| import Foundation |
|
|
| struct GeminiJSONLParser: ImportParser { |
| func parse(sourceURL: URL, dedupIndex: DedupIndex) throws -> ImportParseResult { |
| let files = try ImportFiles.immediateFiles( |
| at: sourceURL, |
| expected: "eine Gemini-.jsonl-Datei oder einen Ordner mit .jsonl-Dateien", |
| directFileMatches: { $0.pathExtension.lowercased() == "jsonl" }, |
| directoryFileMatches: { $0.pathExtension.lowercased() == "jsonl" } |
| ) |
| var accumulator = ImportAccumulator( |
| minimumLength: 20, |
| maximumLength: 10_000, |
| dedupIndex: dedupIndex |
| ) |
|
|
| for fileURL in files { |
| let notebook = fileURL.deletingPathExtension().lastPathComponent |
| for line in try ImportFiles.lines(from: fileURL) { |
| let root: Any |
| do { |
| root = try ImportJSON.object(fromLine: line) |
| } catch { |
| accumulator.skipMalformed() |
| continue |
| } |
| guard let object = root as? [String: Any] else { |
| accumulator.skipMalformed() |
| continue |
| } |
| guard object["role"] as? String == "user" else { continue } |
| let text: String |
| if object["content"] == nil { |
| |
| |
| text = "" |
| } else if let content = object["content"] as? String { |
| text = content |
| } else { |
| accumulator.skipMalformed() |
| continue |
| } |
|
|
| accumulator.accept( |
| text, |
| sourcePrefix: "gemini-raw", |
| tags: "gemini,\(notebook.lowercased())", |
| promptedAt: firstTimestamp(in: object) |
| ) |
| } |
| } |
|
|
| return accumulator.result() |
| } |
|
|
| private func firstTimestamp(in object: [String: Any]) -> String? { |
| for key in ["timestamp", "date", "created_at", "create_time"] { |
| if let timestamp = ImportTimestamp.normalize(object[key]) { |
| return timestamp |
| } |
| } |
| return nil |
| } |
| } |
|
|
