KorAP_schemas/span.rng

<?xml version="1.0" encoding="UTF-8"?>

<grammar xmlns="http://relaxng.org/ns/structure/1.0" datatypeLibrary="http://www.w3.org/2001/XMLSchema-datatypes" ns="http://ids-mannheim.de/ns/KorAP" xmlns:a="http://relaxng.org/ns/compatibility/annotations/1.0">
  <!-- $Id$ -->
  <start>
    <element name="layer">
      <optional>
        <attribute name="type">
          <a:documentation>As of Sep 2013, this isn't used anywhere.</a:documentation>
          <data type="NCName"/>
        </attribute>
      </optional>
      <attribute name="version">
        <value>KorAP-0.4</value>
      </attribute>
      <attribute name="docid">
        <data type="NCName"/>
      </attribute>
      <optional>
        <attribute name="unit" a:defaultValue="char">
          <documentation xmlns="http://relaxng.org/ns/compatibility/annotations/1.0">Default: character</documentation>
          <choice>
            <value type="NCName">char</value>
            <value type="NCName">msec</value>
            <value type="NCName">sec</value>
          </choice>
        </attribute>
      </optional>
      <oneOrMore>
        <element name="spanList">
          <!-- shouldn't this come straight from XML?
            <optional>
            <attribute>
              <documentation xmlns="http://relaxng.org/ns/compatibility/annotations/1.0">If you ask me about why we need xml:base here, the answer is probably: I forgot to remove it and maybe I should -P</documentation>
              <name ns="http://www.w3.org/XML/1998/namespace">base</name>
              <data type="anyURI"/>
            </attribute>
          </optional>-->
          <optional>
            <attribute name="fragmented">
              <a:documentation>For spanLists non-exhaustively covered by the spans therein.</a:documentation>
              <data type="boolean"/>
            </attribute>
          </optional>
          <optional>
            <attribute name="speaker-id">
              <documentation xmlns="http://relaxng.org/ns/compatibility/annotations/1.0">This is for speech corpora; the datatype restriction may be too strict, I assumed it is a real ID in the source</documentation>
              <data type="NCName"/>
            </attribute>
          </optional>
          <optional>
            <attribute name="id">
              <data type="ID"/>
            </attribute>
          </optional>
          <optional>
            <attribute name="type">
              <data type="NCName"/>
            </attribute>
          </optional>
          <oneOrMore>
            <ref name="span"/>
          </oneOrMore>
        </element>
      </oneOrMore>
    </element>
  </start>
  <define name="span">
    <element name="span">
      <optional>
        <attribute name="id">
          <data type="ID"/>
        </attribute>
      </optional>
      <!--<optional>
        <attribute name="type">
          <data type="NCName"/>
        </attribute>
      </optional>-->
      <optional>
        <attribute name="l">
          <data type="integer"/>
        </attribute>
      </optional>
      <attribute name="from">
        <choice>
          <data type="integer"/>
        </choice>
      </attribute>
      <attribute name="to">
        <choice>
          <data type="integer"/>
        </choice>
      </attribute>
      <optional>
        <documentation xmlns="http://relaxng.org/ns/compatibility/annotations/1.0">This attribute is used in the base foundry tokenization files only, and only for a while, for debugging purposes. It may be that the time has come to comment it out.</documentation>
        <attribute name="word">
          <data type="string"/>
        </attribute>
      </optional>
      <optional>
        <attribute name="prev">
          <data type="integer"/>
        </attribute>
      </optional>
      <optional>
        <attribute name="next">
          <data type="integer"/>
        </attribute>
      </optional>
      <optional>
        <attribute name="prev_id">
          <data type="IDREF"/>
        </attribute>
      </optional>
      <optional>
        <attribute name="next_id">
          <data type="IDREF"/>
        </attribute>
      </optional>
      
      <optional>
        <ref name="fs"/>
      </optional>

      <zeroOrMore>
        <ref name="rel"/>
      </zeroOrMore>
    </element>
  </define>
  
  <define name="rel">
    <element name="rel">
      <documentation xmlns="http://relaxng.org/ns/compatibility/annotations/1.0">The possible children are <fs/> and <span/>, in any order. <fs/> or the @label attribute are used to encode the 'labelling' information. Instead of <span/>, one can use either the @target attribute (for referring to IDs within the same file), or the @uri attribute (for URIs), or the attribute @type with the values set to either "unary" (for e.g. the malformed dependencies that XIP produces) or "refl" for reflexive relationships.</documentation>
      <!--<optional>
        <attribute name="n">
          <documentation xmlns="http://relaxng.org/ns/compatibility/annotations/1.0">This is a kludgey attribute, currently used for dependencies that target e.g. a compound or an infinitive together with "zu"</documentation>
          <data type="integer"/>
        </attribute>
      </optional>-->
      <interleave>
        <choice>
          <attribute name="label">
            <text/>
          </attribute>
          <ref name="fs"/>
        </choice>
        <choice>
          <choice>
            <attribute name="target">
              <data type="IDREF"/>
            </attribute>
            <attribute name="uri">
              <data type="anyURI"/>
            </attribute>
          </choice>
          <oneOrMore><ref name="span"/></oneOrMore>
<!-- oneOrMore was introduced here for time spans; this could be handled by separate <rel> elements, of course-->
          <attribute name="type">
            <choice>
              <value type="NCName">unary</value>
              <value type="NCName">refl</value>
            </choice>
          </attribute>
        </choice>
      </interleave>
    </element>
  </define>
  
  <include href="fsr.rng"/>
</grammar>