support words before and after the FailurePart

SMI · Aug 7, 2023 · 8ceef12 · 8ceef12
1 parent a86f02a
commit 8ceef12
Show file tree

Hide file tree

Showing 2 changed files with 45 additions and 10 deletions.
diff --git a/IsIdentifiable/Reporting/Reports/FailureStoreReport.cs b/IsIdentifiable/Reporting/Reports/FailureStoreReport.cs
@@ -153,6 +153,7 @@ public static IEnumerable<Failure> Deserialize(IFileInfo oldFile, Action<int> lo
             token.ThrowIfCancellationRequested();
             lineNumber++;
             var problemField = r["ProblemField"];
+            var problemValue = r["ProblemValue"];
             var words = r["PartWords"].Split(Separator);
             var classes = r["PartClassifications"].Split(Separator);
             var offsets = r["PartOffsets"].Split(Separator);
@@ -172,7 +173,7 @@ 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)))
+                foreach (var part in parts.Where(x => partRule.Covers(x, problemValue)))
                     toRemove.Add(part);
             }
             parts = parts.Except(toRemove);
@@ -184,7 +185,7 @@ public static IEnumerable<Failure> Deserialize(IFileInfo oldFile, Action<int> lo
                     Resource = r["Resource"],
                     ResourcePrimaryKey = r["ResourcePrimaryKey"],
                     ProblemField = problemField,
-                    ProblemValue = r["ProblemValue"],
+                    ProblemValue = problemValue,
                 };
 
             if (lineNumber % 1000 == 0)

diff --git a/IsIdentifiable/Rules/RegexRule_PartTemp.cs b/IsIdentifiable/Rules/RegexRule_PartTemp.cs
@@ -41,24 +41,58 @@ public override bool CaseSensitive
         }
     }
 
+    public string WordBefore { get; set; }
+
+    public string WordAfter { get; set; }
+
     // TODO(rkm 2023-07-25) Shouldn't be needed when IfPattern is readonly
     private void RebuildPartRegex()
     {
-        if (!_ifPartPatternString.StartsWith("^") || _ifPartPatternString.EndsWith("$"))
+        if (_ifPartPatternString == null)
+            throw new Exception("Illegal rule setup. You must specify IfPartPattern");
+
         if (!_ifPartPatternString.StartsWith("^") || !_ifPartPatternString.EndsWith("$"))
             throw new ArgumentException("IfPartPattern must be enclosed by ^ and $");
-        IfPartPatternRegex = _ifPartPatternString == null ? null : new Regex(_ifPartPatternString, (CaseSensitive ? RegexOptions.None : RegexOptions.IgnoreCase) | RegexOptions.Compiled);
+
+        IfPartPatternRegex = new Regex(_ifPartPatternString, (CaseSensitive ? RegexOptions.None : RegexOptions.IgnoreCase) | RegexOptions.Compiled);
     }
 
-    public bool Covers(FailurePart failurePart)
+    public bool Covers(FailurePart failurePart, string problemValue)
     {
-        if (IfPartPattern == null)
-            throw new Exception("Illegal rule setup. You must specify IfPartPattern");
-
         if (As != failurePart.Classification)
             return false;
 
-        var matches = IfPartPatternRegex.Matches(failurePart.Word);
-        return matches.Any();
+        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($"\\b{WordBefore}\\s+{IfPartPattern.TrimStart('^')}", (CaseSensitive ? RegexOptions.None : RegexOptions.IgnoreCase));
+            matchesBefore = wordBeforeRegex.Matches(problemValueUpToOffset).Any();
+        }
+
+        bool matchesAfter = false;
+        if (!string.IsNullOrWhiteSpace(WordAfter))
+        {
+            var problemValueFromOffset = problemValue[failurePart.Offset..];
+            if (!problemValueFromOffset.StartsWith(failurePart.Word))
+                throw new Exception("Invlaid data: actual word and word at offset did not match");
+
+            var wordAfterRegex = new Regex($"{IfPartPattern.TrimEnd('$')}\\s+{WordAfter}\\b", (CaseSensitive ? RegexOptions.None : RegexOptions.IgnoreCase));
+            matchesAfter = wordAfterRegex.Matches(problemValueFromOffset).Any();
+        }
+
+        if (
+            matchesBefore && string.IsNullOrWhiteSpace(WordAfter) ||
+            matchesAfter && string.IsNullOrWhiteSpace(WordBefore) ||
+            (matchesBefore && matchesAfter)
+        )
+        {
+            return true;
+        }
+
+        return IfPartPatternRegex.Matches(failurePart.Word).Any();
     }
 }