Get v2 -> v1 transform working with files array

src/Sarif.UnitTests/TestData/SarifCurrentToVersionOneVisitor/v1/OneRunWithBasicInvocation.sarif

@@ -27,9 +27,7 @@
         }
       },
       "files": {
-        "collections.rsp": {
-          "uri": "collections.rsp"
-        }
+        "collections.rsp": {}
       },
       "results": []
     }

src/Sarif.UnitTests/TestData/SarifCurrentToVersionOneVisitor/v1/OneRunWithFiles.sarif

@@ -27,7 +27,7 @@
           }
         },
         "file:///home/buildAgent/bin/app.zip#/docs/intro.docx": {
-          "uri": "file:///docs/intro.docx",
+          "uri": "/docs/intro.docx",
           "parentKey": "file:///home/buildAgent/bin/app.zip",
           "offset": 17522,
           "length": 4050,

src/Sarif.UnitTests/TestData/SarifCurrentToVersionOneVisitor/v1/Regions.sarif

@@ -11,16 +11,13 @@
       "invocation": {},
       "files": {
         "file:///src/base/driver.cs": {
-          "uri": "file:///src/base/driver.cs",
           "contents": "YQBiAGMAZAANAAoAZQBmAGcADQAKAGgAaQBqAGsADQAKAGwAbQBuAA=="
         },
         "file:///src/ui/client.cs": {
-          "uri": "file:///src/ui/client.cs",
           "mimeType": "text/x-csharp",
           "contents": "VGhlIHF1aWNrIGJyb3duIGZveA0KanVtcHMgb3ZlciB0aGUgbGFhenkgZG9n"
         },
         "file:///src/ui/unicodeText.cs": {
-          "uri": "file:///src/ui/unicodeText.cs",
           "mimeType": "text/x-csharp",
           "contents": "VGhlIHF1aWNrIGJyb3duIGZveA0KanVtcHMgb3ZlciB0aGUgbGFhenkgZG9n"
         }

src/Sarif.UnitTests/Visitors/SarifCurrentToVersionOneVisitorTests.cs

@@ -14,7 +14,7 @@
 using Xunit;
 using Xunit.Abstractions;
 
-namespace Microsoft.CodeAnalysis.Sarif.UnitTests.Transformers
+namespace Microsoft.CodeAnalysis.Sarif.UnitTests.Visitors
 {
     public class SarifCurrentToVersionOneVisitorTests : FileDiffingTests
     {
@@ -69,9 +69,7 @@ private void VerifyCurrentToVersionOneTransformationFromResource(string v2InputR
                 File.WriteAllText(expectedFilePath, v1ExpectedLogText);
                 File.WriteAllText(actualFilePath, v1ActualLogText);
 
-
-                string errorMessage = string.Format(@"V2 conversion from V1 produced unexpected diffs for test: '{0}'.", v2InputResourceName);
-
+                sb.AppendLine($"Conversion from current to V1 produced unexpected diffs for test: '{v2InputResourceName}'.");
                 sb.AppendLine("To compare all difference for this test suite:");
                 sb.AppendLine(GenerateDiffCommand("SarifCurrentToVersionOneVisitor", Path.GetDirectoryName(expectedFilePath), Path.GetDirectoryName(actualFilePath)) + Environment.NewLine);
             }

src/Sarif/Visitors/SarifCurrentToVersionOneVisitor.cs

@@ -22,6 +22,10 @@ public class SarifCurrentToVersionOneVisitor : SarifRewritingVisitor
         private Run _currentV2Run = null;
         private RunVersionOne _currentRun = null;
 
+        // To understand the purpose of these fields, see the comment on CreateFileKeyIndexMappings.
+        private IDictionary<string, int> _fileKeyToIndexDictionary;
+        private IDictionary<int, string> _fileIndexToKeyDictionary;
+
         public bool EmbedVersionTwoContentInPropertyBag { get; set; }
 
         public SarifLogVersionOne SarifLogVersionOne { get; private set; }
@@ -148,15 +152,15 @@ private string GetFileEncodingName(Uri uri)
         {
             string encodingName = null;
             IList<FileData> files = _currentV2Run.Files;
-#if FILES_ARRAY_WORKS
-            FileData fileData;
+
             if (uri != null &&
-                files != null &&
-                files.TryGetValue(uri.OriginalString, out fileData))
+                files != null
+                && _fileKeyToIndexDictionary != null
+                && _fileKeyToIndexDictionary.TryGetValue(uri.OriginalString, out int index))
             {
-                encodingName = fileData.Encoding;
+                encodingName = files[index].Encoding;
             }
-#endif
+
             return encodingName;
         }
 
@@ -173,21 +177,24 @@ private Encoding GetFileEncoding(string encodingName)
             return encoding;
         }
 
-        internal FileDataVersionOne CreateFileData(FileData v2FileData)
+        private FileDataVersionOne CreateFileData(FileData v2FileData)
         {
             FileDataVersionOne fileData = null;
 
             if (v2FileData != null)
             {
+                int parentIndex = v2FileData.ParentIndex;
+                string parentKey = parentIndex == -1
+                    ? null
+                    : _fileIndexToKeyDictionary?[parentIndex];
+
                 fileData = new FileDataVersionOne
                 {
                     Hashes = CreateHashVersionOneListFromV2Hashes(v2FileData.Hashes),
                     Length = v2FileData.Length,
                     MimeType = v2FileData.MimeType,
                     Offset = v2FileData.Offset,
-#if FILES_ARRAY_WORKS
-                    ParentKey = v2FileData.ParentKey,
-#endif
+                    ParentKey = parentKey,
                     Properties = v2FileData.Properties,
                     Uri = v2FileData.FileLocation?.Uri,
                     UriBaseId = v2FileData.FileLocation?.UriBaseId
@@ -496,9 +503,8 @@ internal RegionVersionOne CreateRegion(Region v2Region, Uri uri)
         private int ConvertCharOffsetToByteOffset(int charOffset, Uri uri)
         {
             int byteOffset = 0;
-            Encoding encoding;
 
-            using (StreamReader reader = GetFileStreamReader(uri, out encoding))
+            using (StreamReader reader = GetFileStreamReader(uri, out Encoding encoding))
             {
                 if (reader != null)
                 {
@@ -518,9 +524,8 @@ private int ConvertCharOffsetToByteOffset(int charOffset, Uri uri)
         private int GetRegionByteLength(Region v2Region, Uri uri)
         {
             int byteLength = 0;
-            Encoding encoding;
 
-            using (StreamReader reader = GetFileStreamReader(uri, out encoding))
+            using (StreamReader reader = GetFileStreamReader(uri, out Encoding encoding))
             {
                 if (reader != null)
                 {
@@ -572,9 +577,8 @@ private int GetRegionByteLength(Region v2Region, Uri uri)
         private int GetRegionEndColumn(Region v2Region, Uri uri)
         {
             int endColumn = 0;
-            Encoding encoding;
 
-            using (StreamReader reader = GetFileStreamReader(uri, out encoding))
+            using (StreamReader reader = GetFileStreamReader(uri, out Encoding encoding))
             {
                 if (reader != null)
                 {
@@ -600,13 +604,15 @@ private Stream GetContentStream(Uri uri, out Encoding encoding)
             Stream stream = null;
             encoding = null;
             string failureReason = null;
+            IList<FileData> files = _currentV2Run.Files;
 
-            if (uri != null && _currentV2Run.Files != null)
+            if (uri != null && files != null && _fileKeyToIndexDictionary != null)
             {
-#if FILES_ARRAY_WORKS
-                FileData fileData;
-                if (_currentV2Run.Files.TryGetValue(uri.OriginalString, out fileData))
+                if (_fileKeyToIndexDictionary.TryGetValue(uri.OriginalString, out int index)
+                    && index < files.Count)
                 {
+                    FileData fileData = files[index];
+
                     // We need the encoding because the content might have been transcoded to UTF-8
                     string encodingName = fileData.Encoding ?? _currentV2Run.DefaultFileEncoding;
                     encoding = GetFileEncoding(encodingName);
@@ -654,7 +660,6 @@ private Stream GetContentStream(Uri uri, out Encoding encoding)
                         failureReason = $"Encoding for file '{uri.OriginalString}' could not be determined";
                     }
                 }
-#endif
             }
 
             if (stream == null && failureReason == null)
@@ -731,16 +736,11 @@ internal Dictionary<string, string> CreateResponseFilesDictionary(IList<FileLoca
                 foreach (FileLocation fileLocation in v2ResponseFilesList)
                 {
                     string key = fileLocation.Uri.OriginalString;
-                    string fileContent = null;
-#if FILES_ARRAY_WORKS
-                    FileData responseFile;
-                    if (_currentV2Run.Files != null && _currentV2Run.Files.TryGetValue(key, out responseFile))
+                    if (_fileKeyToIndexDictionary != null && _fileKeyToIndexDictionary.TryGetValue(key, out int responseFileIndex))
                     {
-                        fileContent = responseFile.Contents?.Text;
+                        FileData responseFile = _currentV2Run.Files[responseFileIndex];
+                        responseFiles.Add(key, responseFile.Contents?.Text);
                     }
-#endif
-
-                    responseFiles.Add(key, fileContent);
                 }
             }
 
@@ -825,8 +825,7 @@ internal RuleVersionOne CreateRule(IRule v2IRule)
         {
             RuleVersionOne rule = null;
 
-            Rule v2Rule = v2IRule as Rule;
-            var properties = v2Rule != null ? v2Rule.Properties : null;
+            var properties = v2IRule is Rule v2Rule ? v2Rule.Properties : null;
 
             if (v2IRule != null)
             {
@@ -872,10 +871,10 @@ internal RunVersionOne CreateRun(Run v2Run)
                     run = new RunVersionOne();
                     _currentRun = run;
 
+                    CreateFileKeyIndexMappings(v2Run.Files, out _fileKeyToIndexDictionary, out _fileIndexToKeyDictionary);
+
                     run.BaselineId = v2Run.BaselineInstanceGuid;
-#if FILES_ARRAY_WORKS
-                    run.Files = v2Run.Files?.ToDictionary(v => v.Key, v => CreateFileData(v.Value));
-#endif
+                    run.Files = CreateV1FileDictionary();
                     run.Id = v2Run.Id?.InstanceGuid;
                     run.AutomationId = v2Run.AggregateIds?.FirstOrDefault()?.InstanceId;
 
@@ -904,6 +903,105 @@ internal RunVersionOne CreateRun(Run v2Run)
             return run;
         }
 
+        // In SARIF v1, run.files was a dictionary, whereas in v2 it is an array.
+        // During the v2-to-v1 conversion, it is sometimes necessary to look up
+        // information from a v2 FileData object. Rather than having to search the
+        // v2 run.files array to find the required FileData object, we construct a
+        // dictionary from the array so we can access the required FileData object
+        // directly. It turns out that we need mappings in both directions: from
+        // array index to dictionary key, and vice versa.
+        private static void CreateFileKeyIndexMappings(
+            IList<FileData> v2Files,
+            out IDictionary<string, int> fileKeyToIndexDictionary,
+            out IDictionary<int, string> fileIndexToKeyDictionary)
+        {
+            fileKeyToIndexDictionary = null;
+            fileIndexToKeyDictionary = null;
+
+            if (v2Files != null)
+            {
+                fileKeyToIndexDictionary = new Dictionary<string, int>();
+                fileIndexToKeyDictionary = new Dictionary<int, string>();
+
+                for (int index = 0; index < v2Files.Count; ++index)
+                {
+                    FileData v2File = v2Files[index];
+                    string key = CreateFileDictionaryKey(v2File, v2Files);
+
+                    fileKeyToIndexDictionary[key] = index;
+                    fileIndexToKeyDictionary[index] = key;
+                }
+            }
+        }
+
+        // Given a v2 FileData object, synthesize a key that will be used to locate that object
+        // in the v2 file data dictionary. We will use the same key to locate the corresponding
+        // v1 FileData object when we create the v1 run.files dictionary.
+        //
+        // This method is necessary because, although ideally we would use the FileData
+        // object's URI as the dictionary key, it is possible for two FileData objects to
+        // have the same URI. For example:
+        //
+        //    1. Nested files with the same name might exist in two different containers.
+        //    2. There might be two files with the same URI but difference uriBaseIds.
+        //
+        // To deal with Case #1, we synthesize a key as follows:
+        //
+        //    <root file URI>#<nested file path>/<level 2 nested file path>/...
+        //
+        // To deal with Case #2, we would follow the chain or uriBaseIds, prepending
+        // the value of each one to the URI.
+        //
+        // WE DO NOT YET HANDLE CASE #2.
+        private static string CreateFileDictionaryKey(FileData v2File, IList<FileData> v2Files)
+        {
+            var sb = new StringBuilder(v2File.FileLocation.Uri.OriginalString);
+            while (v2File.ParentIndex != -1)
+            {
+                FileData parentFile = v2Files[v2File.ParentIndex];
+
+                // The convention for building the key is as follows:
+                // The root file URI is separated from the chain of nested files by '#';
+                // The nested file URIs are separated from each other by '/'.
+                string separator = parentFile.ParentIndex == -1 ? "#" : "/";
+
+                sb.Insert(0, separator);
+                sb.Insert(0, parentFile.FileLocation.Uri.OriginalString);
+
+                v2File = parentFile;
+            }
+
+            return sb.ToString();
+        }
+
+        private IDictionary<string, FileDataVersionOne> CreateV1FileDictionary()
+        {
+            Dictionary<string, FileDataVersionOne> filesVersionOne = null;
+
+            if (_fileKeyToIndexDictionary != null)
+            {
+                filesVersionOne = new Dictionary<string, FileDataVersionOne>();
+                foreach (KeyValuePair<string, int> entry in _fileKeyToIndexDictionary)
+                {
+                    string key = entry.Key;
+                    int index = entry.Value;
+                    FileData v2File = _currentV2Run.Files[index];
+                    FileDataVersionOne fileDataVersionOne = CreateFileData(v2File);
+
+                    // There's no need to repeat the URI in the v1 FileData object
+                    // if it matches the dictionary key.
+                    if (fileDataVersionOne.Uri.OriginalString.Equals(key))
+                    {
+                        fileDataVersionOne.Uri = null;
+                    }
+
+                    filesVersionOne[key] = fileDataVersionOne;
+                }
+            }
+
+            return filesVersionOne;
+        }
+
         private IDictionary<string, LogicalLocationVersionOne> CreateLogicalLocationsDictionary(IList<LogicalLocation> logicalLocations)
         {
             Dictionary<string, LogicalLocationVersionOne> logicalLocationsVersionOne = null;