fix: classify direct report artifacts by status

Treat direct skill report-artifact payloads as task outcomes so partial and empty reports stay successful while blocked and error statuses fail explicitly. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-04-10 17:59:24 +08:00
parent 34035cdc9c
commit 7443b9da7f
4 changed files with 480 additions and 7 deletions
--- a/tests/agent_runtime_test.rs
+++ b/tests/agent_runtime_test.rs
@@ -164,6 +164,57 @@ fn success_browser_response(seq: u64, data: serde_json::Value) -> BrowserMessage
    }
 }

+fn report_artifact_browser_response(
+    seq: u64,
+    status: &str,
+    partial_reasons: &[&str],
+    detail_rows: Vec<serde_json::Value>,
+    summary_rows: Vec<serde_json::Value>,
+) -> BrowserMessage {
+    success_browser_response(
+        seq,
+        serde_json::json!({
+            "text": {
+                "type": "report-artifact",
+                "report_name": "fault-details-report",
+                "period": "2026-03",
+                "selected_range": {
+                    "start": "2026-03-08 16:00:00",
+                    "end": "2026-03-09 16:00:00"
+                },
+                "columns": ["qxdbh"],
+                "rows": detail_rows,
+                "sections": [{
+                    "name": "summary-sheet",
+                    "columns": ["index"],
+                    "rows": summary_rows
+                }],
+                "counts": {
+                    "detail_rows": detail_rows.len(),
+                    "summary_rows": summary_rows.len()
+                },
+                "status": status,
+                "partial_reasons": partial_reasons,
+                "downstream": {
+                    "export": {
+                        "attempted": true,
+                        "success": status != "blocked" && status != "error",
+                        "path": "http://localhost/export.xlsx"
+                    },
+                    "report_log": {
+                        "attempted": true,
+                        "success": partial_reasons.is_empty(),
+                        "error": partial_reasons
+                            .first()
+                            .copied()
+                            .unwrap_or("")
+                    }
+                }
+            }
+        }),
+    )
+}
+
 #[test]
 fn direct_submit_runtime_executes_fault_details_skill_without_provider_path() {
    let skill_root = build_direct_runtime_skill_root();
@@ -204,7 +255,8 @@ fn direct_submit_runtime_executes_fault_details_skill_without_provider_path() {
    )
    .unwrap();

-    assert!(summary.contains("fault_type"));
+    assert!(summary.success);
+    assert!(summary.summary.contains("fault_type"));
    let sent = transport.sent_messages();
    assert!(sent.iter().all(|message| !matches!(message, AgentMessage::LogEntry { level, message } if level == "info" && message.contains("DeepSeek config loaded"))));
    assert!(matches!(
@@ -322,6 +374,162 @@ fn submit_task_rejects_invalid_direct_submit_skill_config_before_routing() {
    assert!(!sent.iter().any(|message| matches!(message, AgentMessage::Command { .. })));
 }

+#[test]
+fn submit_task_treats_partial_report_artifact_as_success_with_warning_summary() {
+    std::env::remove_var("DEEPSEEK_API_KEY");
+    std::env::remove_var("DEEPSEEK_BASE_URL");
+    std::env::remove_var("DEEPSEEK_MODEL");
+
+    let skill_root = build_direct_runtime_skill_root();
+    let runtime_context = direct_submit_runtime_context(&skill_root);
+    let transport = Arc::new(MockTransport::new(vec![report_artifact_browser_response(
+        1,
+        "partial",
+        &["report_log_failed"],
+        vec![serde_json::json!({ "qxdbh": "QX-1" })],
+        vec![serde_json::json!({ "index": 1 })],
+    )]));
+    let browser_tool = BrowserPipeTool::new(
+        transport.clone(),
+        direct_runtime_test_policy(),
+        vec![1, 2, 3, 4, 5, 6, 7, 8],
+    )
+    .with_response_timeout(Duration::from_secs(1));
+
+    handle_browser_message_with_context(
+        transport.as_ref(),
+        &browser_tool,
+        &runtime_context,
+        submit_fault_details_message(),
+    )
+    .unwrap();
+
+    let sent = transport.sent_messages();
+    let completion = direct_submit_completion(&sent).expect("task completion");
+
+    assert!(completion.0, "expected partial artifact to succeed: {sent:?}");
+    assert!(completion.1.contains("fault-details-report"));
+    assert!(completion.1.contains("2026-03"));
+    assert!(completion.1.contains("status=partial"));
+    assert!(completion.1.contains("detail_rows=1"));
+    assert!(completion.1.contains("summary_rows=1"));
+    assert!(completion.1.contains("report_log_failed"));
+}
+
+#[test]
+fn submit_task_treats_empty_report_artifact_as_success() {
+    std::env::remove_var("DEEPSEEK_API_KEY");
+    std::env::remove_var("DEEPSEEK_BASE_URL");
+    std::env::remove_var("DEEPSEEK_MODEL");
+
+    let skill_root = build_direct_runtime_skill_root();
+    let runtime_context = direct_submit_runtime_context(&skill_root);
+    let transport = Arc::new(MockTransport::new(vec![report_artifact_browser_response(
+        1,
+        "empty",
+        &[],
+        vec![],
+        vec![],
+    )]));
+    let browser_tool = BrowserPipeTool::new(
+        transport.clone(),
+        direct_runtime_test_policy(),
+        vec![1, 2, 3, 4, 5, 6, 7, 8],
+    )
+    .with_response_timeout(Duration::from_secs(1));
+
+    handle_browser_message_with_context(
+        transport.as_ref(),
+        &browser_tool,
+        &runtime_context,
+        submit_fault_details_message(),
+    )
+    .unwrap();
+
+    let sent = transport.sent_messages();
+    let completion = direct_submit_completion(&sent).expect("task completion");
+
+    assert!(completion.0, "expected empty artifact to succeed: {sent:?}");
+    assert!(completion.1.contains("status=empty"));
+    assert!(completion.1.contains("detail_rows=0"));
+}
+
+#[test]
+fn submit_task_treats_blocked_report_artifact_as_failure() {
+    std::env::remove_var("DEEPSEEK_API_KEY");
+    std::env::remove_var("DEEPSEEK_BASE_URL");
+    std::env::remove_var("DEEPSEEK_MODEL");
+
+    let skill_root = build_direct_runtime_skill_root();
+    let runtime_context = direct_submit_runtime_context(&skill_root);
+    let transport = Arc::new(MockTransport::new(vec![report_artifact_browser_response(
+        1,
+        "blocked",
+        &["selected_range_unavailable"],
+        vec![],
+        vec![],
+    )]));
+    let browser_tool = BrowserPipeTool::new(
+        transport.clone(),
+        direct_runtime_test_policy(),
+        vec![1, 2, 3, 4, 5, 6, 7, 8],
+    )
+    .with_response_timeout(Duration::from_secs(1));
+
+    handle_browser_message_with_context(
+        transport.as_ref(),
+        &browser_tool,
+        &runtime_context,
+        submit_fault_details_message(),
+    )
+    .unwrap();
+
+    let sent = transport.sent_messages();
+    let completion = direct_submit_completion(&sent).expect("task completion");
+
+    assert!(!completion.0, "expected blocked artifact to fail: {sent:?}");
+    assert!(completion.1.contains("status=blocked"));
+    assert!(completion.1.contains("selected_range_unavailable"));
+}
+
+#[test]
+fn submit_task_treats_error_report_artifact_as_failure() {
+    std::env::remove_var("DEEPSEEK_API_KEY");
+    std::env::remove_var("DEEPSEEK_BASE_URL");
+    std::env::remove_var("DEEPSEEK_MODEL");
+
+    let skill_root = build_direct_runtime_skill_root();
+    let runtime_context = direct_submit_runtime_context(&skill_root);
+    let transport = Arc::new(MockTransport::new(vec![report_artifact_browser_response(
+        1,
+        "error",
+        &["detail_normalization_failed"],
+        vec![],
+        vec![],
+    )]));
+    let browser_tool = BrowserPipeTool::new(
+        transport.clone(),
+        direct_runtime_test_policy(),
+        vec![1, 2, 3, 4, 5, 6, 7, 8],
+    )
+    .with_response_timeout(Duration::from_secs(1));
+
+    handle_browser_message_with_context(
+        transport.as_ref(),
+        &browser_tool,
+        &runtime_context,
+        submit_fault_details_message(),
+    )
+    .unwrap();
+
+    let sent = transport.sent_messages();
+    let completion = direct_submit_completion(&sent).expect("task completion");
+
+    assert!(!completion.0, "expected error artifact to fail: {sent:?}");
+    assert!(completion.1.contains("status=error"));
+    assert!(completion.1.contains("detail_normalization_failed"));
+}
+
 #[test]
 fn direct_skill_mode_logs_direct_skill_primary() {
    std::env::remove_var("DEEPSEEK_API_KEY");
--- a/tests/browser_script_skill_tool_test.rs
+++ b/tests/browser_script_skill_tool_test.rs
@@ -362,6 +362,121 @@ return {
    ));
 }

+#[tokio::test]
+async fn execute_browser_script_tool_preserves_structured_report_artifact_payload() {
+    let skill_dir = unique_temp_dir("sgclaw-browser-script-helper-report-artifact");
+    let scripts_dir = skill_dir.join("scripts");
+    fs::create_dir_all(&scripts_dir).unwrap();
+    fs::write(
+        scripts_dir.join("collect_fault_details.js"),
+        r#"
+return {
+  type: "report-artifact",
+  report_name: "fault-details-report",
+  period: args.period,
+  selected_range: {
+    start: "2026-03-08 16:00:00",
+    end: "2026-03-09 16:00:00"
+  },
+  columns: ["qxdbh"],
+  rows: [{ qxdbh: "QX-1" }],
+  sections: [{ name: "summary-sheet", columns: ["index"], rows: [{ index: 1 }] }],
+  counts: { detail_rows: 1, summary_rows: 1 },
+  status: "partial",
+  partial_reasons: ["report_log_failed"],
+  downstream: {
+    export: { attempted: true, success: true, path: "http://localhost/export.xlsx" },
+    report_log: { attempted: true, success: false, error: "500" }
+  }
+};
+"#,
+    )
+    .unwrap();
+
+    let transport = Arc::new(MockTransport::new(vec![BrowserMessage::Response {
+        seq: 1,
+        success: true,
+        data: json!({
+            "text": {
+                "type": "report-artifact",
+                "report_name": "fault-details-report",
+                "period": "2026-03",
+                "selected_range": {
+                    "start": "2026-03-08 16:00:00",
+                    "end": "2026-03-09 16:00:00"
+                },
+                "columns": ["qxdbh"],
+                "rows": [{ "qxdbh": "QX-1" }],
+                "sections": [{ "name": "summary-sheet", "columns": ["index"], "rows": [{ "index": 1 }] }],
+                "counts": { "detail_rows": 1, "summary_rows": 1 },
+                "status": "partial",
+                "partial_reasons": ["report_log_failed"],
+                "downstream": {
+                    "export": { "attempted": true, "success": true, "path": "http://localhost/export.xlsx" },
+                    "report_log": { "attempted": true, "success": false, "error": "500" }
+                }
+            }
+        }),
+        aom_snapshot: vec![],
+        timing: Timing {
+            queue_ms: 1,
+            exec_ms: 5,
+        },
+    }]));
+    let browser_tool = BrowserPipeTool::new(
+        transport.clone(),
+        test_policy(),
+        vec![1, 2, 3, 4, 5, 6, 7, 8],
+    )
+    .with_response_timeout(Duration::from_secs(1));
+
+    let mut tool_args = HashMap::new();
+    tool_args.insert("period".to_string(), "YYYY-MM period to collect".to_string());
+    let skill_tool = SkillTool {
+        name: "collect_fault_details".to_string(),
+        description: "Collect structured fault details".to_string(),
+        kind: "browser_script".to_string(),
+        command: "scripts/collect_fault_details.js".to_string(),
+        args: tool_args,
+    };
+
+    let result = execute_browser_script_tool(
+        &skill_tool,
+        &skill_dir,
+        browser_tool,
+        json!({
+            "expected_domain": "https://www.zhihu.com/",
+            "period": "2026-03"
+        }),
+    )
+    .await
+    .unwrap();
+
+    assert!(result.success);
+    assert_eq!(
+        serde_json::from_str::<serde_json::Value>(&result.output).unwrap(),
+        json!({
+            "type": "report-artifact",
+            "report_name": "fault-details-report",
+            "period": "2026-03",
+            "selected_range": {
+                "start": "2026-03-08 16:00:00",
+                "end": "2026-03-09 16:00:00"
+            },
+            "columns": ["qxdbh"],
+            "rows": [{ "qxdbh": "QX-1" }],
+            "sections": [{ "name": "summary-sheet", "columns": ["index"], "rows": [{ "index": 1 }] }],
+            "counts": { "detail_rows": 1, "summary_rows": 1 },
+            "status": "partial",
+            "partial_reasons": ["report_log_failed"],
+            "downstream": {
+                "export": { "attempted": true, "success": true, "path": "http://localhost/export.xlsx" },
+                "report_log": { "attempted": true, "success": false, "error": "500" }
+            }
+        })
+    );
+}
+
 fn unique_temp_dir(prefix: &str) -> PathBuf {
    let nanos = SystemTime::now()
        .duration_since(UNIX_EPOCH)