From 8a1c1bc812a19a7ba4b2bfc9c63b60457b87a435 Mon Sep 17 00:00:00 2001
From: Cyprian Laskowski <cyp@cjvt.si>
Date: Fri, 16 Oct 2020 15:40:39 +0200
Subject: [PATCH] IssueID #1224: added support and example for morphological
 lexicon

---
 resources/schema/inventory.xsd             | 127 +++++++++++++++++----
 resources/schema/morphological_lexicon.xsd |   6 +-
 resources/schema/top_dictionaries.xsd      |   5 +-
 3 files changed, 112 insertions(+), 26 deletions(-)

diff --git a/resources/schema/inventory.xsd b/resources/schema/inventory.xsd
index 945cd31..1b94a52 100644
--- a/resources/schema/inventory.xsd
+++ b/resources/schema/inventory.xsd
@@ -23,6 +23,19 @@
             </xsd:restriction>
           </xsd:simpleType>
         </xsd:attribute>
+        <xsd:attribute name="orthography_id" type="xsd:integer"/>
+        <xsd:attribute name="dynamic_accentuation_id" type="xsd:integer"/>
+        <xsd:attribute name="tonemic_accentuation_id" type="xsd:integer"/>
+        <xsd:attribute name="pronunciation_id" type="xsd:integer"/>
+        <xsd:attribute name="hyphenation_id" type="xsd:integer"/>
+        <xsd:attribute name="norm">
+          <xsd:simpleType>
+            <xsd:restriction base="xsd:string">
+              <xsd:enumeration value="non-standard"/>
+            </xsd:restriction>
+          </xsd:simpleType>
+        </xsd:attribute>
+
       </xsd:extension>
     </xsd:simpleContent>
   </xsd:complexType>
@@ -101,8 +114,8 @@
       </xsd:choice>
     </xsd:sequence>
     <xsd:attribute name="id" type="xsd:int"/>
-    <xsd:attribute name="sloleks_id" type="xsd:string"/>
-    <xsd:attribute name="sloleks_key" type="xsd:string"/>
+    <xsd:attribute name="sloleksId" type="xsd:string"/>
+    <xsd:attribute name="sloleksKey" type="xsd:string"/>
     <xsd:attribute name="type" use="required"> <!-- types of headwords allowed; hmm, probably should standardise allowed settings -->
       <xsd:simpleType>
         <xsd:restriction base="xsd:string">
@@ -507,29 +520,101 @@
     </xsd:simpleContent>
   </xsd:complexType>
 
-  <!-- Normalisation types for form representations  -->
-  <xsd:simpleType name="formNormType">
-    <xsd:restriction base="xsd:string">
-      <xsd:enumeration value="standard"/>
-      <xsd:enumeration value="non-standard"/>
-      <xsd:enumeration value="variant"/>
-    </xsd:restriction>
-  </xsd:simpleType>
-
-  <!-- Form representation wrapper -->
-  <xsd:complexType name="formRepresentationType">
+  <!-- Container for all kinds of form representation lists -->
+  <xsd:complexType name="formRepresentationsType">
     <xsd:sequence>
-      <xsd:element name="measureList" type="measureListType"/>
-      <xsd:element name="norm" type="formNormType" minOccurs="0"/>
-      <xsd:element name="form" type="xsd:string"/>
+      <xsd:element name="orthographyList" type="orthographyListType"/>
+      <xsd:element name="accentuationList" type="accentuationListType" minOccurs="0" maxOccurs="2"/>
+      <xsd:element name="pronunciationList" type="pronunciationListType" minOccurs="0"/>
+      <xsd:element name="hyphenationList" type="hyphenationListType" minOccurs="0"/>
     </xsd:sequence>
-    <xsd:attribute name="id" type="xsd:int" use="required"/>
   </xsd:complexType>
 
-  <!-- List of form representations -->
-  <xsd:complexType name="formRepresentationListType">
+  <xsd:complexType name="orthographyType">
     <xsd:sequence>
-      <xsd:element name="formRepresentation" type="formRepresentationType" maxOccurs="unbounded"/>
+      <xsd:element name="form" type="formType"/>
+      <xsd:element name="measureList" type="measureListType" minOccurs="0"/>
+    </xsd:sequence>
+    <xsd:attribute name="id" type="xsd:integer" use="required"/>
+    <xsd:attribute name="morphologyPatterns" type="xsd:string"/>
+    <xsd:attribute name="status" type="xsd:string"/>
+    <xsd:attribute name="norm" type="xsd:string"/> <!-- closed set? -->
+  </xsd:complexType>
+
+  <xsd:complexType name="orthographyListType">
+    <xsd:sequence>
+      <xsd:element name="orthography" type="orthographyType" maxOccurs="unbounded"/>
+    </xsd:sequence>
+  </xsd:complexType>
+
+  <xsd:complexType name="accentuationType">
+    <xsd:sequence>
+      <xsd:element name="form" type="formType"/>
+      <xsd:element name="measureList" type="measureListType" minOccurs="0"/>
+    </xsd:sequence>
+    <xsd:attribute name="id" type="xsd:integer" use="required"/>
+    <xsd:attribute name="orthography_id" type="xsd:integer" use="required"/>
+    <xsd:attribute name="stressPatterns" type="xsd:string"/> 
+    <xsd:attribute name="status" type="xsd:string"/> 
+    <xsd:attribute name="norm" type="xsd:string"/>
+  </xsd:complexType>
+
+  <xsd:complexType name="accentuationListType">
+    <xsd:sequence>
+      <xsd:element name="accentuation" type="accentuationType" maxOccurs="unbounded"/>
+    </xsd:sequence>
+    <xsd:attribute name="type">
+      <xsd:simpleType>
+        <xsd:restriction base="xsd:string">
+          <xsd:enumeration value="dynamic"/>
+          <xsd:enumeration value="tonemic"/>
+        </xsd:restriction>
+      </xsd:simpleType>
+    </xsd:attribute>
+  </xsd:complexType>
+
+  <xsd:simpleType name="formType">
+    <xsd:restriction base="xsd:string"/>
+  </xsd:simpleType>
+
+  <xsd:complexType name="pronunciationFormType">
+    <xsd:simpleContent>
+      <xsd:extension base="xsd:string">
+        <xsd:attribute name="script" type="xsd:string" use="required"/>
+      </xsd:extension>
+    </xsd:simpleContent>
+  </xsd:complexType>
+
+  <xsd:complexType name="pronunciationType">
+    <xsd:sequence>
+      <xsd:element name="form" type="pronunciationFormType" maxOccurs="unbounded"/>
+      <xsd:element name="measureList" type="measureListType" minOccurs="0"/>
+    </xsd:sequence>
+    <xsd:attribute name="id" type="xsd:integer" use="required"/>
+    <xsd:attribute name="dynamic_accentuation_id" type="xsd:integer" use="required"/>
+    <xsd:attribute name="tonemic_accentuation_id" type="xsd:integer"/>
+    <xsd:attribute name="status" type="xsd:string"/> 
+    <xsd:attribute name="norm" type="xsd:string"/>
+  </xsd:complexType>
+
+  <xsd:complexType name="pronunciationListType">
+    <xsd:sequence>
+      <xsd:element name="pronunciation" type="pronunciationType" maxOccurs="unbounded"/>
+    </xsd:sequence>
+  </xsd:complexType>
+
+  <xsd:complexType name="hyphenationType">
+    <xsd:sequence>
+      <xsd:element name="form" type="formType"/>
+      <xsd:element name="measureList" type="measureListType" minOccurs="0"/>
+    </xsd:sequence>
+    <xsd:attribute name="id" type="xsd:integer" use="required"/>
+    <xsd:attribute name="orthography_id" type="xsd:integer" use="required"/>
+  </xsd:complexType>
+
+  <xsd:complexType name="hyphenationListType">
+    <xsd:sequence>
+      <xsd:element name="hyphenation" type="hyphenationType" maxOccurs="unbounded"/>
     </xsd:sequence>
   </xsd:complexType>
 
@@ -538,7 +623,7 @@
     <xsd:sequence>
       <xsd:element name="msd" type="msdType" maxOccurs="unbounded"/>
       <xsd:element name="grammarFeatureList" type="grammarFeatureListType" minOccurs="0"/>
-      <xsd:element name="formRepresentationList" type="formRepresentationListType"/>
+      <xsd:element name="formRepresentations" type="formRepresentationsType"/>
     </xsd:sequence>
     <xsd:attribute name="id" type="xsd:int" use="required"/>
   </xsd:complexType>
diff --git a/resources/schema/morphological_lexicon.xsd b/resources/schema/morphological_lexicon.xsd
index 4746d6e..6ee932c 100644
--- a/resources/schema/morphological_lexicon.xsd
+++ b/resources/schema/morphological_lexicon.xsd
@@ -16,9 +16,9 @@
                   <xsd:sequence>
                     <xsd:element name="status" type="statusType" minOccurs="0"/>
 	            <xsd:element name="headword" type="headwordType"/>
-	            <xsd:element name="lexicalUnit" type="lexicalUnitType" minOccurs="0"/>
-                    <xsd:element name="grammar" type="grammarType" minOccurs="0"/>
-                    <xsd:element name="measureList" type="measureListType" minOccurs="0"/>
+	            <xsd:element name="lexicalUnit" type="lexicalUnitType"/>
+                    <xsd:element name="grammar" type="grammarType"/>
+                    <xsd:element name="measureList" type="measureListType"/>
                     <xsd:element name="relatedEntryList" type="relatedEntryListType" minOccurs="0"/>
                     <xsd:element name="labelList" type="labelListType" minOccurs="0"/>
 	            <xsd:element name="comment" type="commentType" minOccurs="0"/>
diff --git a/resources/schema/top_dictionaries.xsd b/resources/schema/top_dictionaries.xsd
index 493958c..3099dfb 100644
--- a/resources/schema/top_dictionaries.xsd
+++ b/resources/schema/top_dictionaries.xsd
@@ -6,10 +6,11 @@ validated both with this schema and with an appropriate more specific schema bas
   <xsd:import namespace="http://www.w3.org/XML/1998/namespace" schemaLocation="http://www.w3.org/2001/xml.xsd" />
 
   <!-- top-level element is a dictionary -->
-  <xsd:element name="dictionary" type="dictionaryType"/>
+  <xsd:element name="dictionary" type="rootType"/>
+  <xsd:element name="lexicon" type="rootType"/>
 
   <!-- top-level type consists of sequence of entries -->
-  <xsd:complexType name="dictionaryType">
+  <xsd:complexType name="rootType">
     <xsd:sequence>
       <xsd:element name="entry" type="entryType" minOccurs="0" maxOccurs="unbounded"/>
     </xsd:sequence>