| import Foundation |
|
|
| struct ChatGPTRawConversationsParser: ImportParser { |
| func parse(sourceURL: URL, dedupIndex: DedupIndex) throws -> ImportParseResult { |
| let files = try ImportFiles.immediateFiles( |
| at: sourceURL, |
| expected: "eine ChatGPT-conversations-.json-Datei oder einen passenden Exportordner", |
| directFileMatches: { $0.pathExtension.lowercased() == "json" }, |
| directoryFileMatches: { |
| $0.pathExtension.lowercased() == "json" |
| && $0.lastPathComponent.hasPrefix("conversations") |
| } |
| ) |
| var accumulator = ImportAccumulator( |
| minimumLength: 20, |
| maximumLength: 10_000, |
| dedupIndex: dedupIndex |
| ) |
|
|
| for fileURL in files { |
| let root: Any |
| do { |
| root = try ImportJSON.object(from: fileURL) |
| } catch { |
| accumulator.skipMalformed() |
| continue |
| } |
| guard let conversations = root as? [Any] else { |
| accumulator.skipMalformed() |
| continue |
| } |
|
|
| for conversationValue in conversations { |
| guard let conversation = conversationValue as? [String: Any] else { |
| accumulator.skipMalformed() |
| continue |
| } |
| let model = conversation["model_slug"] as? String |
| guard let mapping = conversation["mapping"] as? [String: Any] else { |
| accumulator.skipMalformed() |
| continue |
| } |
|
|
| for nodeValue in mapping.values { |
| guard |
| let node = nodeValue as? [String: Any], |
| let message = node["message"] as? [String: Any] |
| else { |
| continue |
| } |
| let author = message["author"] as? [String: Any] |
| guard author?["role"] as? String == "user" else { continue } |
| guard |
| let content = message["content"] as? [String: Any], |
| let parts = content["parts"] as? [Any], |
| let text = parts.first as? String |
| else { |
| accumulator.skipMalformed() |
| continue |
| } |
|
|
| accumulator.accept( |
| text, |
| sourcePrefix: "chatgpt-raw", |
| tags: "chatgpt", |
| model: model, |
| promptedAt: ImportTimestamp.normalize(message["create_time"]) |
| ) |
| } |
| } |
| } |
|
|
| return accumulator.result() |
| } |
| } |
|
|
