Merge pull request #1456 from Microsoft/mergeMarkers

Make the compiler resilient to encountering merge conflict markers in a source code file.

src/compiler/diagnosticInformationMap.generated.ts

@@ -144,6 +144,7 @@ module ts {
         Array_element_destructuring_pattern_expected: { code: 1181, category: DiagnosticCategory.Error, key: "Array element destructuring pattern expected." },
         A_destructuring_declaration_must_have_an_initializer: { code: 1182, category: DiagnosticCategory.Error, key: "A destructuring declaration must have an initializer." },
         Destructuring_declarations_are_not_allowed_in_ambient_contexts: { code: 1183, category: DiagnosticCategory.Error, key: "Destructuring declarations are not allowed in ambient contexts." },
+        Merge_conflict_marker_encountered: { code: 1184, category: DiagnosticCategory.Error, key: "Merge conflict marker encountered." },
         Duplicate_identifier_0: { code: 2300, category: DiagnosticCategory.Error, key: "Duplicate identifier '{0}'." },
         Initializer_of_instance_member_variable_0_cannot_reference_identifier_1_declared_in_the_constructor: { code: 2301, category: DiagnosticCategory.Error, key: "Initializer of instance member variable '{0}' cannot reference identifier '{1}' declared in the constructor." },
         Static_members_cannot_reference_class_type_parameters: { code: 2302, category: DiagnosticCategory.Error, key: "Static members cannot reference class type parameters." },

src/compiler/diagnosticMessages.json

@@ -568,6 +568,10 @@
         "category": "Error",
         "code": 1183
     },
+    "Merge conflict marker encountered.": {
+        "category": "Error",
+        "code": 1184
+    },
 
     "Duplicate identifier '{0}'.": {
         "category": "Error",

src/compiler/scanner.ts

@@ -333,10 +333,14 @@ module ts {
             var ch = text.charCodeAt(pos);
             switch (ch) {
                 case CharacterCodes.carriageReturn:
-                    if (text.charCodeAt(pos + 1) === CharacterCodes.lineFeed) pos++;
+                    if (text.charCodeAt(pos + 1) === CharacterCodes.lineFeed) {
+                        pos++;
+                    }
                 case CharacterCodes.lineFeed:
                     pos++;
-                    if (stopAfterLineBreak) return pos;
+                    if (stopAfterLineBreak) {
+                        return pos;
+                    }
                     continue;
                 case CharacterCodes.tab:
                 case CharacterCodes.verticalTab:
@@ -367,6 +371,16 @@ module ts {
                         continue;
                     }
                     break;
+
+                case CharacterCodes.lessThan:
+                case CharacterCodes.equals:
+                case CharacterCodes.greaterThan:
+                    if (isConflictMarkerTrivia(text, pos)) {
+                        pos = scanConflictMarkerTrivia(text, pos);
+                        continue;
+                    }
+                    break;
+
                 default:
                     if (ch > CharacterCodes.maxAsciiCharacter && (isWhiteSpace(ch) || isLineBreak(ch))) {
                         pos++;
@@ -378,6 +392,39 @@ module ts {
         }
     }
 
+    function isConflictMarkerTrivia(text: string, pos: number) {
+        // Conflict markers must be at the start of a line.
+        if (pos > 0 && isLineBreak(text.charCodeAt(pos - 1))) {
+            var ch = text.charCodeAt(pos);
+
+            // All conflict markers consist of the same character repeated seven times.  If it is 
+            // a <<<<<<< or >>>>>>> marker then it is also followd by a space.
+            var markerLength = "<<<<<<<".length;
+
+            if ((pos + markerLength) < text.length) {
+                for (var i = 0, n = markerLength; i < n; i++) {
+                    if (text.charCodeAt(pos + i) !== ch) {
+                        return false;
+                    }
+                }
+
+                return ch === CharacterCodes.equals ||
+                    text.charCodeAt(pos + markerLength) === CharacterCodes.space;
+            }
+        }
+
+        return false;
+    }
+
+    function scanConflictMarkerTrivia(text: string, pos: number) {
+        var len = text.length;
+        while (pos < len && !isLineBreak(text.charCodeAt(pos))) {
+            pos++;
+        }
+
+        return pos;
+    }
+
     // Extract comments from the given source text starting at the given position. If trailing is false, whitespace is skipped until
     // the first line break and comments between that location and the next token are returned. If trailing is true, comments occurring
     // between the given position and the next line break are returned. The return value is an array containing a TextRange for each
@@ -1010,6 +1057,17 @@ module ts {
                     case CharacterCodes.semicolon:
                         return pos++, token = SyntaxKind.SemicolonToken;
                     case CharacterCodes.lessThan:
+                        if (isConflictMarkerTrivia(text, pos)) {
+                            error(Diagnostics.Merge_conflict_marker_encountered);
+                            pos = scanConflictMarkerTrivia(text, pos);
+                            if (skipTrivia) {
+                                continue;
+                            }
+                            else {
+                                return token = SyntaxKind.ConflictMarkerTrivia;
+                            }
+                        }
+
                         if (text.charCodeAt(pos + 1) === CharacterCodes.lessThan) {
                             if (text.charCodeAt(pos + 2) === CharacterCodes.equals) {
                                 return pos += 3, token = SyntaxKind.LessThanLessThanEqualsToken;
@@ -1021,6 +1079,17 @@ module ts {
                         }
                         return pos++, token = SyntaxKind.LessThanToken;
                     case CharacterCodes.equals:
+                        if (isConflictMarkerTrivia(text, pos)) {
+                            error(Diagnostics.Merge_conflict_marker_encountered);
+                            pos = scanConflictMarkerTrivia(text, pos);
+                            if (skipTrivia) {
+                                continue;
+                            }
+                            else {
+                                return token = SyntaxKind.ConflictMarkerTrivia;
+                            }
+                        }
+
                         if (text.charCodeAt(pos + 1) === CharacterCodes.equals) {
                             if (text.charCodeAt(pos + 2) === CharacterCodes.equals) {
                                 return pos += 3, token = SyntaxKind.EqualsEqualsEqualsToken;
@@ -1032,6 +1101,17 @@ module ts {
                         }
                         return pos++, token = SyntaxKind.EqualsToken;
                     case CharacterCodes.greaterThan:
+                        if (isConflictMarkerTrivia(text, pos)) {
+                            error(Diagnostics.Merge_conflict_marker_encountered);
+                            pos = scanConflictMarkerTrivia(text, pos);
+                            if (skipTrivia) {
+                                continue;
+                            }
+                            else {
+                                return token = SyntaxKind.ConflictMarkerTrivia;
+                            }
+                        }
+
                         return pos++, token = SyntaxKind.GreaterThanToken;
                     case CharacterCodes.question:
                         return pos++, token = SyntaxKind.QuestionToken;

src/compiler/types.ts

@@ -18,6 +18,7 @@ module ts {
         MultiLineCommentTrivia,
         NewLineTrivia,
         WhitespaceTrivia,
+        ConflictMarkerTrivia,
         // Literals
         NumericLiteral,
         StringLiteral,
@@ -269,7 +270,7 @@ module ts {
         FirstToken = Unknown,
         LastToken = TypeKeyword,
         FirstTriviaToken = SingleLineCommentTrivia,
-        LastTriviaToken = WhitespaceTrivia,
+        LastTriviaToken = ConflictMarkerTrivia,
         FirstLiteralToken = NumericLiteral,
         LastLiteralToken = NoSubstitutionTemplateLiteral,
         FirstTemplateToken = NoSubstitutionTemplateLiteral,

src/services/services.ts

@@ -5920,6 +5920,7 @@ module ts {
                     return TokenClass.StringLiteral;
                 case SyntaxKind.RegularExpressionLiteral:
                     return TokenClass.RegExpLiteral;
+                case SyntaxKind.ConflictMarkerTrivia:
                 case SyntaxKind.MultiLineCommentTrivia:
                 case SyntaxKind.SingleLineCommentTrivia:
                     return TokenClass.Comment;

tests/baselines/reference/conflictMarkerTrivia1.errors.txt

@@ -0,0 +1,25 @@
+tests/cases/compiler/conflictMarkerTrivia1.ts(2,1): error TS1184: Merge conflict marker encountered.
+tests/cases/compiler/conflictMarkerTrivia1.ts(4,1): error TS1184: Merge conflict marker encountered.
+tests/cases/compiler/conflictMarkerTrivia1.ts(6,1): error TS1184: Merge conflict marker encountered.
+tests/cases/compiler/conflictMarkerTrivia1.ts(3,5): error TS2300: Duplicate identifier 'v'.
+tests/cases/compiler/conflictMarkerTrivia1.ts(5,5): error TS2300: Duplicate identifier 'v'.
+
+
+==== tests/cases/compiler/conflictMarkerTrivia1.ts (5 errors) ====
+    class C {
+    <<<<<<< HEAD
+    
+!!! error TS1184: Merge conflict marker encountered.
+        v = 1;
+        ~
+!!! error TS2300: Duplicate identifier 'v'.
+    =======
+    
+!!! error TS1184: Merge conflict marker encountered.
+        v = 2;
+        ~
+!!! error TS2300: Duplicate identifier 'v'.
+    >>>>>>> Branch-a
+    
+!!! error TS1184: Merge conflict marker encountered.
+    }

tests/cases/compiler/conflictMarkerTrivia1.ts

@@ -0,0 +1,7 @@
+class C {
+<<<<<<< HEAD
+    v = 1;
+=======
+    v = 2;
+>>>>>>> Branch-a
+}

tests/cases/fourslash/formatConflictMarker1.ts

@@ -0,0 +1,18 @@
+/// <reference path='fourslash.ts' />
+
+////class C {
+////<<<<<<< HEAD
+////    v = 1;
+////=======
+////v = 2;
+////>>>>>>> Branch - a
+////}
+
+format.document();
+verify.currentFileContentIs("class C {\r\n\
+<<<<<<< HEAD\r\n\
+    v = 1;\r\n\
+=======\r\n\
+    v = 2;\r\n\
+>>>>>>> Branch - a\r\n\
+}");

tests/cases/unittests/services/colorization.ts

@@ -317,7 +317,9 @@ describe('Colorization', function () {
                 operator("<"),
                 identifier("number"),
                 finalEndOfLineState(ts.EndOfLineState.Start));
+        });
 
+        it("ClassifiesConflictTokens", () => {
             // no longer in something that looks generic.
             test("Foo<Foo> number",
                 ts.EndOfLineState.Start,
@@ -327,6 +329,33 @@ describe('Colorization', function () {
                 operator(">"),
                 keyword("number"),
                 finalEndOfLineState(ts.EndOfLineState.Start));
+
+            // Test conflict markers.
+            test(
+"class C {\r\n\
+<<<<<<< HEAD\r\n\
+    v = 1;\r\n\
+=======\r\n\
+    v = 2;\r\n\
+>>>>>>> Branch - a\r\n\
+}",
+                ts.EndOfLineState.Start,
+                keyword("class"),
+                identifier("C"),
+                punctuation("{"),
+                comment("<<<<<<< HEAD"),
+                identifier("v"),
+                operator("="),
+                numberLiteral("1"),
+                punctuation(";"),
+                comment("======="),
+                identifier("v"),
+                operator("="),
+                numberLiteral("2"),
+                punctuation(";"),
+                comment(">>>>>>> Branch - a"),
+                punctuation("}"),
+                finalEndOfLineState(ts.EndOfLineState.Start));
         });
     });
 });