testing maps

[djbpitt]

The following test succeeds when run from the command line using XSpec pulled from the master branch of the GitHub repo (Latest commit 4c50caa; SAXON HE 9.9.1.4), but fails in <oXygen/> 21.1, build 2019101513 with version XSpec 1.2.1 of the add-on (both Framework and Helper view). The error message, from Ant, reads "FOTY0013: Maps cannot be atomized". The test is:

<x:scenario label="Scenario for testing function tokenize_input with two one-word inputs">
    <x:call function="djb:tokenize_input">
        <x:param name="s1" select="'kittens'"/>
        <x:param name="s2" select="'sitting'"/>
    </x:call>
    <x:expect label="Two one-word inputs are okay"
        select='            
        map{
        "left":("s","i","t","t","i","n","g"),
        "top":("k","i","t","t","e","n","s"),
        "type":"characters"
        }'
    />
</x:scenario>

and the function is:

<xsl:function name="djb:tokenize_input" as="map(xs:string, item()+)">
    <xsl:param name="top" as="xs:string"/>
    <xsl:param name="left" as="xs:string"/>
    <!-- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* -->
    <!-- validate input                                        -->
    <!-- no null strings                                       -->
    <!-- both strings must be either single  or multiple-word  -->
    <!-- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* -->
    <!-- normalize whitespace first                            -->
    <!-- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* -->
    <xsl:variable name="top_n" as="xs:string" select="normalize-space($top)"/>
    <xsl:variable name="left_n" as="xs:string" select="normalize-space($left)"/>

    <!-- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* -->
    <!-- check for mismatch parameter types                    -->
    <!-- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* -->
    <xsl:if test="(string-length($top_n), string-length($left_n)) = 0">
        <xsl:message select="'Null strings are not permitted'" terminate="yes"/>
    </xsl:if>
    <xsl:if
        test="
        not(
        (matches($top_n, '\s') and matches($left_n, '\s'))
        or
        not(matches($top_n, '\s')) and not(matches($left_n, '\s'))
        )">
        <xsl:message
            select="'Either both strings must be single words or both strings must be multiple words'"
            terminate="yes"/>
    </xsl:if>


    <!-- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* -->
    <!-- split the inputs                                      -->
    <!-- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* -->
    <xsl:variable name="top_out" as="xs:string+"
        select="
        if (matches($top_n, '\s')) then
        tokenize($top_n, '\s+')
        else
        for $c in string-to-codepoints($top_n)
        return
        codepoints-to-string($c)"/>
    <xsl:variable name="left_out" as="xs:string+"
        select="
        if (matches($left_n, '\s')) then
        tokenize($left_n, '\s+')
        else
        for $c in string-to-codepoints($left_n)
        return
        codepoints-to-string($c)"/>

    <!-- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* -->
    <!-- are we returning characters or words?                 -->
    <!-- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* -->
    <xsl:variable name="input_type" as="xs:string+"
        select="
        if (matches($top_n, '\s')) then
        'words'
        else
        'characters'"/>

    <!-- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* -->
    <!-- return tokenized sequences and type in map            -->
    <!-- *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* -->
    <xsl:sequence
        select="
        map {
        'top': $top_out,
        'left': $left_out,
        'type': $input_type
        }"
    />
</xsl:function>

Synopsis: The point of the function is to accept either two one-word inputs or two multi-word inputs (both as single strings; the multi-word ones are identified by internal whitespace). With one-word inputs, it should split the string into single-character strings. With multi-word inputs, it should word-tokenize the inputs on white space. It returns a map with three items: the sequences for each of the two inputs and a report on whether it split into characters or words. It should trap any input that isn't two single words or two multi-word strings, either as a type error on the parameter or by checking performed inside the function.

[AirQuick]

Testing map (or array or a derivation of function(*)) was fixed in XSpec v1.3.0 for XSLT and in v1.4.0 for XQuery. (xspec/xspec#355)

@AlexJitianu
You may possibly want to update XSpec Framework for XSpec Helper view.
Also, looking at this repository, I noticed you needed to patch XSpec compiler. Would it be worth integrating some of the patches into the main repository?

As a side note, I have submitted a pull request to integrate xspec.framework into the XSpec repository (xspec/xspec#713).

[djbpitt]

@AirQuick Thank you for the quick response. Am I correct in thinking that you're advising me to wait until SyncroSoft pushes new versions of the two XSpec add-ins?

[AirQuick]

Yes, at least XSpec Framework for XSpec Helper view needs to be updated. Looks like it's still based on v1.2.0.

[djbpitt]

@AirQuick Thanks again. I'll make do with command-line testing for now.