// SPDX-License-Identifier: GPL-3.0-or-later import Foundation /// Immutable LWASM listing. Source filenames are descriptive metadata only; /// importing a listing never opens those paths or loads its bytes into the CPU. struct DebugSourceListing: Sendable { struct Line: Identifiable, Sendable { let id: Int let text: String let sourceFile: String? let sourceLine: Int? let address: UInt16? let bytes: [UInt8] var location: String { guard let sourceLine else { return "Listing line \(id + 1)" } return "\(sourceFile ?? "Source"):\(sourceLine)" } } enum ImportError: LocalizedError { case tooLarge, encoding, format, notFile, relocatable var errorDescription: String? { switch self { case .tooLarge: return "Use a listing smaller than 8 MB, with at most 50,000 lines and 4,096 characters per line." case .encoding: return "The listing must be UTF-8 or BOM-marked UTF-16 text." case .format: return "No addressed LWASM byte lines were found. Generate a listing with LWASM --list and use final logical CPU addresses." case .notFile: return "Choose a regular listing file." case .relocatable: return "This listing uses relocatable sections. Its addresses need a linker map before they can match CPU addresses. Open a final absolute LWASM listing; CMOC/link-map integration is planned separately." } } } static let maximumBytes = 8 * 1024 * 1024 let name: String let lines: [Line] let byAddress: [UInt16: [Int]] var addressedLineCount: Int { byAddress.values.reduce(0) { $0 + $1.count } } static func read(_ url: URL) throws -> Self { let access = url.startAccessingSecurityScopedResource() defer { if access { url.stopAccessingSecurityScopedResource() } } guard try url.resourceValues(forKeys: [.isRegularFileKey]).isRegularFile == true else { throw ImportError.notFile } let file = try FileHandle(forReadingFrom: url) defer { try? file.close() } let data = try file.read(upToCount: maximumBytes + 1) ?? Data() return try parse(data, name: url.lastPathComponent) } static func parse(_ data: Data, name: String) throws -> Self { guard data.count <= maximumBytes else { throw ImportError.tooLarge } let encoding: String.Encoding = data.starts(with: [0xff, 0xfe]) || data.starts(with: [0xfe, 0xff]) ? .utf16 : .utf8 guard var text = String(data: data, encoding: encoding), !text.contains("\0") else { throw ImportError.encoding } if text.hasPrefix("\u{feff}") { text.removeFirst() } text = text.replacingOccurrences(of: "\r\n", with: "\n").replacingOccurrences(of: "\r", with: "\n") let rawLines = text.split(separator: "\n", omittingEmptySubsequences: false) guard rawLines.count <= 50_000 else { throw ImportError.tooLarge } // LWASM list.c: address/marker + 16 byte digits + space, followed by // (17-column filename):line or --list-nofiles' decimal line number. // Cycle-count columns stay in the displayed text, without interpretation. let metadata = try NSRegularExpression(pattern: #"^(?:\((.{0,17})\):([0-9]{5,})|([0-9]{5,})) (.*)$"#) let section = try NSRegularExpression(pattern: #"(?i)^\s*(?:[a-z_.$@?][a-z0-9_.$@?]*\s+)?(?:section|endsection)\b"#) var lines: [Line] = [] var addresses: [UInt16: [Int]] = [:] for (index, raw) in rawLines.enumerated() { guard raw.count <= 4096 else { throw ImportError.tooLarge } let string = String(raw), octets = Array(string.utf8) var sourceFile: String?, sourceLine: Int?, address: UInt16? var sourceText = string, bytes: [UInt8] = [] if octets.count >= 22, octets[21] == 32 { let remainder = String(decoding: octets.dropFirst(22), as: UTF8.self) let ns = remainder as NSString if let match = metadata.firstMatch(in: remainder, range: NSRange(location: 0, length: ns.length)) { func group(_ i: Int) -> String? { match.range(at: i).location == NSNotFound ? nil : ns.substring(with: match.range(at: i)) } sourceFile = group(1)?.trimmingCharacters(in: .whitespaces) sourceLine = (group(2) ?? group(3)).flatMap(Int.init) sourceText = group(4) ?? "" if section.firstMatch(in: sourceText, range: NSRange(sourceText.startIndex..., in: sourceText)) != nil { throw ImportError.relocatable } let digits = String(decoding: octets[5..<21], as: UTF8.self).trimmingCharacters(in: .whitespaces) // A space denotes a logical address. '.' (OS-9 data) and // 's' (structure offset) are not CPU-code associations. if octets[4] == 32, digits.count > 0, digits.count % 2 == 0, digits.allSatisfy({ $0.isASCII && $0.isHexDigit }), octets.prefix(4).allSatisfy({ (48...57).contains($0) || (65...70).contains($0) || (97...102).contains($0) }), let value = UInt16(String(decoding: octets.prefix(4), as: UTF8.self), radix: 16) { address = value let chars = Array(digits) bytes = stride(from: 0, to: chars.count, by: 2).compactMap { UInt8(String(chars[$0...($0 + 1)]), radix: 16) } addresses[value, default: []].append(index) } } } lines.append(Line(id: index, text: sourceText, sourceFile: sourceFile, sourceLine: sourceLine, address: address, bytes: bytes)) } guard !addresses.isEmpty else { throw ImportError.format } return Self(name: name, lines: lines, byAddress: addresses) } func line(at address: UInt16, preferring selected: Int?) -> Int? { guard let candidates = byAddress[address] else { return nil } return selected.flatMap { candidates.contains($0) ? $0 : nil } ?? candidates.first } func find(_ query: String, after selected: Int?) -> Int? { guard !query.isEmpty else { return nil } let start = selected.map { ($0 + 1) % lines.count } ?? 0 for offset in 0..