Bug fix/document assembler paragraph properties not copied from template

Clippit.Tests/Word/DocumentAssemblerTests.cs

@@ -179,6 +179,89 @@ out var returnedTemplateError
             Assert.Equal(err, returnedTemplateError);
         }
 
+        [Theory]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-no-inline-styles.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-bold.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-strong.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-italic.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-emphasis.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-underline.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-bold-underline.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-bold-italic.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-italic-underline.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-subscript.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-superscript.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-strikethrough.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-hyperlink.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-hyperlink-bold.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-hyperlink-italic.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-hyperlink-underline.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-hyperlink-no-protocol.xml", 1, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-multi-paragraph.xml", 3, false)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-invalid.xml", 0, true)]
+        [InlineData("DA289-xhtml-formatting.docx", "DA289-not-well-formed.xml", 0, true)]
+        public void DA289(string name, string data, int parasInContent, bool err)
+        {
+            var templateDocx = new FileInfo(Path.Combine(_sourceDir.FullName, name));
+            var dataFile = new FileInfo(Path.Combine(_sourceDir.FullName, data));
+
+            var wmlTemplate = new WmlDocument(templateDocx.FullName);
+            var xmlData = XElement.Load(dataFile.FullName);
+
+            var wmlResult = DocumentAssembler.AssembleDocument(wmlTemplate, xmlData, out var returnedTemplateError);
+            var assembledDocx = new FileInfo(
+                Path.Combine(TempDir, data.Replace(".xml", "-processed-by-DocumentAssembler.docx"))
+            );
+            wmlResult.SaveAs(assembledDocx.FullName);
+
+            Validate(assembledDocx);
+            Assert.Equal(err, returnedTemplateError);
+
+            // if we are not expecting an error then verify that we have the same number of paragraphs and that
+            // the paragraph properties from source and target are the same
+            if (!err)
+            {
+                IList<XElement> sourceParas = wmlTemplate.MainDocumentPart.Element(W.body).Descendants(W.p).ToList();
+                IList<XElement> targetParas = wmlResult.MainDocumentPart.Element(W.body).Descendants(W.p).ToList();
+
+                // Check we have the expected number of paragraphs
+                // Expected document structure is:
+                //   Heading paragraph (1 line)
+                //   Empty paragraph (1 line)
+                //   Escaped HTML paragraph (potential multi-line)
+                //   CDATA paragraph (potential multi-line)
+
+                int expectedParas = sourceParas.Count + (2 * parasInContent) - 2;
+                Assert.Equal(expectedParas, targetParas.Count);
+
+                var equalityComparer = new XNodeEqualityComparer();
+                int paraOffset = 0;
+
+                for (var i = 0; i < sourceParas.Count(); i++)
+                {
+                    var parasToCompare = i <= 1 ? 1 : parasInContent;
+                    var sourceProps = sourceParas[i].Element(W.pPr);
+
+                    for (var j = i + paraOffset; j < i + paraOffset + parasToCompare; j++)
+                    {
+                        var targetProps = targetParas[j].Element(W.pPr);
+                        if (sourceProps == null && targetProps == null)
+                        {
+                            continue;
+                        }
+
+                        Assert.True(equalityComparer.Equals(sourceProps, targetProps));
+                    }
+
+                    // update paragraph offset versus source when we have processed multi-line content
+                    if (parasToCompare > 1)
+                    {
+                        paraOffset += parasToCompare - 1;
+                    }
+                }
+            }
+        }
+
         [Theory]
         [InlineData("DA259-MultiLineContents.docx", "DA-Data.xml", false)]
         public void DA259(string name, string data, bool err)

Clippit/PtOpenXmlUtil.cs

@@ -5870,7 +5870,7 @@ public static class XhtmlNoNamespace
         public static XName ol = "ol";
         public static XName ul = "ul";
         public static XName li = "li";
-        public static XName strong = "Bold";
+        public static XName strong = "strong";
         public static XName em = "em";
         public static XName tbody = "tbody";
         public static XName valign = "valign";

Clippit/Word/DocumentAssembler.cs

@@ -1298,7 +1298,8 @@ OpenXmlPart part
                     return element.CreateContextErrorMessage($"Content: {ex.Message}", templateError);
                 }
 
-                // get XElements and ensure all but the first element is in a
+                // get XElements and ensure all but the first element is in a paragraph
+
                 List<XElement> elements = new List<XElement>();
                 for (int i = 0; i < content.Count; i++)
                 {
@@ -1308,12 +1309,25 @@ OpenXmlPart part
                         var objEl = obj as XElement;
                         if (i > 0 && objEl.Name == W.r || objEl.Name == W.hyperlink)
                         {
-                            elements.Add(new XElement(W.p, currentParaProps, content[i]));
+                            objEl = new XElement(W.p, currentParaProps, content[i]);
                         }
-                        else
+                        else if (objEl.Name == W.p)
                         {
-                            elements.Add(objEl);
+                            // get the processed paragraph properties
+                            XElement pProps = objEl.Descendants(W.pPr).FirstOrDefault();
+                            if (pProps != null)
+                            {
+                                pProps.Remove();
+                            }
+
+                            // add the current paragraph properties
+                            if (currentParaProps != null)
+                            {
+                                objEl.AddFirst(currentParaProps);
+                            }
                         }
+
+                        elements.Add(objEl);
                     }
                 }
 

TestFiles/DA/DA289-bold-italic.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Bold and Italic</TestCase>
+	<EscapedHTML>&lt;p&gt;This content is escaped &lt;b&gt;&lt;i&gt;HTML&lt;/i&gt;&lt;/b&gt;&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>This content comes from <b><i>CDATA</i></b></p>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-bold-underline.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Bold and Underline</TestCase>
+	<EscapedHTML>&lt;p&gt;This content is escaped &lt;b&gt;&lt;u&gt;HTML&lt;/u&gt;&lt;/b&gt;&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>This content comes from <b><u>CDATA</u></b></p>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-bold.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Bold</TestCase>
+	<EscapedHTML>&lt;p&gt;This content is escaped &lt;b&gt;HTML&lt;/b&gt;&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>This content comes from <b>CDATA</b></p>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-emphasis.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Emphasis</TestCase>
+	<EscapedHTML>&lt;p&gt;This content is escaped &lt;em&gt;HTML&lt;/em&gt;&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>This content comes from <em>CDATA</em></p>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-hyperlink-bold.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Hyperlink and Bold</TestCase>
+	<EscapedHTML>&lt;b&gt;&lt;a href="https://www.bbc.co.uk"&gt;BBC&lt;/a&gt;&lt;/b&gt;</EscapedHTML>
+	<CDATA><![CDATA[<b><a href="https://www.bbc.co.uk">BBC</a></b>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-hyperlink-italic.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Hyperlink and Italic</TestCase>
+	<EscapedHTML>&lt;i&gt;&lt;a href="https://www.bbc.co.uk"&gt;BBC&lt;/a&gt;&lt;/i&gt;</EscapedHTML>
+	<CDATA><![CDATA[<i><a href="https://www.bbc.co.uk">BBC</a></i>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-hyperlink-no-protocol.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Hyperlink with No Protocol</TestCase>
+	<EscapedHTML>&lt;a href="www.bbc.co.uk"&gt;BBC&lt;/a&gt;</EscapedHTML>
+	<CDATA><![CDATA[<a href="www.bbc.co.uk">BBC</a>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-hyperlink-underline.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Hyperlink and Underline</TestCase>
+	<EscapedHTML>&lt;u&gt;&lt;a href="https://www.bbc.co.uk"&gt;BBC&lt;/a&gt;&lt;/u&gt;</EscapedHTML>
+	<CDATA><![CDATA[<u><a href="https://www.bbc.co.uk">BBC</a></u>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-hyperlink.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Hyperlink</TestCase>
+	<EscapedHTML>&lt;a href="https://www.bbc.co.uk"&gt;BBC&lt;/a&gt;</EscapedHTML>
+	<CDATA><![CDATA[<a href="https://www.bbc.co.uk">BBC</a>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-invalid.xml

@@ -0,0 +1,6 @@
+<Doc>
+    <TestCase>Invalid XHTML</TestCase>
+	<EscapedHTML>This is NOT valid XHTML.&lt;br&gt;But is Valid HTML.</EscapedHTML>
+	<!-- Invalid XHTML -->
+	<CDATA><![CDATA[This is NOT valid XHTML.<br>But is Valid HTML.]]></CDATA>
+</Doc>

TestFiles/DA/DA289-italic-underline.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Italic and Underline</TestCase>
+	<EscapedHTML>&lt;p&gt;This content is escaped &lt;i&gt;&lt;u&gt;HTML&lt;/u&gt;&lt;/i&gt;&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>This content comes from <i><u>CDATA</u></i></p>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-italic.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Italic</TestCase>
+	<EscapedHTML>&lt;p&gt;This content is escaped &lt;i&gt;HTML&lt;/i&gt;&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>This content comes from <i>CDATA</i></p>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-multi-paragraph.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Multiple Paragraphs</TestCase>
+	<EscapedHTML>&lt;p&gt;Boggarts lavender robes, Hermione Granger &lt;i&gt;Fantastic Beasts&lt;/i&gt; and Where to Find Them.&lt;/p&gt;&lt;p&gt;Bee in your bonnet Hand of Glory elder wand, spectacles &lt;b&gt;House Cup&lt;/b&gt; Bertie Bott’s Every Flavor Beans Impedimenta.&lt;/p&gt;&lt;p&gt;Stunning spells tap-dancing spider Slytherin’s Heir &lt;u&gt;mewing kittens&lt;/u&gt; &lt;s&gt;Remus Lupin&lt;/s&gt;.&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>Boggarts lavender robes, Hermione Granger <i>Fantastic Beasts</i> and Where to Find Them.</p><p>Bee in your bonnet Hand of Glory elder wand, spectacles <b>House Cup</b> Bertie Bott’s Every Flavor Beans Impedimenta.</p><p>Stunning spells tap-dancing spider Slytherin’s Heir <u>mewing kittens</u> <s>Remus Lupin</s>.</p>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-no-inline-styles.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>No Inline Styles</TestCase>
+	<EscapedHTML>&lt;p&gt;This content is escaped HTML&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>This content comes from CDATA</p>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-not-well-formed.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Not Well Formed XHTML</TestCase>
+	<EscapedHTML>&lt;b&gt;&lt;i&gt;This is NOT well formed XHTML.&lt;/b&gt;&lt;/i&gt;</EscapedHTML>
+	<CDATA><![CDATA[<b><i>This is NOT well formed XHTML.</b></i>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-strikethrough.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Strikethrough</TestCase>
+	<EscapedHTML>&lt;p&gt;This content is escaped &lt;s&gt;HTML&lt;/s&gt;&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>This content comes from <s>CDATA</s></p>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-strong.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Strong</TestCase>
+	<EscapedHTML>&lt;p&gt;This content is escaped &lt;strong&gt;HTML&lt;/strong&gt;&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>This content comes from <strong>CDATA</strong></p>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-subscript.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Subscript</TestCase>
+	<EscapedHTML>&lt;p&gt;This content is escaped &lt;sub&gt;HTML&lt;/sub&gt;&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>This content comes from <sub>CDATA</sub></p>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-superscript.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Superscript</TestCase>
+	<EscapedHTML>&lt;p&gt;This content is escaped &lt;sup&gt;HTML&lt;/sup&gt;&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>This content comes from <sup>CDATA</sup></p>]]></CDATA>
+</Doc>

TestFiles/DA/DA289-underline.xml

@@ -0,0 +1,5 @@
+<Doc>
+	<TestCase>Underline</TestCase>
+	<EscapedHTML>&lt;p&gt;This content is escaped &lt;u&gt;HTML&lt;/u&gt;&lt;/p&gt;</EscapedHTML>
+	<CDATA><![CDATA[<p>This content comes from <u>CDATA</u></p>]]></CDATA>
+</Doc>