| 1 | use super::constants::{ |
| 2 | TOOL_OUTPUT_HEAD_LINES, TOOL_OUTPUT_LINE_LIMIT, TOOL_OUTPUT_TAIL_LINES, |
| 3 | TOOL_SUCCESS_OUTPUT_PREVIEW_LINES, |
| 4 | }; |
| 5 | use super::{ |
| 6 | ASSISTANT_GLYPH, ExecCell, ExecSource, GenericToolCell, HistoryCell, McpToolCell, |
| 7 | PlanUpdateCell, REASONING_CURSOR, REASONING_OPENER, REASONING_RAIL, TOOL_RUNNING_SYMBOLS, |
| 8 | TOOL_STATUS_SYMBOL_MS, ToolCell, ToolStatus, TranscriptRenderOptions, USER_GLYPH, |
| 9 | WebSearchCell, assistant_label_style_for, extract_reasoning_summary, |
| 10 | render_spillover_annotation, render_thinking, render_thinking_with_highlight, |
| 11 | running_status_label_with_elapsed, |
| 12 | }; |
| 13 | use crate::deepseek_theme::Theme; |
| 14 | use crate::models::{ContentBlock, Message}; |
| 15 | use crate::palette; |
| 16 | use crate::tools::plan::{PlanSnapshot, StepStatus}; |
| 17 | use crate::tui::motion::MotionMode; |
| 18 | use crate::tui::ui_text::{line_to_plain, slice_text, text_display_width}; |
| 19 | use ratatui::style::Modifier; |
| 20 | use std::time::{Duration, Instant}; |
| 21 | |
| 22 | #[test] |
| 23 | fn web_search_cell_renders_receipt_source_degradation_and_citations() { |
| 24 | let cell = WebSearchCell { |
| 25 | query: "current release".to_string(), |
| 26 | status: ToolStatus::Success, |
| 27 | summary: Some("Found 2 results".to_string()), |
| 28 | source: Some("provider-native/xai/grok-4.5".to_string()), |
| 29 | degraded: Some("provider_native -> duckduckgo".to_string()), |
| 30 | ref_count: 2, |
| 31 | }; |
| 32 | let rendered = cell |
| 33 | .lines_with_motion(120, true) |
| 34 | .iter() |
| 35 | .map(line_to_plain) |
| 36 | .collect::<Vec<_>>() |
| 37 | .join("\n"); |
| 38 | |
| 39 | assert!(rendered.contains("source")); |
| 40 | assert!(rendered.contains("provider-native/xai/grok-4.5")); |
| 41 | assert!(rendered.contains("degraded")); |
| 42 | assert!(rendered.contains("provider_native -> duckduckgo")); |
| 43 | assert!(rendered.contains("citations")); |
| 44 | } |
| 45 | |
| 46 | // ---- elapsed-seconds badge for long-running tools ---- |
| 47 | // |
| 48 | // Below 3s the label stays "running" — quick reads/greps shouldn't |
| 49 | // visually churn. From 3s onward the badge appears and ticks each |
| 50 | // second so the user can tell the call hasn't hung. |
| 51 | // ---- #4619 adaptive evidence UI receipt ---- |
| 52 | // |
| 53 | // When a tool result carries a `spillover_path` (set by the |
| 54 | // tool-routing layer when the tool's `metadata.spillover_path` is |
| 55 | // populated), expanded live detail appends a calm path-free receipt. |
| 56 | // Transcript-mode replay leaves the hint off because the full output is |
| 57 | // already inline. |
| 58 | |
| 59 | #[test] |
| 60 | fn compact_live_row_does_not_leak_evidence_path() { |
| 61 | use std::path::PathBuf; |
| 62 | let cell = GenericToolCell { |
| 63 | name: "read_file".to_string(), |
| 64 | status: ToolStatus::Success, |
| 65 | input_summary: Some("cmd: cargo build --release".to_string()), |
| 66 | output: Some("very large output...".to_string()), |
| 67 | prompts: None, |
| 68 | spillover_path: Some(PathBuf::from( |
| 69 | "/Users/dev/.deepseek/tool_outputs/call-abc12.txt", |
| 70 | )), |
| 71 | output_summary: None, |
| 72 | is_diff: false, |
| 73 | }; |
| 74 | let lines = cell.lines_with_mode(120, true, super::RenderMode::Live); |
| 75 | let joined: String = lines |
| 76 | .iter() |
| 77 | .flat_map(|l| l.spans.iter().map(|s| s.content.as_ref())) |
| 78 | .collect(); |
| 79 | assert!( |
| 80 | joined.contains("read done · cmd: cargo build --release"), |
| 81 | "expected compact live summary: {joined:?}" |
| 82 | ); |
| 83 | assert!( |
| 84 | !joined.contains("full output:"), |
| 85 | "spillover paths stay out of compact live rows: {joined:?}" |
| 86 | ); |
| 87 | } |
| 88 | |
| 89 | #[test] |
| 90 | fn render_spillover_annotation_omitted_in_transcript_mode() { |
| 91 | use std::path::PathBuf; |
| 92 | // Transcript mode is for replay; the full output is already |
| 93 | // inline so the annotation would just be redundant. |
| 94 | let cell = GenericToolCell { |
| 95 | name: "read_file".to_string(), |
| 96 | status: ToolStatus::Success, |
| 97 | input_summary: None, |
| 98 | output: Some("output".to_string()), |
| 99 | prompts: None, |
| 100 | spillover_path: Some(PathBuf::from("/tmp/spill.txt")), |
| 101 | output_summary: None, |
| 102 | is_diff: false, |
| 103 | }; |
| 104 | let lines = cell.lines_with_mode(120, true, super::RenderMode::Transcript); |
| 105 | let joined: String = lines |
| 106 | .iter() |
| 107 | .flat_map(|l| l.spans.iter().map(|s| s.content.as_ref())) |
| 108 | .collect(); |
| 109 | assert!( |
| 110 | !joined.contains("full output:"), |
| 111 | "annotation should be omitted in transcript mode: {joined:?}" |
| 112 | ); |
| 113 | } |
| 114 | |
| 115 | #[test] |
| 116 | fn workflow_tool_renders_run_card_instead_of_generic_oneliner() { |
| 117 | let output = serde_json::json!({ |
| 118 | "run_id": "workflow_2400c600", |
| 119 | "status": "completed", |
| 120 | "workflow_goal": "audit the FLEET and WORKFLOW docs", |
| 121 | "child_ids": ["a1", "a2", "a3"], |
| 122 | "progress": ["phase: Scan", "log: 3 findings"], |
| 123 | "events": [ |
| 124 | { |
| 125 | "type": "task_started", |
| 126 | "task_id": "a1", |
| 127 | "label": "scan-docs", |
| 128 | "workflow_run_id": "workflow_2400c600", |
| 129 | "workflow_phase_id": "Scan", |
| 130 | "workflow_task_label": "scan-docs", |
| 131 | "workflow_child_index": 0, |
| 132 | }, |
| 133 | { |
| 134 | "type": "task_started", |
| 135 | "task_id": "a2", |
| 136 | "workflow_task_label": "check-fleet", |
| 137 | "workflow_run_id": "workflow_2400c600", |
| 138 | "workflow_child_index": 1, |
| 139 | }, |
| 140 | { |
| 141 | "type": "task_started", |
| 142 | "task_id": "a3", |
| 143 | "label": "summarize", |
| 144 | "workflow_run_id": "workflow_2400c600", |
| 145 | "workflow_child_index": 2, |
| 146 | }, |
| 147 | ], |
| 148 | "schema_errors": [], |
| 149 | }) |
| 150 | .to_string(); |
| 151 | let cell = GenericToolCell { |
| 152 | name: "workflow".to_string(), |
| 153 | status: ToolStatus::Success, |
| 154 | input_summary: Some("action: run".to_string()), |
| 155 | output: Some(output), |
| 156 | prompts: None, |
| 157 | spillover_path: None, |
| 158 | output_summary: None, |
| 159 | is_diff: false, |
| 160 | }; |
| 161 | let joined: String = cell |
| 162 | .lines_with_mode(120, true, super::RenderMode::Live) |
| 163 | .iter() |
| 164 | .flat_map(|l| l.spans.iter().map(|s| s.content.as_ref())) |
| 165 | .collect(); |
| 166 | // Compact (#4122): lifecycle, children, phases, failures, elapsed. |
| 167 | assert!( |
| 168 | joined.contains("3 children") || joined.contains("children"), |
| 169 | "child count: {joined:?}" |
| 170 | ); |
| 171 | assert!( |
| 172 | joined.contains("success") || joined.contains("done"), |
| 173 | "header lifecycle: {joined:?}" |
| 174 | ); |
| 175 | assert!(joined.contains("phase"), "phase count: {joined:?}"); |
| 176 | assert!(joined.contains("fail"), "failure count present: {joined:?}"); |
| 177 | assert!( |
| 178 | joined.contains('s') || joined.contains('m'), |
| 179 | "elapsed: {joined:?}" |
| 180 | ); |
| 181 | assert!( |
| 182 | !joined.contains("status:"), |
| 183 | "body must not repeat the header lifecycle: {joined:?}" |
| 184 | ); |
| 185 | } |
| 186 | |
| 187 | #[test] |
| 188 | fn workflow_tool_expanded_card_shows_phase_child_result_and_failures() { |
| 189 | let output = serde_json::json!({ |
| 190 | "run_id": "workflow_exp", |
| 191 | "status": "failed", |
| 192 | "workflow_goal": "ship v0.8.68", |
| 193 | "started_at_ms": 1000, |
| 194 | "completed_at_ms": 5000, |
| 195 | "source_path": "workflows/demo.workflow.js", |
| 196 | "error": "phase Verify failed", |
| 197 | "result": {"summary": "2 of 3 children ok"}, |
| 198 | "events": [ |
| 199 | { |
| 200 | "type": "run_started", |
| 201 | "at_ms": 1000, |
| 202 | "run_id": "workflow_exp", |
| 203 | "workflow_goal": "ship v0.8.68" |
| 204 | }, |
| 205 | {"type": "phase_started", "at_ms": 1100, "title": "Verify"}, |
| 206 | { |
| 207 | "type": "task_started", |
| 208 | "at_ms": 1200, |
| 209 | "task_id": "t1", |
| 210 | "label": "run tests", |
| 211 | "workflow_task_label": "run tests", |
| 212 | "profile": "implementer" |
| 213 | }, |
| 214 | { |
| 215 | "type": "task_completed", |
| 216 | "at_ms": 4000, |
| 217 | "task_id": "t1", |
| 218 | "status": "failed" |
| 219 | }, |
| 220 | { |
| 221 | "type": "run_completed", |
| 222 | "at_ms": 5000, |
| 223 | "status": "failed", |
| 224 | "error": "phase Verify failed" |
| 225 | } |
| 226 | ] |
| 227 | }) |
| 228 | .to_string(); |
| 229 | let cell = GenericToolCell { |
| 230 | name: "workflow".to_string(), |
| 231 | status: ToolStatus::Failed, |
| 232 | input_summary: Some("action: run".to_string()), |
| 233 | output: Some(output), |
| 234 | prompts: None, |
| 235 | spillover_path: Some(std::path::PathBuf::from("/tmp/wf-artifact.json")), |
| 236 | output_summary: None, |
| 237 | is_diff: false, |
| 238 | }; |
| 239 | let joined: String = cell |
| 240 | .lines_with_mode(140, true, super::RenderMode::Transcript) |
| 241 | .iter() |
| 242 | .flat_map(|l| l.spans.iter().map(|s| s.content.as_ref())) |
| 243 | .collect::<Vec<_>>() |
| 244 | .join("\n"); |
| 245 | assert!(joined.contains("ship v0.8.68"), "goal: {joined}"); |
| 246 | assert!( |
| 247 | joined.contains("phases:") || joined.contains("Verify"), |
| 248 | "phase: {joined}" |
| 249 | ); |
| 250 | assert!( |
| 251 | joined.contains("children:") || joined.contains("child"), |
| 252 | "child: {joined}" |
| 253 | ); |
| 254 | assert!(joined.contains("run tests"), "child label: {joined}"); |
| 255 | assert!( |
| 256 | joined.contains("result:") || joined.contains("2 of 3"), |
| 257 | "final result: {joined}" |
| 258 | ); |
| 259 | assert!( |
| 260 | joined.contains("artifact:") |
| 261 | || joined.contains("source:") |
| 262 | || joined.contains("transcript:"), |
| 263 | "links: {joined}" |
| 264 | ); |
| 265 | assert!( |
| 266 | joined.contains("error:") || joined.contains("phase Verify failed"), |
| 267 | "failure details: {joined}" |
| 268 | ); |
| 269 | } |
| 270 | |
| 271 | #[test] |
| 272 | fn workflow_tool_renders_status_list_card() { |
| 273 | let output = serde_json::json!({ |
| 274 | "action": "status", |
| 275 | "count": 2, |
| 276 | "runs": [ |
| 277 | {"run_id": "workflow_aaa", "status": "running", "child_count": 4}, |
| 278 | {"run_id": "workflow_bbb", "status": "completed", "child_count": 1}, |
| 279 | ], |
| 280 | }) |
| 281 | .to_string(); |
| 282 | let cell = GenericToolCell { |
| 283 | name: "workflow".to_string(), |
| 284 | status: ToolStatus::Success, |
| 285 | input_summary: Some("action: status".to_string()), |
| 286 | output: Some(output), |
| 287 | prompts: None, |
| 288 | spillover_path: None, |
| 289 | output_summary: None, |
| 290 | is_diff: false, |
| 291 | }; |
| 292 | let joined: String = cell |
| 293 | .lines_with_mode(120, true, super::RenderMode::Live) |
| 294 | .iter() |
| 295 | .flat_map(|l| l.spans.iter().map(|s| s.content.as_ref())) |
| 296 | .collect(); |
| 297 | assert!(joined.contains("2 run(s)"), "count header: {joined:?}"); |
| 298 | assert!(joined.contains("workflow_aaa"), "first run row: {joined:?}"); |
| 299 | assert!(joined.contains("running"), "run status: {joined:?}"); |
| 300 | assert!( |
| 301 | joined.contains("workflow_bbb"), |
| 302 | "second run row: {joined:?}" |
| 303 | ); |
| 304 | } |
| 305 | |
| 306 | #[test] |
| 307 | fn render_spillover_annotation_omitted_when_no_path_set() { |
| 308 | // The common case: most tool results don't trigger spillover. |
| 309 | let cell = GenericToolCell { |
| 310 | name: "read_file".to_string(), |
| 311 | status: ToolStatus::Success, |
| 312 | input_summary: None, |
| 313 | output: Some("contents".to_string()), |
| 314 | prompts: None, |
| 315 | spillover_path: None, |
| 316 | output_summary: None, |
| 317 | is_diff: false, |
| 318 | }; |
| 319 | let lines = cell.lines_with_mode(80, true, super::RenderMode::Live); |
| 320 | let joined: String = lines |
| 321 | .iter() |
| 322 | .flat_map(|l| l.spans.iter().map(|s| s.content.as_ref())) |
| 323 | .collect(); |
| 324 | assert!(!joined.contains("full output:"), "{joined:?}"); |
| 325 | } |
| 326 | |
| 327 | #[test] |
| 328 | fn summarize_tool_args_ignores_control_only_defaults() { |
| 329 | let summary = super::summarize_tool_args(&serde_json::json!({ |
| 330 | "max_count": 15, |
| 331 | "timeout_ms": 30_000 |
| 332 | })); |
| 333 | |
| 334 | assert_eq!(summary, None); |
| 335 | } |
| 336 | |
| 337 | #[test] |
| 338 | fn summarize_tool_args_falls_back_to_meaningful_unknown_key() { |
| 339 | let summary = super::summarize_tool_args(&serde_json::json!({ |
| 340 | "max_count": 15, |
| 341 | "branch": "main" |
| 342 | })); |
| 343 | |
| 344 | assert_eq!(summary.as_deref(), Some("branch: main")); |
| 345 | } |
| 346 | |
| 347 | #[test] |
| 348 | fn compact_git_tool_header_names_tool_not_control_default() { |
| 349 | let cell = GenericToolCell { |
| 350 | name: "git_log".to_string(), |
| 351 | status: ToolStatus::Success, |
| 352 | input_summary: Some("max_count: 15".to_string()), |
| 353 | output: None, |
| 354 | prompts: None, |
| 355 | spillover_path: None, |
| 356 | output_summary: None, |
| 357 | is_diff: false, |
| 358 | }; |
| 359 | |
| 360 | let lines = cell.lines_with_mode(120, true, super::RenderMode::Live); |
| 361 | let joined: String = lines |
| 362 | .iter() |
| 363 | .flat_map(|l| l.spans.iter().map(|s| s.content.as_ref())) |
| 364 | .collect(); |
| 365 | |
| 366 | assert_eq!(lines.len(), 1); |
| 367 | assert!( |
| 368 | joined.contains("read done · git_log"), |
| 369 | "expected exact tool name in compact row: {joined:?}" |
| 370 | ); |
| 371 | assert!( |
| 372 | !joined.contains("max_count"), |
| 373 | "control defaults should not become the visible tool summary: {joined:?}" |
| 374 | ); |
| 375 | } |
| 376 | |
| 377 | #[test] |
| 378 | fn compact_unknown_tool_header_names_tool_not_control_default() { |
| 379 | let cell = GenericToolCell { |
| 380 | name: "future_private_tool".to_string(), |
| 381 | status: ToolStatus::Success, |
| 382 | input_summary: Some("max_count: 15".to_string()), |
| 383 | output: None, |
| 384 | prompts: None, |
| 385 | spillover_path: None, |
| 386 | output_summary: None, |
| 387 | is_diff: false, |
| 388 | }; |
| 389 | |
| 390 | let lines = cell.lines_with_mode(120, true, super::RenderMode::Live); |
| 391 | let joined: String = lines |
| 392 | .iter() |
| 393 | .flat_map(|l| l.spans.iter().map(|s| s.content.as_ref())) |
| 394 | .collect(); |
| 395 | |
| 396 | assert_eq!(lines.len(), 1); |
| 397 | assert!( |
| 398 | joined.contains("tool done · future_private_tool"), |
| 399 | "expected exact tool name in compact row: {joined:?}" |
| 400 | ); |
| 401 | assert!( |
| 402 | !joined.contains("max_count"), |
| 403 | "control defaults should not become the visible tool summary: {joined:?}" |
| 404 | ); |
| 405 | } |
| 406 | |
| 407 | #[test] |
| 408 | fn render_spillover_annotation_truncates_to_width() { |
| 409 | use std::path::PathBuf; |
| 410 | let long_path = "/Users/dev/.deepseek/tool_outputs/this-is-a-very-long-tool-call-id-that-will-not-fit-in-narrow-widths.txt"; |
| 411 | let cell = GenericToolCell { |
| 412 | name: "read_file".to_string(), |
| 413 | status: ToolStatus::Success, |
| 414 | input_summary: None, |
| 415 | output: Some("output".to_string()), |
| 416 | prompts: None, |
| 417 | spillover_path: Some(PathBuf::from(long_path)), |
| 418 | output_summary: None, |
| 419 | is_diff: false, |
| 420 | }; |
| 421 | let width = 80; |
| 422 | let lines = cell.lines_with_mode(width, true, super::RenderMode::Live); |
| 423 | let rendered: String = lines |
| 424 | .iter() |
| 425 | .flat_map(|line| line.spans.iter().map(|span| span.content.as_ref())) |
| 426 | .collect(); |
| 427 | assert!( |
| 428 | rendered.contains("Output shortened"), |
| 429 | "compact live rows should expose the calm expand affordance: {rendered:?}" |
| 430 | ); |
| 431 | assert!( |
| 432 | rendered.contains(":output"), |
| 433 | "expected cap:verb affordance, got {rendered:?}" |
| 434 | ); |
| 435 | assert!(!rendered.contains("opens full output"), "{rendered:?}"); |
| 436 | assert!(text_display_width(&rendered) <= usize::from(width)); |
| 437 | assert!(!rendered.contains(long_path)); |
| 438 | } |
| 439 | |
| 440 | #[test] |
| 441 | fn specialized_bash_and_mcp_cells_share_the_calm_expand_affordance() { |
| 442 | let receipt = "head\n\n… 200 KiB of output omitted — view full output in the tool details view\n\n…\ntail"; |
| 443 | let bash = ExecCell { |
| 444 | command: "cargo test".to_string(), |
| 445 | status: ToolStatus::Failed, |
| 446 | output: Some(receipt.to_string()), |
| 447 | live_output: None, |
| 448 | shell_task_id: None, |
| 449 | owner_agent_id: None, |
| 450 | owner_agent_name: None, |
| 451 | started_at: None, |
| 452 | duration_ms: Some(50), |
| 453 | stale_elapsed_since_output_ms: None, |
| 454 | source: ExecSource::Assistant, |
| 455 | interaction: None, |
| 456 | output_summary: None, |
| 457 | }; |
| 458 | let mcp = McpToolCell { |
| 459 | tool: "mcp_fixture".to_string(), |
| 460 | status: ToolStatus::Failed, |
| 461 | content: Some(receipt.to_string()), |
| 462 | is_image: false, |
| 463 | }; |
| 464 | for rendered in [ |
| 465 | lines_text(&ToolCell::Exec(bash).lines_with_motion(80, true)), |
| 466 | lines_text(&ToolCell::Mcp(mcp).lines_with_motion(80, true)), |
| 467 | ] { |
| 468 | assert!(rendered.contains("Output shortened"), "{rendered}"); |
| 469 | assert!( |
| 470 | rendered.contains(":output"), |
| 471 | "expected cap:verb affordance, got {rendered}" |
| 472 | ); |
| 473 | assert!(!rendered.contains("opens full output"), "{rendered}"); |
| 474 | // The chord is a global details affordance, not a per-card stamp (#4718). |
| 475 | assert!(!rendered.contains("Option+V to inspect"), "{rendered}"); |
| 476 | assert!(!rendered.contains("retrieve_tool_result"), "{rendered}"); |
| 477 | assert!(!rendered.contains("head"), "{rendered}"); |
| 478 | } |
| 479 | } |
| 480 | |
| 481 | #[test] |
| 482 | fn adaptive_evidence_affordance_is_calm_path_free_and_width_bounded() { |
| 483 | use std::path::Path; |
| 484 | |
| 485 | let secret_path = Path::new("/Users/private/.codewhale/sessions/session-a/artifacts/hash.txt"); |
| 486 | let expected = format!( |
| 487 | "Output shortened — {}", |
| 488 | crate::tui::key_shortcuts::tool_details_shortcut_action_hint("output") |
| 489 | ); |
| 490 | for width in [18_u16, 40, 80, 120] { |
| 491 | let rendered = line_to_plain(&render_spillover_annotation(width)); |
| 492 | assert!( |
| 493 | text_display_width(&rendered) <= usize::from(width), |
| 494 | "affordance exceeds width {width}: {rendered:?}" |
| 495 | ); |
| 496 | assert!(!rendered.contains("/Users")); |
| 497 | assert!(!rendered.contains("hash.txt")); |
| 498 | assert!(!rendered.contains("Option+V")); |
| 499 | if usize::from(width) >= text_display_width(&expected) { |
| 500 | assert_eq!(rendered, expected); |
| 501 | } |
| 502 | } |
| 503 | // The path parameter is gone: the annotation never exposes storage paths. |
| 504 | let _ = secret_path; |
| 505 | } |
| 506 | |
| 507 | #[test] |
| 508 | fn activity_group_renders_as_single_metadata_line() { |
| 509 | let cell = GenericToolCell { |
| 510 | name: "activity_group".to_string(), |
| 511 | status: ToolStatus::Success, |
| 512 | input_summary: Some("Explored 2 files, 1 search".to_string()), |
| 513 | output: None, |
| 514 | prompts: None, |
| 515 | spillover_path: None, |
| 516 | output_summary: None, |
| 517 | is_diff: false, |
| 518 | }; |
| 519 | |
| 520 | let lines = cell.lines_with_mode(120, true, super::RenderMode::Live); |
| 521 | let joined: String = lines |
| 522 | .iter() |
| 523 | .flat_map(|line| line.spans.iter().map(|span| span.content.as_ref())) |
| 524 | .collect(); |
| 525 | |
| 526 | assert_eq!(lines.len(), 1); |
| 527 | assert_eq!(joined, "Explored 2 files, 1 search"); |
| 528 | assert!(!joined.contains("activity_group")); |
| 529 | } |
| 530 | |
| 531 | // ---- Compact agent rendering ---- |
| 532 | // |
| 533 | // The DelegateCard owns live state for spawned sub-agents; the |
| 534 | // generic tool block previously duplicated that signal at 3-4 lines |
| 535 | // per spawn. In live mode we now render a single compact line that |
| 536 | // points at the spawned agent id; transcript-mode replay keeps the |
| 537 | // full block so debug history is intact. |
| 538 | |
| 539 | #[test] |
| 540 | fn extract_agent_id_pulls_id_from_json_output() { |
| 541 | let output = |
| 542 | r#"{"agent_id": "agent-abc12", "nickname": "Beluga", "model": "deepseek-v4-flash"}"#; |
| 543 | assert_eq!(super::extract_agent_id(output), Some("agent-abc12")); |
| 544 | } |
| 545 | |
| 546 | #[test] |
| 547 | fn extract_agent_id_handles_extra_whitespace() { |
| 548 | let output = r#"{ |
| 549 | "agent_id" : "agent-xyz", |
| 550 | "model": "x" |
| 551 | }"#; |
| 552 | assert_eq!(super::extract_agent_id(output), Some("agent-xyz")); |
| 553 | } |
| 554 | |
| 555 | #[test] |
| 556 | fn extract_agent_id_returns_none_when_missing() { |
| 557 | let output = r#"{"nickname": "Orca", "model": "x"}"#; |
| 558 | assert!(super::extract_agent_id(output).is_none()); |
| 559 | assert!(super::extract_agent_id("(not json)").is_none()); |
| 560 | assert!(super::extract_agent_id("").is_none()); |
| 561 | } |
| 562 | |
| 563 | #[test] |
| 564 | fn extract_agent_id_returns_none_for_empty_id() { |
| 565 | let output = r#"{"agent_id": "", "model": "x"}"#; |
| 566 | assert!(super::extract_agent_id(output).is_none()); |
| 567 | } |
| 568 | |
| 569 | #[test] |
| 570 | fn agent_spawn_suppresses_generic_card_in_live_mode() { |
| 571 | // #4133: spawn cards yield entirely to DelegateCard — no generic tool row. |
| 572 | let cell = GenericToolCell { |
| 573 | name: "agent".to_string(), |
| 574 | status: ToolStatus::Running, |
| 575 | input_summary: Some("prompt: do thing".to_string()), |
| 576 | output: Some( |
| 577 | r#"{"agent_id": "agent-abc12", "nickname": "Beluga", "model": "deepseek-v4-flash"}"# |
| 578 | .to_string(), |
| 579 | ), |
| 580 | prompts: None, |
| 581 | spillover_path: None, |
| 582 | output_summary: None, |
| 583 | is_diff: false, |
| 584 | }; |
| 585 | let lines = cell.lines_with_mode(80, true, super::RenderMode::Live); |
| 586 | assert!( |
| 587 | lines.is_empty(), |
| 588 | "spawn generic tool card must be suppressed: {lines:?}" |
| 589 | ); |
| 590 | } |
| 591 | |
| 592 | #[test] |
| 593 | fn agent_inspection_renders_single_compact_line_in_live_mode() { |
| 594 | let cell = GenericToolCell { |
| 595 | name: "agent".to_string(), |
| 596 | status: ToolStatus::Running, |
| 597 | input_summary: Some("action: peek agent_id: agent-abc12".to_string()), |
| 598 | output: Some( |
| 599 | r#"{"agent_id": "agent-abc12", "nickname": "Beluga", "model": "deepseek-v4-flash"}"# |
| 600 | .to_string(), |
| 601 | ), |
| 602 | prompts: None, |
| 603 | spillover_path: None, |
| 604 | output_summary: None, |
| 605 | is_diff: false, |
| 606 | }; |
| 607 | let lines = cell.lines_with_mode(80, true, super::RenderMode::Live); |
| 608 | assert_eq!(lines.len(), 1, "expected exactly 1 line, got {lines:?}"); |
| 609 | let rendered: String = lines[0].spans.iter().map(|s| s.content.as_ref()).collect(); |
| 610 | assert!( |
| 611 | rendered.contains("agent-abc12"), |
| 612 | "expected agent id in header: {rendered:?}" |
| 613 | ); |
| 614 | assert!( |
| 615 | rendered.contains("checking"), |
| 616 | "expected inspection status in header: {rendered:?}" |
| 617 | ); |
| 618 | assert!( |
| 619 | !rendered.contains("args"), |
| 620 | "args should be hidden: {rendered:?}" |
| 621 | ); |
| 622 | } |
| 623 | |
| 624 | #[test] |
| 625 | fn agent_pending_inspection_uses_fallback_token() { |
| 626 | // Pending inspection (no agent_id yet) still renders a compact check line. |
| 627 | let cell = GenericToolCell { |
| 628 | name: "agent".to_string(), |
| 629 | status: ToolStatus::Running, |
| 630 | input_summary: Some("action: peek prompt: do thing".to_string()), |
| 631 | output: None, |
| 632 | prompts: None, |
| 633 | spillover_path: None, |
| 634 | output_summary: None, |
| 635 | is_diff: false, |
| 636 | }; |
| 637 | let lines = cell.lines_with_mode(80, true, super::RenderMode::Live); |
| 638 | assert_eq!(lines.len(), 1, "inspection must stay compact: {lines:?}"); |
| 639 | let rendered: String = lines[0].spans.iter().map(|s| s.content.as_ref()).collect(); |
| 640 | assert!( |
| 641 | rendered.contains("checking") || rendered.contains("subagent"), |
| 642 | "{rendered:?}" |
| 643 | ); |
| 644 | assert!(!rendered.contains('\u{2026}'), "{rendered:?}"); |
| 645 | } |
| 646 | |
| 647 | #[test] |
| 648 | fn agent_spawn_suppresses_generic_card_in_transcript_mode() { |
| 649 | // #4133: spawn cards are suppressed in both Live and Transcript; DelegateCard |
| 650 | // is the sole visible spawn artifact. |
| 651 | let cell = GenericToolCell { |
| 652 | name: "agent".to_string(), |
| 653 | status: ToolStatus::Success, |
| 654 | input_summary: Some("prompt: do thing".to_string()), |
| 655 | output: Some(r#"{"agent_id": "agent-abc12", "model": "deepseek-v4-flash"}"#.to_string()), |
| 656 | prompts: None, |
| 657 | spillover_path: None, |
| 658 | output_summary: None, |
| 659 | is_diff: false, |
| 660 | }; |
| 661 | let lines = cell.lines_with_mode(80, true, super::RenderMode::Transcript); |
| 662 | assert!( |
| 663 | lines.is_empty(), |
| 664 | "spawn generic tool card must be suppressed in transcript: {lines:?}" |
| 665 | ); |
| 666 | } |
| 667 | |
| 668 | #[test] |
| 669 | fn other_tools_are_unaffected_by_agent_compact_path() { |
| 670 | // Live-mode tool rows are compact by default; raw detail remains |
| 671 | // available through the detail pager. |
| 672 | let cell = GenericToolCell { |
| 673 | name: "read_file".to_string(), |
| 674 | status: ToolStatus::Success, |
| 675 | input_summary: Some("path: foo.rs".to_string()), |
| 676 | output: Some("first line\nsecond line\nthird line".to_string()), |
| 677 | prompts: None, |
| 678 | spillover_path: None, |
| 679 | output_summary: None, |
| 680 | is_diff: false, |
| 681 | }; |
| 682 | let lines = cell.lines_with_mode(80, true, super::RenderMode::Live); |
| 683 | assert_eq!(lines.len(), 1, "live tools should use compact rows"); |
| 684 | } |
| 685 | |
| 686 | #[test] |
| 687 | fn agent_compact_header_omits_unknown_child_fallback() { |
| 688 | // #4148: an inspection whose identity can't be resolved must not leak the |
| 689 | // raw internal "unknown child" token into the default transcript. |
| 690 | let cell = GenericToolCell { |
| 691 | name: "agent".to_string(), |
| 692 | status: ToolStatus::Running, |
| 693 | input_summary: Some("action: peek agent_type: delegate".to_string()), |
| 694 | output: None, |
| 695 | prompts: None, |
| 696 | spillover_path: None, |
| 697 | output_summary: None, |
| 698 | is_diff: false, |
| 699 | }; |
| 700 | let lines = cell.lines_with_mode(80, true, super::RenderMode::Live); |
| 701 | assert_eq!(lines.len(), 1, "inspection must stay compact: {lines:?}"); |
| 702 | let rendered: String = lines[0].spans.iter().map(|s| s.content.as_ref()).collect(); |
| 703 | assert!( |
| 704 | !rendered.contains("unknown child"), |
| 705 | "raw fallback token must not leak: {rendered:?}" |
| 706 | ); |
| 707 | assert!( |
| 708 | rendered.contains("subagent"), |
| 709 | "friendly fallback label should be shown: {rendered:?}" |
| 710 | ); |
| 711 | } |
| 712 | |
| 713 | #[test] |
| 714 | fn agent_compact_header_does_not_duplicate_delegate_verb() { |
| 715 | // #4148: when the resolved identity collapses to the "delegate" verb, the |
| 716 | // compact inspection header must not render a redundant "delegate · delegate". |
| 717 | let cell = GenericToolCell { |
| 718 | name: "agent".to_string(), |
| 719 | status: ToolStatus::Running, |
| 720 | input_summary: Some("action: peek role: delegate".to_string()), |
| 721 | output: None, |
| 722 | prompts: None, |
| 723 | spillover_path: None, |
| 724 | output_summary: None, |
| 725 | is_diff: false, |
| 726 | }; |
| 727 | let lines = cell.lines_with_mode(80, true, super::RenderMode::Live); |
| 728 | assert_eq!(lines.len(), 1, "inspection must stay compact: {lines:?}"); |
| 729 | let rendered: String = lines[0].spans.iter().map(|s| s.content.as_ref()).collect(); |
| 730 | assert!( |
| 731 | !rendered.contains("delegate delegate"), |
| 732 | "no adjacent duplicate: {rendered:?}" |
| 733 | ); |
| 734 | assert_eq!( |
| 735 | rendered.matches("delegate").count(), |
| 736 | 1, |
| 737 | "verb must not be echoed by the summary: {rendered:?}" |
| 738 | ); |
| 739 | } |
| 740 | |
| 741 | // ---- #403 concise todo / checklist update rendering ---- |
| 742 | // |
| 743 | // The tool emits an "Updated todo #N to STATUS" leading line plus a |
| 744 | // JSON snapshot. The renderer should detect the prefix and produce |
| 745 | // a compact one-line state-change card instead of dumping the full |
| 746 | // item list every time. |
| 747 | |
| 748 | #[test] |
| 749 | fn parse_update_prefix_recognises_todo_form() { |
| 750 | let parsed = super::parse_update_prefix("Updated todo #3 to in_progress\n{ \"items\": [...] }"); |
| 751 | assert_eq!( |
| 752 | parsed, |
| 753 | Some(super::ChecklistChange { |
| 754 | id: 3, |
| 755 | status: "in_progress".to_string(), |
| 756 | }), |
| 757 | ); |
| 758 | } |
| 759 | |
| 760 | #[test] |
| 761 | fn parse_update_prefix_recognises_checklist_form() { |
| 762 | let parsed = super::parse_update_prefix("Updated checklist #7 to completed\n{ \"items\": [] }"); |
| 763 | assert_eq!( |
| 764 | parsed, |
| 765 | Some(super::ChecklistChange { |
| 766 | id: 7, |
| 767 | status: "completed".to_string(), |
| 768 | }), |
| 769 | ); |
| 770 | } |
| 771 | |
| 772 | #[test] |
| 773 | fn parse_update_prefix_returns_none_for_writes() { |
| 774 | // `todo_write` / `checklist_write` outputs don't start with |
| 775 | // "Updated …" — they should fall through to the full-card path. |
| 776 | assert!(super::parse_update_prefix("{ \"items\": [] }").is_none()); |
| 777 | assert!(super::parse_update_prefix("Wrote 5 todos\n{}").is_none()); |
| 778 | } |
| 779 | |
| 780 | #[test] |
| 781 | fn parse_update_prefix_returns_none_for_malformed() { |
| 782 | // Missing arrow/status → fall through. |
| 783 | assert!(super::parse_update_prefix("Updated todo #3\n").is_none()); |
| 784 | // Non-numeric id → fall through. |
| 785 | assert!(super::parse_update_prefix("Updated todo #foo to done\n").is_none()); |
| 786 | } |
| 787 | |
| 788 | #[test] |
| 789 | fn render_checklist_change_card_shows_only_changed_item() { |
| 790 | // Build a snapshot with three items; render the change for #2. |
| 791 | let snapshot = super::ChecklistSnapshot { |
| 792 | items: vec![ |
| 793 | super::ChecklistItemSnapshot { |
| 794 | content: "Read the spec".to_string(), |
| 795 | status: "completed".to_string(), |
| 796 | }, |
| 797 | super::ChecklistItemSnapshot { |
| 798 | content: "Write the test".to_string(), |
| 799 | status: "in_progress".to_string(), |
| 800 | }, |
| 801 | super::ChecklistItemSnapshot { |
| 802 | content: "Land the PR".to_string(), |
| 803 | status: "pending".to_string(), |
| 804 | }, |
| 805 | ], |
| 806 | completion_pct: 33, |
| 807 | completed: 1, |
| 808 | total: 3, |
| 809 | }; |
| 810 | let change = super::ChecklistChange { |
| 811 | id: 2, |
| 812 | status: "in_progress".to_string(), |
| 813 | }; |
| 814 | let lines = super::render_checklist_change_card( |
| 815 | "todo_update", |
| 816 | ToolStatus::Success, |
| 817 | &snapshot, |
| 818 | &change, |
| 819 | 80, |
| 820 | true, |
| 821 | ); |
| 822 | // Header + change line + summary affordance = 3 lines. |
| 823 | assert!(lines.len() >= 3, "expected ≥3 lines, got {}", lines.len()); |
| 824 | |
| 825 | // The change line should mention the title and the new status, |
| 826 | // and should NOT include the other two item titles (that's the |
| 827 | // whole point — concise rendering). |
| 828 | let change_line: String = lines[1].spans.iter().map(|s| s.content.as_ref()).collect(); |
| 829 | assert!(change_line.contains("#2"), "missing id: {change_line:?}"); |
| 830 | assert!( |
| 831 | change_line.contains("Write the test"), |
| 832 | "missing title: {change_line:?}" |
| 833 | ); |
| 834 | assert!( |
| 835 | change_line.contains("in_progress"), |
| 836 | "missing status: {change_line:?}" |
| 837 | ); |
| 838 | assert!( |
| 839 | !change_line.contains("Land the PR"), |
| 840 | "should not show other items: {change_line:?}" |
| 841 | ); |
| 842 | assert!( |
| 843 | !change_line.contains("Read the spec"), |
| 844 | "should not show other items: {change_line:?}" |
| 845 | ); |
| 846 | |
| 847 | // The summary line carries the count + explicit details-pager hint. |
| 848 | let summary_line: String = lines |
| 849 | .last() |
| 850 | .unwrap() |
| 851 | .spans |
| 852 | .iter() |
| 853 | .map(|s| s.content.as_ref()) |
| 854 | .collect(); |
| 855 | assert!(summary_line.contains("3 items"), "{summary_line:?}"); |
| 856 | let expected_hint = crate::tui::key_shortcuts::tool_details_shortcut_action_hint("list"); |
| 857 | assert!(summary_line.contains(&expected_hint), "{summary_line:?}"); |
| 858 | } |
| 859 | |
| 860 | #[test] |
| 861 | fn render_checklist_change_card_handles_missing_title_gracefully() { |
| 862 | // If the change targets an out-of-range id, the title falls |
| 863 | // back to a placeholder rather than crashing. |
| 864 | let snapshot = super::ChecklistSnapshot { |
| 865 | items: vec![super::ChecklistItemSnapshot { |
| 866 | content: "only item".to_string(), |
| 867 | status: "pending".to_string(), |
| 868 | }], |
| 869 | completion_pct: 0, |
| 870 | completed: 0, |
| 871 | total: 1, |
| 872 | }; |
| 873 | let change = super::ChecklistChange { |
| 874 | id: 99, |
| 875 | status: "completed".to_string(), |
| 876 | }; |
| 877 | let lines = super::render_checklist_change_card( |
| 878 | "todo_update", |
| 879 | ToolStatus::Success, |
| 880 | &snapshot, |
| 881 | &change, |
| 882 | 80, |
| 883 | true, |
| 884 | ); |
| 885 | let change_line: String = lines[1].spans.iter().map(|s| s.content.as_ref()).collect(); |
| 886 | assert!(change_line.contains("#99")); |
| 887 | assert!(change_line.contains("(missing title)")); |
| 888 | } |
| 889 | |
| 890 | #[test] |
| 891 | fn running_status_label_omits_elapsed_below_threshold() { |
| 892 | assert_eq!(running_status_label_with_elapsed(0), "running"); |
| 893 | assert_eq!(running_status_label_with_elapsed(1), "running"); |
| 894 | assert_eq!(running_status_label_with_elapsed(2), "running"); |
| 895 | } |
| 896 | |
| 897 | #[test] |
| 898 | fn running_status_label_appends_elapsed_at_three_seconds() { |
| 899 | assert_eq!(running_status_label_with_elapsed(3), "running (3s)"); |
| 900 | assert_eq!(running_status_label_with_elapsed(7), "running (7s)"); |
| 901 | assert_eq!(running_status_label_with_elapsed(120), "running (120s)"); |
| 902 | } |
| 903 | |
| 904 | #[test] |
| 905 | fn extract_reasoning_summary_prefers_summary_block() { |
| 906 | let text = "Thinking...\nSummary: First line\nSecond line\n\nTail"; |
| 907 | let summary = extract_reasoning_summary(text).expect("summary should exist"); |
| 908 | assert_eq!(summary, "First line\nSecond line"); |
| 909 | } |
| 910 | |
| 911 | #[test] |
| 912 | fn extract_reasoning_summary_falls_back_to_full_text() { |
| 913 | let text = "Line one\nLine two"; |
| 914 | let summary = extract_reasoning_summary(text).expect("summary should exist"); |
| 915 | assert_eq!(summary, "Line one\nLine two"); |
| 916 | } |
| 917 | |
| 918 | #[test] |
| 919 | fn archived_context_metadata_preserves_spaces_in_attributes() { |
| 920 | let msg = Message { |
| 921 | role: "assistant".to_string(), |
| 922 | content: vec![ContentBlock::Text { |
| 923 | text: "<archived_context level=\"1\" range=\"msg 0-128\" tokens=\"2499\" density=\"~2,500 tokens\" model=\"deepseek-v4-flash\" timestamp=\"2026-04-28T00:00:00Z\">\nSummary body\n</archived_context>".to_string(), |
| 924 | cache_control: None, |
| 925 | }], |
| 926 | }; |
| 927 | |
| 928 | let cells = super::history_cells_from_message(&msg); |
| 929 | assert_eq!(cells.len(), 1); |
| 930 | let HistoryCell::ArchivedContext { |
| 931 | level, |
| 932 | range, |
| 933 | tokens, |
| 934 | density, |
| 935 | model, |
| 936 | timestamp, |
| 937 | summary, |
| 938 | } = &cells[0] |
| 939 | else { |
| 940 | panic!("expected archived context cell"); |
| 941 | }; |
| 942 | |
| 943 | assert_eq!(*level, 1); |
| 944 | assert_eq!(range, "msg 0-128"); |
| 945 | assert_eq!(tokens, "2499"); |
| 946 | assert_eq!(density, "~2,500 tokens"); |
| 947 | assert_eq!(model, "deepseek-v4-flash"); |
| 948 | assert_eq!(timestamp, "2026-04-28T00:00:00Z"); |
| 949 | assert_eq!(summary, "Summary body"); |
| 950 | } |
| 951 | |
| 952 | #[test] |
| 953 | fn tool_history_repair_receipt_renders_as_system_history() { |
| 954 | let msg = Message { |
| 955 | role: "assistant".to_string(), |
| 956 | content: vec![ContentBlock::Text { |
| 957 | text: "[tool_history_repair] Repaired 1 crashed tool call(s); quarantined 0 duplicate and 0 orphan terminal result(s).".to_string(), |
| 958 | cache_control: None, |
| 959 | }], |
| 960 | }; |
| 961 | |
| 962 | let cells = super::history_cells_from_message(&msg); |
| 963 | |
| 964 | assert!(matches!( |
| 965 | cells.as_slice(), |
| 966 | [HistoryCell::System { content }] if content.starts_with("[tool_history_repair]") |
| 967 | )); |
| 968 | } |
| 969 | |
| 970 | #[test] |
| 971 | fn user_history_hides_only_the_trailing_turn_metadata_block() { |
| 972 | let visible = "Explain this literal: <turn_meta>example</turn_meta>"; |
| 973 | let turn_meta = concat!( |
| 974 | "<turn_meta>\n", |
| 975 | "Current local date: 2026-07-22\n", |
| 976 | "Input provenance: external_user\n", |
| 977 | "Input authority: external_current_turn\n", |
| 978 | "</turn_meta>", |
| 979 | ); |
| 980 | let msg = Message { |
| 981 | role: "user".to_string(), |
| 982 | content: vec![ |
| 983 | ContentBlock::Text { |
| 984 | text: visible.to_string(), |
| 985 | cache_control: None, |
| 986 | }, |
| 987 | ContentBlock::Text { |
| 988 | text: turn_meta.to_string(), |
| 989 | cache_control: None, |
| 990 | }, |
| 991 | ], |
| 992 | }; |
| 993 | |
| 994 | let cells = super::history_cells_from_message(&msg); |
| 995 | |
| 996 | assert!(matches!( |
| 997 | cells.as_slice(), |
| 998 | [HistoryCell::User { content }] if content == visible |
| 999 | )); |
| 1000 | |
| 1001 | let literal_only = Message { |
| 1002 | role: "user".to_string(), |
| 1003 | content: vec![ContentBlock::Text { |
| 1004 | text: "<turn_meta>user-authored example</turn_meta>".to_string(), |
| 1005 | cache_control: None, |
| 1006 | }], |
| 1007 | }; |
| 1008 | let literal_cells = super::history_cells_from_message(&literal_only); |
| 1009 | assert!(matches!( |
| 1010 | literal_cells.as_slice(), |
| 1011 | [HistoryCell::User { content }] |
| 1012 | if content == "<turn_meta>user-authored example</turn_meta>" |
| 1013 | )); |
| 1014 | } |
| 1015 | |
| 1016 | #[test] |
| 1017 | fn history_replays_update_plan_tool_use_as_plan_card() { |
| 1018 | let msg = Message { |
| 1019 | role: "assistant".to_string(), |
| 1020 | content: vec![ContentBlock::ToolUse { |
| 1021 | id: "plan-1".to_string(), |
| 1022 | name: "update_plan".to_string(), |
| 1023 | input: serde_json::json!({ |
| 1024 | "objective": "Make Plan mode reviewable", |
| 1025 | "sources_used": ["gh issue view 2691"], |
| 1026 | "critical_files": ["crates/tui/src/tools/plan.rs"], |
| 1027 | "plan": [ |
| 1028 | { "step": "render replay card", "status": "completed" } |
| 1029 | ] |
| 1030 | }), |
| 1031 | caller: None, |
| 1032 | }], |
| 1033 | }; |
| 1034 | |
| 1035 | let cells = super::history_cells_from_message(&msg); |
| 1036 | assert_eq!(cells.len(), 1); |
| 1037 | let HistoryCell::Tool(ToolCell::PlanUpdate(cell)) = &cells[0] else { |
| 1038 | panic!("expected update_plan replay cell"); |
| 1039 | }; |
| 1040 | |
| 1041 | assert_eq!(cell.status, ToolStatus::Success); |
| 1042 | assert_eq!( |
| 1043 | cell.snapshot.objective.as_deref(), |
| 1044 | Some("Make Plan mode reviewable") |
| 1045 | ); |
| 1046 | assert_eq!(cell.snapshot.sources_used, vec!["gh issue view 2691"]); |
| 1047 | assert_eq!(cell.snapshot.items[0].status, StepStatus::Completed); |
| 1048 | } |
| 1049 | |
| 1050 | #[test] |
| 1051 | fn render_thinking_collapsed_shows_details_affordance() { |
| 1052 | let lines = render_thinking( |
| 1053 | "Summary: First line\nSecond line\nThird line\nFourth line\nFifth line", |
| 1054 | 80, |
| 1055 | false, |
| 1056 | Some(2.0), |
| 1057 | true, |
| 1058 | false, |
| 1059 | ); |
| 1060 | let text = lines |
| 1061 | .iter() |
| 1062 | .flat_map(|line| line.spans.iter().map(|span| span.content.as_ref())) |
| 1063 | .collect::<String>(); |
| 1064 | assert!(text.contains("Ctrl+O:detail"), "{text}"); |
| 1065 | assert!(text.contains("Space:expand"), "{text}"); |
| 1066 | // Pin the actual header shape ("… reasoning done") — a bare |
| 1067 | // `contains("reasoning")` is already satisfied by the Ctrl+O |
| 1068 | // affordance line above and would never fail on its own. |
| 1069 | let header = lines |
| 1070 | .first() |
| 1071 | .map(|line| { |
| 1072 | line.spans |
| 1073 | .iter() |
| 1074 | .map(|span| span.content.as_ref()) |
| 1075 | .collect::<String>() |
| 1076 | }) |
| 1077 | .unwrap_or_default(); |
| 1078 | assert!( |
| 1079 | header.starts_with(REASONING_OPENER), |
| 1080 | "header opens with the dotted opener: {header:?}" |
| 1081 | ); |
| 1082 | assert!( |
| 1083 | header.contains("reasoning done"), |
| 1084 | "header carries the reasoning title and done status: {header:?}" |
| 1085 | ); |
| 1086 | } |
| 1087 | |
| 1088 | #[test] |
| 1089 | fn render_thinking_streaming_collapsed_shows_live_content() { |
| 1090 | // #861 RC4 / #1324: during a live thinking block in collapsed view, |
| 1091 | // the body must NOT be blanked out. Users want to watch the model |
| 1092 | // think; the previous behaviour stalled on a "thinking..." spinner |
| 1093 | // until ThinkingComplete fired. |
| 1094 | let lines = render_thinking( |
| 1095 | "Step 1: read the code\nStep 2: trace the call\nStep 3: form a hypothesis", |
| 1096 | 80, |
| 1097 | true, // streaming |
| 1098 | None, // no duration yet |
| 1099 | true, // collapsed |
| 1100 | true, // low_motion (no cursor noise to grep) |
| 1101 | ); |
| 1102 | let text = lines |
| 1103 | .iter() |
| 1104 | .flat_map(|line| line.spans.iter().map(|span| span.content.as_ref())) |
| 1105 | .collect::<String>(); |
| 1106 | assert!( |
| 1107 | text.contains("Step 3: form a hypothesis"), |
| 1108 | "the most recent thinking line must be visible during streaming, got: {text}" |
| 1109 | ); |
| 1110 | // "thinking..." placeholder must not be the only thing rendered. |
| 1111 | assert!( |
| 1112 | !text.contains("thinking..."), |
| 1113 | "raw content present means the placeholder line should not be drawn, got: {text}" |
| 1114 | ); |
| 1115 | } |
| 1116 | |
| 1117 | #[test] |
| 1118 | fn render_hidden_streaming_thinking_shows_activity_without_content() { |
| 1119 | let cell = HistoryCell::Thinking { |
| 1120 | content: "private chain of thought that must not be shown".to_string(), |
| 1121 | streaming: true, |
| 1122 | duration_secs: None, |
| 1123 | }; |
| 1124 | |
| 1125 | let lines = cell.lines_with_options( |
| 1126 | 80, |
| 1127 | TranscriptRenderOptions { |
| 1128 | show_thinking: false, |
| 1129 | low_motion: true, |
| 1130 | ..TranscriptRenderOptions::default() |
| 1131 | }, |
| 1132 | ); |
| 1133 | let text = lines_text(&lines); |
| 1134 | |
| 1135 | assert!( |
| 1136 | text.contains("reasoning hidden"), |
| 1137 | "hidden live thinking should still show progress: {text}" |
| 1138 | ); |
| 1139 | assert_eq!( |
| 1140 | lines.len(), |
| 1141 | 1, |
| 1142 | "hidden reasoning should have one compact status treatment: {text}" |
| 1143 | ); |
| 1144 | assert!( |
| 1145 | !text.contains("reasoning live") && !text.contains("model is still working"), |
| 1146 | "hidden reasoning should not stack duplicate live-state copy: {text}" |
| 1147 | ); |
| 1148 | assert!( |
| 1149 | !text.contains("private chain of thought"), |
| 1150 | "hidden live thinking must not reveal content: {text}" |
| 1151 | ); |
| 1152 | } |
| 1153 | |
| 1154 | #[test] |
| 1155 | fn render_hidden_completed_thinking_stays_hidden() { |
| 1156 | let cell = HistoryCell::Thinking { |
| 1157 | content: "completed hidden reasoning".to_string(), |
| 1158 | streaming: false, |
| 1159 | duration_secs: Some(1.0), |
| 1160 | }; |
| 1161 | |
| 1162 | let lines = cell.lines_with_options( |
| 1163 | 80, |
| 1164 | TranscriptRenderOptions { |
| 1165 | show_thinking: false, |
| 1166 | ..TranscriptRenderOptions::default() |
| 1167 | }, |
| 1168 | ); |
| 1169 | |
| 1170 | assert!( |
| 1171 | lines.is_empty(), |
| 1172 | "completed hidden thinking should stay out of the transcript" |
| 1173 | ); |
| 1174 | } |
| 1175 | |
| 1176 | #[test] |
| 1177 | fn render_thinking_streaming_truncated_shows_continues_affordance() { |
| 1178 | // #861 RC4: when a streaming thinking block exceeds the line cap, |
| 1179 | // surface a live affordance pointing at Ctrl+O. The earlier code |
| 1180 | // suppressed the affordance unless `!streaming`. |
| 1181 | let long = (1..=16) |
| 1182 | .map(|i| format!("Reasoning line {i}")) |
| 1183 | .collect::<Vec<_>>() |
| 1184 | .join("\n"); |
| 1185 | let lines = render_thinking(&long, 80, true, None, true, true); |
| 1186 | let text = lines |
| 1187 | .iter() |
| 1188 | .flat_map(|line| line.spans.iter().map(|span| span.content.as_ref())) |
| 1189 | .collect::<String>(); |
| 1190 | assert!( |
| 1191 | text.contains("Ctrl+O:more"), |
| 1192 | "streaming-truncation affordance missing, got: {text}" |
| 1193 | ); |
| 1194 | // The most recent line must be the visible tail (head dropped). |
| 1195 | assert!( |
| 1196 | text.contains("Reasoning line 16"), |
| 1197 | "tail line missing, got: {text}" |
| 1198 | ); |
| 1199 | assert!( |
| 1200 | !text.contains("Reasoning line 1\n"), |
| 1201 | "head should be clipped, got: {text}" |
| 1202 | ); |
| 1203 | } |
| 1204 | |
| 1205 | #[test] |
| 1206 | fn tool_lines_with_options_respects_low_motion_in_default_path() { |
| 1207 | // Use a 2× cycle offset so the animated frame lands on index 2, |
| 1208 | // which is maximally far from index 0. This avoids flaky failures on |
| 1209 | // platforms with coarse timer resolution (Windows ≈ 15.6 ms) and |
| 1210 | // gives several frame intervals of headroom before the index could |
| 1211 | // wrap back to 0. |
| 1212 | let started_at = Some( |
| 1213 | Instant::now() |
| 1214 | - Duration::from_millis( |
| 1215 | crate::tui::spinner::LIVE_MARKER_DELAY_MS + TOOL_STATUS_SYMBOL_MS * 2, |
| 1216 | ), |
| 1217 | ); |
| 1218 | let cell = HistoryCell::Tool(ToolCell::Exec(ExecCell { |
| 1219 | command: "echo hi".to_string(), |
| 1220 | status: ToolStatus::Running, |
| 1221 | output: None, |
| 1222 | live_output: None, |
| 1223 | shell_task_id: None, |
| 1224 | owner_agent_id: None, |
| 1225 | owner_agent_name: None, |
| 1226 | started_at, |
| 1227 | duration_ms: None, |
| 1228 | stale_elapsed_since_output_ms: None, |
| 1229 | source: ExecSource::Assistant, |
| 1230 | interaction: None, |
| 1231 | output_summary: None, |
| 1232 | })); |
| 1233 | |
| 1234 | let animated = cell.lines_with_options(80, TranscriptRenderOptions::default()); |
| 1235 | let low_motion = cell.lines_with_options( |
| 1236 | 80, |
| 1237 | TranscriptRenderOptions { |
| 1238 | low_motion: true, |
| 1239 | motion_mode: MotionMode::Reduced, |
| 1240 | ..TranscriptRenderOptions::default() |
| 1241 | }, |
| 1242 | ); |
| 1243 | let still = cell.lines_with_options( |
| 1244 | 80, |
| 1245 | TranscriptRenderOptions { |
| 1246 | low_motion: true, |
| 1247 | motion_mode: MotionMode::Still, |
| 1248 | ..TranscriptRenderOptions::default() |
| 1249 | }, |
| 1250 | ); |
| 1251 | |
| 1252 | // Index 0 is card-rail glyph (╭); the animated symbol is at index 1. |
| 1253 | let animated_symbol = animated[0].spans[1].content.trim(); |
| 1254 | let low_motion_symbol = low_motion[0].spans[1].content.trim(); |
| 1255 | let still_symbol = still[0].spans[1].content.trim(); |
| 1256 | |
| 1257 | // Reduced motion freezes at a filled, legible bubble rather than an |
| 1258 | // invisible blank braille cell. |
| 1259 | assert_eq!(low_motion_symbol, "⣤"); |
| 1260 | assert_eq!(still_symbol, "›"); |
| 1261 | // The animated path should be on a different frame (index 2). |
| 1262 | assert_ne!(animated_symbol, TOOL_RUNNING_SYMBOLS[0]); |
| 1263 | } |
| 1264 | |
| 1265 | #[test] |
| 1266 | fn reduced_verify_marker_uses_the_shared_calm_glyph() { |
| 1267 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 1268 | name: "run_verifiers".to_string(), |
| 1269 | status: ToolStatus::Running, |
| 1270 | input_summary: None, |
| 1271 | output: None, |
| 1272 | prompts: None, |
| 1273 | spillover_path: None, |
| 1274 | output_summary: None, |
| 1275 | is_diff: false, |
| 1276 | })); |
| 1277 | let lines = cell.lines_with_options( |
| 1278 | 80, |
| 1279 | TranscriptRenderOptions { |
| 1280 | low_motion: true, |
| 1281 | motion_mode: MotionMode::Reduced, |
| 1282 | ..TranscriptRenderOptions::default() |
| 1283 | }, |
| 1284 | ); |
| 1285 | |
| 1286 | assert_eq!(lines[0].spans[1].content.trim(), "⣤"); |
| 1287 | } |
| 1288 | |
| 1289 | #[test] |
| 1290 | fn still_fanout_marker_uses_the_shared_chevron() { |
| 1291 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 1292 | name: "workflow".to_string(), |
| 1293 | status: ToolStatus::Running, |
| 1294 | input_summary: Some("action: run".to_string()), |
| 1295 | output: None, |
| 1296 | prompts: None, |
| 1297 | spillover_path: None, |
| 1298 | output_summary: None, |
| 1299 | is_diff: false, |
| 1300 | })); |
| 1301 | let lines = cell.lines_with_options( |
| 1302 | 80, |
| 1303 | TranscriptRenderOptions { |
| 1304 | low_motion: true, |
| 1305 | motion_mode: MotionMode::Still, |
| 1306 | ..TranscriptRenderOptions::default() |
| 1307 | }, |
| 1308 | ); |
| 1309 | |
| 1310 | assert_eq!(lines[0].spans[1].content.trim(), "›"); |
| 1311 | assert_eq!(lines[0].spans[2].content.trim(), "⋮⋮"); |
| 1312 | } |
| 1313 | |
| 1314 | #[test] |
| 1315 | fn still_marker_rewrite_never_consumes_braille_tool_output() { |
| 1316 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 1317 | name: "read_file".to_string(), |
| 1318 | status: ToolStatus::Running, |
| 1319 | input_summary: None, |
| 1320 | output: Some("⣿".to_string()), |
| 1321 | prompts: None, |
| 1322 | spillover_path: None, |
| 1323 | output_summary: None, |
| 1324 | is_diff: false, |
| 1325 | })); |
| 1326 | let lines = cell.lines_with_options( |
| 1327 | 80, |
| 1328 | TranscriptRenderOptions { |
| 1329 | low_motion: true, |
| 1330 | motion_mode: MotionMode::Still, |
| 1331 | ..TranscriptRenderOptions::default() |
| 1332 | }, |
| 1333 | ); |
| 1334 | |
| 1335 | assert_eq!(lines[0].spans[1].content.trim(), "›"); |
| 1336 | assert!( |
| 1337 | lines |
| 1338 | .iter() |
| 1339 | .flat_map(|line| line.spans.iter()) |
| 1340 | .any(|span| span.content.as_ref() == "⣿"), |
| 1341 | "tool output must survive the typed-header marker pass: {lines:?}" |
| 1342 | ); |
| 1343 | } |
| 1344 | |
| 1345 | // === Speaker glyph tests (v0.6.6 UI redesign) === |
| 1346 | // |
| 1347 | // The literal "Assistant" / "You" labels are replaced by the calmer |
| 1348 | // bullet/bar glyphs (`●` / `▎`). Only the assistant glyph pulses, and |
| 1349 | // only while the cell is streaming — finished turns sit at the source |
| 1350 | // sky color so the transcript reads as solid history. |
| 1351 | |
| 1352 | #[test] |
| 1353 | fn user_cell_renders_with_bar_glyph_not_literal_label() { |
| 1354 | let cell = HistoryCell::User { |
| 1355 | content: "hello".to_string(), |
| 1356 | }; |
| 1357 | let lines = cell.lines(80); |
| 1358 | let head = &lines[0]; |
| 1359 | assert_eq!(head.spans[0].content.as_ref(), USER_GLYPH); |
| 1360 | assert_eq!(head.spans[0].style.fg, Some(palette::USER_BODY)); |
| 1361 | assert_eq!(head.style.bg, Some(palette::SURFACE_ELEVATED)); |
| 1362 | assert_eq!(head.width(), 80); |
| 1363 | assert!( |
| 1364 | head.spans.iter().any(|span| span.style.bg.is_none()), |
| 1365 | "content spans should keep their own styles and inherit the line background" |
| 1366 | ); |
| 1367 | // No "You" literal anywhere in the rendered head line. |
| 1368 | let visible: String = head |
| 1369 | .spans |
| 1370 | .iter() |
| 1371 | .map(|s| s.content.as_ref()) |
| 1372 | .collect::<String>(); |
| 1373 | assert!(!visible.contains("You"), "user label dropped: {visible:?}"); |
| 1374 | assert!(visible.contains("hello")); |
| 1375 | } |
| 1376 | |
| 1377 | #[test] |
| 1378 | fn user_cell_wraps_fill_transcript_rows() { |
| 1379 | let cell = HistoryCell::User { |
| 1380 | content: "hello world this prompt wraps onto multiple transcript lines".to_string(), |
| 1381 | }; |
| 1382 | let lines = cell.lines(18); |
| 1383 | |
| 1384 | assert!(lines.len() > 1, "expected wrapped user message"); |
| 1385 | assert!( |
| 1386 | lines |
| 1387 | .iter() |
| 1388 | .all(|line| line.style.bg == Some(palette::SURFACE_ELEVATED)), |
| 1389 | "wrapped user message lines should keep the highlighted block background" |
| 1390 | ); |
| 1391 | assert!( |
| 1392 | lines.iter().all(|line| line.width() == 18), |
| 1393 | "wrapped user message lines should fill the rendered row width" |
| 1394 | ); |
| 1395 | } |
| 1396 | |
| 1397 | #[test] |
| 1398 | fn user_transcript_lines_do_not_append_visual_padding() { |
| 1399 | let cell = HistoryCell::User { |
| 1400 | content: "hello".to_string(), |
| 1401 | }; |
| 1402 | let lines = cell.transcript_lines(80); |
| 1403 | let head = &lines[0]; |
| 1404 | let visible: String = head.spans.iter().map(|s| s.content.as_ref()).collect(); |
| 1405 | |
| 1406 | assert_eq!(visible, format!("{USER_GLYPH} hello")); |
| 1407 | assert!(head.width() < 80); |
| 1408 | assert_eq!(head.style.bg, None); |
| 1409 | } |
| 1410 | |
| 1411 | #[test] |
| 1412 | fn user_cell_renders_plain_text_without_markdown_interpretation() { |
| 1413 | let cell = HistoryCell::User { |
| 1414 | content: " # heading\n- item\n \nhello world".to_string(), |
| 1415 | }; |
| 1416 | let visible: Vec<String> = cell.lines(80).iter().map(line_text).collect(); |
| 1417 | |
| 1418 | assert_eq!(visible[0].trim_end(), format!("{USER_GLYPH} # heading")); |
| 1419 | assert!( |
| 1420 | visible[1].trim_end().ends_with("- item"), |
| 1421 | "dash-prefixed text must remain literal: {visible:?}" |
| 1422 | ); |
| 1423 | assert!( |
| 1424 | visible[2].ends_with(" "), |
| 1425 | "whitespace-only lines must survive: {visible:?}" |
| 1426 | ); |
| 1427 | assert!( |
| 1428 | visible[3].trim_end().ends_with("hello world"), |
| 1429 | "internal spacing must remain literal: {visible:?}" |
| 1430 | ); |
| 1431 | assert!( |
| 1432 | !visible.iter().any(|line| line.contains('\u{2500}')), |
| 1433 | "plain user heading must not add markdown heading rule: {visible:?}" |
| 1434 | ); |
| 1435 | } |
| 1436 | |
| 1437 | #[test] |
| 1438 | fn assistant_cell_renders_with_bullet_glyph_not_literal_label() { |
| 1439 | let cell = HistoryCell::Assistant { |
| 1440 | content: "ready".to_string(), |
| 1441 | streaming: false, |
| 1442 | }; |
| 1443 | let lines = cell.lines(80); |
| 1444 | let head = &lines[0]; |
| 1445 | assert_eq!(head.spans[0].content.as_ref(), ASSISTANT_GLYPH); |
| 1446 | let visible: String = head |
| 1447 | .spans |
| 1448 | .iter() |
| 1449 | .map(|s| s.content.as_ref()) |
| 1450 | .collect::<String>(); |
| 1451 | assert!( |
| 1452 | !visible.contains("Assistant"), |
| 1453 | "assistant label dropped: {visible:?}" |
| 1454 | ); |
| 1455 | assert!(visible.contains("ready")); |
| 1456 | assert_ne!(head.style.bg, Some(palette::SURFACE_ELEVATED)); |
| 1457 | } |
| 1458 | |
| 1459 | #[test] |
| 1460 | fn copy_metadata_strips_tool_receipt_chrome_but_keeps_text() { |
| 1461 | let cell = HistoryCell::Tool(ToolCell::Exec(ExecCell { |
| 1462 | command: "printf 'receipt'".to_string(), |
| 1463 | status: ToolStatus::Success, |
| 1464 | output: Some("receipt".to_string()), |
| 1465 | live_output: None, |
| 1466 | shell_task_id: None, |
| 1467 | owner_agent_id: None, |
| 1468 | owner_agent_name: None, |
| 1469 | started_at: None, |
| 1470 | duration_ms: None, |
| 1471 | stale_elapsed_since_output_ms: None, |
| 1472 | source: ExecSource::Assistant, |
| 1473 | interaction: None, |
| 1474 | output_summary: None, |
| 1475 | })); |
| 1476 | let rendered = cell.lines_with_copy_metadata(80, TranscriptRenderOptions::default()); |
| 1477 | let header = rendered.first().expect("tool receipt header"); |
| 1478 | assert!( |
| 1479 | header.copy_prefix_width >= 4, |
| 1480 | "missing status/family chrome width" |
| 1481 | ); |
| 1482 | assert!( |
| 1483 | header |
| 1484 | .line |
| 1485 | .spans |
| 1486 | .iter() |
| 1487 | .any(|span| span.content.contains("receipt")), |
| 1488 | "receipt text must remain in the rendered copy source" |
| 1489 | ); |
| 1490 | let header_text = line_to_plain(&ratatui::text::Line::from( |
| 1491 | header |
| 1492 | .line |
| 1493 | .spans |
| 1494 | .iter() |
| 1495 | .skip(1) |
| 1496 | .cloned() |
| 1497 | .collect::<Vec<_>>(), |
| 1498 | )); |
| 1499 | let copied = slice_text( |
| 1500 | &header_text, |
| 1501 | header.copy_prefix_width, |
| 1502 | text_display_width(&header_text), |
| 1503 | ); |
| 1504 | assert!( |
| 1505 | !copied.contains('✓'), |
| 1506 | "status chrome leaked into copy: {copied:?}" |
| 1507 | ); |
| 1508 | assert!( |
| 1509 | !copied.contains('●'), |
| 1510 | "family chrome leaked into copy: {copied:?}" |
| 1511 | ); |
| 1512 | assert!( |
| 1513 | copied.contains("run done"), |
| 1514 | "receipt text was clipped: {copied:?}" |
| 1515 | ); |
| 1516 | } |
| 1517 | |
| 1518 | #[test] |
| 1519 | fn copy_metadata_tracks_wrapped_assistant_code_prefix_in_display_columns() { |
| 1520 | let cell = HistoryCell::Assistant { |
| 1521 | content: "```text\n 中文 = 1\n```".to_string(), |
| 1522 | streaming: false, |
| 1523 | }; |
| 1524 | let rendered = cell.lines_with_copy_metadata(24, TranscriptRenderOptions::default()); |
| 1525 | let code_line = rendered |
| 1526 | .iter() |
| 1527 | .find(|line| { |
| 1528 | line.line |
| 1529 | .spans |
| 1530 | .iter() |
| 1531 | .any(|span| span.content.contains("中文")) |
| 1532 | }) |
| 1533 | .expect("wrapped fenced code line"); |
| 1534 | assert_eq!( |
| 1535 | code_line.copy_prefix_width, 2, |
| 1536 | "code continuation prefix uses the role marker's two display columns" |
| 1537 | ); |
| 1538 | let code = line_to_plain(&code_line.line); |
| 1539 | let copied = slice_text( |
| 1540 | &code, |
| 1541 | code_line.copy_prefix_width, |
| 1542 | text_display_width(&code), |
| 1543 | ); |
| 1544 | assert!( |
| 1545 | copied.contains("中文 = 1"), |
| 1546 | "code text was clipped: {copied:?}" |
| 1547 | ); |
| 1548 | assert!( |
| 1549 | copied.starts_with(" 中文"), |
| 1550 | "code indentation or visual prefix was wrong: {copied:?}" |
| 1551 | ); |
| 1552 | } |
| 1553 | |
| 1554 | #[test] |
| 1555 | fn copy_metadata_keeps_fenced_code_indentation_after_prefix_removal() { |
| 1556 | let cell = HistoryCell::Assistant { |
| 1557 | content: "```rust\n let answer = 42;\n```".to_string(), |
| 1558 | streaming: false, |
| 1559 | }; |
| 1560 | let rendered = cell.lines_with_copy_metadata(40, TranscriptRenderOptions::default()); |
| 1561 | let code_line = rendered |
| 1562 | .iter() |
| 1563 | .find(|line| { |
| 1564 | line.line |
| 1565 | .spans |
| 1566 | .iter() |
| 1567 | .any(|span| span.content.contains("answer")) |
| 1568 | }) |
| 1569 | .expect("fenced code body"); |
| 1570 | let text = line_to_plain(&code_line.line); |
| 1571 | let content = slice_text( |
| 1572 | &text, |
| 1573 | code_line.copy_prefix_width, |
| 1574 | text_display_width(&text), |
| 1575 | ); |
| 1576 | assert!( |
| 1577 | content.contains(" let answer = 42;"), |
| 1578 | "code indentation was not preserved: {content:?}" |
| 1579 | ); |
| 1580 | for glyph in ['╎', '▎', '●', '│', '┃'] { |
| 1581 | assert!( |
| 1582 | !content.contains(glyph), |
| 1583 | "decorative glyph leaked: {content:?}" |
| 1584 | ); |
| 1585 | } |
| 1586 | } |
| 1587 | |
| 1588 | #[test] |
| 1589 | fn whitespace_only_assistant_cell_renders_nothing() { |
| 1590 | // Regression: a stray newline/space streamed between reasoning and a |
| 1591 | // tool call produced a whitespace-only Assistant cell that rendered as |
| 1592 | // a bare, orphaned role glyph — the "blue dot with nothing after it" |
| 1593 | // artifact. It must collapse to zero lines instead. |
| 1594 | for content in ["", " ", "\n", "\n\n", " \t \n"] { |
| 1595 | for streaming in [false, true] { |
| 1596 | let cell = HistoryCell::Assistant { |
| 1597 | content: content.to_string(), |
| 1598 | streaming, |
| 1599 | }; |
| 1600 | assert!( |
| 1601 | cell.lines(80).is_empty(), |
| 1602 | "whitespace-only assistant content {content:?} (streaming={streaming}) \ |
| 1603 | must render no lines", |
| 1604 | ); |
| 1605 | } |
| 1606 | } |
| 1607 | |
| 1608 | // Sanity: real prose still renders the role glyph as its first span. |
| 1609 | let cell = HistoryCell::Assistant { |
| 1610 | content: "hi".to_string(), |
| 1611 | streaming: false, |
| 1612 | }; |
| 1613 | assert_eq!( |
| 1614 | cell.lines(80)[0].spans[0].content.as_ref(), |
| 1615 | ASSISTANT_GLYPH, |
| 1616 | "non-empty assistant content must still render the role glyph", |
| 1617 | ); |
| 1618 | } |
| 1619 | |
| 1620 | #[test] |
| 1621 | fn assistant_cell_still_renders_markdown() { |
| 1622 | let cell = HistoryCell::Assistant { |
| 1623 | content: "# Heading\n\n- item".to_string(), |
| 1624 | streaming: false, |
| 1625 | }; |
| 1626 | let visible: Vec<String> = cell.lines(80).iter().map(line_text).collect(); |
| 1627 | |
| 1628 | assert!( |
| 1629 | visible[0].contains("Heading"), |
| 1630 | "assistant heading text should render: {visible:?}" |
| 1631 | ); |
| 1632 | assert!( |
| 1633 | !visible[0].contains("# Heading"), |
| 1634 | "assistant heading should still be parsed as markdown: {visible:?}" |
| 1635 | ); |
| 1636 | assert!( |
| 1637 | visible.iter().any(|line| line.contains('\u{2500}')), |
| 1638 | "assistant h1 markdown should still add a heading rule: {visible:?}" |
| 1639 | ); |
| 1640 | } |
| 1641 | |
| 1642 | #[test] |
| 1643 | fn assistant_code_block_lines_do_not_get_transcript_rail() { |
| 1644 | let cell = HistoryCell::Assistant { |
| 1645 | content: "SQL:\n```sql\nSELECT\nFROM customers\n```".to_string(), |
| 1646 | streaming: false, |
| 1647 | }; |
| 1648 | let visible: Vec<String> = cell |
| 1649 | .lines(80) |
| 1650 | .iter() |
| 1651 | .map(|line| { |
| 1652 | line.spans |
| 1653 | .iter() |
| 1654 | .map(|span| span.content.as_ref()) |
| 1655 | .collect::<String>() |
| 1656 | }) |
| 1657 | .collect(); |
| 1658 | |
| 1659 | assert_eq!(visible[0], format!("{ASSISTANT_GLYPH} SQL:")); |
| 1660 | for line in visible |
| 1661 | .iter() |
| 1662 | .filter(|line| line.contains("SELECT") || line.contains("FROM customers")) |
| 1663 | { |
| 1664 | assert!( |
| 1665 | !line.contains('\u{258F}'), |
| 1666 | "code block line should not inherit the transcript rail: {line:?}" |
| 1667 | ); |
| 1668 | } |
| 1669 | } |
| 1670 | |
| 1671 | /// Issue #1212 repro: a multi-line SQL fence rendered after a short |
| 1672 | /// intro paragraph. Every code-block line — not just the first or last — |
| 1673 | /// must avoid the `▏` rail. |
| 1674 | #[test] |
| 1675 | fn assistant_long_code_block_keeps_every_line_rail_free() { |
| 1676 | let cell = HistoryCell::Assistant { |
| 1677 | content: "Here's the query:\n```sql\nSELECT\n c.customer_id,\n c.name,\n COUNT(o.order_id) AS order_count\nFROM customers c\nJOIN orders o ON c.customer_id = o.customer_id;\n```".to_string(), |
| 1678 | streaming: false, |
| 1679 | }; |
| 1680 | let visible: Vec<String> = cell |
| 1681 | .lines(80) |
| 1682 | .iter() |
| 1683 | .map(|line| { |
| 1684 | line.spans |
| 1685 | .iter() |
| 1686 | .map(|span| span.content.as_ref()) |
| 1687 | .collect::<String>() |
| 1688 | }) |
| 1689 | .collect(); |
| 1690 | |
| 1691 | let code_markers = ["SELECT", "customer_id", "name,", "COUNT", "FROM", "JOIN"]; |
| 1692 | for marker in code_markers { |
| 1693 | let line = visible |
| 1694 | .iter() |
| 1695 | .find(|line| line.contains(marker)) |
| 1696 | .unwrap_or_else(|| panic!("expected code line containing {marker:?}")); |
| 1697 | assert!( |
| 1698 | !line.contains('\u{258F}'), |
| 1699 | "code block line containing {marker:?} must not have the transcript rail: {line:?}" |
| 1700 | ); |
| 1701 | } |
| 1702 | } |
| 1703 | |
| 1704 | /// Edge case: a blank line inside a fence is still a code line; it must |
| 1705 | /// not regress to the rail because the empty body falls through a |
| 1706 | /// different wrap branch. |
| 1707 | #[test] |
| 1708 | fn assistant_code_block_blank_line_keeps_no_rail() { |
| 1709 | let cell = HistoryCell::Assistant { |
| 1710 | content: "```\nfn one() {}\n\nfn two() {}\n```".to_string(), |
| 1711 | streaming: false, |
| 1712 | }; |
| 1713 | for line in cell.lines(80).iter().skip(1) { |
| 1714 | let text: String = line.spans.iter().map(|s| s.content.as_ref()).collect(); |
| 1715 | assert!( |
| 1716 | !text.contains('\u{258F}'), |
| 1717 | "fence body line must stay rail-free: {text:?}" |
| 1718 | ); |
| 1719 | } |
| 1720 | } |
| 1721 | |
| 1722 | /// Wrapped code lines (a single source line longer than the viewport) |
| 1723 | /// emit multiple rendered lines from one `Block::Code`. None of them |
| 1724 | /// should leak the rail. |
| 1725 | #[test] |
| 1726 | fn assistant_wrapped_code_lines_keep_no_rail() { |
| 1727 | let long = "let x = ".to_string() + &"abcdef ".repeat(40); |
| 1728 | let content = format!("```\n{long}\n```"); |
| 1729 | let cell = HistoryCell::Assistant { |
| 1730 | content, |
| 1731 | streaming: false, |
| 1732 | }; |
| 1733 | for line in cell.lines(40).iter().skip(1) { |
| 1734 | let text: String = line.spans.iter().map(|s| s.content.as_ref()).collect(); |
| 1735 | assert!( |
| 1736 | !text.contains('\u{258F}'), |
| 1737 | "wrapped code line must stay rail-free: {text:?}" |
| 1738 | ); |
| 1739 | } |
| 1740 | } |
| 1741 | |
| 1742 | #[test] |
| 1743 | fn assistant_glyph_holds_full_brightness_when_idle() { |
| 1744 | // Idle (streaming=false) and low_motion both pin the colour to the |
| 1745 | // source sky — pulse only fires when actively streaming. |
| 1746 | let idle = assistant_label_style_for(false, false); |
| 1747 | let low_motion = assistant_label_style_for(true, true); |
| 1748 | assert_eq!(idle.fg, Some(palette::WHALE_INFO)); |
| 1749 | assert_eq!(low_motion.fg, Some(palette::WHALE_INFO)); |
| 1750 | } |
| 1751 | |
| 1752 | #[test] |
| 1753 | fn assistant_glyph_pulses_when_streaming_and_motion_allowed() { |
| 1754 | // The streaming path runs through `pulse_brightness`, which yields |
| 1755 | // an RGB colour scaled within 30%..100% of the source. Sample twice |
| 1756 | // — at least one of the samples must fall below 100% brightness, or |
| 1757 | // the test wouldn't be exercising the pulse at all. (We can't pin |
| 1758 | // the value because the function reads SystemTime::now().) |
| 1759 | use ratatui::style::Color; |
| 1760 | let mut saw_dimmed = false; |
| 1761 | for _ in 0..50 { |
| 1762 | if let Some(Color::Rgb(_, _, b)) = assistant_label_style_for(true, false).fg { |
| 1763 | let Color::Rgb(_, _, src_b) = palette::WHALE_INFO else { |
| 1764 | panic!("WHALE_INFO must be RGB"); |
| 1765 | }; |
| 1766 | if b < src_b { |
| 1767 | saw_dimmed = true; |
| 1768 | break; |
| 1769 | } |
| 1770 | } |
| 1771 | std::thread::sleep(std::time::Duration::from_millis(20)); |
| 1772 | } |
| 1773 | assert!( |
| 1774 | saw_dimmed, |
| 1775 | "expected the streaming pulse to dip below source brightness at least once", |
| 1776 | ); |
| 1777 | } |
| 1778 | |
| 1779 | // === Tool-card verb-glyph tests (v0.6.6 UI redesign) === |
| 1780 | |
| 1781 | #[test] |
| 1782 | fn exec_cell_header_uses_run_verb_glyph_and_label() { |
| 1783 | let cell = ExecCell { |
| 1784 | command: "ls".to_string(), |
| 1785 | status: ToolStatus::Success, |
| 1786 | output: Some("a\nb\n".to_string()), |
| 1787 | live_output: None, |
| 1788 | shell_task_id: None, |
| 1789 | owner_agent_id: None, |
| 1790 | owner_agent_name: None, |
| 1791 | started_at: None, |
| 1792 | duration_ms: Some(10), |
| 1793 | stale_elapsed_since_output_ms: None, |
| 1794 | source: ExecSource::Assistant, |
| 1795 | interaction: None, |
| 1796 | output_summary: None, |
| 1797 | }; |
| 1798 | let header = &cell.lines_with_motion(80, true)[0]; |
| 1799 | let visible: String = header |
| 1800 | .spans |
| 1801 | .iter() |
| 1802 | .map(|s| s.content.as_ref()) |
| 1803 | .collect::<String>(); |
| 1804 | assert!( |
| 1805 | visible.contains('\u{25B6}'), |
| 1806 | "Run glyph `▶` present: {visible:?}" |
| 1807 | ); |
| 1808 | assert!(visible.contains(" run "), "verb label `run`: {visible:?}"); |
| 1809 | // Old literal title must be gone. |
| 1810 | assert!( |
| 1811 | !visible.contains("Shell"), |
| 1812 | "old `Shell` literal is gone: {visible:?}" |
| 1813 | ); |
| 1814 | } |
| 1815 | |
| 1816 | #[test] |
| 1817 | fn exec_cell_header_includes_compact_command_summary() { |
| 1818 | let cell = ExecCell { |
| 1819 | command: "cargo test --workspace --all-features".to_string(), |
| 1820 | status: ToolStatus::Running, |
| 1821 | output: None, |
| 1822 | live_output: None, |
| 1823 | shell_task_id: None, |
| 1824 | owner_agent_id: None, |
| 1825 | owner_agent_name: None, |
| 1826 | started_at: None, |
| 1827 | duration_ms: None, |
| 1828 | stale_elapsed_since_output_ms: None, |
| 1829 | source: ExecSource::Assistant, |
| 1830 | interaction: None, |
| 1831 | output_summary: None, |
| 1832 | }; |
| 1833 | |
| 1834 | let header = &cell.lines_with_motion(80, true)[0]; |
| 1835 | let visible: String = header |
| 1836 | .spans |
| 1837 | .iter() |
| 1838 | .map(|s| s.content.as_ref()) |
| 1839 | .collect::<String>(); |
| 1840 | assert!(visible.contains("run running")); |
| 1841 | assert!( |
| 1842 | visible.contains("Ctrl+B"), |
| 1843 | "foreground wait header should expose Ctrl+B hint, not command: {visible:?}" |
| 1844 | ); |
| 1845 | assert!( |
| 1846 | !visible.contains("cargo test"), |
| 1847 | "foreground wait live header must not repeat command target: {visible:?}" |
| 1848 | ); |
| 1849 | |
| 1850 | let transcript_visible: String = HistoryCell::Tool(ToolCell::Exec(ExecCell { |
| 1851 | command: "cargo test --workspace --all-features".to_string(), |
| 1852 | status: ToolStatus::Running, |
| 1853 | output: None, |
| 1854 | live_output: None, |
| 1855 | shell_task_id: None, |
| 1856 | owner_agent_id: None, |
| 1857 | owner_agent_name: None, |
| 1858 | started_at: None, |
| 1859 | duration_ms: None, |
| 1860 | stale_elapsed_since_output_ms: None, |
| 1861 | source: ExecSource::Assistant, |
| 1862 | interaction: None, |
| 1863 | output_summary: None, |
| 1864 | })) |
| 1865 | .transcript_lines(80)[0] |
| 1866 | .spans |
| 1867 | .iter() |
| 1868 | .map(|s| s.content.as_ref()) |
| 1869 | .collect::<String>(); |
| 1870 | assert!( |
| 1871 | transcript_visible.contains("Ctrl+B"), |
| 1872 | "transcript compact wait should expose Ctrl+B hint: {transcript_visible:?}" |
| 1873 | ); |
| 1874 | assert!( |
| 1875 | !transcript_visible.contains("cargo test --workspace --all-features"), |
| 1876 | "transcript compact wait must not repeat command target: {transcript_visible:?}" |
| 1877 | ); |
| 1878 | } |
| 1879 | |
| 1880 | #[test] |
| 1881 | fn generic_tool_cell_picks_family_from_tool_name() { |
| 1882 | // Use an inspection call so the compact Delegate header still renders; |
| 1883 | // spawn cards are suppressed entirely (#4133). |
| 1884 | let cell = GenericToolCell { |
| 1885 | name: "agent".to_string(), |
| 1886 | status: ToolStatus::Running, |
| 1887 | input_summary: Some("action: peek foo".to_string()), |
| 1888 | output: None, |
| 1889 | prompts: None, |
| 1890 | spillover_path: None, |
| 1891 | output_summary: None, |
| 1892 | is_diff: false, |
| 1893 | }; |
| 1894 | let lines = cell.lines_with_mode(80, true, super::RenderMode::Live); |
| 1895 | assert_eq!(lines.len(), 1, "inspection must stay compact: {lines:?}"); |
| 1896 | let header_visible: String = lines[0] |
| 1897 | .spans |
| 1898 | .iter() |
| 1899 | .map(|s| s.content.as_ref()) |
| 1900 | .collect::<String>(); |
| 1901 | // agent → Delegate family (◐ delegate). |
| 1902 | assert!( |
| 1903 | header_visible.contains('\u{25D0}'), |
| 1904 | "Delegate glyph `◐`: {header_visible:?}" |
| 1905 | ); |
| 1906 | assert!( |
| 1907 | header_visible.contains(" delegate "), |
| 1908 | "verb label `delegate`: {header_visible:?}" |
| 1909 | ); |
| 1910 | } |
| 1911 | |
| 1912 | #[test] |
| 1913 | fn generic_tool_cell_renders_rlm_with_rlm_label_not_swarm() { |
| 1914 | let cell = GenericToolCell { |
| 1915 | name: "rlm".to_string(), |
| 1916 | status: ToolStatus::Running, |
| 1917 | input_summary: Some("task: compare source trees".to_string()), |
| 1918 | output: None, |
| 1919 | prompts: None, |
| 1920 | spillover_path: None, |
| 1921 | output_summary: None, |
| 1922 | is_diff: false, |
| 1923 | }; |
| 1924 | let lines = cell.lines_with_mode(80, true, super::RenderMode::Live); |
| 1925 | let header_visible: String = lines[0] |
| 1926 | .spans |
| 1927 | .iter() |
| 1928 | .map(|s| s.content.as_ref()) |
| 1929 | .collect::<String>(); |
| 1930 | |
| 1931 | assert!( |
| 1932 | header_visible.contains(" rlm "), |
| 1933 | "RLM card should identify RLM work: {header_visible:?}" |
| 1934 | ); |
| 1935 | assert!( |
| 1936 | !header_visible.contains("swarm"), |
| 1937 | "RLM card must not use removed swarm wording: {header_visible:?}" |
| 1938 | ); |
| 1939 | } |
| 1940 | |
| 1941 | #[test] |
| 1942 | fn exploring_card_search_reads_as_find_not_read() { |
| 1943 | // #4145: a completed grep grouped under the exploration card must not |
| 1944 | // render `read done · Searching …`; the header verb has to agree with the |
| 1945 | // `Searching for …` label. |
| 1946 | let cell = super::ExploringCell { |
| 1947 | entries: vec![super::ExploringEntry { |
| 1948 | label: "Searching for `TranscriptScroll`".to_string(), |
| 1949 | status: ToolStatus::Success, |
| 1950 | }], |
| 1951 | }; |
| 1952 | let header: String = cell.lines_with_motion(80, true)[0] |
| 1953 | .spans |
| 1954 | .iter() |
| 1955 | .map(|s| s.content.as_ref()) |
| 1956 | .collect::<String>(); |
| 1957 | assert!( |
| 1958 | header.contains("find done"), |
| 1959 | "search card header should read `find done`: {header:?}" |
| 1960 | ); |
| 1961 | assert!( |
| 1962 | !header.contains("read done"), |
| 1963 | "search card must not pair `read done` with a search label: {header:?}" |
| 1964 | ); |
| 1965 | assert!( |
| 1966 | header.contains("Searching for `TranscriptScroll`"), |
| 1967 | "search label should remain intact: {header:?}" |
| 1968 | ); |
| 1969 | } |
| 1970 | |
| 1971 | #[test] |
| 1972 | fn exploring_card_read_keeps_read_verb() { |
| 1973 | // The fix only re-verbs search-only cards — a plain read stays `read`. |
| 1974 | let cell = super::ExploringCell { |
| 1975 | entries: vec![super::ExploringEntry { |
| 1976 | label: "Reading src/foo.rs".to_string(), |
| 1977 | status: ToolStatus::Success, |
| 1978 | }], |
| 1979 | }; |
| 1980 | let header: String = cell.lines_with_motion(80, true)[0] |
| 1981 | .spans |
| 1982 | .iter() |
| 1983 | .map(|s| s.content.as_ref()) |
| 1984 | .collect::<String>(); |
| 1985 | assert!( |
| 1986 | header.contains("read done"), |
| 1987 | "read card header should read `read done`: {header:?}" |
| 1988 | ); |
| 1989 | } |
| 1990 | |
| 1991 | // === Reasoning treatment tests (v0.6.6 UI redesign) === |
| 1992 | |
| 1993 | #[test] |
| 1994 | fn render_thinking_uses_dotted_opener_in_header() { |
| 1995 | let lines = render_thinking("Step one\nStep two", 80, false, Some(2.0), false, true); |
| 1996 | let header = &lines[0]; |
| 1997 | // First span carries `…` followed by a space. |
| 1998 | assert!( |
| 1999 | header.spans[0].content.starts_with(REASONING_OPENER), |
| 2000 | "header opener: {:?}", |
| 2001 | header.spans[0].content |
| 2002 | ); |
| 2003 | } |
| 2004 | |
| 2005 | #[test] |
| 2006 | fn render_thinking_body_lines_use_dashed_rail_and_italic() { |
| 2007 | let lines = render_thinking( |
| 2008 | "concrete reasoning content", |
| 2009 | 80, |
| 2010 | /*streaming*/ false, |
| 2011 | Some(1.0), |
| 2012 | /*collapsed*/ false, |
| 2013 | /*low_motion*/ true, |
| 2014 | ); |
| 2015 | // Header is index 0; first body line is index 1. |
| 2016 | assert!(lines.len() >= 2, "expected at least one body line"); |
| 2017 | let body = &lines[1]; |
| 2018 | assert_eq!( |
| 2019 | body.spans[0].content.as_ref(), |
| 2020 | REASONING_RAIL, |
| 2021 | "body rail must be the dashed `╎ ` glyph" |
| 2022 | ); |
| 2023 | // The body span should carry italic. |
| 2024 | let italic_seen = body |
| 2025 | .spans |
| 2026 | .iter() |
| 2027 | .skip(1) |
| 2028 | .any(|span| span.style.add_modifier.contains(Modifier::ITALIC)); |
| 2029 | assert!(italic_seen, "body content should carry italic modifier"); |
| 2030 | } |
| 2031 | |
| 2032 | #[test] |
| 2033 | fn render_thinking_can_omit_background_highlight() { |
| 2034 | let lines = render_thinking_with_highlight( |
| 2035 | "reasoning without a filled surface", |
| 2036 | 80, |
| 2037 | false, |
| 2038 | Some(1.0), |
| 2039 | false, |
| 2040 | true, |
| 2041 | false, |
| 2042 | ); |
| 2043 | |
| 2044 | assert!( |
| 2045 | lines |
| 2046 | .iter() |
| 2047 | .flat_map(|line| line.spans.iter()) |
| 2048 | .all(|span| span.style.bg.is_none()), |
| 2049 | "disabled thinking highlight must not apply a background to any span" |
| 2050 | ); |
| 2051 | } |
| 2052 | |
| 2053 | #[test] |
| 2054 | fn render_thinking_streaming_appends_cursor_when_motion_allowed() { |
| 2055 | let lines = render_thinking( |
| 2056 | "ongoing reasoning...", |
| 2057 | 80, |
| 2058 | /*streaming*/ true, |
| 2059 | None, |
| 2060 | /*collapsed*/ false, |
| 2061 | /*low_motion*/ false, |
| 2062 | ); |
| 2063 | // Last line is the most recent body line — cursor lives there. |
| 2064 | let last = lines.last().expect("body line present"); |
| 2065 | let last_span = last.spans.last().expect("trailing span present"); |
| 2066 | assert!( |
| 2067 | last_span.content.contains(REASONING_CURSOR), |
| 2068 | "expected trailing cursor `▎` on last streaming body line, got {:?}", |
| 2069 | last_span.content |
| 2070 | ); |
| 2071 | } |
| 2072 | |
| 2073 | #[test] |
| 2074 | fn render_thinking_streaming_omits_cursor_when_low_motion() { |
| 2075 | let lines = render_thinking( |
| 2076 | "ongoing reasoning...", |
| 2077 | 80, |
| 2078 | /*streaming*/ true, |
| 2079 | None, |
| 2080 | /*collapsed*/ false, |
| 2081 | /*low_motion*/ true, |
| 2082 | ); |
| 2083 | let last = lines.last().expect("body line present"); |
| 2084 | let visible: String = last |
| 2085 | .spans |
| 2086 | .iter() |
| 2087 | .map(|s| s.content.as_ref()) |
| 2088 | .collect::<String>(); |
| 2089 | assert!( |
| 2090 | !visible.contains(REASONING_CURSOR), |
| 2091 | "low_motion must suppress the streaming cursor: {visible:?}" |
| 2092 | ); |
| 2093 | } |
| 2094 | |
| 2095 | // === Theme parity tests === |
| 2096 | // |
| 2097 | // These lock the visible color/style choices for one plan cell and one |
| 2098 | // tool cell against `deepseek_theme::Theme::dark()`. The render path is |
| 2099 | // unchanged in shape; the assertions just guarantee a future skin swap |
| 2100 | // (or accidental drift) is caught here instead of at runtime. |
| 2101 | |
| 2102 | #[test] |
| 2103 | fn plan_update_cell_renders_with_dark_theme_tokens() { |
| 2104 | let theme = Theme::dark(); |
| 2105 | let cell = PlanUpdateCell { |
| 2106 | snapshot: PlanSnapshot { |
| 2107 | items: vec![ |
| 2108 | crate::tools::plan::PlanItemArg { |
| 2109 | step: "scan repo".to_string(), |
| 2110 | status: StepStatus::Completed, |
| 2111 | }, |
| 2112 | crate::tools::plan::PlanItemArg { |
| 2113 | step: "extract theme".to_string(), |
| 2114 | status: StepStatus::InProgress, |
| 2115 | }, |
| 2116 | crate::tools::plan::PlanItemArg { |
| 2117 | step: "land tests".to_string(), |
| 2118 | status: StepStatus::Pending, |
| 2119 | }, |
| 2120 | ], |
| 2121 | ..PlanSnapshot::default() |
| 2122 | }, |
| 2123 | status: ToolStatus::Running, |
| 2124 | }; |
| 2125 | |
| 2126 | let lines = cell.lines_with_motion(80, true); |
| 2127 | |
| 2128 | // Header: "<spinner> <family-glyph> <verb> <state>" (v0.6.6 layout). |
| 2129 | // PlanUpdate has no canonical family yet, so it falls into the |
| 2130 | // Generic bullet glyph + "tool" verb. The shape and colour wiring |
| 2131 | // is what matters for the theme parity; the verb text moves with |
| 2132 | // the redesign. |
| 2133 | // PlanUpdate does NOT use card-rail wrapping (separate render path). |
| 2134 | let header = &lines[0]; |
| 2135 | let symbol_span = &header.spans[0]; |
| 2136 | let glyph_span = &header.spans[1]; |
| 2137 | let title_span = &header.spans[2]; |
| 2138 | let state_span = &header.spans[4]; |
| 2139 | |
| 2140 | assert_eq!( |
| 2141 | symbol_span.style.fg, |
| 2142 | Some(theme.tool_running_accent), |
| 2143 | "running header symbol should use the dark theme running accent" |
| 2144 | ); |
| 2145 | assert_eq!( |
| 2146 | glyph_span.style.fg, |
| 2147 | Some(theme.tool_running_accent), |
| 2148 | "family glyph rides the same status colour as the spinner" |
| 2149 | ); |
| 2150 | assert_eq!( |
| 2151 | title_span.content.as_ref(), |
| 2152 | "tool", |
| 2153 | "PlanUpdate routes to Generic family → 'tool' verb", |
| 2154 | ); |
| 2155 | assert_eq!(title_span.style.fg, Some(theme.tool_title_color)); |
| 2156 | assert!( |
| 2157 | title_span.style.add_modifier.contains(Modifier::BOLD), |
| 2158 | "tool title should be bold" |
| 2159 | ); |
| 2160 | assert_eq!( |
| 2161 | state_span.content.as_ref(), |
| 2162 | "running", |
| 2163 | "running PlanUpdate should label state as 'running'" |
| 2164 | ); |
| 2165 | assert_eq!(state_span.style.fg, Some(theme.tool_running_accent)); |
| 2166 | |
| 2167 | // Each step row: ["▏ ", "<marker>:", " ", "<step>"] |
| 2168 | let step_line = &lines[1]; |
| 2169 | let label_span = &step_line.spans[1]; |
| 2170 | let value_span = &step_line.spans[3]; |
| 2171 | assert_eq!( |
| 2172 | label_span.style.fg, |
| 2173 | Some(theme.tool_label_color), |
| 2174 | "step label should use theme.tool_label_color" |
| 2175 | ); |
| 2176 | assert_eq!( |
| 2177 | value_span.style.fg, |
| 2178 | Some(theme.tool_value_color), |
| 2179 | "step value should use theme.tool_value_color" |
| 2180 | ); |
| 2181 | |
| 2182 | // Plain content stays identical so visible output does not move. |
| 2183 | let visible = lines |
| 2184 | .iter() |
| 2185 | .map(|l| { |
| 2186 | l.spans |
| 2187 | .iter() |
| 2188 | .map(|s| s.content.as_ref()) |
| 2189 | .collect::<String>() |
| 2190 | }) |
| 2191 | .collect::<Vec<_>>(); |
| 2192 | assert_eq!(visible[1].trim_end(), "▏ done: scan repo"); |
| 2193 | assert_eq!(visible[2].trim_end(), "▏ live: extract theme"); |
| 2194 | assert_eq!(visible[3].trim_end(), "▏ next: land tests"); |
| 2195 | } |
| 2196 | |
| 2197 | #[test] |
| 2198 | fn plan_update_cell_renders_rich_artifact_metadata() { |
| 2199 | let cell = PlanUpdateCell { |
| 2200 | snapshot: PlanSnapshot { |
| 2201 | objective: Some("Make Plan mode reviewable".to_string()), |
| 2202 | context_summary: Some("Grounded in issue #2691".to_string()), |
| 2203 | sources_used: vec!["gh issue view 2691".to_string()], |
| 2204 | critical_files: vec!["crates/tui/src/tools/plan.rs".to_string()], |
| 2205 | constraints: vec!["Keep To-do primary".to_string()], |
| 2206 | recommended_approach: Some( |
| 2207 | "Enrich update_plan without breaking legacy calls".to_string(), |
| 2208 | ), |
| 2209 | verification_plan: Some("Run focused renderer tests".to_string()), |
| 2210 | risks_and_unknowns: Some("Metadata-only plans can disappear".to_string()), |
| 2211 | handoff_packet: Some("Next agent should inspect relay output".to_string()), |
| 2212 | items: vec![crate::tools::plan::PlanItemArg { |
| 2213 | step: "Render artifact sections".to_string(), |
| 2214 | status: StepStatus::InProgress, |
| 2215 | }], |
| 2216 | ..PlanSnapshot::default() |
| 2217 | }, |
| 2218 | status: ToolStatus::Success, |
| 2219 | }; |
| 2220 | |
| 2221 | let visible = cell |
| 2222 | .lines_with_motion(120, true) |
| 2223 | .into_iter() |
| 2224 | .map(|line| { |
| 2225 | line.spans |
| 2226 | .into_iter() |
| 2227 | .map(|span| span.content.into_owned()) |
| 2228 | .collect::<String>() |
| 2229 | }) |
| 2230 | .collect::<Vec<_>>() |
| 2231 | .join("\n"); |
| 2232 | |
| 2233 | assert!(visible.contains("objective:")); |
| 2234 | assert!(visible.contains("Make Plan mode reviewable")); |
| 2235 | assert!(visible.contains("source:")); |
| 2236 | assert!(visible.contains("gh issue view 2691")); |
| 2237 | assert!(visible.contains("file:")); |
| 2238 | assert!(visible.contains("verify:")); |
| 2239 | assert!(visible.contains("handoff:")); |
| 2240 | assert!(visible.contains("Render artifact sections")); |
| 2241 | } |
| 2242 | |
| 2243 | #[test] |
| 2244 | fn exec_cell_failed_status_renders_with_dark_theme_tokens() { |
| 2245 | let theme = Theme::dark(); |
| 2246 | let cell = ExecCell { |
| 2247 | command: "false".to_string(), |
| 2248 | status: ToolStatus::Failed, |
| 2249 | output: Some("boom".to_string()), |
| 2250 | live_output: None, |
| 2251 | shell_task_id: None, |
| 2252 | owner_agent_id: None, |
| 2253 | owner_agent_name: None, |
| 2254 | started_at: None, |
| 2255 | duration_ms: Some(42), |
| 2256 | stale_elapsed_since_output_ms: None, |
| 2257 | source: ExecSource::Assistant, |
| 2258 | interaction: None, |
| 2259 | output_summary: None, |
| 2260 | }; |
| 2261 | |
| 2262 | let lines = cell.lines_with_motion(80, true); |
| 2263 | |
| 2264 | let header = &lines[0]; |
| 2265 | let symbol_span = &header.spans[1]; |
| 2266 | let glyph_span = &header.spans[2]; |
| 2267 | let title_span = &header.spans[3]; |
| 2268 | let state_span = &header.spans[5]; |
| 2269 | |
| 2270 | assert_eq!( |
| 2271 | symbol_span.style.fg, |
| 2272 | Some(theme.tool_failed_accent), |
| 2273 | "failed exec header symbol should use the dark theme failed accent" |
| 2274 | ); |
| 2275 | // ExecCell is family Run → glyph `▶ ` and verb `run`. |
| 2276 | assert!( |
| 2277 | glyph_span.content.starts_with('\u{25B6}'), |
| 2278 | "Run family glyph: {:?}", |
| 2279 | glyph_span.content |
| 2280 | ); |
| 2281 | assert_eq!( |
| 2282 | title_span.content.as_ref(), |
| 2283 | "run", |
| 2284 | "ExecCell routes to Run family → 'run' verb", |
| 2285 | ); |
| 2286 | assert_eq!(title_span.style.fg, Some(theme.tool_title_color)); |
| 2287 | assert!(title_span.style.add_modifier.contains(Modifier::BOLD)); |
| 2288 | assert_eq!(state_span.content.as_ref(), "issue"); |
| 2289 | assert_eq!(state_span.style.fg, Some(theme.tool_failed_accent)); |
| 2290 | } |
| 2291 | |
| 2292 | // === display_lines (lines_with_options) vs transcript_lines parity === |
| 2293 | // |
| 2294 | // These lock the contract for CX#8: live view keeps reasoning compact |
| 2295 | // and caps tool output, transcript view shows the full body. Completed |
| 2296 | // reasoning without an explicit Summary stays out of the main flow so it |
| 2297 | // cannot masquerade as user text. |
| 2298 | |
| 2299 | fn line_text(line: &ratatui::text::Line<'static>) -> String { |
| 2300 | line.spans |
| 2301 | .iter() |
| 2302 | .map(|span| span.content.as_ref()) |
| 2303 | .collect() |
| 2304 | } |
| 2305 | |
| 2306 | fn lines_text(lines: &[ratatui::text::Line<'static>]) -> String { |
| 2307 | lines.iter().map(line_text).collect::<Vec<_>>().join("\n") |
| 2308 | } |
| 2309 | |
| 2310 | #[test] |
| 2311 | fn exec_cell_renders_live_shell_output_before_final_output() { |
| 2312 | let cell = ExecCell { |
| 2313 | command: "cargo test".to_string(), |
| 2314 | status: ToolStatus::Running, |
| 2315 | output: None, |
| 2316 | live_output: Some("running line 1\nrunning line 2".to_string()), |
| 2317 | shell_task_id: Some("shell_live".to_string()), |
| 2318 | owner_agent_id: None, |
| 2319 | owner_agent_name: None, |
| 2320 | started_at: None, |
| 2321 | duration_ms: None, |
| 2322 | stale_elapsed_since_output_ms: None, |
| 2323 | source: ExecSource::Assistant, |
| 2324 | interaction: None, |
| 2325 | output_summary: None, |
| 2326 | }; |
| 2327 | |
| 2328 | let live_text = lines_text(&cell.lines_with_motion(80, true)); |
| 2329 | assert!( |
| 2330 | !live_text.contains("running line 1"), |
| 2331 | "foreground shell live output belongs in sidebar/jobs, not main transcript: {live_text}" |
| 2332 | ); |
| 2333 | assert!( |
| 2334 | live_text.contains("Ctrl+B"), |
| 2335 | "compact foreground wait must keep Ctrl+B hint: {live_text}" |
| 2336 | ); |
| 2337 | assert!(!live_text.contains("command:")); |
| 2338 | assert!(!live_text.contains("Ctrl+B backgrounds this command")); |
| 2339 | assert!(!live_text.contains("Ctrl+B moves this shell wait to /jobs")); |
| 2340 | |
| 2341 | let transcript_text = lines_text(&HistoryCell::Tool(ToolCell::Exec(cell)).transcript_lines(80)); |
| 2342 | assert!( |
| 2343 | !transcript_text.contains("running line 1"), |
| 2344 | "foreground shell live output belongs in sidebar/jobs, not transcript: {transcript_text}" |
| 2345 | ); |
| 2346 | assert!(!transcript_text.contains("command:")); |
| 2347 | assert!(transcript_text.contains("Ctrl+B")); |
| 2348 | } |
| 2349 | |
| 2350 | #[test] |
| 2351 | fn exec_cell_prefers_final_output_over_live_shell_tail() { |
| 2352 | let cell = ExecCell { |
| 2353 | command: "cargo test".to_string(), |
| 2354 | status: ToolStatus::Success, |
| 2355 | output: Some("final output".to_string()), |
| 2356 | live_output: Some("stale live tail".to_string()), |
| 2357 | shell_task_id: Some("shell_live".to_string()), |
| 2358 | owner_agent_id: None, |
| 2359 | owner_agent_name: None, |
| 2360 | started_at: None, |
| 2361 | duration_ms: None, |
| 2362 | stale_elapsed_since_output_ms: None, |
| 2363 | source: ExecSource::Assistant, |
| 2364 | interaction: None, |
| 2365 | output_summary: None, |
| 2366 | }; |
| 2367 | |
| 2368 | let text = lines_text(&cell.lines_with_motion(80, true)); |
| 2369 | |
| 2370 | assert!(text.contains("cargo test")); |
| 2371 | assert!(!text.contains("stale live tail")); |
| 2372 | } |
| 2373 | |
| 2374 | #[test] |
| 2375 | fn long_thinking_display_is_shorter_than_transcript() { |
| 2376 | // Build a multi-paragraph thinking body so the live view has |
| 2377 | // something to compress. Without an explicit Summary block, the live |
| 2378 | // surface should show a bounded preview plus affordance; Ctrl+O |
| 2379 | // remains the path to the full body. |
| 2380 | let body = "First paragraph lede.\n\ |
| 2381 | Second sentence of the first paragraph.\n\n\ |
| 2382 | Second paragraph: deeper analysis follows.\n\ |
| 2383 | More detail in paragraph two.\n\n\ |
| 2384 | Third paragraph: even more reasoning.\n\ |
| 2385 | With another line.\n\n\ |
| 2386 | Fourth paragraph: the conclusion.\n\ |
| 2387 | And one more line for good measure.\n\n\ |
| 2388 | Fifth paragraph: final verification.\n\ |
| 2389 | One last supporting detail."; |
| 2390 | let cell = HistoryCell::Thinking { |
| 2391 | content: body.to_string(), |
| 2392 | streaming: false, |
| 2393 | duration_secs: Some(3.2), |
| 2394 | }; |
| 2395 | |
| 2396 | let live = cell.lines_with_options( |
| 2397 | 80, |
| 2398 | TranscriptRenderOptions { |
| 2399 | low_motion: true, |
| 2400 | ..TranscriptRenderOptions::default() |
| 2401 | }, |
| 2402 | ); |
| 2403 | let transcript = cell.transcript_lines(80); |
| 2404 | |
| 2405 | assert!( |
| 2406 | live.len() < transcript.len(), |
| 2407 | "live thinking should compress (live = {} lines, transcript = {} lines)", |
| 2408 | live.len(), |
| 2409 | transcript.len() |
| 2410 | ); |
| 2411 | |
| 2412 | let live_text = lines_text(&live); |
| 2413 | let transcript_text = lines_text(&transcript); |
| 2414 | |
| 2415 | assert!( |
| 2416 | transcript_text.contains("First paragraph lede"), |
| 2417 | "transcript thinking must keep the lede" |
| 2418 | ); |
| 2419 | assert!( |
| 2420 | live_text.contains("First paragraph lede"), |
| 2421 | "live thinking should preview completed reasoning: {live_text}" |
| 2422 | ); |
| 2423 | assert!( |
| 2424 | transcript_text.contains("Fifth paragraph"), |
| 2425 | "transcript thinking must keep the full body" |
| 2426 | ); |
| 2427 | assert!( |
| 2428 | !live_text.contains("Fifth paragraph"), |
| 2429 | "live thinking must drop the tail when collapsed" |
| 2430 | ); |
| 2431 | assert!( |
| 2432 | live_text.contains("Ctrl+O:detail"), |
| 2433 | "live thinking must offer the pager affordance" |
| 2434 | ); |
| 2435 | assert!( |
| 2436 | !transcript_text.contains("Ctrl+O:detail"), |
| 2437 | "transcript thinking must not include the live affordance" |
| 2438 | ); |
| 2439 | } |
| 2440 | |
| 2441 | #[test] |
| 2442 | fn completed_short_thinking_without_summary_stays_visible_in_live_view() { |
| 2443 | // Short completed reasoning should not become a dead "Full reasoning |
| 2444 | // in Ctrl+O" card. The reasoning rail and tint already distinguish it |
| 2445 | // from the user's prompt, so show the useful body inline. |
| 2446 | let cell = HistoryCell::Thinking { |
| 2447 | content: "One brief reasoning step.".to_string(), |
| 2448 | streaming: false, |
| 2449 | duration_secs: Some(0.4), |
| 2450 | }; |
| 2451 | |
| 2452 | let live = cell.lines_with_options( |
| 2453 | 80, |
| 2454 | TranscriptRenderOptions { |
| 2455 | low_motion: true, |
| 2456 | ..TranscriptRenderOptions::default() |
| 2457 | }, |
| 2458 | ); |
| 2459 | let transcript = cell.transcript_lines(80); |
| 2460 | |
| 2461 | let live_text = lines_text(&live); |
| 2462 | let transcript_text = lines_text(&transcript); |
| 2463 | |
| 2464 | assert!( |
| 2465 | live_text.contains("One brief reasoning step."), |
| 2466 | "live thinking must preview short completed reasoning: {live_text}" |
| 2467 | ); |
| 2468 | assert!( |
| 2469 | transcript_text.contains("One brief reasoning step."), |
| 2470 | "transcript thinking must keep the full reasoning body" |
| 2471 | ); |
| 2472 | assert!( |
| 2473 | !live_text.contains("Ctrl+O:detail"), |
| 2474 | "complete short reasoning should not need the detail affordance: {live_text}" |
| 2475 | ); |
| 2476 | } |
| 2477 | |
| 2478 | #[test] |
| 2479 | fn completed_reasoning_receipt_shows_verbatim_body_and_expands() { |
| 2480 | // The old #4146/#4148 scrub could not tell CodeWhale's identifiers from |
| 2481 | // the user's, and in a coding harness the user's dominate: it rendered |
| 2482 | // `short_dated_radar.py` as `….py`, `data/market_data/` as `data/…/`, and |
| 2483 | // every env var and module name as a bare `…`, which made the default |
| 2484 | // reasoning view unreadable. It also protected nothing — the full body |
| 2485 | // was always one keypress away on Space/Ctrl+O. A reasoning receipt now |
| 2486 | // shows the model's own words verbatim; only the line budget truncates. |
| 2487 | let cell = HistoryCell::Thinking { |
| 2488 | content: "I will call refresh_catalog_cache to refresh the model list.".to_string(), |
| 2489 | streaming: false, |
| 2490 | duration_secs: Some(1.0), |
| 2491 | }; |
| 2492 | |
| 2493 | // Default collapsed view: the identifier is shown, not scrubbed, and a |
| 2494 | // short body needs no expand affordance. |
| 2495 | let collapsed = cell.lines_with_options( |
| 2496 | 80, |
| 2497 | TranscriptRenderOptions { |
| 2498 | low_motion: true, |
| 2499 | ..TranscriptRenderOptions::default() |
| 2500 | }, |
| 2501 | ); |
| 2502 | let collapsed_text = lines_text(&collapsed); |
| 2503 | assert!( |
| 2504 | collapsed_text.contains("refresh_catalog_cache"), |
| 2505 | "reasoning must be verbatim in the collapsed receipt: {collapsed_text}" |
| 2506 | ); |
| 2507 | assert!( |
| 2508 | collapsed_text.contains("refresh the model list"), |
| 2509 | "surrounding prose must still read: {collapsed_text}" |
| 2510 | ); |
| 2511 | assert!( |
| 2512 | !collapsed_text.contains("Ctrl+O:detail"), |
| 2513 | "a short completed receipt fits the budget and needs no affordance: {collapsed_text}" |
| 2514 | ); |
| 2515 | |
| 2516 | // A long body truncates at the line budget and offers the expand |
| 2517 | // affordance; expanding restores every line, identifiers intact. |
| 2518 | let long_body = (1..=20) |
| 2519 | .map(|i| format!("step {i:02}: refresh_catalog_cache iteration")) |
| 2520 | .collect::<Vec<_>>() |
| 2521 | .join("\n"); |
| 2522 | let long_cell = HistoryCell::Thinking { |
| 2523 | content: long_body.clone(), |
| 2524 | streaming: false, |
| 2525 | duration_secs: Some(1.0), |
| 2526 | }; |
| 2527 | let long_collapsed = long_cell.lines_with_options( |
| 2528 | 80, |
| 2529 | TranscriptRenderOptions { |
| 2530 | low_motion: true, |
| 2531 | ..TranscriptRenderOptions::default() |
| 2532 | }, |
| 2533 | ); |
| 2534 | let long_collapsed_text = lines_text(&long_collapsed); |
| 2535 | assert!( |
| 2536 | long_collapsed_text.contains("Space:expand · Ctrl+O:detail"), |
| 2537 | "a truncated receipt must offer the expand affordance: {long_collapsed_text}" |
| 2538 | ); |
| 2539 | assert!( |
| 2540 | long_collapsed_text.contains("refresh_catalog_cache"), |
| 2541 | "the shown head must keep identifiers verbatim: {long_collapsed_text}" |
| 2542 | ); |
| 2543 | |
| 2544 | // Expanded view (Space toggles the fold relative to the default): every |
| 2545 | // line is restored. |
| 2546 | let expanded = long_cell.lines_with_options_folded( |
| 2547 | 80, |
| 2548 | TranscriptRenderOptions { |
| 2549 | low_motion: true, |
| 2550 | ..TranscriptRenderOptions::default() |
| 2551 | }, |
| 2552 | true, |
| 2553 | ); |
| 2554 | let expanded_text = lines_text(&expanded); |
| 2555 | for i in 1..=20 { |
| 2556 | assert!( |
| 2557 | expanded_text.contains(&format!("step {i:02}: refresh_catalog_cache iteration")), |
| 2558 | "expanded reasoning must restore every line ({i}): {expanded_text}" |
| 2559 | ); |
| 2560 | } |
| 2561 | } |
| 2562 | |
| 2563 | #[test] |
| 2564 | fn thinking_default_expanded_inverts_but_preserves_the_space_toggle() { |
| 2565 | // A 20-line body guarantees the collapsed fold actually truncates, so the |
| 2566 | // Space toggle is observable: default-expanded shows everything, Space |
| 2567 | // collapses to the 10-line budget with the expand affordance, and both |
| 2568 | // states show the model's identifiers verbatim (no #4146/#4148 scrub). |
| 2569 | let long_body = (1..=20) |
| 2570 | .map(|i| format!("step {i:02}: refresh_catalog_cache iteration")) |
| 2571 | .collect::<Vec<_>>() |
| 2572 | .join("\n"); |
| 2573 | let cell = HistoryCell::Thinking { |
| 2574 | content: long_body.clone(), |
| 2575 | streaming: false, |
| 2576 | duration_secs: Some(1.0), |
| 2577 | }; |
| 2578 | let options = TranscriptRenderOptions { |
| 2579 | thinking_default_expanded: true, |
| 2580 | low_motion: true, |
| 2581 | ..TranscriptRenderOptions::default() |
| 2582 | }; |
| 2583 | |
| 2584 | let expanded = cell.lines_with_options_folded(80, options, false); |
| 2585 | let expanded_text = lines_text(&expanded); |
| 2586 | for i in 1..=20 { |
| 2587 | assert!( |
| 2588 | expanded_text.contains(&format!("step {i:02}: refresh_catalog_cache iteration")), |
| 2589 | "the configured default must show the full reasoning body ({i}): {expanded_text}" |
| 2590 | ); |
| 2591 | } |
| 2592 | |
| 2593 | let collapsed = cell.lines_with_options_folded(80, options, true); |
| 2594 | let collapsed_text = lines_text(&collapsed); |
| 2595 | assert!( |
| 2596 | collapsed_text.contains("refresh_catalog_cache"), |
| 2597 | "Space must still collapse a default-expanded reasoning cell, verbatim: {collapsed_text}" |
| 2598 | ); |
| 2599 | assert!( |
| 2600 | collapsed_text.contains("Space:expand · Ctrl+O:detail"), |
| 2601 | "the collapsed state must retain the full-reasoning affordance" |
| 2602 | ); |
| 2603 | assert!( |
| 2604 | !collapsed_text.contains("step 20:"), |
| 2605 | "the collapsed fold must truncate the long body" |
| 2606 | ); |
| 2607 | } |
| 2608 | |
| 2609 | /// The live card must spend the whole output budget it advertises. |
| 2610 | /// |
| 2611 | /// `selected_output_indices` fills head + tail, then tops up from lines that |
| 2612 | /// look important (error / warning / path). Plain output — a list of names, a |
| 2613 | /// table, a clean build log — matches none of those, so the top-up found |
| 2614 | /// nothing and the card silently forfeited the rest of its budget: it showed |
| 2615 | /// `head + tail` rows and reported the remainder as "omitted". That is the |
| 2616 | /// "even truncated mode over-truncates" complaint. |
| 2617 | #[test] |
| 2618 | fn live_tool_output_spends_its_whole_line_budget_on_unremarkable_output() { |
| 2619 | let total_output_lines = 40usize; |
| 2620 | // Deliberately bland: no error/warning keywords, no slashes, no dots, so |
| 2621 | // `output_importance_rank` returns None for every single line. |
| 2622 | let output = (0..total_output_lines) |
| 2623 | .map(|i| format!("row {i:02} plain content")) |
| 2624 | .collect::<Vec<_>>() |
| 2625 | .join("\n"); |
| 2626 | |
| 2627 | let cell = HistoryCell::Tool(ToolCell::Exec(ExecCell { |
| 2628 | command: "list_things".to_string(), |
| 2629 | status: ToolStatus::Failed, |
| 2630 | output: Some(output), |
| 2631 | live_output: None, |
| 2632 | shell_task_id: None, |
| 2633 | owner_agent_id: None, |
| 2634 | owner_agent_name: None, |
| 2635 | started_at: None, |
| 2636 | duration_ms: Some(120), |
| 2637 | stale_elapsed_since_output_ms: None, |
| 2638 | source: ExecSource::Assistant, |
| 2639 | interaction: None, |
| 2640 | output_summary: None, |
| 2641 | })); |
| 2642 | |
| 2643 | let live = cell.lines_with_options( |
| 2644 | 80, |
| 2645 | TranscriptRenderOptions { |
| 2646 | low_motion: true, |
| 2647 | ..TranscriptRenderOptions::default() |
| 2648 | }, |
| 2649 | ); |
| 2650 | let live_text = lines_text(&live); |
| 2651 | let shown = (0..total_output_lines) |
| 2652 | .filter(|i| live_text.contains(&format!("row {i:02} plain content"))) |
| 2653 | .count(); |
| 2654 | |
| 2655 | assert_eq!( |
| 2656 | shown, TOOL_OUTPUT_LINE_LIMIT, |
| 2657 | "a live card promising {TOOL_OUTPUT_LINE_LIMIT} output rows must show \ |
| 2658 | {TOOL_OUTPUT_LINE_LIMIT}, not stop at head+tail: {live_text}" |
| 2659 | ); |
| 2660 | // The shown region stays readable: a contiguous head, then the tail. |
| 2661 | for i in 0..TOOL_OUTPUT_HEAD_LINES { |
| 2662 | assert!( |
| 2663 | live_text.contains(&format!("row {i:02} plain content")), |
| 2664 | "head row {i} missing: {live_text}" |
| 2665 | ); |
| 2666 | } |
| 2667 | for i in (total_output_lines - TOOL_OUTPUT_TAIL_LINES)..total_output_lines { |
| 2668 | assert!( |
| 2669 | live_text.contains(&format!("row {i:02} plain content")), |
| 2670 | "tail row {i} missing: {live_text}" |
| 2671 | ); |
| 2672 | } |
| 2673 | } |
| 2674 | |
| 2675 | #[test] |
| 2676 | fn tool_exec_live_caps_failed_output_transcript_does_not() { |
| 2677 | // A *failed* exec keeps its output in live mode, capped to head+tail |
| 2678 | // with a "lines omitted" marker. Transcript mode emits it uncapped. |
| 2679 | let total_output_lines = 30usize; |
| 2680 | let output = (0..total_output_lines) |
| 2681 | .map(|i| format!("output line {i:02}")) |
| 2682 | .collect::<Vec<_>>() |
| 2683 | .join("\n"); |
| 2684 | |
| 2685 | let cell = HistoryCell::Tool(ToolCell::Exec(ExecCell { |
| 2686 | command: "noisy_script.sh".to_string(), |
| 2687 | status: ToolStatus::Failed, |
| 2688 | output: Some(output), |
| 2689 | live_output: None, |
| 2690 | shell_task_id: None, |
| 2691 | owner_agent_id: None, |
| 2692 | owner_agent_name: None, |
| 2693 | started_at: None, |
| 2694 | duration_ms: Some(120), |
| 2695 | stale_elapsed_since_output_ms: None, |
| 2696 | source: ExecSource::Assistant, |
| 2697 | interaction: None, |
| 2698 | output_summary: None, |
| 2699 | })); |
| 2700 | |
| 2701 | let live = cell.lines_with_options( |
| 2702 | 80, |
| 2703 | TranscriptRenderOptions { |
| 2704 | low_motion: true, |
| 2705 | ..TranscriptRenderOptions::default() |
| 2706 | }, |
| 2707 | ); |
| 2708 | let transcript = cell.transcript_lines(80); |
| 2709 | |
| 2710 | let live_text = lines_text(&live); |
| 2711 | let transcript_text = lines_text(&transcript); |
| 2712 | |
| 2713 | assert!( |
| 2714 | live.len() < transcript.len(), |
| 2715 | "live exec output must be shorter than transcript exec output (live={}, transcript={})", |
| 2716 | live.len(), |
| 2717 | transcript.len() |
| 2718 | ); |
| 2719 | assert!( |
| 2720 | live_text.contains("lines omitted"), |
| 2721 | "live failed-exec output must surface the omission marker: {live_text}" |
| 2722 | ); |
| 2723 | assert!( |
| 2724 | !transcript_text.contains("lines omitted"), |
| 2725 | "transcript exec output must not include the omission marker" |
| 2726 | ); |
| 2727 | assert!(transcript_text.contains("output line 00")); |
| 2728 | // The middle should only appear in the transcript, since the live |
| 2729 | // view truncates the head/tail around the cap. |
| 2730 | assert!( |
| 2731 | transcript_text.contains("output line 15"), |
| 2732 | "transcript must include the middle of the exec output" |
| 2733 | ); |
| 2734 | // Last line should appear in both because the live view shows |
| 2735 | // head + tail around an omission marker. |
| 2736 | let last = format!("output line {:02}", total_output_lines - 1); |
| 2737 | assert!(transcript_text.contains(&last)); |
| 2738 | } |
| 2739 | |
| 2740 | #[test] |
| 2741 | fn tool_exec_live_previews_successful_command_without_its_full_body() { |
| 2742 | // A *successful* exec does not earn its full body in live mode — no |
| 2743 | // command echo, and only `TOOL_SUCCESS_OUTPUT_PREVIEW_LINES` of output. |
| 2744 | // It used to collapse to the bare header, which meant a run card told you |
| 2745 | // a command finished and nothing whatsoever about what it produced. |
| 2746 | // Transcript mode still records everything for the pager/clipboard. |
| 2747 | let output = (0..30usize) |
| 2748 | .map(|i| format!("output line {i:02}")) |
| 2749 | .collect::<Vec<_>>() |
| 2750 | .join("\n"); |
| 2751 | let cell = HistoryCell::Tool(ToolCell::Exec(ExecCell { |
| 2752 | command: "noisy_script.sh".to_string(), |
| 2753 | status: ToolStatus::Success, |
| 2754 | output: Some(output), |
| 2755 | live_output: None, |
| 2756 | shell_task_id: None, |
| 2757 | owner_agent_id: None, |
| 2758 | owner_agent_name: None, |
| 2759 | started_at: None, |
| 2760 | duration_ms: Some(120), |
| 2761 | stale_elapsed_since_output_ms: None, |
| 2762 | source: ExecSource::Assistant, |
| 2763 | interaction: None, |
| 2764 | output_summary: None, |
| 2765 | })); |
| 2766 | |
| 2767 | let live_text = lines_text(&cell.lines_with_options( |
| 2768 | 80, |
| 2769 | TranscriptRenderOptions { |
| 2770 | low_motion: true, |
| 2771 | ..TranscriptRenderOptions::default() |
| 2772 | }, |
| 2773 | )); |
| 2774 | let transcript_text = lines_text(&cell.transcript_lines(80)); |
| 2775 | |
| 2776 | // Live: a bounded preview from the top of the output. |
| 2777 | let previewed = (0..30usize) |
| 2778 | .filter(|i| live_text.contains(&format!("output line {i:02}"))) |
| 2779 | .count(); |
| 2780 | assert_eq!( |
| 2781 | previewed, TOOL_SUCCESS_OUTPUT_PREVIEW_LINES, |
| 2782 | "a successful exec should preview exactly \ |
| 2783 | {TOOL_SUCCESS_OUTPUT_PREVIEW_LINES} output rows: {live_text}" |
| 2784 | ); |
| 2785 | assert!( |
| 2786 | live_text.contains("output line 00"), |
| 2787 | "the preview reads from the top of the output: {live_text}" |
| 2788 | ); |
| 2789 | assert!( |
| 2790 | !live_text.contains("output line 29"), |
| 2791 | "a successful exec must not render its full body in live mode: {live_text}" |
| 2792 | ); |
| 2793 | assert!( |
| 2794 | !live_text.contains("command:"), |
| 2795 | "a successful exec still skips the command echo; the header carries \ |
| 2796 | the summary: {live_text}" |
| 2797 | ); |
| 2798 | // Transcript still has the full output. |
| 2799 | assert!(transcript_text.contains("output line 00")); |
| 2800 | assert!(transcript_text.contains("output line 29")); |
| 2801 | } |
| 2802 | |
| 2803 | #[test] |
| 2804 | fn generic_tool_cell_renders_prompts_as_indexed_rows() { |
| 2805 | // When prompts are populated by a fan-out tool, each child shows on |
| 2806 | // its own row instead of the inline `args:` summary so the user can |
| 2807 | // read what each child was asked. |
| 2808 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 2809 | name: "read_file".to_string(), |
| 2810 | status: ToolStatus::Running, |
| 2811 | input_summary: Some("prompts: <3 items>".to_string()), |
| 2812 | output: None, |
| 2813 | prompts: Some(vec![ |
| 2814 | "Summarize the README".to_string(), |
| 2815 | "List the public types in client.rs".to_string(), |
| 2816 | "Diff this commit against main".to_string(), |
| 2817 | ]), |
| 2818 | spillover_path: None, |
| 2819 | output_summary: None, |
| 2820 | is_diff: false, |
| 2821 | })); |
| 2822 | let text = lines_text(&cell.lines(80)); |
| 2823 | |
| 2824 | assert!(text.contains("[0] Summarize the README")); |
| 2825 | assert!(text.contains("[1] List the public types in client.rs")); |
| 2826 | assert!(text.contains("[2] Diff this commit against main")); |
| 2827 | // The inline args summary must not also be emitted — we replaced it |
| 2828 | // with the per-child rows. |
| 2829 | assert!( |
| 2830 | !text.contains("args: prompts:"), |
| 2831 | "inline `args:` summary must be suppressed when per-prompt rows render" |
| 2832 | ); |
| 2833 | } |
| 2834 | |
| 2835 | #[test] |
| 2836 | fn generic_tool_cell_falls_back_to_args_when_prompts_none() { |
| 2837 | // Non-fan-out tools keep the existing `args:` summary so behavior |
| 2838 | // doesn't drift for everything else. |
| 2839 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 2840 | name: "file_search".to_string(), |
| 2841 | status: ToolStatus::Running, |
| 2842 | input_summary: Some("query: foo".to_string()), |
| 2843 | output: None, |
| 2844 | prompts: None, |
| 2845 | spillover_path: None, |
| 2846 | output_summary: None, |
| 2847 | is_diff: false, |
| 2848 | })); |
| 2849 | let text = lines_text(&cell.lines(80)); |
| 2850 | assert!(text.contains("query: foo")); |
| 2851 | } |
| 2852 | |
| 2853 | #[test] |
| 2854 | fn known_generic_tool_hides_raw_name_in_live_mode() { |
| 2855 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 2856 | name: "run_verifiers".to_string(), |
| 2857 | status: ToolStatus::Running, |
| 2858 | input_summary: Some("profile: auto, level: quick".to_string()), |
| 2859 | output: None, |
| 2860 | prompts: None, |
| 2861 | spillover_path: None, |
| 2862 | output_summary: None, |
| 2863 | is_diff: false, |
| 2864 | })); |
| 2865 | |
| 2866 | let text = lines_text(&cell.lines(80)); |
| 2867 | assert!(text.contains("verify running"), "{text}"); |
| 2868 | assert!( |
| 2869 | !text.contains("name: run_verifiers"), |
| 2870 | "live card should not spend a row on internal tool id: {text}" |
| 2871 | ); |
| 2872 | assert!( |
| 2873 | !text.contains("run_verifiers"), |
| 2874 | "known tool id should not leak into compact live card: {text}" |
| 2875 | ); |
| 2876 | } |
| 2877 | |
| 2878 | #[test] |
| 2879 | fn known_generic_tool_keeps_raw_name_in_transcript_mode() { |
| 2880 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 2881 | name: "run_verifiers".to_string(), |
| 2882 | status: ToolStatus::Running, |
| 2883 | input_summary: Some("profile: auto, level: quick".to_string()), |
| 2884 | output: None, |
| 2885 | prompts: None, |
| 2886 | spillover_path: None, |
| 2887 | output_summary: None, |
| 2888 | is_diff: false, |
| 2889 | })); |
| 2890 | |
| 2891 | let text = lines_text(&cell.transcript_lines(80)); |
| 2892 | assert!(text.contains("verify running"), "{text}"); |
| 2893 | assert!( |
| 2894 | text.contains("name: run_verifiers"), |
| 2895 | "transcript replay should preserve exact tool id: {text}" |
| 2896 | ); |
| 2897 | } |
| 2898 | |
| 2899 | #[test] |
| 2900 | fn unknown_generic_tool_keeps_raw_name_in_live_mode() { |
| 2901 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 2902 | name: "future_private_tool".to_string(), |
| 2903 | status: ToolStatus::Running, |
| 2904 | input_summary: Some("query: foo".to_string()), |
| 2905 | output: None, |
| 2906 | prompts: None, |
| 2907 | spillover_path: None, |
| 2908 | output_summary: None, |
| 2909 | is_diff: false, |
| 2910 | })); |
| 2911 | |
| 2912 | let text = lines_text(&cell.lines(80)); |
| 2913 | // Unknown/Generic tools collapse to a single header line in live mode. |
| 2914 | assert!( |
| 2915 | !text.is_empty(), |
| 2916 | "collapsed header must still render: {text}" |
| 2917 | ); |
| 2918 | } |
| 2919 | |
| 2920 | #[test] |
| 2921 | fn generic_tool_cell_preserves_multi_line_output_in_transcript() { |
| 2922 | // Repro for #80: a `git diff --stat`-shaped tool result should keep |
| 2923 | // its newlines on the transcript surface — one file per row, not |
| 2924 | // squashed into a single line. |
| 2925 | let diff_stat = "Cargo.lock | 1 +\n\ |
| 2926 | crates/cli/Cargo.toml | 1 +\n\ |
| 2927 | crates/cli/src/main.rs | 47 ++++++\n\ |
| 2928 | crates/config/src/lib.rs | 27 ++++\n\ |
| 2929 | crates/tui/src/mcp.rs | 384 +++++"; |
| 2930 | |
| 2931 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 2932 | name: "read_file".to_string(), |
| 2933 | status: ToolStatus::Success, |
| 2934 | input_summary: Some("command: git diff --stat".to_string()), |
| 2935 | output: Some(diff_stat.to_string()), |
| 2936 | prompts: None, |
| 2937 | spillover_path: None, |
| 2938 | output_summary: None, |
| 2939 | is_diff: false, |
| 2940 | })); |
| 2941 | |
| 2942 | let transcript_text = lines_text(&cell.transcript_lines(80)); |
| 2943 | |
| 2944 | // Each file path must appear on its own row in the transcript. |
| 2945 | for needle in [ |
| 2946 | "Cargo.lock", |
| 2947 | "crates/cli/Cargo.toml", |
| 2948 | "crates/cli/src/main.rs", |
| 2949 | "crates/config/src/lib.rs", |
| 2950 | "crates/tui/src/mcp.rs", |
| 2951 | ] { |
| 2952 | assert!( |
| 2953 | transcript_text.contains(needle), |
| 2954 | "transcript missing '{needle}': {transcript_text}" |
| 2955 | ); |
| 2956 | } |
| 2957 | // The pre-fix bug: result line containing |
| 2958 | // "Cargo.lock | 1 + crates/cli/Cargo.toml" — joined into one row. |
| 2959 | // With the fix, the diff-stat pipes are still present per-line, but |
| 2960 | // adjacent file paths are on separate rendered rows. Assert that the |
| 2961 | // first file's line ends before the second begins. |
| 2962 | let lines: Vec<&str> = transcript_text.lines().collect(); |
| 2963 | let cargo_lock_line = lines |
| 2964 | .iter() |
| 2965 | .find(|l| l.contains("Cargo.lock")) |
| 2966 | .expect("Cargo.lock row must exist"); |
| 2967 | assert!( |
| 2968 | !cargo_lock_line.contains("crates/cli/Cargo.toml"), |
| 2969 | "Cargo.lock row must not also contain the second file: {cargo_lock_line}" |
| 2970 | ); |
| 2971 | } |
| 2972 | |
| 2973 | #[test] |
| 2974 | fn generic_tool_cell_expands_failed_multi_line_output_in_live() { |
| 2975 | // Failed tools should auto-expand in live mode so the command/input summary |
| 2976 | // and full error output remain immediately visible. |
| 2977 | let total = 30usize; |
| 2978 | let output = (0..total) |
| 2979 | .map(|i| format!("row {i:02}: payload")) |
| 2980 | .collect::<Vec<_>>() |
| 2981 | .join("\n"); |
| 2982 | |
| 2983 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 2984 | name: "read_file".to_string(), |
| 2985 | status: ToolStatus::Failed, |
| 2986 | input_summary: Some("command: ls".to_string()), |
| 2987 | output: Some(output), |
| 2988 | prompts: None, |
| 2989 | spillover_path: None, |
| 2990 | output_summary: None, |
| 2991 | is_diff: false, |
| 2992 | })); |
| 2993 | |
| 2994 | let live = cell.lines_with_options(80, TranscriptRenderOptions::default()); |
| 2995 | let transcript = cell.transcript_lines(80); |
| 2996 | let live_text = lines_text(&live); |
| 2997 | let transcript_text = lines_text(&transcript); |
| 2998 | |
| 2999 | assert!(live_text.contains("command: ls"), "{live_text}"); |
| 3000 | assert!( |
| 3001 | !live_text.contains("lines omitted"), |
| 3002 | "failed output must not be hidden behind an omission marker: {live_text}" |
| 3003 | ); |
| 3004 | assert!(transcript_text.contains("row 29")); |
| 3005 | assert!(live_text.contains("row 29")); |
| 3006 | } |
| 3007 | |
| 3008 | #[test] |
| 3009 | fn generic_tool_failed_output_live_renders_card_rail() { |
| 3010 | let output = (0..24usize) |
| 3011 | .map(|i| format!("line {i:02}")) |
| 3012 | .collect::<Vec<_>>() |
| 3013 | .join("\n"); |
| 3014 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 3015 | name: "read_file".to_string(), |
| 3016 | status: ToolStatus::Failed, |
| 3017 | input_summary: Some("command: noisy".to_string()), |
| 3018 | output: Some(output), |
| 3019 | prompts: None, |
| 3020 | spillover_path: None, |
| 3021 | output_summary: None, |
| 3022 | is_diff: false, |
| 3023 | })); |
| 3024 | |
| 3025 | let live_text = lines_text(&cell.lines_with_options(80, TranscriptRenderOptions::default())); |
| 3026 | |
| 3027 | // Card-rail wrapping: first line starts with ╭, last with ╰. |
| 3028 | assert!( |
| 3029 | live_text.starts_with('\u{256D}'), |
| 3030 | "live view must start with card-rail top glyph ╭: {live_text}" |
| 3031 | ); |
| 3032 | assert!(!live_text.contains("lines omitted"), "{live_text}"); |
| 3033 | assert!(live_text.contains("line 00")); |
| 3034 | assert!(live_text.contains("line 23")); |
| 3035 | } |
| 3036 | |
| 3037 | #[test] |
| 3038 | fn hidden_tool_details_keeps_failed_generic_output_expanded() { |
| 3039 | let output = (0..30usize) |
| 3040 | .map(|i| format!("row {i:02}: payload")) |
| 3041 | .collect::<Vec<_>>() |
| 3042 | .join("\n"); |
| 3043 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 3044 | name: "read_file".to_string(), |
| 3045 | status: ToolStatus::Failed, |
| 3046 | input_summary: Some("command: noisy".to_string()), |
| 3047 | output: Some(output), |
| 3048 | prompts: None, |
| 3049 | spillover_path: None, |
| 3050 | output_summary: None, |
| 3051 | is_diff: false, |
| 3052 | })); |
| 3053 | |
| 3054 | let live_text = lines_text(&cell.lines_with_options( |
| 3055 | 80, |
| 3056 | TranscriptRenderOptions { |
| 3057 | show_tool_details: false, |
| 3058 | ..TranscriptRenderOptions::default() |
| 3059 | }, |
| 3060 | )); |
| 3061 | |
| 3062 | assert!( |
| 3063 | !live_text.contains("lines omitted") && !live_text.contains("details"), |
| 3064 | "failed output must not be hidden behind a details affordance: {live_text}" |
| 3065 | ); |
| 3066 | assert!(live_text.contains("row 29"), "{live_text}"); |
| 3067 | } |
| 3068 | |
| 3069 | #[test] |
| 3070 | fn calm_mode_keeps_failed_generic_output_expanded() { |
| 3071 | let output = (0..30usize) |
| 3072 | .map(|i| format!("row {i:02}: payload")) |
| 3073 | .collect::<Vec<_>>() |
| 3074 | .join("\n"); |
| 3075 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 3076 | name: "read_file".to_string(), |
| 3077 | status: ToolStatus::Failed, |
| 3078 | input_summary: Some("command: noisy".to_string()), |
| 3079 | output: Some(output), |
| 3080 | prompts: None, |
| 3081 | spillover_path: None, |
| 3082 | output_summary: None, |
| 3083 | is_diff: false, |
| 3084 | })); |
| 3085 | |
| 3086 | let live_text = lines_text(&cell.lines_with_options( |
| 3087 | 80, |
| 3088 | TranscriptRenderOptions { |
| 3089 | calm_mode: true, |
| 3090 | ..TranscriptRenderOptions::default() |
| 3091 | }, |
| 3092 | )); |
| 3093 | |
| 3094 | assert!( |
| 3095 | !live_text.contains("lines omitted") && !live_text.contains("details"), |
| 3096 | "failed output must not be hidden behind a details affordance: {live_text}" |
| 3097 | ); |
| 3098 | assert!(live_text.contains("row 29"), "{live_text}"); |
| 3099 | } |
| 3100 | |
| 3101 | #[test] |
| 3102 | fn generic_tool_success_live_collapses_output_transcript_keeps_it() { |
| 3103 | let output = (0..24usize) |
| 3104 | .map(|i| format!("row {i:02}: payload")) |
| 3105 | .collect::<Vec<_>>() |
| 3106 | .join("\n"); |
| 3107 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 3108 | name: "read_file".to_string(), |
| 3109 | status: ToolStatus::Success, |
| 3110 | input_summary: Some("path: crates/tui/src/main.rs".to_string()), |
| 3111 | output: Some(output), |
| 3112 | prompts: None, |
| 3113 | spillover_path: None, |
| 3114 | output_summary: None, |
| 3115 | is_diff: false, |
| 3116 | })); |
| 3117 | |
| 3118 | let live_text = lines_text(&cell.lines_with_options(80, TranscriptRenderOptions::default())); |
| 3119 | let transcript_text = lines_text(&cell.transcript_lines(80)); |
| 3120 | |
| 3121 | assert!( |
| 3122 | !live_text.contains("row 00"), |
| 3123 | "successful generic tool output should be hidden live: {live_text}" |
| 3124 | ); |
| 3125 | assert!( |
| 3126 | !live_text.contains("lines omitted"), |
| 3127 | "collapsed success should not spend a row on an omission marker: {live_text}" |
| 3128 | ); |
| 3129 | assert!(transcript_text.contains("row 00")); |
| 3130 | assert!(transcript_text.contains("row 23")); |
| 3131 | } |
| 3132 | |
| 3133 | #[test] |
| 3134 | fn tool_output_live_preserves_error_card_rail() { |
| 3135 | let output = [ |
| 3136 | "start", |
| 3137 | "still starting", |
| 3138 | "middle noise 1", |
| 3139 | "fatal: failed to read /tmp/deepseek/config.toml", |
| 3140 | "middle noise 2", |
| 3141 | "see https://example.test/build/log for details", |
| 3142 | "middle noise 3", |
| 3143 | "almost done", |
| 3144 | "final line", |
| 3145 | ] |
| 3146 | .join("\n"); |
| 3147 | let cell = HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 3148 | name: "read_file".to_string(), |
| 3149 | status: ToolStatus::Failed, |
| 3150 | input_summary: Some("command: tool".to_string()), |
| 3151 | output: Some(output), |
| 3152 | prompts: None, |
| 3153 | spillover_path: None, |
| 3154 | output_summary: Some("Error: failed to read config".to_string()), |
| 3155 | is_diff: false, |
| 3156 | })); |
| 3157 | |
| 3158 | let live_text = lines_text(&cell.lines_with_options(80, TranscriptRenderOptions::default())); |
| 3159 | |
| 3160 | assert!( |
| 3161 | !live_text.contains("lines omitted"), |
| 3162 | "failed output must not be hidden behind an omission marker: {live_text}" |
| 3163 | ); |
| 3164 | assert!( |
| 3165 | live_text.contains("Error:") || live_text.contains("fatal:"), |
| 3166 | "live summary should capture error text: {live_text}" |
| 3167 | ); |
| 3168 | assert!(live_text.contains("final line"), "{live_text}"); |
| 3169 | } |
| 3170 | |
| 3171 | // === ErrorEnvelope severity → cell color tests (#66) === |
| 3172 | |
| 3173 | /// Snapshot: an `Error`-severity cell uses the red status palette token |
| 3174 | /// for both the leading "Error" label glyph and the body. This is the |
| 3175 | /// load-bearing visual signal that distinguishes an error cell from a |
| 3176 | /// neutral system note. |
| 3177 | #[test] |
| 3178 | fn error_severity_cell_renders_in_red() { |
| 3179 | let cell = HistoryCell::Error { |
| 3180 | message: "Authentication failed: invalid API key".to_string(), |
| 3181 | severity: crate::error_taxonomy::ErrorSeverity::Error, |
| 3182 | }; |
| 3183 | let lines = cell.lines(80); |
| 3184 | assert!( |
| 3185 | !lines.is_empty(), |
| 3186 | "error cell must render at least one line" |
| 3187 | ); |
| 3188 | |
| 3189 | let head = &lines[0]; |
| 3190 | let label_span = &head.spans[0]; |
| 3191 | assert_eq!(label_span.content.as_ref(), "Error"); |
| 3192 | assert_eq!(label_span.style.fg, Some(palette::STATUS_ERROR)); |
| 3193 | assert!(label_span.style.add_modifier.contains(Modifier::BOLD)); |
| 3194 | |
| 3195 | // The body carries the error message and is rendered in the same red. |
| 3196 | let body_text = lines |
| 3197 | .iter() |
| 3198 | .flat_map(|line| line.spans.iter().map(|span| span.content.as_ref())) |
| 3199 | .collect::<String>(); |
| 3200 | assert!(body_text.contains("Authentication failed")); |
| 3201 | // Find a span whose text contains "Authentication" and verify its color. |
| 3202 | let body_span = lines |
| 3203 | .iter() |
| 3204 | .flat_map(|line| line.spans.iter()) |
| 3205 | .find(|span| span.content.contains("Authentication")) |
| 3206 | .expect("error body span must exist"); |
| 3207 | assert_eq!(body_span.style.fg, Some(palette::STATUS_ERROR)); |
| 3208 | } |
| 3209 | |
| 3210 | /// `Warning`-severity uses amber, not red — distinguishes a transient |
| 3211 | /// retry hiccup from a hard failure. |
| 3212 | #[test] |
| 3213 | fn warning_severity_cell_renders_in_amber() { |
| 3214 | let cell = HistoryCell::Error { |
| 3215 | message: "Stream stalled: no data received for 60s, closing stream".to_string(), |
| 3216 | severity: crate::error_taxonomy::ErrorSeverity::Warning, |
| 3217 | }; |
| 3218 | let lines = cell.lines(80); |
| 3219 | let label_span = &lines[0].spans[0]; |
| 3220 | assert_eq!(label_span.content.as_ref(), "Warn"); |
| 3221 | assert_eq!(label_span.style.fg, Some(palette::STATUS_WARNING)); |
| 3222 | } |
| 3223 | |
| 3224 | /// `Critical` severity collapses to the same red as `Error` — both flip |
| 3225 | /// offline mode and both should read as the loudest signal in the |
| 3226 | /// transcript. |
| 3227 | #[test] |
| 3228 | fn critical_severity_cell_renders_in_red() { |
| 3229 | let cell = HistoryCell::Error { |
| 3230 | message: "API key expired".to_string(), |
| 3231 | severity: crate::error_taxonomy::ErrorSeverity::Critical, |
| 3232 | }; |
| 3233 | let lines = cell.lines(80); |
| 3234 | let label_span = &lines[0].spans[0]; |
| 3235 | assert_eq!(label_span.content.as_ref(), "Error"); |
| 3236 | assert_eq!(label_span.style.fg, Some(palette::STATUS_ERROR)); |
| 3237 | } |
| 3238 | |
| 3239 | /// `Info` severity stays neutral / dim so it doesn't draw the eye away |
| 3240 | /// from real failures sitting alongside it in the transcript. |
| 3241 | #[test] |
| 3242 | fn info_severity_cell_renders_in_dim() { |
| 3243 | let cell = HistoryCell::Error { |
| 3244 | message: "Reconnected".to_string(), |
| 3245 | severity: crate::error_taxonomy::ErrorSeverity::Info, |
| 3246 | }; |
| 3247 | let lines = cell.lines(80); |
| 3248 | let label_span = &lines[0].spans[0]; |
| 3249 | assert_eq!(label_span.content.as_ref(), "Info"); |
| 3250 | assert_eq!(label_span.style.fg, Some(palette::TEXT_DIM)); |
| 3251 | } |
| 3252 | |
| 3253 | fn success_generic_tool(name: &str) -> HistoryCell { |
| 3254 | HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 3255 | name: name.to_string(), |
| 3256 | status: ToolStatus::Success, |
| 3257 | input_summary: Some(format!("args for {name}")), |
| 3258 | output: Some(format!("output for {name}")), |
| 3259 | prompts: None, |
| 3260 | spillover_path: None, |
| 3261 | output_summary: None, |
| 3262 | is_diff: false, |
| 3263 | })) |
| 3264 | } |
| 3265 | |
| 3266 | fn failed_generic_tool(name: &str) -> HistoryCell { |
| 3267 | HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 3268 | name: name.to_string(), |
| 3269 | status: ToolStatus::Failed, |
| 3270 | input_summary: None, |
| 3271 | output: Some("failed".to_string()), |
| 3272 | prompts: None, |
| 3273 | spillover_path: None, |
| 3274 | output_summary: None, |
| 3275 | is_diff: false, |
| 3276 | })) |
| 3277 | } |
| 3278 | |
| 3279 | fn running_generic_tool(name: &str) -> HistoryCell { |
| 3280 | HistoryCell::Tool(ToolCell::Generic(GenericToolCell { |
| 3281 | name: name.to_string(), |
| 3282 | status: ToolStatus::Running, |
| 3283 | input_summary: None, |
| 3284 | output: None, |
| 3285 | prompts: None, |
| 3286 | spillover_path: None, |
| 3287 | output_summary: None, |
| 3288 | is_diff: false, |
| 3289 | })) |
| 3290 | } |
| 3291 | |
| 3292 | fn shell_tool(command: &str) -> HistoryCell { |
| 3293 | HistoryCell::Tool(ToolCell::Exec(ExecCell { |
| 3294 | command: command.to_string(), |
| 3295 | status: ToolStatus::Success, |
| 3296 | output: Some("ok".to_string()), |
| 3297 | live_output: None, |
| 3298 | shell_task_id: None, |
| 3299 | owner_agent_id: None, |
| 3300 | owner_agent_name: None, |
| 3301 | started_at: None, |
| 3302 | duration_ms: None, |
| 3303 | stale_elapsed_since_output_ms: None, |
| 3304 | source: ExecSource::Assistant, |
| 3305 | interaction: None, |
| 3306 | output_summary: None, |
| 3307 | })) |
| 3308 | } |
| 3309 | |
| 3310 | #[test] |
| 3311 | fn detect_tool_runs_finds_contiguous_successful_safe_tools() { |
| 3312 | let history = vec![ |
| 3313 | HistoryCell::User { |
| 3314 | content: "go".to_string(), |
| 3315 | }, |
| 3316 | success_generic_tool("read_file"), |
| 3317 | success_generic_tool("list_dir"), |
| 3318 | success_generic_tool("web_search"), |
| 3319 | HistoryCell::Assistant { |
| 3320 | content: "done".to_string(), |
| 3321 | streaming: false, |
| 3322 | }, |
| 3323 | ]; |
| 3324 | |
| 3325 | let runs = super::detect_tool_runs(&history, 3); |
| 3326 | |
| 3327 | assert_eq!(runs.len(), 1); |
| 3328 | assert_eq!(runs[0].start, 1); |
| 3329 | assert_eq!(runs[0].count, 3); |
| 3330 | assert_eq!( |
| 3331 | runs[0].tool_families, |
| 3332 | vec!["read_file", "list_dir", "web_search"] |
| 3333 | ); |
| 3334 | assert_eq!(runs[0].activity.files, 2); |
| 3335 | assert_eq!(runs[0].activity.searches, 1); |
| 3336 | } |
| 3337 | |
| 3338 | #[test] |
| 3339 | fn detect_tool_runs_honors_threshold_and_boundaries() { |
| 3340 | let short = vec![ |
| 3341 | success_generic_tool("read_file"), |
| 3342 | success_generic_tool("list_dir"), |
| 3343 | ]; |
| 3344 | assert!(super::detect_tool_runs(&short, 3).is_empty()); |
| 3345 | |
| 3346 | let with_assistant_boundary = vec![ |
| 3347 | success_generic_tool("read_file"), |
| 3348 | HistoryCell::Assistant { |
| 3349 | content: "pause".to_string(), |
| 3350 | streaming: false, |
| 3351 | }, |
| 3352 | success_generic_tool("list_dir"), |
| 3353 | success_generic_tool("web_search"), |
| 3354 | ]; |
| 3355 | assert!(super::detect_tool_runs(&with_assistant_boundary, 3).is_empty()); |
| 3356 | } |
| 3357 | |
| 3358 | #[test] |
| 3359 | fn detect_tool_runs_keeps_failed_running_and_shell_cells_visible() { |
| 3360 | let history = vec![ |
| 3361 | success_generic_tool("read_file"), |
| 3362 | success_generic_tool("list_dir"), |
| 3363 | failed_generic_tool("web_search"), |
| 3364 | success_generic_tool("read_file"), |
| 3365 | success_generic_tool("list_dir"), |
| 3366 | running_generic_tool("web_search"), |
| 3367 | success_generic_tool("read_file"), |
| 3368 | success_generic_tool("list_dir"), |
| 3369 | shell_tool("rm -rf target"), |
| 3370 | success_generic_tool("read_file"), |
| 3371 | success_generic_tool("list_dir"), |
| 3372 | success_generic_tool("web_search"), |
| 3373 | ]; |
| 3374 | |
| 3375 | let runs = super::detect_tool_runs(&history, 3); |
| 3376 | |
| 3377 | assert_eq!(runs.len(), 1); |
| 3378 | assert_eq!(runs[0].start, 9); |
| 3379 | assert_eq!(runs[0].count, 3); |
| 3380 | } |
| 3381 | |
| 3382 | #[test] |
| 3383 | fn detect_tool_runs_summarizes_safe_command_tools() { |
| 3384 | let history = vec![ |
| 3385 | success_generic_tool("run_tests"), |
| 3386 | success_generic_tool("run_verifiers"), |
| 3387 | success_generic_tool("validate_data"), |
| 3388 | ]; |
| 3389 | |
| 3390 | let runs = super::detect_tool_runs(&history, 3); |
| 3391 | |
| 3392 | assert_eq!(runs.len(), 1); |
| 3393 | assert_eq!(runs[0].start, 0); |
| 3394 | assert_eq!(runs[0].count, 3); |
| 3395 | assert_eq!(runs[0].activity.commands, 3); |
| 3396 | assert_eq!( |
| 3397 | runs[0].tool_families, |
| 3398 | vec!["run_tests", "run_verifiers", "validate_data"] |
| 3399 | ); |
| 3400 | assert_eq!( |
| 3401 | super::tool_run_summary(&runs[0]), |
| 3402 | "Ran 3 commands: run_tests, run_verifiers, validate_data" |
| 3403 | ); |
| 3404 | } |
| 3405 | |
| 3406 | #[test] |
| 3407 | fn tool_run_summary_reports_compact_success_group() { |
| 3408 | let run = super::ToolRun { |
| 3409 | start: 4, |
| 3410 | count: 5, |
| 3411 | tool_families: vec!["read_file".to_string(), "list_dir".to_string()], |
| 3412 | activity: super::ToolRunActivitySummary { |
| 3413 | files: 4, |
| 3414 | searches: 1, |
| 3415 | ..Default::default() |
| 3416 | }, |
| 3417 | }; |
| 3418 | |
| 3419 | let summary = super::tool_run_summary(&run); |
| 3420 | |
| 3421 | assert_eq!(summary, "Explored 4 files, 1 search: read_file, list_dir"); |
| 3422 | } |
| 3423 | |
| 3424 | #[test] |
| 3425 | fn tool_run_summary_keeps_git_history_tools_visible() { |
| 3426 | let history = vec![ |
| 3427 | success_generic_tool("git_log"), |
| 3428 | success_generic_tool("git_show"), |
| 3429 | success_generic_tool("git_blame"), |
| 3430 | ]; |
| 3431 | |
| 3432 | let runs = super::detect_tool_runs(&history, 3); |
| 3433 | |
| 3434 | assert_eq!(runs.len(), 1); |
| 3435 | assert_eq!(runs[0].activity.files, 3); |
| 3436 | assert_eq!( |
| 3437 | super::tool_run_summary(&runs[0]), |
| 3438 | "Explored 3 files: git_log, git_show, git_blame" |
| 3439 | ); |
| 3440 | } |
| 3441 | |
| 3442 | #[test] |
| 3443 | fn tool_run_summary_lists_only_command_families_for_command_clause() { |
| 3444 | let run = super::ToolRun { |
| 3445 | start: 4, |
| 3446 | count: 4, |
| 3447 | tool_families: vec![ |
| 3448 | "read_file".to_string(), |
| 3449 | "run_tests".to_string(), |
| 3450 | "validate_data".to_string(), |
| 3451 | ], |
| 3452 | activity: super::ToolRunActivitySummary { |
| 3453 | files: 2, |
| 3454 | commands: 2, |
| 3455 | ..Default::default() |
| 3456 | }, |
| 3457 | }; |
| 3458 | |
| 3459 | assert_eq!( |
| 3460 | super::tool_run_summary(&run), |
| 3461 | "Explored 2 files: read_file, ran 2 commands: run_tests, validate_data" |
| 3462 | ); |
| 3463 | } |
| 3464 | |
| 3465 | #[test] |
| 3466 | fn tool_run_summary_uses_metadata_fallback_for_unknown_groups() { |
| 3467 | let run = super::ToolRun { |
| 3468 | start: 4, |
| 3469 | count: 2, |
| 3470 | tool_families: vec!["session_sync".to_string()], |
| 3471 | activity: super::ToolRunActivitySummary { |
| 3472 | other: 2, |
| 3473 | ..Default::default() |
| 3474 | }, |
| 3475 | }; |
| 3476 | |
| 3477 | assert_eq!(super::tool_run_summary(&run), "Updated metadata"); |
| 3478 | } |
| 3479 | |
| 3480 | // ---- #4112 / dogfood A5: transcript noise ---- |
| 3481 | |
| 3482 | fn agent_cell( |
| 3483 | action_summary: Option<&str>, |
| 3484 | status: ToolStatus, |
| 3485 | output: Option<&str>, |
| 3486 | ) -> GenericToolCell { |
| 3487 | GenericToolCell { |
| 3488 | name: "agent".to_string(), |
| 3489 | status, |
| 3490 | input_summary: action_summary.map(str::to_string), |
| 3491 | output: output.map(str::to_string), |
| 3492 | prompts: None, |
| 3493 | spillover_path: None, |
| 3494 | output_summary: None, |
| 3495 | is_diff: false, |
| 3496 | } |
| 3497 | } |
| 3498 | |
| 3499 | fn joined_lines(cell: &GenericToolCell, mode: super::RenderMode) -> String { |
| 3500 | cell.lines_with_mode(120, true, mode) |
| 3501 | .iter() |
| 3502 | .flat_map(|l| l.spans.iter().map(|s| s.content.as_ref().to_string())) |
| 3503 | .collect() |
| 3504 | } |
| 3505 | |
| 3506 | #[test] |
| 3507 | fn unknown_tool_failure_collapses_to_one_line() { |
| 3508 | let cell = GenericToolCell { |
| 3509 | name: "item".to_string(), |
| 3510 | status: ToolStatus::Failed, |
| 3511 | input_summary: Some("status: pending".to_string()), |
| 3512 | output: Some( |
| 3513 | "Tool 'item' is not available in the current tool catalog. \ |
| 3514 | Checklist entries are not separate tool calls." |
| 3515 | .to_string(), |
| 3516 | ), |
| 3517 | prompts: None, |
| 3518 | spillover_path: None, |
| 3519 | output_summary: None, |
| 3520 | is_diff: false, |
| 3521 | }; |
| 3522 | for mode in [super::RenderMode::Live, super::RenderMode::Transcript] { |
| 3523 | let lines = cell.lines_with_mode(120, true, mode); |
| 3524 | assert_eq!( |
| 3525 | lines.len(), |
| 3526 | 1, |
| 3527 | "unknown-tool failure should be a single header line in {mode:?}: {lines:?}" |
| 3528 | ); |
| 3529 | let joined = joined_lines(&cell, mode); |
| 3530 | assert!( |
| 3531 | joined.contains("Tool 'item' is not available"), |
| 3532 | "the catalog error is the useful part: {joined:?}" |
| 3533 | ); |
| 3534 | assert!( |
| 3535 | !joined.contains("name: item"), |
| 3536 | "no name:/args:/result: block for unknown tools: {joined:?}" |
| 3537 | ); |
| 3538 | } |
| 3539 | } |
| 3540 | |
| 3541 | #[test] |
| 3542 | fn agent_peek_renders_checked_not_done() { |
| 3543 | let cell = agent_cell( |
| 3544 | Some("action: peek agent_id: agent_scout_1"), |
| 3545 | ToolStatus::Success, |
| 3546 | Some(r#"{"agent_id":"agent_scout_1","status":"running"}"#), |
| 3547 | ); |
| 3548 | let joined = joined_lines(&cell, super::RenderMode::Live); |
| 3549 | assert!( |
| 3550 | joined.contains("checked") && joined.contains("agent_scout_1"), |
| 3551 | "peek should read as a check, not a completed delegate: {joined:?}" |
| 3552 | ); |
| 3553 | assert!( |
| 3554 | !joined.contains("delegate done"), |
| 3555 | "peek must not draw the spawn-completion line: {joined:?}" |
| 3556 | ); |
| 3557 | } |
| 3558 | |
| 3559 | #[test] |
| 3560 | fn agent_wait_renders_waited_label() { |
| 3561 | let cell = agent_cell( |
| 3562 | Some("action: wait"), |
| 3563 | ToolStatus::Success, |
| 3564 | Some(r#"{"action":"wait","settled":[{"agent_id":"agent_scout_1"}]}"#), |
| 3565 | ); |
| 3566 | let joined = joined_lines(&cell, super::RenderMode::Live); |
| 3567 | assert!( |
| 3568 | joined.contains("waited"), |
| 3569 | "wait cells should read as a join: {joined:?}" |
| 3570 | ); |
| 3571 | } |
| 3572 | |
| 3573 | #[test] |
| 3574 | fn agent_inspection_stays_compact_in_transcript_mode() { |
| 3575 | let cell = agent_cell( |
| 3576 | Some("action: status agent_id: agent_scout_1"), |
| 3577 | ToolStatus::Success, |
| 3578 | Some(r#"{"agent_id":"agent_scout_1","status":"running","terminal":false}"#), |
| 3579 | ); |
| 3580 | let lines = cell.lines_with_mode(120, true, super::RenderMode::Transcript); |
| 3581 | assert_eq!( |
| 3582 | lines.len(), |
| 3583 | 1, |
| 3584 | "status checks should not dump full projections in the pager: {lines:?}" |
| 3585 | ); |
| 3586 | } |
| 3587 | |
| 3588 | #[test] |
| 3589 | fn agent_spawn_suppresses_generic_card_in_favor_of_delegate_card() { |
| 3590 | let cell = agent_cell( |
| 3591 | Some("prompt: map the repo"), |
| 3592 | ToolStatus::Success, |
| 3593 | Some(r#"{"agent_id":"agent_scout_1","status":"running"}"#), |
| 3594 | ); |
| 3595 | for mode in [super::RenderMode::Live, super::RenderMode::Transcript] { |
| 3596 | let lines = cell.lines_with_mode(120, true, mode); |
| 3597 | assert!( |
| 3598 | lines.is_empty(), |
| 3599 | "spawn generic tool card must yield to DelegateCard (#4133): {mode:?} {lines:?}" |
| 3600 | ); |
| 3601 | } |
| 3602 | } |
| 3603 |