|
| 1 | +import XCTest |
| 2 | +@testable import SwiftParser |
| 3 | + |
| 4 | +final class ParserPipelineTests: XCTestCase { |
| 5 | + enum SimpleTokenElement: String, CaseIterable, CodeTokenElement { |
| 6 | + case number |
| 7 | + case plus |
| 8 | + case eof |
| 9 | + } |
| 10 | + |
| 11 | + struct SimpleToken: CodeToken { |
| 12 | + typealias Element = SimpleTokenElement |
| 13 | + let element: SimpleTokenElement |
| 14 | + let text: String |
| 15 | + let range: Range<String.Index> |
| 16 | + } |
| 17 | + |
| 18 | + struct NumberTokenBuilder: CodeTokenBuilder { |
| 19 | + typealias Token = SimpleTokenElement |
| 20 | + func build(from context: inout CodeTokenContext<Token>) -> Bool { |
| 21 | + guard context.consuming < context.source.endIndex, |
| 22 | + context.source[context.consuming].isNumber else { return false } |
| 23 | + var end = context.consuming |
| 24 | + while end < context.source.endIndex && context.source[end].isNumber { |
| 25 | + end = context.source.index(after: end) |
| 26 | + } |
| 27 | + let range = context.consuming..<end |
| 28 | + let token = SimpleToken(element: .number, |
| 29 | + text: String(context.source[range]), |
| 30 | + range: range) |
| 31 | + context.tokens.append(token) |
| 32 | + context.consuming = end |
| 33 | + return true |
| 34 | + } |
| 35 | + } |
| 36 | + |
| 37 | + struct PlusTokenBuilder: CodeTokenBuilder { |
| 38 | + typealias Token = SimpleTokenElement |
| 39 | + func build(from context: inout CodeTokenContext<Token>) -> Bool { |
| 40 | + guard context.consuming < context.source.endIndex, |
| 41 | + context.source[context.consuming] == "+" else { return false } |
| 42 | + let start = context.consuming |
| 43 | + context.consuming = context.source.index(after: start) |
| 44 | + let token = SimpleToken(element: .plus, text: "+", range: start..<context.consuming) |
| 45 | + context.tokens.append(token) |
| 46 | + return true |
| 47 | + } |
| 48 | + } |
| 49 | + |
| 50 | + struct WhitespaceTokenBuilder: CodeTokenBuilder { |
| 51 | + typealias Token = SimpleTokenElement |
| 52 | + func build(from context: inout CodeTokenContext<Token>) -> Bool { |
| 53 | + guard context.consuming < context.source.endIndex, |
| 54 | + context.source[context.consuming].isWhitespace else { return false } |
| 55 | + context.consuming = context.source.index(after: context.consuming) |
| 56 | + return true |
| 57 | + } |
| 58 | + } |
| 59 | + |
| 60 | + enum SimpleNodeElement: String, CaseIterable, CodeNodeElement { |
| 61 | + case root |
| 62 | + case number |
| 63 | + } |
| 64 | + |
| 65 | + struct NumberNodeBuilder: CodeNodeBuilder { |
| 66 | + typealias Node = SimpleNodeElement |
| 67 | + typealias Token = SimpleTokenElement |
| 68 | + func build(from context: inout CodeConstructContext<Node, Token>) -> Bool { |
| 69 | + guard context.consuming < context.tokens.count, |
| 70 | + let token = context.tokens[context.consuming] as? SimpleToken, |
| 71 | + token.element == .number else { return false } |
| 72 | + let node = CodeNode<Node>(element: .number) |
| 73 | + context.current.append(node) |
| 74 | + context.consuming += 1 |
| 75 | + return true |
| 76 | + } |
| 77 | + } |
| 78 | + |
| 79 | + struct SimpleLanguage: CodeLanguage { |
| 80 | + typealias Node = SimpleNodeElement |
| 81 | + typealias Token = SimpleTokenElement |
| 82 | + |
| 83 | + var tokens: [any CodeTokenBuilder<Token>] { |
| 84 | + [WhitespaceTokenBuilder(), NumberTokenBuilder(), PlusTokenBuilder()] |
| 85 | + } |
| 86 | + var nodes: [any CodeNodeBuilder<Node, Token>] { [NumberNodeBuilder()] } |
| 87 | + |
| 88 | + func root() -> CodeNode<Node> { CodeNode<Node>(element: .root) } |
| 89 | + func state() -> (any CodeConstructState<Node, Token>)? { nil } |
| 90 | + func state() -> (any CodeTokenState<Token>)? { nil } |
| 91 | + // rely on default eof implementation |
| 92 | + } |
| 93 | + |
| 94 | + func testTokenizerProducesTokensAndErrors() { |
| 95 | + let tokenizer = CodeTokenizer( |
| 96 | + builders: [NumberTokenBuilder(), PlusTokenBuilder()], |
| 97 | + state: { nil }, |
| 98 | + eof: { SimpleToken(element: .eof, text: "", range: $0) } |
| 99 | + ) |
| 100 | + |
| 101 | + let (tokens, errors) = tokenizer.tokenize("1+a") |
| 102 | + XCTAssertEqual(tokens.count, 3) // number, plus, eof |
| 103 | + XCTAssertEqual((tokens[0] as? SimpleToken)?.text, "1") |
| 104 | + XCTAssertEqual((tokens[1] as? SimpleToken)?.element, .plus) |
| 105 | + XCTAssertEqual((tokens[2] as? SimpleToken)?.element, .eof) |
| 106 | + XCTAssertEqual(errors.count, 1) |
| 107 | + } |
| 108 | + |
| 109 | + func testConstructorBuildsNodesAndErrors() { |
| 110 | + let oneRange = "1".startIndex..<"1".endIndex |
| 111 | + let plusRange = "+".startIndex..<"+".endIndex |
| 112 | + let twoRange = "2".startIndex..<"2".endIndex |
| 113 | + let tokens: [any CodeToken<SimpleTokenElement>] = [ |
| 114 | + SimpleToken(element: .number, text: "1", range: oneRange), |
| 115 | + SimpleToken(element: .plus, text: "+", range: plusRange), |
| 116 | + SimpleToken(element: .number, text: "2", range: twoRange) |
| 117 | + ] |
| 118 | + let root = CodeNode<SimpleNodeElement>(element: .root) |
| 119 | + let constructor = CodeConstructor(builders: [NumberNodeBuilder()], state: { nil }) |
| 120 | + let (parsed, errors) = constructor.parse(tokens, root: root) |
| 121 | + XCTAssertEqual(parsed.children.count, 2) |
| 122 | + XCTAssertEqual(errors.count, 1) // plus token unrecognized |
| 123 | + } |
| 124 | + |
| 125 | + func testParserNormalizesAndParses() { |
| 126 | + let language = SimpleLanguage() |
| 127 | + let parser = CodeParser(language: language) |
| 128 | + let result = parser.parse("1\r\n2\r", language: language) |
| 129 | + XCTAssertEqual(result.root.children.count, 2) |
| 130 | + XCTAssertTrue(result.errors.isEmpty) |
| 131 | + XCTAssertEqual(result.tokens.count, 2) |
| 132 | + // ensure default eof returns nil |
| 133 | + XCTAssertNil(language.eof(at: "".startIndex..<"".endIndex)) |
| 134 | + } |
| 135 | + |
| 136 | + func testContextAndErrorInitialization() { |
| 137 | + let tokenContext = CodeTokenContext<SimpleTokenElement>(source: "1") |
| 138 | + XCTAssertEqual(tokenContext.source, "1") |
| 139 | + XCTAssertTrue(tokenContext.tokens.isEmpty) |
| 140 | + XCTAssertTrue(tokenContext.errors.isEmpty) |
| 141 | + |
| 142 | + let root = CodeNode<SimpleNodeElement>(element: .root) |
| 143 | + let constructContext = CodeConstructContext<SimpleNodeElement, SimpleTokenElement>( |
| 144 | + current: root, tokens: [] |
| 145 | + ) |
| 146 | + XCTAssertTrue(constructContext.current === root) |
| 147 | + XCTAssertEqual(constructContext.tokens.count, 0) |
| 148 | + XCTAssertEqual(constructContext.consuming, 0) |
| 149 | + XCTAssertTrue(constructContext.errors.isEmpty) |
| 150 | + |
| 151 | + let error = CodeError("msg", range: "a".startIndex..<"a".startIndex) |
| 152 | + XCTAssertEqual(error.message, "msg") |
| 153 | + XCTAssertNotNil(error.range) |
| 154 | + } |
| 155 | +} |
| 156 | + |
0 commit comments