fix-name-capitalization.xq

xquery version "3.0";

declare namespace fn="http://www.w3.org/2005/xpath-functions";

(: Fix problems with mis-capitalized names. For example: 
    Before: MACARTHUR, Douglas II
    After:  MacArthur, Douglas II
:)
declare function local:fix-name-capitalization($name as xs:string) {
    (: 
    We'll use analyze-string() to split the name string up into "words". 
    We're defining "words" as strings of one-or-more upper- or lower-case letters and hyphens. 
    E.g.: 
        In "MAO Tse-tung", "MAO" and "Tse-tung" are the two words.
        In "MCCLURKIN, Robert J. G.", the words are "MCCLURKIN", "Robert", "J", and "G".
    The analyze-string() function will return results like:
        <fn:analyze-string-result xmlns:fn="http://www.w3.org/2005/xpath-functions">
            <fn:match>MCCLURKIN</fn:match>
            <fn:non-match>, </fn:non-match>
            <fn:match>Robert</fn:match>
            <fn:non-match> </fn:non-match>
            <fn:match>J</fn:match>
            <fn:non-match>. </fn:non-match>
            <fn:match>G</fn:match>
            <fn:non-match>.</fn:non-match>
        </fn:analyze-string-result>
    We'll keep the "non-matches" unchanged, only looking carefully at each "match" to make
    sure we apply the right capitalization rules.
    :)
    let $word-vs-non-word-pattern := '[A-Za-z-]+' 
    let $analyze-string-result := analyze-string($name, $word-vs-non-word-pattern)
    return
        string-join(
            for $node in $analyze-string-result/*
            return
                (: let punctuation, spaces, etc. through unchanged :)
                if ($node/self::fn:non-match) then 
                    $node/string()
                (: examine :)
                else
                    (: MACARTHUR -> MacArthur :)
                    if (starts-with(lower-case($node), 'mac')) then 
                        concat('Mac', upper-case(substring($node, 4, 1)), lower-case(substring($node, 5)))
                    (: MCCARTHY -> McCarthy :)
                    else if (starts-with(lower-case($node), 'mc')) then 
                        concat('Mc', upper-case(substring($node, 3, 1)), lower-case(substring($node, 4)))
                    (: II -> II :)
                    else if (matches($node, '^[IVX]+$')) then 
                        $node/string()
                    (: otherwise, just capitalize the word :)
                    else
                        concat(upper-case(substring($node, 1, 1)), lower-case(substring($node, 2)))
        )
};

let $names := 
    (
    'MCCARTHY, Senator Joseph R.' (: potential problem because of "Mc" prefix :),
    'MACARTHUR, Douglas II' (: potential problem because of "Mac" prefix and generational name "II" :),
    'O’CONNOR, Roderic L' (: potential problem because of apostrophe in surname :),
    'VAN Hollen, Christopher' (: potential problem because the last name is in two parts :),
    'CHERWELL, Lord (Frederick Alexander Lindemann)' (: potential problem because of the parantheses :),
    'LINDEMANN, Frederick Alexander.' (: potential problem because of the period :),
    'MAO TSE-TUNG' (: potential problem because names are not comma-delimited :)
    )
return
    element results {
        for $name in $names
        return 
            element result {
                element source { $name },
                element repair { local:fix-name-capitalization($name) }
            }
    }

fix-name-capitalization_results.xml

<results>
    <result>
        <source>MCCARTHY, Senator Joseph R.</source>
        <repair>McCarthy, Senator Joseph R.</repair>
    </result>
    <result>
        <source>MACARTHUR, Douglas II</source>
        <repair>MacArthur, Douglas II</repair>
    </result>
    <result>
        <source>O’CONNOR, Roderic L</source>
        <repair>O’Connor, Roderic L</repair>
    </result>
    <result>
        <source>VAN Hollen, Christopher</source>
        <repair>Van Hollen, Christopher</repair>
    </result>
    <result>
        <source>CHERWELL, Lord (Frederick Alexander Lindemann)</source>
        <repair>Cherwell, Lord (Frederick Alexander Lindemann)</repair>
    </result>
    <result>
        <source>LINDEMANN, Frederick Alexander.</source>
        <repair>Lindemann, Frederick Alexander.</repair>
    </result>
    <result>
        <source>MAO TSE-TUNG</source>
        <repair>Mao Tse-tung</repair>
    </result>
</results>