a new work part for this element (e.g., in NCF, for <div> elements). "sometimes"
        means to create a new work part only if it is necessary  (e.g., in NCF, for <trailer>
        and <epigraph> elements, which sometimes occur in "stranded" contexts where there is
        no active work part into which we are able to generate text.)
        
        footnote = true to treat as a footnote. Default = false. Footnotes are represented by 
        footnote numbers in the main text, and the numbered footnotes proper are generated at 
        the end of each work part.
        
        ignoreRend = true to ignore any rend attributes that might be present on this element.
        Default = false.
        
        genBefore = plain text string to generate before processing the element. Default =
        nothing.
        
        genAfter = plain text string to generate after processing the element. Default =
        nothing.
        
        <w> and <c> elements are generated by MorphAdorner. These elements are processed
        specially by ConvertMorph, and result in WordHoard <w> and <punc> elements. None of the
        other elements are processed specially in any way other than as specified by the rules
        enumerated here.
        
        ConvertMorph generates an error message if it encounters an element while processing
        the text which is not definied by a rule.
    -->
    
    <textElementRules>
        <textElementRule name="add"/>
        <textElementRule name="argument" parBreak="true" indent="20"/>
        <textElementRule name="back"/>
        <textElementRule name="bibl" parBreak="true" lineStyle="right"/>
        <textElementRule name="body"/>
        <textElementRule name="c" ignoreChildren="true"/>
        <textElementRule name="cell" genBefore=" [" genAfter="] " ignoreRend="true"/>
        <textElementRule name="closer" parBreak="true"/>
        <textElementRule name="div" createPart="always"/>
        <textElementRule name="epigraph" parBreak="true" indent="20" createPart="sometimes"/>
        <textElementRule name="figure" ignoreChildren="true"/>
        <textElementRule name="foreign" wordStyles="italic"/>
        <textElementRule name="front"/>
        <textElementRule name="gap" ignoreChildren="true" genBefore=" "/>
        <textElementRule name="head" parBreak="true" lineStyle="center" wordStyles="bold"/>
        <textElementRule name="hi"/>
        <textElementRule name="insertDoc" parBreak="true" indent="20"/>
        <textElementRule name="item" parBreak="true" indent="20"/>
        <textElementRule name="l" lineBreak="true"/>
        <textElementRule name="label" parBreak="true" lineStyle="center" wordStyles="italic"/>
        <textElementRule name="lb" lineBreak="true" ignoreChildren="true"/>
        <textElementRule name="letter" parBreak="true" indent="20"/>
        <textElementRule name="lg" parBreak="true"/>
        <textElementRule name="list" parBreak="true"/>
        <textElementRule name="milestone" ignoreChildren="true"/>
        <textElementRule name="note" parBreak="true" footnote="true"/>
        <textElementRule name="opener" parBreak="true"/>
        <textElementRule name="p" parBreak="true"/>
        <textElementRule name="pb" ignoreChildren="true"/>
        <textElementRule name="q" parBreak="true" indent="20"/>
        <textElementRule name="row" lineBreak="true"/>
        <textElementRule name="salute" parBreak="true"/>
        <textElementRule name="seg"/>
        <textElementRule name="signed" lineBreak="true"/>
        <textElementRule name="sp" parBreak="true" indent="20"/>
        <textElementRule name="speaker" parBreak="true" indent="-20"/>
        <textElementRule name="stage" parBreak="true" lineStyle="center" wordStyles="italic"/>
        <textElementRule name="table" parBreak="true" indent="20"/>
        <textElementRule name="text"/>
        <textElementRule name="title" wordStyles="italic"/>
        <textElementRule name="trailer" parBreak="true" lineStyle="center" createPart="sometimes"/>
        <textElementRule name="w" ignoreChildren="true"/>
    </textElementRules>
    
    <!--
        Work part title rules.
        This element is required.
        
        There are four kinds of rules that can be used for getting work part titles. The rules 
        are tried in the order listed until one works. If none of the rules work, an error 
        message is issued and the title is set to "Untitled".
        
        useFirstChild: Use the first child element with a specified name. For NCF, we use the 
        first <head> child element, which works reasonably well in most cases. All of the text
        of the child element is used, except for any embedded footnotes and any embedded
        descendant elements which have ignoreChildren set to true in their text element rule.
        This is important, because in NCF there are indeed some <head> elements which have
        these kinds of descendants.
        
        useAttributeValue: Use an attribute value, optionally converting the first letter
        to upper case. For NCF we use the "type" attribute. For example, in some NCF
        works, there are <div type="dedication"> elements which have no <head> children. In
        this case, the work part title is set to "Dedication".
        
        useElementName: Use the element name, optionally converting the first letter to
        upper case. In NCF, this rule catches quite a few <trailer> elements which become
        work parts with the title "Trailer".
        
        useAttributeValuePair: Uses a pair of attribute values separated by a space,
        optionally converting the first letter of the first attribute value to upper case.
        This rule is not used for NCF, but it might be useful for other collections. For
        example, a <div n="3" type="chapter"> element under this rule might result in the
        work title "Chapter 3". The rule for this example would be:
        
           <useAttributeValuePair name1="type" name2="n" capitalizeFirstLetter="true"/>
        
        For many NCF works, perhaps most of them, these rules work quite well, even
        surprisingly well. For many other works, however, the table of contents formed by
        the work part hierarchy and the titles generated by these rules ends up being, shall we 
        say, a bit goofy, ugly, and rather short of optimal. 
    -->
    
    <workPartTitleRules>
        <useFirstChild name="head"/>
        <useAttributeValue name="type" capitalizeFirstLetter="true"/>
        <useElementName capitalizeFirstLetter="true"/>
    </workPartTitleRules>
    
    <!--
        Rend attribute rules.
        
        These rules map the "rend" attribute values used in NCF to WordHoard "rend" attribute 
        values.
        
        Rend attributes are processed wherever they occur, on any element, unless the element
        rule specifies ignoreRend="true".
        
        Each rend rule can contain optional lineStyle, indent, and wordStyles attributes that work
        the same way as in element rules.
    -->
    
    <rendAttributeRules>
        <rendAttributeRule attrName="rend">
            <rendAttributeMapping value="b(1)" wordStyles="bold"/>
            <rendAttributeMapping value="i(1)" wordStyles="italic"/>
            <rendAttributeMapping value="i(2)" wordStyles="italic"/>
            <rendAttributeMapping value="italics" wordStyles="italic"/>
            <rendAttributeMapping value="align(c)" lineStyle="center"/>
            <rendAttributeMapping value="align(r)" lineStyle="right"/>
            <rendAttributeMapping value="indent(1)" indent="20"/>
            <rendAttributeMapping value="indent(2)" indent="40"/>
            <rendAttributeMapping value="indent(3)" indent="60"/>
            <rendAttributeMapping value="indent(5)" indent="100"/>
            <rendAttributeMapping value="sc(1)"/>
            <rendAttributeMapping value="sc(2)"/>
            <rendAttributeMapping value="small(1)"/>
            <rendAttributeMapping value="small(2)"/>
            <rendAttributeMapping value="sub(1)" wordStyles="subscript"/>
            <rendAttributeMapping value="sub(2)" wordStyles="subscript"/>
            <rendAttributeMapping value="sup(1)" wordStyles="superscript"/>
            <rendAttributeMapping value="sup(2)" wordStyles="superscript"/>
            <rendAttributeMapping value="roman(1)" wordStyles="plain"/>
            <rendAttributeMapping value="roman(2)" wordStyles="plain"/>
            <rendAttributeMapping value="speaker"/>
            <rendAttributeMapping value="caption - pb"/>
            <rendAttributeMapping value="caption - div"/>
        </rendAttributeRule>
    </rendAttributeRules>
    
    <!--
        Footnote rules.
        This element is optional. The default values are as shown below.
        
        Footnotes are rendered at the ends of work parts, with superscript references in
        the main text.
    -->
    
    <footnoteRules>
        <footnoteRefStyle wordStyles="superscript"/>
        <footnoteStyle indent="20"/>
    </footnoteRules>

</ConvertMorphRules>

<?xml version="1.0" encoding="utf-8"?>

<!-- 
    ConvertMorph rules for the Stein collection. 
-->

<ConvertMorphRules>

    <corpusTag>stein</corpusTag>
    
    <!-- We currently have no title page rules for Stein. -->
    
    <!--
        File rules.
        
        Unlike the NCF collection, in the Stein collection full bibliographic data is
        not encoded in the TEI files. Only the work titles are encoded in the files. We
        use the <fileRules> section here to enumerate the other bibliographic information
        for each file in the Stein collection.
    -->
    
    <fileRules>
        <fileRule>
            <name>threelives-1.0.xml</name>
            <workTag>tli</workTag>
            <author>Stein, Gertrude</author>
            <pubDateStart>1909</pubDateStart>
        </fileRule>
        <fileRule>
            <name>moa-1.1.xml</name>
            <workTag>moa</workTag>
            <author>Stein, Gertrude</author>
            <pubDateStart>1925</pubDateStart>
        </fileRule>
    </fileRules>
    
    <!--
        Header rules.
        
        Note that file rules override header rules. For example, suppose a file rule
        specifies a publication date of 1832 for a work, and a header rule extracts a
        publication date of 1847 for the same work. In this case, the WordHoard publication
        date is set to 1832, from the file rule for the file, and the value in the header
        is ignored. This is not an issue for Stein, where the only header rule we specify
        is for titles.
    -->

    <headerRules>
        <headerRule>
            <path>TEI.2/teiHeader/fileDesc/titleStmt/title</path>
            <pattern>
                <match>.*</match>
                <extract item="title">$0</extract>
            </pattern>
        </headerRule>
    </headerRules>
    
    <!-- 
        We have no author rules for Stein - author namess are given by the file rules above,
        and author attributes are specified in the WordHoard authors.xml file. 
    -->
    
    <textPath>TEI.2/text</textPath>
    
    <!--
        Text element rules.
        
        Note that we generate work parts for <div2> elements. This works well for "Three Lives",
        but results in some goofy "Space-break" parts in "Making of Americans". We could
        specify createPart="never" for <div2> to fix this, but then we'd need separate
        rule files for the two works. There's no way to say use one rule for one work and
        use a different rule for other works.
    -->
    
    <textElementRules>
        <textElementRule name="bibl" parBreak="true" lineStyle="right"/>
        <textElementRule name="body"/>
        <textElementRule name="c" ignoreChildren="true"/>
        <textElementRule name="div0" createPart="always"/>
        <textElementRule name="div1" createPart="always"/>
        <textElementRule name="div2" createPart="always"/>
        <textElementRule name="epigraph" parBreak="true" indent="20"/>
        <textElementRule name="head" parBreak="true" lineStyle="center" wordStyles="bold"/>
        <textElementRule name="name"/>
        <textElementRule name="note" parBreak="true" footnote="true"/>
        <textElementRule name="p" parBreak="true"/>
        <textElementRule name="pb" ignoreChildren="true"/>
        <textElementRule name="text"/>
        <textElementRule name="trailer" parBreak="true" lineStyle="center"/>
        <textElementRule name="w" ignoreChildren="true"/>
    </textElementRules>
    
    <!--
        Work part title rules.
        
        The combination and order of the work part title rules below was determined by
        trial and error. It seems to result in the most reasonable titles for the two Stein 
        novels.
    -->
    
    <workPartTitleRules>
        <useAttributeValuePair name1="type" name2="n" capitalizeFirstLetter="true"/>
        <useFirstChild name="head"/>
    </workPartTitleRules>
    
    <!-- We have no rend attribute rules. Stein doesn't have any style formatting! -->

</ConvertMorphRules>