Add fuzz testing for parser, fix two bugs (#414)

Author: jakebailey

internal/parser/jsdoc.go

@@ -509,7 +509,7 @@ loop:
 			whitespace := p.scanner.TokenText()
 			// if the whitespace crosses the margin, take only the whitespace that passes the margin
 			if margin > -1 && indent+len(whitespace) > margin {
-				comments = append(comments, whitespace[margin-indent:])
+				comments = append(comments, whitespace[max(margin-indent, 0):])
 				state = jsdocStateSavingComments
 			}
 			indent += len(whitespace)

internal/parser/parser_test.go

@@ -126,3 +126,59 @@ func allParsableFiles(tb testing.TB, root string) iter.Seq[parsableFile] {
 		assert.NilError(tb, err)
 	}
 }
+
+func FuzzParser(f *testing.F) {
+	repo.SkipIfNoTypeScriptSubmodule(f)
+
+	tests := []string{
+		"src",
+		"scripts",
+		"Herebyfile.mjs",
+		// "tests/cases",
+	}
+
+	var extensions core.Set[string]
+	for _, es := range tspath.AllSupportedExtensionsWithJson {
+		for _, e := range es {
+			extensions.Add(e)
+		}
+	}
+
+	for _, test := range tests {
+		root := filepath.Join(repo.TypeScriptSubmodulePath, test)
+
+		for file := range allParsableFiles(f, root) {
+			sourceText, err := os.ReadFile(file.path)
+			assert.NilError(f, err)
+			extension := tspath.TryGetExtensionFromPath(file.path)
+			f.Add(extension, string(sourceText), int(core.ScriptTargetESNext), int(scanner.JSDocParsingModeParseAll))
+		}
+	}
+
+	f.Fuzz(func(t *testing.T, extension string, sourceText string, scriptTarget_ int, jsdocParsingMode_ int) {
+		scriptTarget := core.ScriptTarget(scriptTarget_)
+		jsdocParsingMode := scanner.JSDocParsingMode(jsdocParsingMode_)
+
+		if !extensions.Has(extension) {
+			t.Skip()
+		}
+
+		if scriptTarget < core.ScriptTargetNone || scriptTarget > core.ScriptTargetLatest {
+			t.Skip()
+		}
+
+		if jsdocParsingMode < scanner.JSDocParsingModeParseAll || jsdocParsingMode > scanner.JSDocParsingModeParseNone {
+			t.Skip()
+		}
+
+		fileName := "/index" + extension
+		path := tspath.Path(fileName)
+
+		if extension == ".json" {
+			ParseJSONText(fileName, path, sourceText)
+			return
+		}
+
+		ParseSourceFile(fileName, path, sourceText, scriptTarget, jsdocParsingMode)
+	})
+}

internal/parser/testdata/fuzz/FuzzParser/02b74efe61495c2a

@@ -0,0 +1,5 @@
+go test fuzz v1
+string(".ts")
+string("/**@0\n     * */0")
+int(99)
+int(0)

internal/parser/testdata/fuzz/FuzzParser/9ce2d994c65c7bfe

@@ -0,0 +1,5 @@
+go test fuzz v1
+string(".ts")
+string("/")
+int(99)
+int(1)

internal/scanner/scanner.go

@@ -2300,7 +2300,7 @@ func iterateCommentRanges(f *ast.NodeFactory, text string, pos int, trailing boo
 			ch, size := utf8.DecodeRuneInString(text[pos:])
 			switch ch {
 			case '\r':
-				if text[pos+1] == '\n' {
+				if pos+1 < len(text) && text[pos+1] == '\n' {
 					pos++
 				}
 				fallthrough
@@ -2320,7 +2320,10 @@ func iterateCommentRanges(f *ast.NodeFactory, text string, pos int, trailing boo
 				pos++
 				continue
 			case '/':
-				nextChar := text[pos+1]
+				var nextChar byte
+				if pos+1 < len(text) {
+					nextChar = text[pos+1]
+				}
 				hasTrailingNewLine := false
 				if nextChar == '/' || nextChar == '*' {
 					var kind ast.Kind
@@ -2343,11 +2346,12 @@ func iterateCommentRanges(f *ast.NodeFactory, text string, pos int, trailing boo
 						}
 					} else {
 						for pos < len(text) {
-							if text[pos] == '*' && text[pos+1] == '/' {
+							c, s := utf8.DecodeRuneInString(text[pos:])
+							if c == '*' && pos+1 < len(text) && text[pos+1] == '/' {
 								pos += 2
 								break
 							}
-							pos++
+							pos += s
 						}
 					}