Joins and Query Use with XML - Use case SEQ: queries based on sequence.

This use case illustrates queries based on the sequence in which elements appear in a document. Although sequence is not significant in most traditional database systems or object systems, it can be important in structured documents. This use case presents a series of queries based on a medical report:

  <!DOCTYPE report[
    <!ELEMENT report (section*)>
    <!ELEMENT section (section.title, section.content)>
    <!ELEMENT section.title (#PCDATA )>
    <!ELEMENT section.content (#PCDATA | anesthesia | prep
                  | incision | action | observation )*>
    <!ELEMENT anesthesia (#PCDATA)>
    <!ELEMENT prep ( (#PCDATA | action)* )>
    <!ELEMENT incision ( (#PCDATA | geography | instrument)* )>
    <!ELEMENT action ( (#PCDATA | instrument )* )>
    <!ELEMENT observation (#PCDATA)>
    <!ELEMENT geography (#PCDATA)>
    <!ELEMENT instrument (#PCDATA)>
]>
<report>
  <section> 
    <section.title>Procedure</section. title>
     <section.content>
      The patient was taken to the operating room where she was placed
      in supine position and
      <anesthesia>induced under general anesthesia.</anesthesia>
      <prep>
        <action>A Foley catheter was placed to decompress the bladder</action>
        and the abdomen was then prepped and draped in sterile fashion.
      </prep>
      <incision>
        A curvilinear incision was made
        <geography>in the midline immediately infraumbilical</geography>
        and the subcutaneous tissue was divided
        <instrument>using electrocautery.</instrument>
      </incision>
      The fascia was identified and
      <action>#2 0 Maxon stay sutures were placed on each side of the midline.
      </action>
      <incision>
        The fascia was divided using 
        <instrument>electrocautery </instrument>
        and the peritoneum was entered.
      </incision>
      <observation>The small bowel was identified.</observation>
      and
      <action>
        the
        <instrument>Hasson trocar</instrument>
        was placed under direct visualization.
      </action>
      <action>
        The
        <instrument>trocar</instrument>
        was secured to the fascia using the stay sutures.
      </action>
     </section.content>
  </section>
</report>
Question 1. In the Procedure section of Report1, what instruments were used in the second incision?
<xsl:template match="section[section.title = 'Procedure']"> <xsl:copy-of select="(.//incision)[2]/instrument"/>
</xsl:template>
Question 2. In the Procedure section of Report1, what are the first two instruments to be used?
<xsl:template match="section[section.title = 'Procedure']"> <xsl:copy-of select="(.//instrument)[position() &lt;= 2]"/>
</xsl:template>
Question 3. In Report1, what instruments were used in the first two actions after
the second incision?
<xsl:template match="report">
<!-- i2 = Second incision in the entire report -->
<xsl:variable name="i2" select="(.//incision)[2]"/>
<!-- Of all the actions following i2
     get the instruments used in the first two -->
<xsl:copy-of
     select="($i2/following::action)[position() &lt;= 2]/instrument"/>
</xsl:template>
Question 4. In Report1, find "Procedure" sections for which no anesthesia element
occurs before the first incision: <xsl:template match="section[section.title = 'Procedure']">
  <xsl:variable name="i1" select="(.//incision)[1]"/>
  <xsl:if test=".//anesthesia[preceding::incision = $i1]">
    <xsl:copy-of select="current()"/>
  </xsl:if>
</xsl:template>
If the result is not empty, then a major lawsuit is soon to follow! Question 5. In Report1, what happened between the first and second incision? <xsl:template match="report"> <critical_sequence>
  <!-- i1 = First incision in the entire report -->
  <xsl:variable name="i1" select="(.//incision)[1]"/>
  <!-- i2 = Second incision in the entire report -->
  <xsl:variable name="i2" select="(.//incision)[2]"/>
  <!-- copy all sibling nodes following i1
    that don't have a preceding element i2 and are not themeseves i2 -->
  <xsl:for-each select="$i1/following-sibling::node()
             [not(./preceding::incision = $i2) and not(. = $i2)]">
    <xsl:copy-of select="."/>
  </xsl:for-each>
</critical_sequence>
</xsl:template>

In Questions 4 and 5, I assume that the string values of incision elements are unique. This is true in the sample data, but may not be true in the most general case. To be precise, you should apply Recipe 4.2. For example, in Question 4, the test should be:

  test=".//anesthesia[count(./preceding::incision | $i1) =
  count(./preceding::incision)]"

Please check back  next week for the conclusion to this article.