Add iMessage built-in crawler

README.md

@@ -16,6 +16,7 @@ Built-in manifests ship for:
 - `slacrawl`
 - `discrawl`
 - `telecrawl`
+- `imsgcrawl`
 - `notcrawl`
 - `gogcli` through the `gog` executable
 - `wacli`

Sources/CrawlBarCore/BuiltInApps/BuiltInCrawlApps+Imsgcrawl.swift

@@ -0,0 +1,36 @@
+import Foundation
+
+public extension BuiltInCrawlApps {
+    static let imsgcrawl = CrawlAppManifest(
+        id: Self.imsgcrawlID,
+        displayName: "iMessage",
+        description: "Local-first iMessage archive crawler",
+        binary: .init(name: "imsgcrawl"),
+        branding: .init(
+            symbolName: "message.fill",
+            accentColor: "#34C759",
+            bundleIdentifier: "com.apple.MobileSMS"),
+        paths: .init(
+            defaultDatabase: "~/.imsgcrawl/archive.db",
+            defaultCache: "~/.imsgcrawl/cache",
+            defaultLogs: "~/.imsgcrawl/logs"),
+        commands: [
+            "metadata": ["--json", "metadata"],
+            "status": ["--json", "status"],
+            "refresh": ["--json", "sync"],
+            "search": ["--json", "search"],
+        ],
+        capabilities: [.status, .refresh, .search],
+        statusRequiresSecrets: false,
+        privacy: .init(
+            containsPrivateMessages: true,
+            exportsSecrets: false,
+            localOnlyScopes: [
+                "apple-messages",
+                "sqlite",
+                "contact-handles",
+                "message-archive",
+                "message-text-search",
+            ]))
+        .withSuggestion(Self.appSuggest("Messages", ["com.apple.MobileSMS"]))
+}

Sources/CrawlBarCore/BuiltInApps/BuiltInCrawlApps.swift

@@ -5,6 +5,7 @@ public enum BuiltInCrawlApps {
     public static let slacrawlID = CrawlAppID(rawValue: "slacrawl")
     public static let discrawlID = CrawlAppID(rawValue: "discrawl")
     public static let telecrawlID = CrawlAppID(rawValue: "telecrawl")
+    public static let imsgcrawlID = CrawlAppID(rawValue: "imsgcrawl")
     public static let notcrawlID = CrawlAppID(rawValue: "notcrawl")
     public static let gogcliID = CrawlAppID(rawValue: "gogcli")
     public static let wacliID = CrawlAppID(rawValue: "wacli")
@@ -16,6 +17,7 @@ public enum BuiltInCrawlApps {
         Self.slacrawl,
         Self.discrawl,
         Self.telecrawl,
+        Self.imsgcrawl,
         Self.notcrawl,
         Self.gogcli,
         Self.wacli,

Sources/CrawlBarCore/StatusMapperCrawlKit.swift

@@ -27,7 +27,9 @@ extension CrawlStatusMapper {
             share: self.shareStatus(in: object),
             remote: remote,
             sqliteObject: self.sqliteObjectStatus(in: object),
-            sqliteBundle: self.sqliteBundleStatus(in: object))
+            sqliteBundle: self.sqliteBundleStatus(in: object),
+            warnings: self.stringValues(["warnings"], in: object),
+            errors: self.stringValues(["errors"], in: object))
     }
 
     func isCrawlKitStatus(_ object: [String: Any]) -> Bool {

Sources/CrawlBarCore/StatusMapperJSON.swift

@@ -53,13 +53,21 @@ extension CrawlStatusMapper {
             return .current
         case "warn", "warning", "degraded":
             return .stale
-        case "failed", "failure":
+        case "failed", "failure", "source_error", "archive_error":
             return .error
         default:
             return nil
         }
     }
 
+    func stringValues(_ keys: [String], in object: [String: Any]) -> [String] {
+        for key in keys {
+            guard let values = self.firstValue(key, in: object) as? [Any] else { continue }
+            return values.compactMap { ($0 as? String)?.nilIfBlank }
+        }
+        return []
+    }
+
     func dateValue(_ keys: [String], in object: [String: Any]) -> Date? {
         for key in keys {
             guard let value = self.firstValue(key, in: object) else { continue }

Sources/CrawlBarSelfTest/SelfTestConfig.swift

@@ -160,6 +160,16 @@ extension CrawlBarSelfTest {
         try Self.expect(BuiltInCrawlApps.telecrawl.install?.package == "steipete/tap/telecrawl", "telecrawl install metadata exists")
         try Self.expect(BuiltInCrawlApps.telecrawl.paths.defaultConfig == "~/.telecrawl/backup.json", "telecrawl config path maps")
         try Self.expect(BuiltInCrawlApps.graincrawl.availability == .available, "graincrawl is available")
+        try Self.expect(BuiltInCrawlApps.imsgcrawl.displayName == "iMessage", "imsgcrawl uses user-facing iMessage name")
+        try Self.expect(BuiltInCrawlApps.imsgcrawl.commands["status"] == ["--json", "status"], "imsgcrawl uses crawlkit status command")
+        try Self.expect(BuiltInCrawlApps.imsgcrawl.commands["refresh"] == ["--json", "sync"], "imsgcrawl sync is wired as refresh")
+        try Self.expect(BuiltInCrawlApps.imsgcrawl.commands["search"] == ["--json", "search"], "imsgcrawl search is wired")
+        try Self.expect(BuiltInCrawlApps.imsgcrawl.commands["chats"] == nil, "imsgcrawl chats stay outside persisted actions")
+        try Self.expect(BuiltInCrawlApps.imsgcrawl.commands["messages"] == nil, "imsgcrawl messages stay outside persisted actions")
+        try Self.expect(BuiltInCrawlApps.imsgcrawl.commands["contact-export"] == nil, "imsgcrawl contact export stays outside persisted actions")
+        try Self.expect(BuiltInCrawlApps.imsgcrawl.privacy.containsPrivateMessages, "imsgcrawl privacy metadata flags iMessage data")
+        try Self.expect(BuiltInCrawlApps.imsgcrawl.branding.bundleIdentifier == "com.apple.MobileSMS", "imsgcrawl uses native Messages app icon")
+        try Self.expect(BuiltInCrawlApps.imsgcrawl.suggestion?.name == "Messages", "imsgcrawl suggests from the native Messages app")
         try Self.expect(BuiltInCrawlApps.graincrawl.commands["status"] == ["status", "--json"], "graincrawl uses crawlkit status command")
         try Self.expect(
             BuiltInCrawlApps.graincrawl.commands["refresh"] == ["sync", "--json"],

Sources/CrawlBarSelfTest/SelfTestStatusMapping.swift

@@ -315,6 +315,30 @@ extension CrawlBarSelfTest {
         let failedStatus = CrawlStatusMapper().status(from: failedResult, manifest: BuiltInCrawlApps.graincrawl)
         try Self.expect(failedStatus.state == .error, "crawlkit failed state maps to error")
 
+        let imsgcrawlSourceErrorResult = CrawlCommandResult(
+            appID: BuiltInCrawlApps.imsgcrawlID,
+            action: "status",
+            exitCode: 0,
+            stdout: """
+            {
+              "schema_version": "crawlkit.control.v1",
+              "app_id": "imsgcrawl",
+              "state": "source_error",
+              "summary": "Messages source could not be read.",
+              "warnings": ["archive has not been synced"],
+              "errors": ["Messages database access was denied"]
+            }
+            """,
+            stderr: "",
+            startedAt: Date(),
+            finishedAt: Date())
+        let imsgcrawlSourceErrorStatus = CrawlStatusMapper().status(
+            from: imsgcrawlSourceErrorResult,
+            manifest: BuiltInCrawlApps.imsgcrawl)
+        try Self.expect(imsgcrawlSourceErrorStatus.state == .error, "crawlkit source errors map to error")
+        try Self.expect(imsgcrawlSourceErrorStatus.warnings == ["archive has not been synced"], "crawlkit warnings are preserved")
+        try Self.expect(imsgcrawlSourceErrorStatus.errors == ["Messages database access was denied"], "crawlkit errors are preserved")
+
         let githubAuthMessage = """
         [github] request GET /repos/openclaw/openclaw
         github GET /repos/openclaw/openclaw failed with status 401: {