Merge pull request #1524 from Kazbek/master

Fix XWPFDocument FindAndReplaceTextInParagraph

ooxml/XWPF/Usermodel/XWPFDocument.cs

@@ -1882,44 +1882,72 @@ namespace NPOI.XWPF.UserModel
             return this;
         }
 
-        private static void FindAndReplaceTextInParagraph(XWPFParagraph paragraph, string oldValue, string newValue)
+        private static void FindAndReplaceTextInParagraph(XWPFParagraph paragraph, string oldValue, string newValue, int startPos = 0)
         {
-            var index = paragraph?.Text.IndexOf(oldValue) ?? -1;
-            if (index != -1)
-            {
-                int? firstIndex = null;
-                int? lastIndex = null;
-                for (var i = 0; i < paragraph.Runs.Count; ++i)
-                {
-                    if (index < paragraph.Runs[i].Text.Length)
-                    {
-                        if (-index <= oldValue.Length)
+            if(paragraph == null)
+                return;
+
+            string paragraphText = string.Concat(paragraph.Runs.Select(p => p.Text));
+
+            var startIndex = paragraphText.IndexOf(oldValue, startPos);
+            if(startIndex == -1)
+                return;
+
+            int firstRun = -1;
+            int firstIndex = -1;
+
+            int lastRun = -1;
+            int lastIndex = -1;
+
+            int processedRuns = 0;
+            int processedChars = 0;
+
+            for(; processedRuns < paragraph.Runs.Count; processedRuns++)
             {
-                            if (firstIndex == null)
+                var text = paragraph.Runs[processedRuns].Text;
+                if(processedChars + text.Length > startIndex)
                 {
-                                firstIndex = i;
+                    firstRun = processedRuns;
+                    firstIndex = startIndex - processedChars;
+                    break;
                 }
+
+                processedChars += text.Length;
             }
-                        else
+
+            int endIndex = startIndex + oldValue.Length;
+
+            for(; processedRuns < paragraph.Runs.Count; processedRuns++)
+            {
+                var text = paragraph.Runs[processedRuns].Text;
+                if(processedChars + text.Length > endIndex)
                 {
-                            lastIndex = i;
+                    lastRun = processedRuns;
+                    lastIndex = endIndex - processedChars;
                     break;
                 }
+
+                processedChars += text.Length;
             }
-                    index -= paragraph.Runs[i].Text.Length;
-                }
-                if (lastIndex == null)
+
+            var initialFirstText = paragraph.Runs[firstRun].Text;
+            if(firstRun == lastRun)
             {
-                    lastIndex = paragraph.Runs.Count - 1;
+                paragraph.Runs[firstRun].SetText(initialFirstText.Substring(0, firstIndex) + newValue + initialFirstText.Substring(lastIndex));
             }
-                var newText = String.Concat(paragraph.Runs.Skip(firstIndex ?? 0).Take((lastIndex ?? 0) - (firstIndex ?? 0) + 1).Select(_ => _.Text));
-                newText = newText.Replace(oldValue, newValue);
-                paragraph.Runs[firstIndex ?? 0].SetText(newText);
-                for (var i = (firstIndex ?? 0) + 1; i <= lastIndex; ++i)
+            else
             {
-                    paragraph.RemoveRun((firstIndex ?? 0) + 1);
-                }
+                paragraph.Runs[firstRun].SetText(initialFirstText.Substring(0, firstIndex) + newValue);
+
+                if(lastRun != -1)
+                    paragraph.Runs[lastRun].SetText(paragraph.Runs[lastRun].Text.Substring(lastIndex));
             }
+
+            int removeTo = lastRun == -1 ? paragraph.Runs.Count : lastRun;
+            for(int i = firstRun + 1; i < removeTo; i++)
+                paragraph.RemoveRun(firstRun + 1);
+
+            FindAndReplaceTextInParagraph(paragraph, oldValue, newValue, startIndex + newValue.Length);
         }
 
         private static void FindAndReplaceTextInTable(XWPFTable table, string oldValue, string newValue)

testcases/ooxml/XWPF/UserModel/TestXWPFDocument.cs

@@ -150,7 +150,7 @@ namespace TestCases.XWPF.UserModel
             doc.FindAndReplaceText("$replace_cell_text$", "Regel1\nRegel2\nRegel3");
 
             //Save Word Document
-            XWPFDocument outputDocument = outputDocument = XWPFTestDataSamples.WriteOutAndReadBack(doc);
+            XWPFDocument outputDocument = XWPFTestDataSamples.WriteOutAndReadBack(doc);
 
             //Combine all runs of all paragraphs
             StringBuilder builder = new StringBuilder();
@@ -184,6 +184,44 @@ namespace TestCases.XWPF.UserModel
 
         }
 
+        [Test]
+        public void FindAndReplaceTextInParagraph()
+        {
+            XWPFDocument doc = XWPFTestDataSamples.OpenSampleDocument("WordFindAndReplaceTextInParagraph.docx");
+            string initialText = string.Concat(doc.Paragraphs.Select(t => t.Text));
+
+            Dictionary<string, string> replacers = new Dictionary<string, string>
+            {
+                {"$FINALQUALIFYINGWORK_QUESTION_1_ASKING_SHORT$", "Asking1" },
+                {"$FINALQUALIFYINGWORK_QUESTION_1_QUESTION$", "Question1" },
+                {"$FINALQUALIFYINGWORK_QUESTION_2_ASKING_SHORT$", "Asking2" },
+                {"$FINALQUALIFYINGWORK_QUESTION_2_QUESTION$", "Question2" },
+                {"$STUDENT_FULL$", "На русском с пробелами" },
+                {"$FINALQUALIFYINGWORK_GRADE$", "5" },
+                {"$SIMPLE$", "Last text" },
+                {"$DOUBLE$", "Twice" },
+            };
+
+            //This is calling FindAndReplaceTextInParagraph for each paragraph in document
+            foreach(var replacer in replacers)
+                doc.FindAndReplaceText(replacer.Key, replacer.Value);
+            
+            //Save Word Document
+            XWPFDocument outputDocument = XWPFTestDataSamples.WriteOutAndReadBack(doc);
+
+            foreach(var replacer in replacers)
+                initialText = initialText.Replace(replacer.Key, replacer.Value);
+
+            string savedText = string.Concat(outputDocument.Paragraphs.Select(t => t.Text));
+
+            //Check that at least replacing in string equal to result file
+            ClassicAssert.AreEqual(initialText, savedText);
+
+            //Check
+            ClassicAssert.AreEqual("Some initial text на разный манер (inserted) and so on:Asking1: Question1Asking2: Question2Result on:1. Say that На русском с пробелами with a very long sentence and one more replacer in the end for (русский язык) sure 5Triple replace with TwiceTwice и ещё одним Twice повторением.Last text", 
+                savedText);
+        }
+
         [Test]
         public void TestAddPicture()
         {

testcases/test-data/document/WordFindAndReplaceTextInParagraph.docx

@@ -0,0 +1,11 @@
+Some initial text на разный манер (inserted) and so on:
+
+$FINALQUALIFYINGWORK_QUESTION_1_ASKING_SHORT$: $FINALQUALIFYINGWORK_QUESTION_1_QUESTION$
+$FINALQUALIFYINGWORK_QUESTION_2_ASKING_SHORT$: $FINALQUALIFYINGWORK_QUESTION_2_QUESTION$
+
+Result on:
+1. Say that $STUDENT_FULL$ with a very long sentence and one more replacer in the end for (русский язык) sure $FINALQUALIFYINGWORK_GRADE$
+                                       
+Triple replace with $DOUBLE$$DOUBLE$ и ещё одним $DOUBLE$ повторением.
+                                       
+$SIMPLE$