Rename "dirty days" to "updated days" across the codebase

+4 -4

apps/agents/routes.py

··· 19 19 from think.facets import get_facets 20 20 from think.models import calc_agent_cost 21 21 from think.muse import get_muse_configs, get_output_path 22 - from think.utils import dirty_days 22 + from think.utils import updated_days 23 23 24 24 agents_bp = Blueprint( 25 25 "app:agents", ··· 632 632 return jsonify({"count": failed_count}) 633 633 634 634 635 - @agents_bp.route("/api/dirty-days") 636 - def api_dirty_days() -> Any: 635 + @agents_bp.route("/api/updated-days") 636 + def api_updated_days() -> Any: 637 637 """Return journal days with pending reprocessing.""" 638 638 today = date.today().strftime("%Y%m%d") 639 639 try: 640 - return jsonify(dirty_days(exclude={today})) 640 + return jsonify(updated_days(exclude={today})) 641 641 except Exception: 642 642 return jsonify([])

+11 -11

apps/agents/workspace.html

··· 851 851 margin-bottom: 1rem; 852 852 } 853 853 854 - .dirty-banner { 854 + .updated-banner { 855 855 background: #fff8e1; 856 856 border: 1px solid #ffe082; 857 857 border-radius: 6px; ··· 861 861 color: #5d4037; 862 862 } 863 863 864 - .dirty-banner-title { 864 + .updated-banner-title { 865 865 font-weight: 600; 866 866 margin-bottom: 0.25rem; 867 867 } 868 868 869 - .dirty-banner a { 869 + .updated-banner a { 870 870 color: #e65100; 871 871 text-decoration: none; 872 872 font-weight: 500; 873 873 } 874 874 875 - .dirty-banner a:hover { 875 + .updated-banner a:hover { 876 876 text-decoration: underline; 877 877 } 878 878 ··· 890 890 891 891  892 892 <div id="grid-view" style="display: none;"> 893 - <div id="dirty-banner" class="dirty-banner" style="display: none;"></div> 893 + <div id="updated-banner" class="updated-banner" style="display: none;"></div> 894 894 <div id="day-summary" class="day-summary" style="display: none;"></div> 895 895 <div id="agent-groups"></div> 896 896 <div id="empty-state" class="empty-state" style="display: none;"> ··· 1117 1117 document.getElementById('run-detail-view').style.display = 'none'; 1118 1118 } 1119 1119 1120 - function loadDirtyBanner() { 1121 - const banner = document.getElementById('dirty-banner'); 1120 + function loadUpdatedBanner() { 1121 + const banner = document.getElementById('updated-banner'); 1122 1122 const now = new Date(); 1123 1123 const today = String(now.getFullYear()) + String(now.getMonth() + 1).padStart(2, '0') + String(now.getDate()).padStart(2, '0'); 1124 1124 if (currentDay !== today || window.selectedFacet) { 1125 1125 banner.style.display = 'none'; 1126 1126 return; 1127 1127 } 1128 - fetch('api/dirty-days') 1128 + fetch('api/updated-days') 1129 1129 .then(r => r.json()) 1130 1130 .then(days => { 1131 1131 if (!days.length) { ··· 1136 1136 const label = d.slice(0, 4) + '-' + d.slice(4, 6) + '-' + d.slice(6); 1137 1137 return `<a href="${d}">${label}</a>`; 1138 1138 }).join(', '); 1139 - banner.innerHTML = `<div class="dirty-banner-title">Days with pending reprocessing</div>${links}`; 1139 + banner.innerHTML = `<div class="updated-banner-title">Days with pending reprocessing</div>${links}`; 1140 1140 banner.style.display = 'block'; 1141 1141 }) 1142 1142 .catch(() => { banner.style.display = 'none'; }); ··· 2134 2134 2135 2135 // Listen for facet changes 2136 2136 window.addEventListener('facet.switch', () => { 2137 - loadDirtyBanner(); 2137 + loadUpdatedBanner(); 2138 2138 loadAgents(); 2139 2139 }); 2140 2140 2141 2141 // Initial load 2142 - loadDirtyBanner(); 2142 + loadUpdatedBanner(); 2143 2143 loadAgents(); 2144 2144 })(); 2145 2145 </script>

+13 -13

tests/test_dirty_days.py tests/test_updated_days.py

··· 1 1 # SPDX-License-Identifier: AGPL-3.0-only 2 2 # Copyright (c) 2026 sol pbc 3 3 4 - """Tests for dirty_days() utility.""" 4 + """Tests for updated_days() utility.""" 5 5 6 6 import time 7 7 8 8 import think.utils 9 - from think.utils import dirty_days 9 + from think.utils import updated_days 10 10 11 11 12 - def test_dirty_days_fixture(monkeypatch): 13 - """20250101 has stream.updated but no daily.updated — should be dirty.""" 12 + def test_updated_days_fixture(monkeypatch): 13 + """20250101 has stream.updated but no daily.updated — should be updated.""" 14 14 monkeypatch.setenv("JOURNAL_PATH", "tests/fixtures/journal") 15 15 monkeypatch.setattr(think.utils, "_journal_path_cache", None) 16 - days = dirty_days() 16 + days = updated_days() 17 17 assert "20250101" in days 18 18 19 19 20 - def test_dirty_days_exclude(monkeypatch): 20 + def test_updated_days_exclude(monkeypatch): 21 21 """Excluded days should not appear in results.""" 22 22 monkeypatch.setenv("JOURNAL_PATH", "tests/fixtures/journal") 23 23 monkeypatch.setattr(think.utils, "_journal_path_cache", None) 24 - days = dirty_days(exclude={"20250101"}) 24 + days = updated_days(exclude={"20250101"}) 25 25 assert "20250101" not in days 26 26 27 27 28 - def test_dirty_days_clean(tmp_path, monkeypatch): 29 - """Day with daily.updated newer than stream.updated is not dirty.""" 28 + def test_updated_days_clean(tmp_path, monkeypatch): 29 + """Day with daily.updated newer than stream.updated is not updated.""" 30 30 monkeypatch.setenv("JOURNAL_PATH", str(tmp_path)) 31 31 day_dir = tmp_path / "20260101" / "health" 32 32 day_dir.mkdir(parents=True) 33 33 (day_dir / "stream.updated").touch() 34 34 time.sleep(0.05) 35 35 (day_dir / "daily.updated").touch() 36 - assert dirty_days() == [] 36 + assert updated_days() == [] 37 37 38 38 39 - def test_dirty_days_no_stream(tmp_path, monkeypatch): 40 - """Day without stream.updated is not dirty (no stream data).""" 39 + def test_updated_days_no_stream(tmp_path, monkeypatch): 40 + """Day without stream.updated is not updated (no stream data).""" 41 41 monkeypatch.setenv("JOURNAL_PATH", str(tmp_path)) 42 42 (tmp_path / "20260101").mkdir() 43 - assert dirty_days() == [] 43 + assert updated_days() == []

+16 -17

tests/test_supervisor_schedule.py

··· 25 25 26 26 with ( 27 27 patch("think.supervisor.datetime") as mock_datetime, 28 - patch("think.supervisor.dirty_days", return_value=["20250101"]), 28 + patch("think.supervisor.updated_days", return_value=["20250101"]), 29 29 ): 30 30 mock_datetime.now.return_value.date.return_value = date(2025, 1, 2) 31 31 handle_daily_tasks() ··· 78 78 79 79 with ( 80 80 patch("think.supervisor.datetime") as mock_datetime, 81 - patch("think.supervisor.dirty_days", return_value=["20250101"]), 81 + patch("think.supervisor.updated_days", return_value=["20250101"]), 82 82 ): 83 83 mock_datetime.now.return_value.date.return_value = date(2025, 1, 2) 84 84 handle_daily_tasks() ··· 114 114 assert _daily_state["last_day"] == date(2025, 1, 1) 115 115 116 116 117 - def test_handle_daily_tasks_multiple_dirty_days_chronological(mock_callosum): 118 - """Dirty days are submitted oldest-first so yesterday is processed last.""" 117 + def test_handle_daily_tasks_multiple_updated_days_chronological(mock_callosum): 118 + """Updated days are submitted oldest-first so yesterday is processed last.""" 119 119 import think.supervisor as mod 120 120 from think.supervisor import _daily_state, handle_daily_tasks 121 121 ··· 133 133 with ( 134 134 patch("think.supervisor.datetime") as mock_datetime, 135 135 patch( 136 - "think.supervisor.dirty_days", 136 + "think.supervisor.updated_days", 137 137 return_value=["20250103", "20250104", "20250105"], 138 138 ), 139 139 ): ··· 145 145 assert days == ["20250103", "20250104", "20250105"] 146 146 147 147 148 - def test_handle_daily_tasks_caps_at_max_dirty_catchup(mock_callosum): 149 - """Only the newest MAX_DIRTY_CATCHUP days are processed.""" 148 + def test_handle_daily_tasks_caps_at_max_updated_catchup(mock_callosum): 149 + """Only the newest MAX_UPDATED_CATCHUP days are processed.""" 150 150 import think.supervisor as mod 151 151 from think.supervisor import _daily_state, handle_daily_tasks 152 152 ··· 161 161 162 162 mod._task_queue.submit = capture_submit 163 163 164 - all_dirty = [ 164 + all_updated = [ 165 165 "20250104", 166 166 "20250105", 167 167 "20250106", ··· 173 173 174 174 with ( 175 175 patch("think.supervisor.datetime") as mock_datetime, 176 - patch("think.supervisor.dirty_days", return_value=all_dirty), 176 + patch("think.supervisor.updated_days", return_value=all_updated), 177 177 ): 178 178 mock_datetime.now.return_value.date.return_value = date(2025, 1, 11) 179 179 handle_daily_tasks() ··· 184 184 assert days == ["20250107", "20250108", "20250109", "20250110"] 185 185 186 186 187 - def test_handle_daily_tasks_no_dirty_days(mock_callosum): 188 - """No submissions when there are no dirty days.""" 187 + def test_handle_daily_tasks_no_updated_days(mock_callosum): 188 + """No submissions when there are no updated days.""" 189 189 import think.supervisor as mod 190 190 from think.supervisor import _daily_state, handle_daily_tasks 191 191 ··· 202 202 203 203 with ( 204 204 patch("think.supervisor.datetime") as mock_datetime, 205 - patch("think.supervisor.dirty_days", return_value=[]), 205 + patch("think.supervisor.updated_days", return_value=[]), 206 206 ): 207 207 mock_datetime.now.return_value.date.return_value = date(2025, 1, 2) 208 208 handle_daily_tasks() 209 209 210 210 assert len(submitted) == 0 211 - # State still advances even with no dirty days 211 + # State still advances even with no updated days 212 212 assert _daily_state["last_day"] == date(2025, 1, 2) 213 213 214 214 215 215 def test_handle_daily_tasks_excludes_today(mock_callosum): 216 - """Today is excluded from dirty_days query.""" 217 - import think.supervisor as mod 216 + """Today is excluded from updated_days query.""" 218 217 from think.supervisor import _daily_state, handle_daily_tasks 219 218 220 219 _daily_state["last_day"] = date(2025, 1, 1) 221 220 222 221 captured_exclude = {} 223 222 224 - def fake_dirty_days(exclude=None): 223 + def fake_updated_days(exclude=None): 225 224 captured_exclude["value"] = exclude 226 225 return ["20250101"] 227 226 228 227 with ( 229 228 patch("think.supervisor.datetime") as mock_datetime, 230 - patch("think.supervisor.dirty_days", side_effect=fake_dirty_days), 229 + patch("think.supervisor.updated_days", side_effect=fake_updated_days), 231 230 ): 232 231 mock_datetime.now.return_value.date.return_value = date(2025, 1, 2) 233 232 handle_daily_tasks()

+1 -1

think/dream.py

··· 1441 1441 if args.activity and not args.facet: 1442 1442 parser.error("--activity requires --facet") 1443 1443 1444 - # Auto-enable refresh for dirty days (full daily runs only) 1444 + # Auto-enable refresh for updated days (full daily runs only) 1445 1445 if not args.refresh and not args.segment and not args.segments: 1446 1446 health_dir = day_dir / "health" 1447 1447 stream_marker = health_dir / "stream.updated"

+16 -16

think/supervisor.py

··· 23 23 from think.runner import DailyLogWriter 24 24 from think.runner import ManagedProcess as RunnerManagedProcess 25 25 from think.utils import ( 26 - dirty_days, 27 26 find_available_port, 28 27 get_journal, 29 28 get_journal_info, 30 29 get_rev, 31 30 now_ms, 32 31 setup_cli, 32 + updated_days, 33 33 ) 34 34 35 35 DEFAULT_THRESHOLD = 60 36 36 CHECK_INTERVAL = 30 37 - MAX_DIRTY_CATCHUP = 4 37 + MAX_UPDATED_CATCHUP = 4 38 38 39 39 # Global shutdown flag 40 40 shutdown_requested = False ··· 1019 1019 1020 1020 1021 1021 def handle_daily_tasks() -> None: 1022 - """Check for day change and submit daily dream for dirty days (non-blocking). 1022 + """Check for day change and submit daily dream for updated days (non-blocking). 1023 1023 1024 - Triggers once when the day rolls over at midnight. Queries ``dirty_days()`` 1024 + Triggers once when the day rolls over at midnight. Queries ``updated_days()`` 1025 1025 for journal days that have new stream data but haven't completed a daily 1026 - dream yet, then submits up to ``MAX_DIRTY_CATCHUP`` dreams in chronological 1026 + dream yet, then submits up to ``MAX_UPDATED_CATCHUP`` dreams in chronological 1027 1027 order (oldest first, yesterday last) via the TaskQueue. 1028 1028 1029 - Dream auto-detects dirty state and enables ``--refresh`` internally, so we 1029 + Dream auto-detects updated state and enables ``--refresh`` internally, so we 1030 1030 don't pass it here. 1031 1031 1032 1032 Skipped in remote mode (no local data to process). ··· 1058 1058 _check_segment_flush(force=True) 1059 1059 1060 1060 today_str = today.strftime("%Y%m%d") 1061 - all_dirty = dirty_days(exclude={today_str}) 1061 + all_updated = updated_days(exclude={today_str}) 1062 1062 1063 - if not all_dirty: 1064 - logging.info("Day changed to %s, no dirty days to process", today) 1063 + if not all_updated: 1064 + logging.info("Day changed to %s, no updated days to process", today) 1065 1065 return 1066 1066 1067 - # Take the newest MAX_DIRTY_CATCHUP days (already sorted ascending) 1068 - days_to_process = all_dirty[-MAX_DIRTY_CATCHUP:] 1069 - skipped = len(all_dirty) - len(days_to_process) 1067 + # Take the newest MAX_UPDATED_CATCHUP days (already sorted ascending) 1068 + days_to_process = all_updated[-MAX_UPDATED_CATCHUP:] 1069 + skipped = len(all_updated) - len(days_to_process) 1070 1070 1071 1071 if skipped: 1072 1072 logging.warning( 1073 - "Skipping %d older dirty days (max catchup %d): %s", 1073 + "Skipping %d older updated days (max catchup %d): %s", 1074 1074 skipped, 1075 - MAX_DIRTY_CATCHUP, 1076 - all_dirty[:skipped], 1075 + MAX_UPDATED_CATCHUP, 1076 + all_updated[:skipped], 1077 1077 ) 1078 1078 1079 1079 logging.info( 1080 - "Day changed to %s, queuing daily dream for %d dirty day(s): %s", 1080 + "Day changed to %s, queuing daily dream for %d updated day(s): %s", 1081 1081 today, 1082 1082 len(days_to_process), 1083 1083 days_to_process,

+8 -8

think/utils.py

··· 212 212 return days 213 213 214 214 215 - def dirty_days(exclude: set[str] | None = None) -> list[str]: 215 + def updated_days(exclude: set[str] | None = None) -> list[str]: 216 216 """Return journal days with pending stream data not yet processed daily. 217 217 218 - A day is "dirty" when it has a ``health/stream.updated`` marker that is 218 + A day is "updated" when it has a ``health/stream.updated`` marker that is 219 219 newer than its ``health/daily.updated`` marker (or daily.updated is missing). 220 220 Days without ``stream.updated`` are skipped entirely. 221 221 ··· 227 227 Returns 228 228 ------- 229 229 list of str 230 - Sorted list of dirty day strings. 230 + Sorted list of updated day strings. 231 231 """ 232 232 days = day_dirs() 233 - dirty: list[str] = [] 233 + updated: list[str] = [] 234 234 for name, path in days.items(): 235 235 if exclude and name in exclude: 236 236 continue ··· 239 239 continue 240 240 daily = os.path.join(path, "health", "daily.updated") 241 241 if not os.path.isfile(daily): 242 - dirty.append(name) 242 + updated.append(name) 243 243 continue 244 244 if os.path.getmtime(stream) > os.path.getmtime(daily): 245 - dirty.append(name) 246 - dirty.sort() 247 - return dirty 245 + updated.append(name) 246 + updated.sort() 247 + return updated 248 248 249 249 250 250 def segment_path(day: str, segment: str, stream: str) -> Path:

Configure Feed

Configure Feed