fix: Don't duplicate #text elements.

The duplicate #text elements were generated in the case,
when xsd:extension was used for xsd:simpleContent. That was
causing the generated records being invalid. We solve this by
filter the duplicate elements.

.gitignore

@@ -6,3 +6,6 @@ include.mk
 .rebar
 .eunit
 .test
+_build/
+erlsom.plt
+rebar.lock

priv/extension/Makefile

@@ -0,0 +1,5 @@
+all: validate
+
+validate:
+	@xmllint --noout --schema extension.xsd extension.xml
+	@xmllint --noout --schema simpleContentExtension.xsd simpleContentExtension.xml

priv/extension/simpleContentExtension.xml

@@ -0,0 +1,2 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<test xmlns="http://my.uri" a="aaa" b="bbb">myuri</test>

priv/extension/simpleContentExtension.xsd

@@ -0,0 +1,34 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<xsd:schema targetNamespace="http://my.uri"
+	xmlns="http://my.uri"
+	xmlns:tns="http://my.uri"
+	xmlns:xsd="http://www.w3.org/2001/XMLSchema"
+	elementFormDefault="qualified"
+	attributeFormDefault="unqualified"
+	blockDefault="#all"
+	version="0.2">
+	<xsd:complexType name="AttributedString">
+		<xsd:simpleContent>
+			<xsd:extension base="xsd:string">
+				<xsd:attribute name="a" type="xsd:string"/>
+				<xsd:anyAttribute namespace="##other" processContents="lax"/>
+			</xsd:extension>
+		</xsd:simpleContent>
+	</xsd:complexType>
+	<xsd:complexType name="EncodedString">
+		<xsd:simpleContent>
+			<xsd:extension base="tns:AttributedString">
+				<xsd:attribute name="b" type="xsd:string"/>
+				<xsd:attribute name="EncodingType" type="xsd:anyURI"/>
+			</xsd:extension>
+		</xsd:simpleContent>
+	</xsd:complexType>
+	<xsd:complexType name="KeyIdentifierType">
+		<xsd:simpleContent>
+			<xsd:extension base="tns:EncodedString">
+				<xsd:attribute name="ValueType" type="xsd:anyURI"/>
+			</xsd:extension>
+		</xsd:simpleContent>
+	</xsd:complexType>
+    <xsd:element name="test" type="KeyIdentifierType"/>
+</xsd:schema>

src/erlsom_pass2.erl

@@ -251,8 +251,16 @@ translateType(Type = #typeInfo{elements=Elemts, attributes=Attrs, extends = Base
           %% debug(BaseEls),
           %% debug(Attrs),
           %% debug(BaseAttrs),
+          % In the case of extensions with xsd:simpleContent the '#text' element is duplicated causing
+          % generated type records failing to compile. Here we filter the duplicates out.
+          DropDupText = fun
+            DropDupText([#elementInfo{alternatives = [#alternative{tag = "#text"}]} = Head | Tail], false) -> [Head | DropDupText(Tail, true)];
+            DropDupText([#elementInfo{alternatives = [#alternative{tag = "#text"}]} | Tail], true) -> DropDupText(Tail, true);
+            DropDupText([#elementInfo{} = Head | Tail], Found) -> [Head | DropDupText(Tail, Found)];
+            DropDupText([], _Found) -> []
+          end,
           NewAnyAttr = if BaseAnyAttr == undefined -> AnyAttr; true -> BaseAnyAttr end,
-          translateType(Type#typeInfo{elements = BaseEls ++ Elemts, %% TODO: will never be 'undefined'?
+          translateType(Type#typeInfo{elements = DropDupText(BaseEls ++ Elemts, false), %% TODO: will never be 'undefined'?
                                       attributes = mergeAttrs(BaseAttrs, Attrs),
                                       anyAttr = NewAnyAttr,
                                       mixed = case Mixed of undefined -> Mixed2; _ -> Mixed end,

test/erlsom_tests.erl

@@ -5,7 +5,7 @@
 %% ------------------------------------------------------------------
 
 -include_lib("eunit/include/eunit.hrl").
-
+-include("src/erlsom_parse.hrl").
 -compile([nowarn_export_all, export_all]).
 
 gexf_test_() ->
@@ -18,10 +18,29 @@ all_test_() ->
                        [])}.
 
 extension_test_() ->
-    {"Test XSD type extensions.",
-     verify_stability_(["extension", "extension.xsd"],
-                       ["extension", "extension.xml"],
-                       [])}.
+    [{"Test XSD type extensions.",
+      verify_stability_(["extension", "extension.xsd"],
+                        ["extension", "extension.xml"],
+                        [])},
+     {"Test simpleContent extension stable",
+      verify_stability_(["extension", "simpleContentExtension.xsd"],
+                        ["extension", "simpleContentExtension.xml"],
+                        [])},
+     {"Test simpleContent #text not duplicated", fun () ->
+        {ok, #model{tps = Types}} = compile_xsd(["extension", "simpleContentExtension.xsd"], []),
+        ok = lists:foreach(fun (Type = #type{els = Els}) ->
+            IsTextEl = fun
+                (#el{alts = [#alt{tag = '#text'}]}) -> true;
+                (_) -> false
+            end,
+            % No duplicate entries for the #text.
+            % Type bellow is only used to get more informative failure.
+            case {Type, lists:filter(IsTextEl, Els)} of
+                {_, []} -> ok;
+                {_, [_]} -> ok
+            end
+        end, Types)
+      end}].
 
 %% @doc
 %% compile the XSD schema file with the given relative path, example: