public static void Main(string[] args)
{
var code1 = @"using System;
class Hello
{
public static readonly void Main(){
foreach(var h in asdf){
h.Write(test);
}
var foo = new Bar(1,2,3,4,5,SomeProperty,""MyString"");
var foo1 = sdf(x => x.sdf);
var foo1 = sdf((x,y,z) => x.sdf);
var foo2 = sdf();
Assert.Equal(SomeInt, new int[] {1,2,3,4});
Main();
}
public static int SomeInt = 2;
public static string SomeProperty {get; set;};
}
class Hello2 {
public static int sdf = 3;
}
";
var syntaxTree = CSharpSyntaxTree.ParseText(code1);
var nodes = syntaxTree.GetRoot().DescendantNodes().ToList();
Console.WriteLine(Canonicalization.CanonicalizeSyntaxNode(syntaxTree.GetRoot(), extractAllVariablesFirst: true).CanonicalSyntaxNode.GetText());
Console.ReadLine();
}
public static IEnumerable <object> ProcessSingleRevision(string jsonLine, JsonSyntaxTreeHelper jsonSyntaxTreeHelper)
{
var entry = JObject.Parse(jsonLine);
var previousFile = entry["prev_file"].ToString();
var updatedFile = entry["updated_file"].ToString();
// Console.WriteLine($"Processing {entry["id"]}");
// File.WriteAllText("a.original.cs", previousFile);
// File.WriteAllText("b.original.cs", updatedFile);
var previousFileAst = CSharpSyntaxTree.ParseText(previousFile);
var updatedFileAst = CSharpSyntaxTree.ParseText(updatedFile);
(SyntaxNode canonicalPrevFileAst, Dictionary <string, string> prevFileVariableNameMap) = Canonicalization.CanonicalizeSyntaxNode(previousFileAst.GetRoot(), extractAllVariablesFirst: true);
(SyntaxNode canonicalUpdatedFileAst, Dictionary <string, string> updatedFileVariableNameMap) = Canonicalization.CanonicalizeSyntaxNode(updatedFileAst.GetRoot(), prevFileVariableNameMap);
var prevCodeFile = canonicalPrevFileAst.GetText();
var updatedCodeFile = canonicalUpdatedFileAst.GetText();
var prevFileTokens = canonicalPrevFileAst.DescendantTokens().ToList();
var updatedFileTokens = canonicalUpdatedFileAst.DescendantTokens().ToList();
var changesInRevision = GetChangesBetweenAsts(canonicalPrevFileAst.SyntaxTree, canonicalUpdatedFileAst.SyntaxTree);
// File.WriteAllText("a.canonical.cs", canonicalPrevFileAst.GetText().ToString());
// File.WriteAllText("b.canonical.cs", canonicalUpdatedFileAst.GetText().ToString());
var prevTokenIndex = new TokenIndex(prevFileTokens);
var updatedTokenIndex = new TokenIndex(updatedFileTokens);
var changeId = 0;
foreach (var change in changesInRevision)
{
var prevCodeChunkLineSpan = canonicalPrevFileAst.SyntaxTree.GetLineSpan(change.BeforeSpan.ChangeSpan);
var updatedCodeChunkLineSpan = canonicalUpdatedFileAst.SyntaxTree.GetLineSpan(change.AfterSpan.ChangeSpan);
var prevCodeChunkLineSpanStart = prevCodeFile.Lines[prevCodeChunkLineSpan.StartLinePosition.Line].Span.Start;
var prevCodeChunkSpanEnd = prevCodeFile.Lines[prevCodeChunkLineSpan.EndLinePosition.Line].Span.End;
var updatedCodeChunkLineSpanStart = updatedCodeFile.Lines[updatedCodeChunkLineSpan.StartLinePosition.Line].Span.Start;
var updatedCodeChunkSpanEnd = updatedCodeFile.Lines[updatedCodeChunkLineSpan.EndLinePosition.Line].Span.End;
// only consider changes of equal number of lines
if (prevCodeChunkLineSpan.EndLinePosition.Line - prevCodeChunkLineSpan.StartLinePosition.Line
!= updatedCodeChunkLineSpan.EndLinePosition.Line - updatedCodeChunkLineSpan.StartLinePosition.Line)
{
continue;
}
// TODO: remove trivial change
// only consider SyntaxKind in allowedSytaxKinds
var prevCodeChunkNodes = GetNodesByLineSpan(canonicalPrevFileAst, prevCodeFile,
prevCodeChunkLineSpan.StartLinePosition.Line, prevCodeChunkLineSpan.EndLinePosition.Line);
if (prevCodeChunkNodes.Any(node => !allowedSytaxKinds.Contains(node.Kind())))
{
continue;
}
var updatedCodeChunkNodes = GetNodesByLineSpan(canonicalUpdatedFileAst, updatedCodeFile,
updatedCodeChunkLineSpan.StartLinePosition.Line, updatedCodeChunkLineSpan.EndLinePosition.Line);
if (updatedCodeChunkNodes.Any(node => !allowedSytaxKinds.Contains(node.Kind())))
{
continue;
}
var previousCodeChunkTokens = prevTokenIndex
.GetTokensInSpan(prevCodeChunkLineSpanStart, prevCodeChunkSpanEnd)
.Select(token => token.ValueText)
.Where(token => !string.IsNullOrWhiteSpace(token) && !string.IsNullOrEmpty(token))
.ToArray();
var updatedsCodeChunkTokens = updatedTokenIndex
.GetTokensInSpan(updatedCodeChunkLineSpanStart, updatedCodeChunkSpanEnd)
.Select(token => token.ValueText)
.Where(token => !string.IsNullOrWhiteSpace(token) && !string.IsNullOrEmpty(token))
.ToArray();
if (previousCodeChunkTokens.Length > 0 && updatedsCodeChunkTokens.Length > 0 &&
IsValidCodeChunkTokens(previousCodeChunkTokens) && IsValidCodeChunkTokens(updatedsCodeChunkTokens) &&
!previousCodeChunkTokens.SequenceEqual(updatedsCodeChunkTokens))
{
var changeSha = entry["id"] + "_" + changeId;
var prevCodeChunkBlockStmt = SyntaxFactory.Block(prevCodeChunkNodes.Select(node => (StatementSyntax)node));
var updatedCodeChunkBlockStmt = SyntaxFactory.Block(updatedCodeChunkNodes.Select(node => (StatementSyntax)node));
IDictionary <string, string> zeroIndexedVariableNameMap;
(prevCodeChunkBlockStmt, updatedCodeChunkBlockStmt, zeroIndexedVariableNameMap) =
zeroIndexVariableNames(prevCodeChunkBlockStmt, updatedCodeChunkBlockStmt);
var prevCodeChunkBlockStmtTokens = prevCodeChunkBlockStmt.DescendantTokens().Skip(1).SkipLast(1).ToArray();
var prevCodeChunkBlackStmtTokensIndex = new TokenIndex(prevCodeChunkBlockStmtTokens).InitInvertedIndex();
var updatedCodeChunkBlockStmtTokens = updatedCodeChunkBlockStmt.DescendantTokens().Skip(1).SkipLast(1).ToArray();
var updatedCodeChunkBlockStmtTokensIndex = new TokenIndex(updatedCodeChunkBlockStmtTokens).InitInvertedIndex();
var prevCodeBlockJObject = jsonSyntaxTreeHelper.GetJObjectForSyntaxNode(prevCodeChunkBlockStmt, prevCodeChunkBlackStmtTokensIndex);
var updatedCodeBlockJObject = jsonSyntaxTreeHelper.GetJObjectForSyntaxNode(updatedCodeChunkBlockStmt, updatedCodeChunkBlockStmtTokensIndex);
var precedingContextTokens = prevTokenIndex.GetTokensInSpan(change.BeforeSpan.SpanOfPrecedingContext);
var succeedingContextTokens = updatedTokenIndex.GetTokensInSpan(change.BeforeSpan.SpanOfSucceedingContext);
precedingContextTokens = zeroIndexVariableNames(precedingContextTokens, zeroIndexedVariableNameMap);
succeedingContextTokens = zeroIndexVariableNames(succeedingContextTokens, zeroIndexedVariableNameMap);
var prevCodeChunkBlockStmtTextTokens =
prevCodeChunkBlockStmtTokens.Select(token => token.ValueText).ToArray();
var updatedCodeChunkBlockStmtTextTokens =
updatedCodeChunkBlockStmtTokens.Select(token => token.ValueText).ToArray();
var prevCodeTextChunk = Utils.ExtractCodeTextFromBraces(prevCodeChunkBlockStmt.GetText().ToString());
prevCodeTextChunk = Utils.RemoveLeadingWhiteSpace(prevCodeTextChunk, naive: true);
var updatedCodeTextChunk = Utils.ExtractCodeTextFromBraces(updatedCodeChunkBlockStmt.GetText().ToString());
updatedCodeTextChunk = Utils.RemoveLeadingWhiteSpace(updatedCodeTextChunk, naive: true);
var precedingContextTextTokens = precedingContextTokens.Select(token => token.ValueText).ToArray();
var succeedingContextTextTokens = succeedingContextTokens.Select(token => token.ValueText).ToArray();
var result = new
{
Id = changeSha,
PrevCodeChunk = prevCodeTextChunk,
UpdatedCodeChunk = updatedCodeTextChunk,
PrevCodeChunkTokens = prevCodeChunkBlockStmtTextTokens,
UpdatedCodeChunkTokens = updatedCodeChunkBlockStmtTextTokens,
PrevCodeAST = prevCodeBlockJObject,
UpdatedCodeAST = updatedCodeBlockJObject,
PrecedingContext = precedingContextTextTokens,
SucceedingContext = succeedingContextTextTokens,
CommitMessage = entry["message"]
};
changeId += 1;
yield return(result);
}
}
}
