clockwork-cli 0.1.2__tar.gz → 0.2.1__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: clockwork-cli
3
- Version: 0.1.2
3
+ Version: 0.2.1
4
4
  Summary: Mine your Claude Code and Codex session logs to estimate active time
5
5
  License-Expression: MIT
6
6
  Project-URL: Homepage, https://github.com/danielsich/clockwork
@@ -217,6 +217,9 @@ one pre-aggregated view), the export is designed to be re-analyzed downstream.
217
217
  # Full-fidelity export you can re-analyze anywhere
218
218
  clockwork claude export > clockwork.json
219
219
 
220
+ # Both tools in one bundle, attributed per provider (see "Provider attribution")
221
+ clockwork both export > clockwork.json
222
+
220
223
  # Smaller / shareable: grouped sessions only, with paths stripped
221
224
  clockwork codex export --detail sessions --anonymize > share.json
222
225
 
@@ -238,13 +241,31 @@ drops the `path`, renames projects to `project-N`, and keeps only a stable
238
241
  hash `id` — so an uploaded file leaks nothing identifying while still letting a
239
242
  tool tell projects apart across exports.
240
243
 
241
- ### Export schema (`clockwork/v1`)
244
+ ### Provider attribution
245
+
246
+ `clockwork both export` carries **both tools in one bundle** without merging
247
+ them: attribution lives at the project level. Every project entry has a
248
+ `provider`, and if the same path was touched by both Claude and Codex it emits
249
+ **one entry per (path, provider)** — each with its own `totals`, `daily`,
250
+ `sessions`, and `prompts`. A viewer can filter by provider or re-merge by path
251
+ for a combined view, so nothing is lost either way.
252
+
253
+ - Top-level `provider` is `"both"` when more than one provider contributed,
254
+ otherwise the single provider name (so a `both` export with only Claude
255
+ activity still reads `"claude"`).
256
+ - `providers` lists the sorted, distinct providers included (always present,
257
+ one element for a single-provider export).
258
+ - `totals.by_provider` breaks the grand totals down per provider; the grand
259
+ `totals` stay the cross-provider sum.
260
+
261
+ ### Export schema (`clockwork/v2`)
242
262
 
243
263
  ```jsonc
244
264
  {
245
- "schema": "clockwork/v1",
265
+ "schema": "clockwork/v2",
246
266
  "generated_at": "2026-07-06T00:30:00+02:00", // local ISO-8601
247
- "provider": "claude",
267
+ "provider": "both", // "claude" | "codex" | "both"
268
+ "providers": ["claude", "codex"], // sorted, distinct; ≥ 1 element
248
269
  "idle_threshold_min": 30,
249
270
  "detail": "raw", // raw | sessions | daily
250
271
  "anonymized": false,
@@ -253,7 +274,8 @@ tool tell projects apart across exports.
253
274
  "until": null,
254
275
  "projects": [
255
276
  {
256
- "id": "0ac6be84", // stable sha1(path) prefix; survives --anonymize
277
+ "id": "0ac6be84", // stable sha1(provider + "\0" + path) prefix; survives --anonymize
278
+ "provider": "claude", // the single tool this entry's activity belongs to
257
279
  "name": "myproject", // basename, or "project-N" when anonymized
258
280
  "path": "/Users/you/dev/myproject", // omitted when anonymized
259
281
  "totals": {
@@ -267,13 +289,20 @@ tool tell projects apart across exports.
267
289
  "prompts": [ 1780521570, 1780521582 ] // detail == raw
268
290
  }
269
291
  ],
270
- "totals": { "projects": 6, "minutes": 3392.97, "prompts": 4588, "sessions": 64 }
292
+ "totals": {
293
+ "projects": 6, "minutes": 3392.97, "prompts": 4588, "sessions": 64,
294
+ "by_provider": {
295
+ "claude": { "projects": 4, "minutes": 2500.0, "prompts": 3000, "sessions": 40 },
296
+ "codex": { "projects": 2, "minutes": 892.97, "prompts": 1588, "sessions": 24 }
297
+ }
298
+ }
271
299
  }
272
300
  ```
273
301
 
274
302
  All instants are **UTC-based epoch seconds** (`new Date(sec * 1000)` in JS), so
275
303
  the consumer picks the display timezone. The `schema` field is the version
276
- contract — bump it if the shape ever changes so tools can guard on it.
304
+ contract — `clockwork/v2` is a breaking bump over `v1` (provider moved to the
305
+ project level), so tools should guard on it.
277
306
 
278
307
  ## License
279
308
 
@@ -199,6 +199,9 @@ one pre-aggregated view), the export is designed to be re-analyzed downstream.
199
199
  # Full-fidelity export you can re-analyze anywhere
200
200
  clockwork claude export > clockwork.json
201
201
 
202
+ # Both tools in one bundle, attributed per provider (see "Provider attribution")
203
+ clockwork both export > clockwork.json
204
+
202
205
  # Smaller / shareable: grouped sessions only, with paths stripped
203
206
  clockwork codex export --detail sessions --anonymize > share.json
204
207
 
@@ -220,13 +223,31 @@ drops the `path`, renames projects to `project-N`, and keeps only a stable
220
223
  hash `id` — so an uploaded file leaks nothing identifying while still letting a
221
224
  tool tell projects apart across exports.
222
225
 
223
- ### Export schema (`clockwork/v1`)
226
+ ### Provider attribution
227
+
228
+ `clockwork both export` carries **both tools in one bundle** without merging
229
+ them: attribution lives at the project level. Every project entry has a
230
+ `provider`, and if the same path was touched by both Claude and Codex it emits
231
+ **one entry per (path, provider)** — each with its own `totals`, `daily`,
232
+ `sessions`, and `prompts`. A viewer can filter by provider or re-merge by path
233
+ for a combined view, so nothing is lost either way.
234
+
235
+ - Top-level `provider` is `"both"` when more than one provider contributed,
236
+ otherwise the single provider name (so a `both` export with only Claude
237
+ activity still reads `"claude"`).
238
+ - `providers` lists the sorted, distinct providers included (always present,
239
+ one element for a single-provider export).
240
+ - `totals.by_provider` breaks the grand totals down per provider; the grand
241
+ `totals` stay the cross-provider sum.
242
+
243
+ ### Export schema (`clockwork/v2`)
224
244
 
225
245
  ```jsonc
226
246
  {
227
- "schema": "clockwork/v1",
247
+ "schema": "clockwork/v2",
228
248
  "generated_at": "2026-07-06T00:30:00+02:00", // local ISO-8601
229
- "provider": "claude",
249
+ "provider": "both", // "claude" | "codex" | "both"
250
+ "providers": ["claude", "codex"], // sorted, distinct; ≥ 1 element
230
251
  "idle_threshold_min": 30,
231
252
  "detail": "raw", // raw | sessions | daily
232
253
  "anonymized": false,
@@ -235,7 +256,8 @@ tool tell projects apart across exports.
235
256
  "until": null,
236
257
  "projects": [
237
258
  {
238
- "id": "0ac6be84", // stable sha1(path) prefix; survives --anonymize
259
+ "id": "0ac6be84", // stable sha1(provider + "\0" + path) prefix; survives --anonymize
260
+ "provider": "claude", // the single tool this entry's activity belongs to
239
261
  "name": "myproject", // basename, or "project-N" when anonymized
240
262
  "path": "/Users/you/dev/myproject", // omitted when anonymized
241
263
  "totals": {
@@ -249,13 +271,20 @@ tool tell projects apart across exports.
249
271
  "prompts": [ 1780521570, 1780521582 ] // detail == raw
250
272
  }
251
273
  ],
252
- "totals": { "projects": 6, "minutes": 3392.97, "prompts": 4588, "sessions": 64 }
274
+ "totals": {
275
+ "projects": 6, "minutes": 3392.97, "prompts": 4588, "sessions": 64,
276
+ "by_provider": {
277
+ "claude": { "projects": 4, "minutes": 2500.0, "prompts": 3000, "sessions": 40 },
278
+ "codex": { "projects": 2, "minutes": 892.97, "prompts": 1588, "sessions": 24 }
279
+ }
280
+ }
253
281
  }
254
282
  ```
255
283
 
256
284
  All instants are **UTC-based epoch seconds** (`new Date(sec * 1000)` in JS), so
257
285
  the consumer picks the display timezone. The `schema` field is the version
258
- contract — bump it if the shape ever changes so tools can guard on it.
286
+ contract — `clockwork/v2` is a breaking bump over `v1` (provider moved to the
287
+ project level), so tools should guard on it.
259
288
 
260
289
  ## License
261
290
 
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
4
4
 
5
5
  [project]
6
6
  name = "clockwork-cli"
7
- version = "0.1.2"
7
+ version = "0.2.1"
8
8
  description = "Mine your Claude Code and Codex session logs to estimate active time"
9
9
  readme = "README.md"
10
10
  license = "MIT"
@@ -55,7 +55,7 @@ import hashlib
55
55
  from datetime import datetime, timedelta, timezone
56
56
  from collections import defaultdict
57
57
 
58
- __version__ = "0.1.2"
58
+ __version__ = "0.2.1"
59
59
 
60
60
 
61
61
  def _resolve_claude_dir():
@@ -228,9 +228,14 @@ def _stdout_target():
228
228
  return None
229
229
 
230
230
 
231
- def _project_id(path):
232
- """Stable short id for a project path (survives anonymization)."""
233
- return hashlib.sha1(path.encode("utf-8")).hexdigest()[:8]
231
+ def _project_id(provider, path):
232
+ """Stable short id for a (provider, path) pair (survives anonymization).
233
+
234
+ Hashing the provider alongside the path keeps ids distinct when the same
235
+ project is touched by more than one tool, so a combined export can carry
236
+ two entries for one path without id collisions.
237
+ """
238
+ return hashlib.sha1(f"{provider}\0{path}".encode("utf-8")).hexdigest()[:8]
234
239
 
235
240
 
236
241
  def _display_name(path):
@@ -265,7 +270,7 @@ def _claude_timestamps_from_files(files):
265
270
  """Parse user/human turn timestamps from Claude .jsonl files."""
266
271
  timestamps = []
267
272
  for f in sorted(files):
268
- with open(f) as fh:
273
+ with open(f, encoding="utf-8") as fh:
269
274
  for line in fh:
270
275
  line = line.strip()
271
276
  if not line:
@@ -334,7 +339,7 @@ def _codex_parse_file(path):
334
339
  canonical = []
335
340
  fallback = []
336
341
  try:
337
- with open(path) as fh:
342
+ with open(path, encoding="utf-8") as fh:
338
343
  for line in fh:
339
344
  line = line.strip()
340
345
  if not line:
@@ -392,7 +397,7 @@ def _codex_cwd_only(path):
392
397
  the `session_meta` record carrying `cwd` is at the top of the rollout.
393
398
  """
394
399
  try:
395
- with open(path) as fh:
400
+ with open(path, encoding="utf-8") as fh:
396
401
  for line in fh:
397
402
  line = line.strip()
398
403
  if not line:
@@ -480,6 +485,15 @@ def _provider_label(provider):
480
485
  return "CLAUDE + CODEX" if provider == "both" else provider.upper()
481
486
 
482
487
 
488
+ def _export_providers(provider):
489
+ """Concrete single-provider backends to walk for an export.
490
+
491
+ `both` fans out to each real provider so a project's activity can be
492
+ attributed per provider; a single provider exports just itself.
493
+ """
494
+ return ["claude", "codex"] if provider == "both" else [provider]
495
+
496
+
483
497
  # --------------------------------------------------------------------------- #
484
498
  # Output modes
485
499
  # --------------------------------------------------------------------------- #
@@ -790,67 +804,102 @@ def export_data(provider, idle_threshold, since, until, detail, anonymize):
790
804
  Higher levels are supersets: a `raw` export still carries the daily and
791
805
  session aggregates so a consumer can start simple and drill down.
792
806
  Timestamps are UTC-based epoch seconds; the `daily` buckets use UTC dates.
793
- """
794
- projects = PROVIDERS[provider]["all"]()
795
807
 
808
+ Provider attribution lives at the project level: every entry carries a
809
+ `provider`, and a `both` export emits one entry per (path, provider) pair
810
+ so a path touched by both tools appears twice — each with its own totals,
811
+ daily, sessions, and prompts. Top-level `provider` is "both" when more than
812
+ one provider contributed; `providers` lists them, and `totals.by_provider`
813
+ breaks the grand totals down per provider.
814
+ """
796
815
  out_projects = []
816
+ by_provider = {}
797
817
  grand_minutes = 0.0
798
818
  grand_prompts = 0
799
819
  grand_sessions = 0
800
-
801
- for i, (label, timestamps) in enumerate(
802
- sorted(projects, key=lambda p: p[0]), start=1):
803
- timestamps = _filter_timestamps(timestamps, since, until)
804
- if not timestamps:
805
- continue
806
- daily, _ = compute_active_hours(timestamps, idle_threshold)
807
- sessions = group_sessions(timestamps, idle_threshold)
808
- total_minutes = sum(daily.values())
809
-
810
- prompts_per_day = defaultdict(int)
811
- for ts in timestamps:
812
- prompts_per_day[ts.date()] += 1
813
-
814
- proj = {
815
- "id": _project_id(label),
816
- "name": f"project-{i}" if anonymize else _display_name(label),
817
- "totals": {
818
- "minutes": round(total_minutes, 2),
819
- "prompts": len(timestamps),
820
- "sessions": len(sessions),
821
- "active_days": len(daily),
822
- "first": int(timestamps[0].timestamp()),
823
- "last": int(timestamps[-1].timestamp()),
824
- },
825
- "daily": [
826
- {"date": str(day),
827
- "minutes": round(daily[day], 2),
828
- "prompts": prompts_per_day[day]}
829
- for day in sorted(daily)
830
- ],
831
- }
832
- if not anonymize:
833
- proj["path"] = label
834
- if detail in ("sessions", "raw"):
835
- proj["sessions"] = [
836
- {"start": int(s.timestamp()),
837
- "end": int(e.timestamp()),
838
- "minutes": round((e - s).total_seconds() / 60, 2),
839
- "prompts": c}
840
- for s, e, c in sessions
841
- ]
842
- if detail == "raw":
843
- proj["prompts"] = [int(ts.timestamp()) for ts in timestamps]
844
-
845
- out_projects.append(proj)
846
- grand_minutes += total_minutes
847
- grand_prompts += len(timestamps)
848
- grand_sessions += len(sessions)
820
+ name_counter = 0
821
+
822
+ for prov in _export_providers(provider):
823
+ projects = PROVIDERS[prov]["all"]()
824
+ p_minutes = 0.0
825
+ p_prompts = 0
826
+ p_sessions = 0
827
+ p_projects = 0
828
+
829
+ for label, timestamps in sorted(projects, key=lambda p: p[0]):
830
+ timestamps = _filter_timestamps(timestamps, since, until)
831
+ if not timestamps:
832
+ continue
833
+ daily, _ = compute_active_hours(timestamps, idle_threshold)
834
+ sessions = group_sessions(timestamps, idle_threshold)
835
+ total_minutes = sum(daily.values())
836
+
837
+ prompts_per_day = defaultdict(int)
838
+ for ts in timestamps:
839
+ prompts_per_day[ts.date()] += 1
840
+
841
+ name_counter += 1
842
+ proj = {
843
+ "id": _project_id(prov, label),
844
+ "provider": prov,
845
+ "name": f"project-{name_counter}" if anonymize else _display_name(label),
846
+ "totals": {
847
+ "minutes": round(total_minutes, 2),
848
+ "prompts": len(timestamps),
849
+ "sessions": len(sessions),
850
+ "active_days": len(daily),
851
+ "first": int(timestamps[0].timestamp()),
852
+ "last": int(timestamps[-1].timestamp()),
853
+ },
854
+ "daily": [
855
+ {"date": str(day),
856
+ "minutes": round(daily[day], 2),
857
+ "prompts": prompts_per_day[day]}
858
+ for day in sorted(daily)
859
+ ],
860
+ }
861
+ if not anonymize:
862
+ proj["path"] = label
863
+ if detail in ("sessions", "raw"):
864
+ proj["sessions"] = [
865
+ {"start": int(s.timestamp()),
866
+ "end": int(e.timestamp()),
867
+ "minutes": round((e - s).total_seconds() / 60, 2),
868
+ "prompts": c}
869
+ for s, e, c in sessions
870
+ ]
871
+ if detail == "raw":
872
+ proj["prompts"] = [int(ts.timestamp()) for ts in timestamps]
873
+
874
+ out_projects.append(proj)
875
+ p_projects += 1
876
+ p_minutes += total_minutes
877
+ p_prompts += len(timestamps)
878
+ p_sessions += len(sessions)
879
+
880
+ if p_projects:
881
+ by_provider[prov] = {
882
+ "projects": p_projects,
883
+ "minutes": round(p_minutes, 2),
884
+ "prompts": p_prompts,
885
+ "sessions": p_sessions,
886
+ }
887
+ grand_minutes += p_minutes
888
+ grand_prompts += p_prompts
889
+ grand_sessions += p_sessions
890
+
891
+ # Report the providers that actually contributed activity; fall back to the
892
+ # requested set so a single-provider export is never emitted with an empty
893
+ # `providers` list even when nothing matched the filters.
894
+ present = sorted(by_provider)
895
+ providers = present if present else sorted(_export_providers(provider))
896
+ provider_field = "both" if len(providers) > 1 else providers[0]
849
897
 
850
898
  _print_json({
851
- "schema": "clockwork/v1",
899
+ "schema": "clockwork/v2",
852
900
  "generated_at": datetime.now().astimezone().isoformat(),
853
- "provider": provider,
901
+ "provider": provider_field,
902
+ "providers": providers,
854
903
  "idle_threshold_min": idle_threshold,
855
904
  "detail": detail,
856
905
  "anonymized": anonymize,
@@ -863,13 +912,14 @@ def export_data(provider, idle_threshold, since, until, detail, anonymize):
863
912
  "minutes": round(grand_minutes, 2),
864
913
  "prompts": grand_prompts,
865
914
  "sessions": grand_sessions,
915
+ "by_provider": by_provider,
866
916
  },
867
917
  })
868
918
 
869
919
  # A friendly receipt on stderr — keeps stdout pure JSON for redirection.
870
920
  dest = _stdout_target()
871
921
  where = dest if dest else "stdout"
872
- tags = [provider, f"detail={detail}"]
922
+ tags = [provider_field, f"detail={detail}"]
873
923
  if anonymize:
874
924
  tags.append("anonymized")
875
925
  print(f"\n ✓ export complete → {where}", file=sys.stderr)
@@ -933,7 +983,25 @@ def _parse_args(argv):
933
983
  return positionals, opts
934
984
 
935
985
 
986
+ def _force_utf8_streams():
987
+ """Write UTF-8 on stdout/stderr regardless of the platform locale.
988
+
989
+ Windows falls back to the locale codepage (usually cp1252) whenever output
990
+ is redirected to a file or pipe, and the bar glyph and status marks used
991
+ below are not encodable there. Reconfiguring keeps output byte-identical
992
+ across platforms; the JSON export is pure ASCII either way.
993
+ """
994
+ for stream in (sys.stdout, sys.stderr):
995
+ try:
996
+ stream.reconfigure(encoding="utf-8")
997
+ except (AttributeError, ValueError):
998
+ # Detached or replaced streams (test capture, some launchers) have
999
+ # nothing to reconfigure; whatever encoding they carry stands.
1000
+ pass
1001
+
1002
+
936
1003
  def main():
1004
+ _force_utf8_streams()
937
1005
  argv = sys.argv[1:]
938
1006
  if "-h" in argv or "--help" in argv:
939
1007
  print(__doc__.strip())
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: clockwork-cli
3
- Version: 0.1.2
3
+ Version: 0.2.1
4
4
  Summary: Mine your Claude Code and Codex session logs to estimate active time
5
5
  License-Expression: MIT
6
6
  Project-URL: Homepage, https://github.com/danielsich/clockwork
@@ -217,6 +217,9 @@ one pre-aggregated view), the export is designed to be re-analyzed downstream.
217
217
  # Full-fidelity export you can re-analyze anywhere
218
218
  clockwork claude export > clockwork.json
219
219
 
220
+ # Both tools in one bundle, attributed per provider (see "Provider attribution")
221
+ clockwork both export > clockwork.json
222
+
220
223
  # Smaller / shareable: grouped sessions only, with paths stripped
221
224
  clockwork codex export --detail sessions --anonymize > share.json
222
225
 
@@ -238,13 +241,31 @@ drops the `path`, renames projects to `project-N`, and keeps only a stable
238
241
  hash `id` — so an uploaded file leaks nothing identifying while still letting a
239
242
  tool tell projects apart across exports.
240
243
 
241
- ### Export schema (`clockwork/v1`)
244
+ ### Provider attribution
245
+
246
+ `clockwork both export` carries **both tools in one bundle** without merging
247
+ them: attribution lives at the project level. Every project entry has a
248
+ `provider`, and if the same path was touched by both Claude and Codex it emits
249
+ **one entry per (path, provider)** — each with its own `totals`, `daily`,
250
+ `sessions`, and `prompts`. A viewer can filter by provider or re-merge by path
251
+ for a combined view, so nothing is lost either way.
252
+
253
+ - Top-level `provider` is `"both"` when more than one provider contributed,
254
+ otherwise the single provider name (so a `both` export with only Claude
255
+ activity still reads `"claude"`).
256
+ - `providers` lists the sorted, distinct providers included (always present,
257
+ one element for a single-provider export).
258
+ - `totals.by_provider` breaks the grand totals down per provider; the grand
259
+ `totals` stay the cross-provider sum.
260
+
261
+ ### Export schema (`clockwork/v2`)
242
262
 
243
263
  ```jsonc
244
264
  {
245
- "schema": "clockwork/v1",
265
+ "schema": "clockwork/v2",
246
266
  "generated_at": "2026-07-06T00:30:00+02:00", // local ISO-8601
247
- "provider": "claude",
267
+ "provider": "both", // "claude" | "codex" | "both"
268
+ "providers": ["claude", "codex"], // sorted, distinct; ≥ 1 element
248
269
  "idle_threshold_min": 30,
249
270
  "detail": "raw", // raw | sessions | daily
250
271
  "anonymized": false,
@@ -253,7 +274,8 @@ tool tell projects apart across exports.
253
274
  "until": null,
254
275
  "projects": [
255
276
  {
256
- "id": "0ac6be84", // stable sha1(path) prefix; survives --anonymize
277
+ "id": "0ac6be84", // stable sha1(provider + "\0" + path) prefix; survives --anonymize
278
+ "provider": "claude", // the single tool this entry's activity belongs to
257
279
  "name": "myproject", // basename, or "project-N" when anonymized
258
280
  "path": "/Users/you/dev/myproject", // omitted when anonymized
259
281
  "totals": {
@@ -267,13 +289,20 @@ tool tell projects apart across exports.
267
289
  "prompts": [ 1780521570, 1780521582 ] // detail == raw
268
290
  }
269
291
  ],
270
- "totals": { "projects": 6, "minutes": 3392.97, "prompts": 4588, "sessions": 64 }
292
+ "totals": {
293
+ "projects": 6, "minutes": 3392.97, "prompts": 4588, "sessions": 64,
294
+ "by_provider": {
295
+ "claude": { "projects": 4, "minutes": 2500.0, "prompts": 3000, "sessions": 40 },
296
+ "codex": { "projects": 2, "minutes": 892.97, "prompts": 1588, "sessions": 24 }
297
+ }
298
+ }
271
299
  }
272
300
  ```
273
301
 
274
302
  All instants are **UTC-based epoch seconds** (`new Date(sec * 1000)` in JS), so
275
303
  the consumer picks the display timezone. The `schema` field is the version
276
- contract — bump it if the shape ever changes so tools can guard on it.
304
+ contract — `clockwork/v2` is a breaking bump over `v1` (provider moved to the
305
+ project level), so tools should guard on it.
277
306
 
278
307
  ## License
279
308
 
@@ -7,4 +7,5 @@ src/clockwork_cli.egg-info/PKG-INFO
7
7
  src/clockwork_cli.egg-info/SOURCES.txt
8
8
  src/clockwork_cli.egg-info/dependency_links.txt
9
9
  src/clockwork_cli.egg-info/entry_points.txt
10
- src/clockwork_cli.egg-info/top_level.txt
10
+ src/clockwork_cli.egg-info/top_level.txt
11
+ tests/test_export.py
@@ -0,0 +1,142 @@
1
+ """Tests for the clockwork/v2 combined-provider export bundle.
2
+
3
+ These exercise ``export_data`` directly, stubbing each provider's ``all``
4
+ backend with fixed (path, timestamps) data so no real ~/.claude or ~/.codex
5
+ files are needed. Stdout is pure JSON (the receipt goes to stderr), so the
6
+ captured stdout parses back with ``json.loads``.
7
+ """
8
+
9
+ import json
10
+ from datetime import datetime, timezone
11
+
12
+ import pytest
13
+
14
+ from clockwork_cli import PROVIDERS, export_data, _project_id
15
+
16
+
17
+ def _ts(*parts):
18
+ """Build a UTC-aware datetime from (year, month, day, hour, minute)."""
19
+ return datetime(*parts, tzinfo=timezone.utc)
20
+
21
+
22
+ def run_export(monkeypatch, capsys, provider, claude=None, codex=None,
23
+ *, idle=30, detail="raw", anonymize=False,
24
+ since=None, until=None):
25
+ """Stub the provider backends, run an export, return the parsed JSON."""
26
+ monkeypatch.setitem(PROVIDERS["claude"], "all", lambda: list(claude or []))
27
+ monkeypatch.setitem(PROVIDERS["codex"], "all", lambda: list(codex or []))
28
+ export_data(provider, idle, since, until, detail, anonymize)
29
+ out = capsys.readouterr().out
30
+ return json.loads(out)
31
+
32
+
33
+ # A couple of prompts on two different days for a project.
34
+ CLAUDE_PROJ = ("/Users/you/dev/alpha", [_ts(2026, 7, 1, 10, 0),
35
+ _ts(2026, 7, 1, 10, 20),
36
+ _ts(2026, 7, 2, 9, 0)])
37
+ CODEX_PROJ = ("/Users/you/dev/beta", [_ts(2026, 7, 3, 14, 0),
38
+ _ts(2026, 7, 3, 14, 15)])
39
+
40
+
41
+ def test_single_provider_shape_is_backward_compatible(monkeypatch, capsys):
42
+ data = run_export(monkeypatch, capsys, "claude", claude=[CLAUDE_PROJ])
43
+
44
+ assert data["schema"] == "clockwork/v2"
45
+ assert data["provider"] == "claude"
46
+ assert data["providers"] == ["claude"]
47
+ assert len(data["projects"]) == 1
48
+ assert data["projects"][0]["provider"] == "claude"
49
+ assert data["projects"][0]["path"] == CLAUDE_PROJ[0]
50
+ assert set(data["totals"]["by_provider"]) == {"claude"}
51
+
52
+
53
+ def test_both_provider_export(monkeypatch, capsys):
54
+ data = run_export(monkeypatch, capsys, "both",
55
+ claude=[CLAUDE_PROJ], codex=[CODEX_PROJ])
56
+
57
+ assert data["provider"] == "both"
58
+ assert data["providers"] == ["claude", "codex"]
59
+
60
+ providers_seen = {p["provider"] for p in data["projects"]}
61
+ assert providers_seen == {"claude", "codex"}
62
+ # Every project entry is attributed to exactly one provider.
63
+ assert all(p["provider"] in ("claude", "codex") for p in data["projects"])
64
+ assert set(data["totals"]["by_provider"]) == {"claude", "codex"}
65
+
66
+
67
+ def test_same_path_two_providers_yields_two_entries(monkeypatch, capsys):
68
+ shared = "/Users/you/dev/shared"
69
+ claude = [(shared, [_ts(2026, 7, 1, 10, 0), _ts(2026, 7, 1, 10, 30)])]
70
+ codex = [(shared, [_ts(2026, 7, 2, 11, 0)])]
71
+
72
+ data = run_export(monkeypatch, capsys, "both", claude=claude, codex=codex)
73
+
74
+ entries = [p for p in data["projects"] if p["path"] == shared]
75
+ assert len(entries) == 2
76
+ assert {e["provider"] for e in entries} == {"claude", "codex"}
77
+ # Distinct, provider-scoped ids for the same path.
78
+ assert entries[0]["id"] != entries[1]["id"]
79
+ assert data["totals"]["projects"] == 2
80
+
81
+
82
+ def test_by_provider_totals_sum_to_grand(monkeypatch, capsys):
83
+ data = run_export(monkeypatch, capsys, "both",
84
+ claude=[CLAUDE_PROJ], codex=[CODEX_PROJ])
85
+
86
+ by = data["totals"]["by_provider"]
87
+ grand = data["totals"]
88
+
89
+ assert by["claude"]["projects"] + by["codex"]["projects"] == grand["projects"]
90
+ assert by["claude"]["prompts"] + by["codex"]["prompts"] == grand["prompts"]
91
+ assert by["claude"]["sessions"] + by["codex"]["sessions"] == grand["sessions"]
92
+ assert (round(by["claude"]["minutes"] + by["codex"]["minutes"], 2)
93
+ == grand["minutes"])
94
+
95
+ # Claude side: 3 prompts, two active days.
96
+ assert by["claude"]["prompts"] == 3
97
+ assert by["codex"]["prompts"] == 2
98
+
99
+
100
+ def test_id_is_provider_path_hash_and_stable_under_anonymize(monkeypatch, capsys):
101
+ plain = run_export(monkeypatch, capsys, "both",
102
+ claude=[CLAUDE_PROJ], codex=[CODEX_PROJ])
103
+ anon = run_export(monkeypatch, capsys, "both",
104
+ claude=[CLAUDE_PROJ], codex=[CODEX_PROJ], anonymize=True)
105
+
106
+ def ids_by_provider(bundle):
107
+ return {p["provider"]: p["id"] for p in bundle["projects"]}
108
+
109
+ assert ids_by_provider(plain) == ids_by_provider(anon)
110
+ # The id is the documented sha1(provider + "\0" + path) prefix.
111
+ assert ids_by_provider(plain)["claude"] == _project_id("claude", CLAUDE_PROJ[0])
112
+ assert ids_by_provider(plain)["codex"] == _project_id("codex", CODEX_PROJ[0])
113
+
114
+ # Anonymized output drops the path and renames projects.
115
+ assert all("path" not in p for p in anon["projects"])
116
+ assert all(p["name"].startswith("project-") for p in anon["projects"])
117
+
118
+
119
+ def test_provider_filtering(monkeypatch, capsys):
120
+ claude_only = run_export(monkeypatch, capsys, "claude",
121
+ claude=[CLAUDE_PROJ], codex=[CODEX_PROJ])
122
+ assert claude_only["providers"] == ["claude"]
123
+ assert {p["provider"] for p in claude_only["projects"]} == {"claude"}
124
+ assert "codex" not in claude_only["totals"]["by_provider"]
125
+
126
+ codex_only = run_export(monkeypatch, capsys, "codex",
127
+ claude=[CLAUDE_PROJ], codex=[CODEX_PROJ])
128
+ assert codex_only["providers"] == ["codex"]
129
+ assert {p["provider"] for p in codex_only["projects"]} == {"codex"}
130
+ assert "claude" not in codex_only["totals"]["by_provider"]
131
+
132
+
133
+ def test_both_with_only_one_provider_present_keeps_single_name(monkeypatch, capsys):
134
+ # Requested `both`, but only claude has activity.
135
+ data = run_export(monkeypatch, capsys, "both", claude=[CLAUDE_PROJ], codex=[])
136
+ assert data["provider"] == "claude"
137
+ assert data["providers"] == ["claude"]
138
+ assert set(data["totals"]["by_provider"]) == {"claude"}
139
+
140
+
141
+ if __name__ == "__main__":
142
+ raise SystemExit(pytest.main([__file__, "-v"]))
File without changes
File without changes