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

<pml_schema xmlns="http://ufal.mff.cuni.cz/pdt/pml/schema/"  version="1.1">
  <revision>1.0.0</revision>
  <description>Data for Treex (formerly TectoMT)</description>
  <!-- Treex PML schema  -->


  <!-- Importing schema submodules -->

  <import type="langcode.type" schema="treex_subschema_langcodes.xml"/>
  <import type="p-nonterminal.type" schema="treex_subschema_p_layer.xml"/>
  <import type="a-root.type" schema="treex_subschema_a_layer.xml"/>
  <import type="t-root.type" schema="treex_subschema_t_layer.xml"/>
  <import type="n-root.type" schema="treex_subschema_n_layer.xml"/>
  <import type="w-doc.type" schema="treex_subschema_w_layer.xml"/>


  <!-- Each document contains "meta-attributes" common for the whole
 document (such as the source text to be translated) and a list of
 tree bundles related to the individual sentences. -->


  <root name="treex_document">
    <structure>
      <member name="meta" required="0" type="meta.type"/>
      <member name="bundles" role="#TREES" required="0">
        <list ordered="1" type="bundle.type"/>
      </member>
    </structure>
  </root>


  <!-- Metainfo (info about the whole document) -->
  <type name="meta.type">
    <structure>

       <member name="description"><cdata format="any"/></member>

       <member name="zones">
         <sequence>
           <element name="zone" type="language_parametrized_doc_zone.type"/>
         </sequence>
       </member>

       <member name="wild_dump"><cdata format="any"/></member>
       
    </structure>
  </type>


  <type name="language_parametrized_doc_zone.type" >
    <structure>
      <member name="language" required="1" as_attribute="1" type="langcode.type"/>
      <member name="selector" as_attribute="1"><cdata format="any"/></member>
      <member name="text"><cdata format="any"/></member>
    </structure>
  </type>



  <type name="bundle.type" >
    <structure role="#NODE" name="bundle-root">
      <member as_attribute="1" name="id" role="#ID" required="1"><cdata format="ID"/></member>

      <!--CzEng meta informations (not stable, may be changed/deleted in future)
	      align_score is a alignment quality score which may be used in filters.
	      CzEng ids are very long, so we don't want to store them in bundle id,
	      because id of each node is derived from the id of its bundle (and zone and tree)
	      and we don't want to have so many long ids.
	  -->
      <member name="czeng">
        <structure>
          <member name="section"><cdata format="any"/></member>
          <member name="blockid"><cdata format="any"/></member>
          <member name="domain"><cdata format="any"/></member>
          <member name="id"><cdata format="any"/></member>
          <member name="missing_sents_before"><cdata format="any"/></member>
          <member name="filter_score"><cdata format="any"/></member>
          <member name="filter_features"><cdata format="any"/></member>
          <member name="align_score"><cdata format="any"/></member>
          <member name="origfile"><cdata format="any"/></member>
        </structure>
      </member>

      <member name="zones">
        <sequence>
          <element name="zone" type="zone.type"/>
        </sequence>
      </member>

      <member name="message_board">
        <list ordered="1">
           <cdata format="any"/>
        </list>
      </member>

      <!-- ptacek: -->
      <member name="dialog">
       		<structure>
       			<member name="from"><cdata format="any"/></member>
       			<member name="to"><cdata format="any"/></member>
       			<member name="body_asr_confidence"><cdata format="any"/></member>
       			<member name="body_gesture"><cdata format="any"/></member>
       			<member name="body_emotion"><cdata format="any"/></member>
       			<member name="dialog_act"><cdata format="any"/></member>
       			<member name="sentence_type"><cdata format="any"/></member>
       			<member name="utterance_datetime"><cdata format="any"/></member>
       			<member name="local_timezone"><cdata format="any"/></member>
       		</structure>
       </member>

       <member name="wild_dump"><cdata format="any"/></member>

    </structure>
  </type>


  <!-- generic (language- and purpose-parametrized) subbundle,
  containing trees and attributes -->

  <type name="zone.type" >
    <structure>
      <member name="language" required="1" as_attribute="1" type="langcode.type"/>
      <member name="selector" as_attribute="1"><cdata format="any"/></member>
      <member name="sentence"><cdata format="any"/></member>

      <member name="trees">
        <structure>
          <member name="a_tree" type="a-root.type"/> <!-- pozor,      bude potreba zmenit a-root.type!!! -->
          <member name="t_tree" type="t-root.type"/>
          <member name="n_tree" type="n-root.type"/>
          <member name="p_tree" type="p-nonterminal.type"/>
        </structure>
      </member>
    </structure>
  </type>

</pml_schema>