[Bug-61354] fix issue with extracting text from Word docs.

OpenXmlFormats/Wordprocessing/Document.cs

@@ -78,10 +78,24 @@ namespace NPOI.OpenXmlFormats.Wordprocessing
             if(node == null)
                 return null;
             CT_Document ctObj = new CT_Document();
+            bool firstBody = true;
             foreach(XmlNode childNode in node.ChildNodes)
             {
                 if(childNode.LocalName == "body")
-                    ctObj.body = CT_Body.Parse(childNode, namespaceManager);
+                {
+                    if(firstBody)
+                    {
+                        ctObj.body = CT_Body.Parse(childNode, namespaceManager);
+                        firstBody = false;
+                    }
+                    else
+                    {
+                        if(ctObj.bodyListField == null)
+                            ctObj.bodyListField = new List<CT_Body> ();
+                        ctObj.bodyListField.Add(CT_Body.Parse(childNode, namespaceManager));
+                    }
+                }
+                    
                 else if(childNode.LocalName == "background")
                     ctObj.background = CT_Background.Parse(childNode, namespaceManager);
             }
@@ -111,6 +125,13 @@ namespace NPOI.OpenXmlFormats.Wordprocessing
             sw.Write("mc:Ignorable=\"w14 w15 w16se w16cid w16 w16cex w16sdtdh wp14\">");
             if(this.body != null)
                 this.body.Write(sw, "body");
+            if(this.bodyList != null)
+            {
+                foreach(var b in this.bodyList)
+                {
+                    b.Write(sw, "body");
+                }
+            }
             if(this.background != null)
                 this.background.Write(sw, "background");
             sw.Write("</w:document>");
@@ -123,6 +144,8 @@ namespace NPOI.OpenXmlFormats.Wordprocessing
             this.bodyField = new CT_Body();
         }
 
+        private List<CT_Body> bodyListField;
+
         [XmlElement(Order = 0)]
         public CT_Body body
         {
@@ -136,6 +159,14 @@ namespace NPOI.OpenXmlFormats.Wordprocessing
             }
         }
 
+        public List<CT_Body> bodyList
+        {
+            get
+            {
+                return bodyListField;
+            }
+        }
+
         public void AddNewBody()
         {
             this.bodyField = new CT_Body();

ooxml/XWPF/Usermodel/XWPFDocument.cs

@@ -103,26 +103,31 @@ namespace NPOI.XWPF.UserModel
                 InitFootnotes();
                 // parse the document with cursor and add
                 //    // the XmlObject to its lists
-
-                foreach (object o in ctDocument.body.Items)
+                List<CT_Body> allBody = [ctDocument.body];
+                if(ctDocument.bodyList != null)
+                    allBody.AddRange(ctDocument.bodyList);
+                foreach(CT_Body body in allBody)
                 {
-                    if (o is CT_P ctP)
-                    {
-                        XWPFParagraph p = new XWPFParagraph(ctP, this);
-                        bodyElements.Add(p);
-                        paragraphs.Add(p);
-                    }
-                    else if (o is CT_Tbl tbl)
+                    foreach (object o in body.Items)
                     {
-                        XWPFTable t = new XWPFTable(tbl, this);
-                        bodyElements.Add(t);
-                        tables.Add(t);
-                    }
-                    else if (o is CT_SdtBlock block)
-                    {
-                        XWPFSDT c = new XWPFSDT(block, this);
-                        bodyElements.Add(c);
-                        contentControls.Add(c);
+                        if (o is CT_P ctP)
+                        {
+                            XWPFParagraph p = new XWPFParagraph(ctP, this);
+                            bodyElements.Add(p);
+                            paragraphs.Add(p);
+                        }
+                        else if (o is CT_Tbl tbl)
+                        {
+                            XWPFTable t = new XWPFTable(tbl, this);
+                            bodyElements.Add(t);
+                            tables.Add(t);
+                        }
+                        else if (o is CT_SdtBlock block)
+                        {
+                            XWPFSDT c = new XWPFSDT(block, this);
+                            bodyElements.Add(c);
+                            contentControls.Add(c);
+                        }
                     }
                 }
                 // Sort out headers and footers

testcases/ooxml/XWPF/Extractor/TestXWPFWordExtractor.cs

@@ -425,5 +425,16 @@ namespace TestCases.XWPF.Extractor
             extractor.Close();
         }
 
+        [Test]
+        public void TestMultipleBodyBug()
+        {
+            XWPFDocument doc = XWPFTestDataSamples.OpenSampleDocument("MultipleBodyBug.docx");
+            XWPFWordExtractor extractor = new XWPFWordExtractor(doc);
+            ClassicAssert.AreEqual("START BODY 1 The quick, brown fox jumps over a lazy dog. END BODY 1.\n"
+                            + "START BODY 2 The quick, brown fox jumps over a lazy dog. END BODY 2.\n"
+                            + "START BODY 3 The quick, brown fox jumps over a lazy dog. END BODY 3.\n",
+                    extractor.Text);
+            extractor.Close();
+        }
     }
 }

testcases/test-data/document/MultipleBodyBug.docx

@@ -0,0 +1,4 @@
+
+START BODY 1 The quick, brown fox jumps over a lazy dog. END BODY 1.
+START BODY 2 The quick, brown fox jumps over a lazy dog. END BODY 2.
+START BODY 3 The quick, brown fox jumps over a lazy dog. END BODY 3.