Saxon

Legacy ID: #7387182 Legacy Poster: Michael Kay (mhkay)

Using regular expressions in this way to parse a recursive grammar is just plain wrong. You're asking how to optimize incorrect code. You need to handle this input properly by writing a recursive top-down parser. Dimitre Novatchev has some examples of how to write parsers in XSLT.

Legacy ID: #7388225 Legacy Poster: Lea Hayes (numberkruncher)

I was not happy with this approach, but it was the only way that I could think of splitting this string. The idea of writing a simple recursive descent parser using XSLT had not crossed my mind. I really should have thought about doing this from the beginning. I have not fully tested this yet, but I think that something like below should do the trick. This simply splits an XPointer fragment into separate schemes which is compatible with my previous XPointer code. Thanks! Lea Hayes <xsl:function name="doc:xptr-fragment"> <xsl:param name="fragment"/> <xsl:sequence select="doc:xptr-scheme($fragment)"/> </xsl:function> <xsl:function name="doc:xptr-scheme-name"> <xsl:param name="fragment"/> <xsl:analyze-string select="$fragment" regex="^\s*([^(]+)("> <xsl:matching-substring> <xsl:value-of select="regex-group(1)"/> </xsl:matching-substring> </xsl:analyze-string> </xsl:function> <xsl:function name="doc:xptr-scheme"> <xsl:param name="fragment"/> <!-- Read scheme name, only proceed if a valid one is found! --> <xsl:variable name="scheme-name" select="doc:xptr-scheme-name($fragment)"/> <xsl:if test="string-length($scheme-name)>0"> <xsl:variable name="remain-content" select="substring($fragment,string-length($scheme-name)+2)"/> <xsl:variable name="scheme-content" select="string-join(doc:xptr-scheme-data($remain-content),'')"/> <doc:scheme type="{$scheme-name}"> <xsl:value-of select="$scheme-content"/> </doc:scheme> <xsl:sequence select="doc:xptr-scheme(substring($remain-content,string-length($scheme-content)+2))"/> </xsl:if> </xsl:function> <xsl:function name="doc:xptr-scheme-data"> <xsl:param name="fragment"/> <!-- Read scheme data... --> <xsl:choose> <xsl:when test="starts-with($fragment,'(')"> <xsl:variable name="content" select="string-join(doc:xptr-scheme-data(substring($fragment,2)),'')"/> <xsl:text>(</xsl:text> <xsl:value-of select="$content"/> <xsl:text>)</xsl:text> <xsl:value-of select="doc:xptr-scheme-data(substring($fragment,string-length($content)+2))"/> </xsl:when> <xsl:when test="not(starts-with($fragment,')'))"> <xsl:variable name="content"> <xsl:analyze-string select="$fragment" regex="^([^()^]|^.)+"> <xsl:matching-substring> <xsl:value-of select="."/> </xsl:matching-substring> </xsl:analyze-string> </xsl:variable> <xsl:value-of select="$content"/> <xsl:variable name="remain-fragment" select="substring($fragment,string-length($content)+1)"/> <xsl:if test="not(starts-with($remain-fragment,')'))"> <xsl:value-of select="doc:xptr-scheme-data($remain-fragment)"/> </xsl:if> </xsl:when> </xsl:choose> </xsl:function> <!-- A simple test element --> <xsl:template match="doc:TEST"> <xsl:sequence select="doc:xptr-fragment('scheme1(a(b^)))scheme2(b(d(e^)f)))')"/> </xsl:template>

Legacy ID: #7388238 Legacy Poster: Michael Kay (mhkay)

Another interesting way to do this would be to replace "(" and ")" by "<a>" and "</a>" respectively, put the result through saxon:parse(), and then process the resulting tree using XPath. (Not quite as easy as that of course, because of special characters).