Add update block error handling

src/SIL.Machine/Corpora/IUsfmUpdateBlockHandler.cs

@@ -0,0 +1,26 @@
+using System;
+
+namespace SIL.Machine.Corpora
+{
+    public interface IUsfmUpdateBlockHandler
+    {
+        UsfmUpdateBlock ProcessBlock(UsfmUpdateBlock block);
+    }
+
+    public class UsfmUpdateBlockHandlerException : Exception
+    {
+        public UsfmUpdateBlock Block { get; }
+
+        public UsfmUpdateBlockHandlerException(string message, UsfmUpdateBlock block)
+            : base(message)
+        {
+            Block = block;
+        }
+
+        public UsfmUpdateBlockHandlerException(string message, Exception exception, UsfmUpdateBlock block)
+            : base(message, exception)
+        {
+            Block = block;
+        }
+    }
+}

src/SIL.Machine/Corpora/PlaceMarkersUsfmUpdateBlockHandler.cs

@@ -112,7 +112,6 @@ public UsfmUpdateBlock ProcessBlock(UsfmUpdateBlock block)
             IReadOnlyList<string> targetTokens = alignmentInfo.TranslationTokens;
             int sourceTokenIndex = 0;
 
-            string sourceSentence = "";
             string targetSentence = "";
             var toPlace = new List<UsfmUpdateBlockElement>();
             var adjacentSourceTokens = new List<int>();
@@ -123,16 +122,9 @@ public UsfmUpdateBlock ProcessBlock(UsfmUpdateBlock block)
             {
                 if (element.Type == UsfmUpdateBlockElementType.Text)
                 {
-                    if (
-                        element.MarkedForRemoval
-                        || (
-                            element.Type == UsfmUpdateBlockElementType.Paragraph
-                            && alignmentInfo.ParagraphBehavior == UpdateUsfmMarkerBehavior.Strip
-                        )
-                    )
+                    if (element.MarkedForRemoval)
                     {
                         string text = element.Tokens[0].ToUsfm();
-                        sourceSentence += text;
 
                         // Track seen tokens
                         while (sourceTokenIndex < sourceTokens.Count && text.Contains(sourceTokens[sourceTokenIndex]))
@@ -152,7 +144,13 @@ public UsfmUpdateBlock ProcessBlock(UsfmUpdateBlock block)
                     }
                 }
 
-                if (element.MarkedForRemoval)
+                if (
+                    element.MarkedForRemoval
+                    || (
+                        element.Type == UsfmUpdateBlockElementType.Paragraph
+                        && alignmentInfo.ParagraphBehavior == UpdateUsfmMarkerBehavior.Strip
+                    )
+                )
                 {
                     ignoredElements.Add(element);
                 }
@@ -174,7 +172,18 @@ public UsfmUpdateBlock ProcessBlock(UsfmUpdateBlock block)
             int prevLength = 0;
             foreach (string token in targetTokens)
             {
-                targetTokenStarts.Add(targetSentence.IndexOf(token, targetTokenStarts.LastOrDefault() + prevLength));
+                int indexOfTargetTokenInSentence = targetSentence.IndexOf(
+                    token,
+                    targetTokenStarts.LastOrDefault() + prevLength
+                );
+                if (indexOfTargetTokenInSentence == -1)
+                {
+                    throw new UsfmUpdateBlockHandlerException(
+                        $"No token \"{token}\" found in text \"{targetSentence}\" at or beyond index {targetTokenStarts.LastOrDefault() + prevLength}. Is the versification correctly specified?",
+                        block
+                    );
+                }
+                targetTokenStarts.Add(indexOfTargetTokenInSentence);
                 prevLength = token.Length;
             }
 

src/SIL.Machine/Corpora/UpdateUsfmParserHandler.cs

@@ -57,6 +57,7 @@ public class UpdateUsfmParserHandler : ScriptureRefUsfmParserHandlerBase
         private readonly Stack<bool> _replace;
         private int _rowIndex;
         private int _tokenIndex;
+        private readonly Func<UsfmUpdateBlockHandlerException, bool> _errorHandler;
 
         public UpdateUsfmParserHandler(
             IReadOnlyList<UpdateUsfmRow> rows = null,
@@ -67,7 +68,8 @@ public UpdateUsfmParserHandler(
             UpdateUsfmMarkerBehavior styleBehavior = UpdateUsfmMarkerBehavior.Strip,
             IEnumerable<string> preserveParagraphStyles = null,
             IEnumerable<IUsfmUpdateBlockHandler> updateBlockHandlers = null,
-            IEnumerable<string> remarks = null
+            IEnumerable<string> remarks = null,
+            Func<UsfmUpdateBlockHandlerException, bool> errorHandler = null
         )
         {
             _rows = rows ?? Array.Empty<UpdateUsfmRow>();
@@ -90,6 +92,9 @@ public UpdateUsfmParserHandler(
                     ? new HashSet<string> { "r", "rem" }
                     : new HashSet<string>(preserveParagraphStyles);
             _remarks = remarks == null ? new List<string>() : remarks.ToList();
+            _errorHandler = errorHandler;
+            if (_errorHandler == null)
+                _errorHandler = (error) => false;
         }
 
         public IReadOnlyList<UsfmToken> Tokens => _tokens;
@@ -576,7 +581,16 @@ private void EndUpdateBlock(UsfmParserState state, IReadOnlyList<ScriptureRef> s
 
             foreach (IUsfmUpdateBlockHandler handler in _updateBlockHandlers)
             {
-                updateBlock = handler.ProcessBlock(updateBlock);
+                try
+                {
+                    updateBlock = handler.ProcessBlock(updateBlock);
+                }
+                catch (UsfmUpdateBlockHandlerException e)
+                {
+                    bool shouldContinue = _errorHandler(e);
+                    if (!shouldContinue)
+                        throw;
+                }
             }
             List<UsfmToken> tokens = updateBlock.GetTokens();
             foreach (UsfmUpdateBlockElement elem in Enumerable.Reverse(paraElems))