Rendering of symbols

[IreneVagionakis]

Since the inclusion in EFES of the latest version of the EpiDoc Stylesheets, symbols are not displayed correctly anymore (and in some cases their presence prevents the inscriptions from being displayed: see iospe-5.238). The problem seems to be in webapps/ROOT/kiln/stylesheets/epidoc/teig.xsl at lines 59 and 101, whose paths to charDecl.xml and to a custom CharDecl are not processed correctly.

Everything works by replacing

<xsl:variable name="chardecl" select="if (//t:charDecl) then //t:charDecl else doc('charDecl.xml')"/>
with
<xsl:variable name="chardecl" select="if (//t:charDecl) then //t:charDecl else doc(concat('file:',system-property('user.dir'),'/webapps/ROOT/kiln/stylesheets/epidoc/charDecl.xml'))"/>

and

<xsl:variable name="externalCharDecl" select="substring-before($parsedRef, '#')"/>
with
<xsl:variable name="externalCharDecl" select="concat('file:',system-property('user.dir'),'/webapps/ROOT/content/xml/authority/', substring-before($parsedRef, '#'))"/>

but this solution is not good since it would only work inside EFES and since I imagine there is a prettier way to specify these paths. @ajenhl and @gabrielbodard, any suggestions?

[IreneVagionakis]

Is anyone aware of a nicer way to write this concat('file:',system-property('user.dir'),'/webapps/ROOT/content/xml/authority/', substring-before($parsedRef, '#')) in a context where relative paths starting with ../../../ etc. do not work?

[ajenhl]

I would try the following approaches:

• Find out where doc("foo.xml") is actually resolving to; if it's somewhere useful, put charDecl.xml there.
• Add a step in the processing of an Epidoc file that incorporates charDecl.xml, so that the test "if (//t:charDecl) then //t:charDecl else doc('charDecl.xml')" will never hit the else, because all of charDecl.xml is already going to match on //t:charDecl. This would be done in internal.xmap#local-preprocess-epidoc-language, to change the map:generate to a map:aggregate, then add in an XSLT that performs the merge.

teig.xsl should probably be reworked regardless so that $chardecl is not evaluated on every instance of tei:g!