fixup offset issues

SMI · Aug 8, 2023 · a91c0b9 · a91c0b9
1 parent 8ceef12
commit a91c0b9
Show file tree

Hide file tree

Showing 2 changed files with 36 additions and 8 deletions.
diff --git a/IsIdentifiable/Reporting/Reports/FailureStoreReport.cs b/IsIdentifiable/Reporting/Reports/FailureStoreReport.cs
@@ -173,8 +173,15 @@ public static IEnumerable<Failure> Deserialize(IFileInfo oldFile, Action<int> lo
                 if (!string.IsNullOrWhiteSpace(partRule.IfColumn) && !string.Equals(partRule.IfColumn, problemField, StringComparison.InvariantCultureIgnoreCase))
                     continue;
 
-                foreach (var part in parts.Where(x => partRule.Covers(x, problemValue)))
-                    toRemove.Add(part);
+                foreach (var part in parts)
+                {
+                    var origOffset = part.Offset;
+                    if (partRule.Covers(part, problemValue))
+                    {
+                        part.Offset = origOffset;
+                        toRemove.Add(part);
+                    }
+                }
             }
             parts = parts.Except(toRemove);
             /* TEMP */

diff --git a/IsIdentifiable/Rules/RegexRule_PartTemp.cs b/IsIdentifiable/Rules/RegexRule_PartTemp.cs
@@ -57,18 +57,38 @@ private void RebuildPartRegex()
         IfPartPatternRegex = new Regex(_ifPartPatternString, (CaseSensitive ? RegexOptions.None : RegexOptions.IgnoreCase) | RegexOptions.Compiled);
     }
 
+    private static void FixupOffset(FailurePart failurePart, string problemValue)
+    {
+        if (problemValue.Substring(failurePart.Offset, failurePart.Word.Length) == failurePart.Word)
+            return;
+
+        // Try looking ahead first, then back
+        var origOffset = failurePart.Offset;
+        try
+        {
+            while (problemValue.Substring(failurePart.Offset, failurePart.Word.Length) != failurePart.Word)
+                failurePart.Offset++;
+        }
+        catch (ArgumentOutOfRangeException)
+        {
+            failurePart.Offset = origOffset;
+            while (problemValue.Substring(failurePart.Offset, failurePart.Word.Length) != failurePart.Word)
+                failurePart.Offset--;
+        }
+    }
+
     public bool Covers(FailurePart failurePart, string problemValue)
     {
         if (As != failurePart.Classification)
             return false;
 
+        // Fixes any offsets that have been mangled by file endings etc.
+        FixupOffset(failurePart, problemValue);
+
         bool matchesBefore = false;
         if (!string.IsNullOrWhiteSpace(WordBefore))
         {
             var problemValueUpToOffset = problemValue[..(failurePart.Offset + failurePart.Word.Length)];
-            if (!problemValueUpToOffset.EndsWith(failurePart.Word))
-                throw new Exception("Invlaid data: actual word and word at offset did not match");
-
             var wordBeforeRegex = new Regex($"\\b{WordBefore}\\s+{IfPartPattern.TrimStart('^')}", (CaseSensitive ? RegexOptions.None : RegexOptions.IgnoreCase));
             matchesBefore = wordBeforeRegex.Matches(problemValueUpToOffset).Any();
         }
@@ -77,9 +97,6 @@ public bool Covers(FailurePart failurePart, string problemValue)
         if (!string.IsNullOrWhiteSpace(WordAfter))
         {
             var problemValueFromOffset = problemValue[failurePart.Offset..];
-            if (!problemValueFromOffset.StartsWith(failurePart.Word))
-                throw new Exception("Invlaid data: actual word and word at offset did not match");
-
             var wordAfterRegex = new Regex($"{IfPartPattern.TrimEnd('$')}\\s+{WordAfter}\\b", (CaseSensitive ? RegexOptions.None : RegexOptions.IgnoreCase));
             matchesAfter = wordAfterRegex.Matches(problemValueFromOffset).Any();
         }
@@ -92,6 +109,10 @@ public bool Covers(FailurePart failurePart, string problemValue)
         {
             return true;
         }
+        else if (!string.IsNullOrWhiteSpace(WordBefore) || !string.IsNullOrWhiteSpace(WordAfter))
+        {
+            return false;
+        }
 
         return IfPartPatternRegex.Matches(failurePart.Word).Any();
     }