| import Foundation |
|
|
| struct ClaudeJSONLParser: ImportParser { |
| private static let agentPrefixes = [ |
| "RESEARCH ONLY", "You are working", "Work in ~/", "Explore the", |
| "Search system", "Check these", "Read ALL", "Find ALL", |
| "do NOT modify", "Make ", "Run ", "Execute ", |
| ] |
|
|
| func parse(sourceURL: URL, dedupIndex: DedupIndex) throws -> ImportParseResult { |
| let files = try ImportFiles.recursiveJSONLFiles(at: sourceURL, excluding: "subagents") |
| var accumulator = ImportAccumulator( |
| minimumLength: 20, |
| maximumLength: 10_000, |
| dedupIndex: dedupIndex |
| ) |
|
|
| for fileURL in files { |
| let project = projectTag(from: fileURL) |
| 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 let message = object["message"] as? [String: Any] else { continue } |
| guard message["role"] as? String == "user" else { continue } |
| |
| |
| let text: String |
| if message["content"] == nil { |
| text = "" |
| } else if let extracted = userText(from: message["content"]) { |
| text = extracted |
| } else { |
| continue |
| } |
|
|
| let trimmed = ImportText.trimmed(text) |
| if Self.agentPrefixes.contains(where: { trimmed.hasPrefix($0) }) { |
| accumulator.skipAgent() |
| continue |
| } |
|
|
| accumulator.accept( |
| trimmed, |
| sourcePrefix: "claude-raw", |
| tags: "claude,\(project)", |
| promptedAt: ImportTimestamp.normalize(object["timestamp"]) |
| ) |
| } |
| } |
|
|
| return accumulator.result() |
| } |
|
|
| private func userText(from content: Any?) -> String? { |
| if let text = content as? String { |
| return text |
| } |
| guard let blocks = content as? [Any] else { return nil } |
|
|
| let textParts = blocks.compactMap { block -> String? in |
| if let text = block as? String { |
| return text |
| } |
| guard |
| let object = block as? [String: Any], |
| object["type"] as? String == "text" |
| else { |
| return nil |
| } |
| return object["text"] as? String ?? "" |
| } |
| return textParts.joined(separator: " ") |
| } |
|
|
| private func projectTag(from fileURL: URL) -> String { |
| for component in fileURL.pathComponents { |
| if let project = decodedProjectComponent(component, homeMarker: "-home-") { |
| return project |
| } |
| if let project = decodedProjectComponent(component, homeMarker: "-Users-") { |
| return project |
| } |
| } |
| return "claude" |
| } |
|
|
| |
| |
| |
| private func decodedProjectComponent(_ component: String, homeMarker: String) -> String? { |
| guard component.hasPrefix(homeMarker) else { return nil } |
| let remainder = component.dropFirst(homeMarker.count) |
| guard let firstDash = remainder.firstIndex(of: "-") else { return nil } |
| let projectStart = remainder.index(after: firstDash) |
| guard projectStart < remainder.endIndex else { return nil } |
| return remainder[projectStart...].lowercased() |
| } |
| } |
|
|
