1use crate::harness::{self, AttributeContext, HarnessAdapter, RegistryHints, SessionSummary, SessionTracker, SpanRetention};
9use crate::model::*;
10use crate::process::{ProcessScanner, RawProc, build_forest};
11use std::collections::{HashMap, HashSet};
12use std::path::{Path, PathBuf};
13use std::time::{Duration, Instant, SystemTime, UNIX_EPOCH};
14
15#[derive(Debug, Clone)]
16pub struct CollectorOptions {
17 pub stopped_window: Duration,
19 pub fs_scan_interval: Duration,
21 pub activity_timeout: Duration,
24}
25
26impl Default for CollectorOptions {
27 fn default() -> Self {
28 CollectorOptions {
29 stopped_window: Duration::from_secs(30 * 60),
30 fs_scan_interval: Duration::from_secs(5),
31 activity_timeout: Duration::from_secs(15 * 60),
32 }
33 }
34}
35
36pub struct Collector {
37 opts: CollectorOptions,
38 scanner: ProcessScanner,
39 adapters: Vec<Box<dyn HarnessAdapter>>,
40 trackers: HashMap<PathBuf, Box<dyn SessionTracker>>,
41 last_fs_scan: Option<Instant>,
42}
43
44impl Collector {
45 pub fn new(opts: CollectorOptions) -> Self {
46 Collector { opts, scanner: ProcessScanner::new(), adapters: harness::adapters(), trackers: HashMap::new(), last_fs_scan: None }
47 }
48
49 fn rescan_fs_if_due(&mut self) {
50 let due = self.last_fs_scan.map(|t| t.elapsed() >= self.opts.fs_scan_interval).unwrap_or(true);
51 if !due {
52 return;
53 }
54 self.last_fs_scan = Some(Instant::now());
55 let since = SystemTime::now().checked_sub(self.opts.stopped_window).unwrap_or(UNIX_EPOCH);
56 for a in &mut self.adapters {
57 a.rescan(since);
58 }
59 }
60
61 fn adapter(&self, harness: Harness) -> Option<&dyn HarnessAdapter> {
62 self.adapters.iter().find(|a| a.harness() == harness).map(|a| a.as_ref())
63 }
64
65 fn tracker_for(&mut self, path: &Path, harness: Harness) -> Option<&mut Box<dyn SessionTracker>> {
67 if !self.trackers.contains_key(path) {
68 let tracker = self.adapter(harness)?.open(path, SpanRetention::Recent);
69 self.trackers.insert(path.to_path_buf(), tracker);
70 }
71 self.trackers.get_mut(path)
72 }
73
74 pub fn collect(&mut self) -> Snapshot {
75 self.scanner.refresh();
76 self.rescan_fs_if_due();
77 let host = self.scanner.host();
78 let procs = self.scanner.processes();
79 let by_pid: HashMap<u32, &RawProc> = procs.iter().map(|p| (p.pid, p)).collect();
80 let (roots, orphans) = build_forest(&procs);
81
82 for a in &mut self.adapters {
84 let mine: Vec<&ProcNode> = roots.iter().filter(|r| r.harness == Some(a.harness())).collect();
85 a.prepare(&mine);
86 }
87
88 let now = SystemTime::now();
89 let mut agents = Vec::new();
90 let mut attached: HashSet<PathBuf> = HashSet::new();
91
92 for root in roots {
93 let raw = by_pid.get(&root.pid).copied();
94 let harness = root.harness.unwrap_or(Harness::Unknown);
95 let proc_start = raw.map(|p| UNIX_EPOCH + Duration::from_secs(p.start_time)).unwrap_or(now);
96 let hints = self.adapter(harness).and_then(|a| a.hints(root.pid));
97 let cwd = root.cwd.clone().or_else(|| hints.as_ref().and_then(|h| h.cwd.clone()));
98
99 let (paths, attribution) = match self.adapter(harness) {
102 Some(a) => {
103 let ctx = AttributeContext {
104 cwd: cwd.as_deref(),
105 proc_start,
106 now,
107 attached: &attached,
108 activity_timeout: self.opts.activity_timeout,
109 };
110 a.attribute(&root, raw, &ctx)
111 }
112 None => (Vec::new(), Attribution::None),
113 };
114
115 let (cpu, rss, count, mcp) = root.totals();
116 let hints = hints.unwrap_or_default();
117
118 if paths.is_empty() {
120 let summary = SessionSummary::default();
121 let state = live_state(&hints, summary.activity, None, cpu, &self.opts);
122 agents.push(Agent {
123 id: format!("pid:{}", root.pid),
124 name: hints.name.clone().unwrap_or_else(|| display_name(harness, cwd.as_deref())),
125 harness,
126 state,
127 activity: summary.activity,
128 pid: Some(root.pid),
129 session_id: hints.session_id.clone(),
130 session_path: None,
131 cwd,
132 model: None,
133 harness_version: hints.version.clone(),
134 usage: summary.usage,
135 cost_usd: 0.0,
136 cost_breakdown: Default::default(),
137 price_source: None,
138 unpriced_tokens: 0,
139 turns: 0,
140 subagent_turns: 0,
141 tool_calls: 0,
142 web_searches: 0,
143 spans: Vec::new(),
144 age_secs: root.age_secs,
145 idle_secs: None,
146 cpu_percent: cpu,
147 rss_bytes: rss,
148 process_count: count,
149 mcp_count: mcp,
150 tree: Some(root),
151 attribution,
152 shares_process: false,
153 parse_warning: None,
154 });
155 continue;
156 }
157
158 for (i, path) in paths.iter().enumerate() {
159 let owns_process = i == 0;
162 let Some(tr) = self.tracker_for(path, harness) else { continue };
163 let _ = tr.refresh();
164 let mut summary = tr.summary().clone();
165 attached.insert(path.clone());
166
167 if summary.session_id.is_none() {
168 summary.session_id = hints.session_id.clone();
169 }
170 if summary.harness_version.is_none() {
171 summary.harness_version = hints.version.clone();
172 }
173
174 let idle_secs = summary.last_activity.and_then(|t| now.duration_since(t).ok()).map(|d| d.as_secs());
175 let state = live_state(&hints, summary.activity, idle_secs, cpu, &self.opts);
176 let name = match (hints.name.clone(), paths.len()) {
179 (Some(n), 1) => n,
180 _ => display_name(harness, summary.cwd.as_deref().or(cwd.as_deref())),
181 };
182 let id = match summary.session_id.as_deref() {
183 Some(sid) => format!("pid:{}:{}", root.pid, sid),
184 None => format!("pid:{}:{}", root.pid, path.file_stem().map(|s| s.to_string_lossy().into_owned()).unwrap_or_default()),
185 };
186
187 agents.push(Agent {
188 id,
189 name,
190 harness,
191 state,
192 activity: summary.activity,
193 pid: Some(root.pid),
194 session_id: summary.session_id.clone(),
195 session_path: Some(path.clone()),
196 cwd: summary.cwd.clone().or_else(|| cwd.clone()),
197 model: summary.model.clone(),
198 harness_version: summary.harness_version.clone(),
199 usage: summary.usage,
200 cost_usd: summary.cost_usd,
201 cost_breakdown: summary.cost_breakdown,
202 price_source: summary.model.as_deref().and_then(|m| crate::pricing::table().source_for(m)),
203 unpriced_tokens: summary.unpriced_tokens,
204 turns: summary.turns,
205 subagent_turns: summary.subagent_turns,
206 tool_calls: summary.tool_calls,
207 web_searches: summary.web_searches,
208 spans: summary.spans.to_vec(),
209 age_secs: root.age_secs,
210 idle_secs,
211 cpu_percent: if owns_process { cpu } else { 0.0 },
212 rss_bytes: if owns_process { rss } else { 0 },
213 process_count: if owns_process { count } else { 0 },
214 mcp_count: if owns_process { mcp } else { 0 },
215 tree: if owns_process { Some(root.clone()) } else { None },
216 attribution,
217 shares_process: !owns_process,
218 parse_warning: parse_warning(&summary, harness),
219 });
220 }
221 }
222
223 let stopped: Vec<(PathBuf, Harness)> =
225 self.adapters.iter().flat_map(|a| a.unowned(&attached).into_iter().map(move |p| (p, a.harness()))).collect();
226 for (p, harness) in stopped {
227 let Some(tr) = self.tracker_for(&p, harness) else { continue };
228 let _ = tr.refresh();
229 let s = tr.summary().clone();
230 if s.turns == 0 && s.usage.total() == 0 {
231 continue;
232 }
233 let idle_secs = s.last_activity.and_then(|t| now.duration_since(t).ok()).map(|d| d.as_secs());
234 let id = s.session_id.clone().unwrap_or_else(|| p.file_stem().map(|x| x.to_string_lossy().into_owned()).unwrap_or_default());
235 agents.push(Agent {
236 id: format!("session:{id}"),
237 name: display_name(harness, s.cwd.as_deref()),
238 harness,
239 state: AgentState::Stopped,
240 activity: s.activity,
241 pid: None,
242 session_id: Some(id),
243 session_path: Some(p),
244 cwd: s.cwd.clone(),
245 model: s.model.clone(),
246 harness_version: s.harness_version.clone(),
247 usage: s.usage,
248 cost_usd: s.cost_usd,
249 cost_breakdown: s.cost_breakdown,
250 price_source: s.model.as_deref().and_then(|m| crate::pricing::table().source_for(m)),
251 unpriced_tokens: s.unpriced_tokens,
252 turns: s.turns,
253 subagent_turns: s.subagent_turns,
254 tool_calls: s.tool_calls,
255 web_searches: s.web_searches,
256 spans: s.spans.to_vec(),
257 age_secs: idle_secs.unwrap_or(0),
258 idle_secs,
259 cpu_percent: 0.0,
260 rss_bytes: 0,
261 process_count: 0,
262 mcp_count: 0,
263 tree: None,
264 attribution: Attribution::TranscriptOnly,
265 shares_process: false,
266 parse_warning: parse_warning(&s, harness),
267 });
268 }
269
270 let keep: HashSet<&PathBuf> = agents.iter().filter_map(|a| a.session_path.as_ref()).collect();
272 self.trackers.retain(|p, _| keep.contains(p));
273
274 let mut snap =
275 Snapshot { schema_version: SNAPSHOT_SCHEMA_VERSION, taken_at: now, host, agents, orphans, totals: Totals::default() };
276 snap.compute_totals();
277 snap
278 }
279}
280
281fn parse_warning(s: &SessionSummary, harness: Harness) -> Option<String> {
285 if !s.health.fields_unrecognised() {
286 return None;
287 }
288 let version = s.harness_version.as_deref().unwrap_or("unknown version");
289 Some(format!(
290 "usage fields not recognised in {} {}: tokens and cost are unreliable, agent-top may need updating",
291 harness.label(),
292 version
293 ))
294}
295
296fn live_state(hints: &RegistryHints, activity: Activity, idle_secs: Option<u64>, cpu: f32, opts: &CollectorOptions) -> AgentState {
297 match hints.status.as_deref() {
300 Some("busy" | "running" | "working" | "shell" | "tool" | "thinking") => return AgentState::Running,
301 Some("idle" | "waiting" | "paused" | "permission" | "blocked") => return AgentState::Idle,
302 _ => {}
303 }
304 match activity {
305 Activity::Working => {
306 if idle_secs.map(|s| s > opts.activity_timeout.as_secs()).unwrap_or(false) {
307 AgentState::Idle
308 } else {
309 AgentState::Running
310 }
311 }
312 Activity::Waiting => AgentState::Idle,
313 Activity::Unknown => {
314 if cpu > 5.0 || idle_secs.map(|s| s < 10).unwrap_or(false) {
315 AgentState::Running
316 } else {
317 AgentState::Idle
318 }
319 }
320 }
321}
322
323fn display_name(harness: Harness, cwd: Option<&Path>) -> String {
324 match cwd.and_then(|c| c.file_name()).map(|f| f.to_string_lossy().into_owned()) {
325 Some(dir) => format!("{}:{}", harness.label(), dir),
326 None => harness.label().to_string(),
327 }
328}
329
330#[cfg(test)]
331mod tests {
332 use super::*;
333
334 #[test]
335 fn the_registry_status_beats_the_transcript_heuristic() {
336 let opts = CollectorOptions::default();
337 let busy = RegistryHints { status: Some("busy".into()), ..Default::default() };
338 assert_eq!(live_state(&busy, Activity::Waiting, Some(0), 0.0, &opts), AgentState::Running);
339 let idle = RegistryHints { status: Some("idle".into()), ..Default::default() };
340 assert_eq!(live_state(&idle, Activity::Working, Some(0), 90.0, &opts), AgentState::Idle);
341 let none = RegistryHints::default();
343 assert_eq!(live_state(&none, Activity::Working, Some(1), 0.0, &opts), AgentState::Running);
344 assert_eq!(live_state(&none, Activity::Working, Some(opts.activity_timeout.as_secs() + 1), 0.0, &opts), AgentState::Idle);
345 assert_eq!(live_state(&none, Activity::Waiting, Some(1), 90.0, &opts), AgentState::Idle);
346 assert_eq!(live_state(&none, Activity::Unknown, Some(3), 0.0, &opts), AgentState::Running);
347 assert_eq!(live_state(&none, Activity::Unknown, Some(300), 0.0, &opts), AgentState::Idle);
348 }
349
350 #[test]
351 fn every_adapter_is_a_distinct_harness_and_recognises_its_own_fixture() {
352 let adapters = harness::adapters();
353 let mut seen = HashSet::new();
354 for a in &adapters {
355 assert!(seen.insert(a.harness()), "two adapters for {:?}", a.harness());
356 }
357 let fixtures = Path::new(env!("CARGO_MANIFEST_DIR")).join("tests/fixtures");
358 for (file, want) in
359 [("claude-2.1.226.jsonl", Harness::Claude), ("codex-0.130.jsonl", Harness::Codex), ("gemini-0.58.jsonl", Harness::Gemini)]
360 {
361 assert_eq!(harness::detect(&fixtures.join(file)), Some(want), "{file}");
362 }
363 }
364}