soothe-cli 1.2.6__tar.gz → 1.2.8__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.
Files changed (151) hide show
  1. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/PKG-INFO +1 -1
  2. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/tool_display.py +24 -0
  3. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/presentation/renderer_base.py +0 -4
  4. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/state/session_stats.py +28 -42
  5. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/token_events_debug.py +19 -2
  6. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/token_usage.py +56 -21
  7. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/turn/prepare.py +5 -0
  8. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_app.py +5 -5
  9. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_model.py +5 -2
  10. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_startup.py +0 -1
  11. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_ui.py +29 -41
  12. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/textual_adapter.py +69 -73
  13. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/context_data.py +13 -0
  14. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/cognition_step_activity.py +13 -8
  15. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/status.py +1 -17
  16. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/.gitignore +0 -0
  17. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/README.md +0 -0
  18. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/pyproject.toml +0 -0
  19. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/__init__.py +0 -0
  20. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/_cli_context.py +0 -0
  21. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/_env_vars.py +0 -0
  22. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/_version.py +0 -0
  23. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/card_wire.py +0 -0
  24. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/__init__.py +0 -0
  25. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/commands/__init__.py +0 -0
  26. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/commands/config_cmd.py +0 -0
  27. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/commands/cron_cmd.py +0 -0
  28. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/commands/loop_cmd.py +0 -0
  29. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/commands/run_cmd.py +0 -0
  30. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/commands/status_cmd.py +0 -0
  31. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/execution/__init__.py +0 -0
  32. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/execution/daemon.py +0 -0
  33. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/execution/daemon_errors.py +0 -0
  34. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/execution/headless.py +0 -0
  35. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/execution/headless_renderer.py +0 -0
  36. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/execution/launcher.py +0 -0
  37. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/main.py +0 -0
  38. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/cli/soothecli.py +0 -0
  39. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/commands/__init__.py +0 -0
  40. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/commands/binding.py +0 -0
  41. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/commands/command_registry.py +0 -0
  42. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/commands/command_router.py +0 -0
  43. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/commands/slash_commands.py +0 -0
  44. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/commands/subagent_routing.py +0 -0
  45. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/config/__init__.py +0 -0
  46. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/config/cli_config.py +0 -0
  47. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/config/loader.py +0 -0
  48. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/config/logging_setup.py +0 -0
  49. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/__init__.py +0 -0
  50. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/card.py +0 -0
  51. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/card_binder.py +0 -0
  52. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/markdown_theme.py +0 -0
  53. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/media_utils.py +0 -0
  54. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/mermaid_render.py +0 -0
  55. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/path_utils.py +0 -0
  56. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/preview_limits.py +0 -0
  57. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/shell_color.py +0 -0
  58. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/spinner_labels.py +0 -0
  59. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/text_extract.py +0 -0
  60. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/theme.py +0 -0
  61. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/tips.py +0 -0
  62. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/display/unicode_security.py +0 -0
  63. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/loops/__init__.py +0 -0
  64. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/loops/sessions.py +0 -0
  65. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/model_config.py +0 -0
  66. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/project_utils.py +0 -0
  67. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/__init__.py +0 -0
  68. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/headless/processor.py +0 -0
  69. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/headless/processor_state.py +0 -0
  70. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/parse/tool_call_resolution.py +0 -0
  71. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/policy/tui_trace_log.py +0 -0
  72. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/presentation/duration_format.py +0 -0
  73. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/presentation/engine.py +0 -0
  74. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/presentation/id_format.py +0 -0
  75. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/presentation/renderer_protocol.py +0 -0
  76. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/presentation/step_id_format.py +0 -0
  77. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/presentation/subagent_task_display.py +0 -0
  78. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/state/file_tracker.py +0 -0
  79. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/state/step_router.py +0 -0
  80. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/state/stream_accumulator.py +0 -0
  81. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/transport/session.py +0 -0
  82. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/wire/chunk_filter.py +0 -0
  83. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/wire/message_text.py +0 -0
  84. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/runtime/wire/messages.py +0 -0
  85. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/settings/__init__.py +0 -0
  86. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/settings/_console_impl.py +0 -0
  87. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/settings/bootstrap.py +0 -0
  88. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/settings/core.py +0 -0
  89. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/settings/glyphs.py +0 -0
  90. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/settings/provider.py +0 -0
  91. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/settings/shell_allow.py +0 -0
  92. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/settings/skills_dirs.py +0 -0
  93. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/settings/stream_config.py +0 -0
  94. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/__init__.py +0 -0
  95. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/__init__.py +0 -0
  96. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_clarification.py +0 -0
  97. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_entrypoints.py +0 -0
  98. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_execution.py +0 -0
  99. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_history.py +0 -0
  100. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_messages_mixin.py +0 -0
  101. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_model_params.py +0 -0
  102. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_terminal.py +0 -0
  103. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_theme_prefs.py +0 -0
  104. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/_types.py +0 -0
  105. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/app/app.tcss +0 -0
  106. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/composer_mode.py +0 -0
  107. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/file_change_notify.py +0 -0
  108. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/file_change_renderers.py +0 -0
  109. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/input.py +0 -0
  110. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/skills/__init__.py +0 -0
  111. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/skills/invocation.py +0 -0
  112. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/skills/load.py +0 -0
  113. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/__init__.py +0 -0
  114. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/_links.py +0 -0
  115. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/autocomplete.py +0 -0
  116. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/chat_input.py +0 -0
  117. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/clipboard.py +0 -0
  118. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/context_viewer.py +0 -0
  119. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/diff.py +0 -0
  120. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/editor.py +0 -0
  121. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/file_change_preview.py +0 -0
  122. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/help_screen.py +0 -0
  123. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/history.py +0 -0
  124. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/loading.py +0 -0
  125. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/loop_selector.py +0 -0
  126. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/mcp_viewer.py +0 -0
  127. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/message_store.py +0 -0
  128. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/__init__.py +0 -0
  129. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/_helpers.py +0 -0
  130. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/app.py +0 -0
  131. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/assistant.py +0 -0
  132. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/cognition_goal_tree.py +0 -0
  133. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/cognition_reason.py +0 -0
  134. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/cognition_step.py +0 -0
  135. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/diff_message.py +0 -0
  136. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/error.py +0 -0
  137. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/skill.py +0 -0
  138. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/structured_ask_user.py +0 -0
  139. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/messages/user.py +0 -0
  140. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/model_selector.py +0 -0
  141. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/notification_settings.py +0 -0
  142. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/pinned_goal_bar.py +0 -0
  143. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/plan_quick_view_overlay.py +0 -0
  144. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/queued_goals_bar.py +0 -0
  145. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/router_profile_selector.py +0 -0
  146. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/theme_selector.py +0 -0
  147. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/tui/widgets/welcome.py +0 -0
  148. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/update_check.py +0 -0
  149. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/ux/__init__.py +0 -0
  150. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/ux/subagent_progress.py +0 -0
  151. {soothe_cli-1.2.6 → soothe_cli-1.2.8}/src/soothe_cli/ux/subagent_wire_display.py +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.5
2
2
  Name: soothe-cli
3
- Version: 1.2.6
3
+ Version: 1.2.8
4
4
  Summary: Soothe CLI client - communicates with daemon via WebSocket
5
5
  Project-URL: Homepage, https://github.com/mirasoth/soothe
6
6
  Project-URL: Documentation, https://soothe.readthedocs.io
@@ -75,6 +75,30 @@ def truncate_to_width(text: str, max_cols: int, *, ellipsis: str = "…") -> str
75
75
  return "".join(out) + ellipsis
76
76
 
77
77
 
78
+ def truncate_to_width_tail(text: str, max_cols: int, *, ellipsis: str = "…") -> str:
79
+ """Truncate *text* to `max_cols` terminal columns, prepending an ellipsis.
80
+
81
+ Mirror of :func:`truncate_to_width` that keeps the *tail* of the string
82
+ (most recent content) instead of the head. Used for reasoning/thinking
83
+ lines where the latest progress is what the user wants to read.
84
+ """
85
+ if max_cols <= 0 or display_width(text) <= max_cols:
86
+ return text
87
+ ellipsis_width = display_width(ellipsis)
88
+ target = max_cols - ellipsis_width
89
+ if target <= 0:
90
+ return ellipsis[:max_cols] if ellipsis_width > max_cols else ellipsis
91
+ out: list[str] = []
92
+ width = 0
93
+ for ch in reversed(text):
94
+ w = _char_width(ch)
95
+ if width + w > target:
96
+ break
97
+ out.append(ch)
98
+ width += w
99
+ return ellipsis + "".join(reversed(out))
100
+
101
+
78
102
  def compact_arg_text(text: str) -> str:
79
103
  """Collapse whitespace/newlines so activity lines stay on one row."""
80
104
  return " ".join(text.split())
@@ -20,10 +20,6 @@ class RendererBase:
20
20
  def repair_concatenated_output(text: str) -> str:
21
21
  """Repair common markdown/text concatenation artifacts.
22
22
 
23
- Consolidates the repair logic previously duplicated in:
24
- - CLI: _repair_concatenated_final_output()
25
- - TUI: _repair_concatenated_output_text()
26
-
27
23
  Fixes issues that arise when streaming text chunks are concatenated:
28
24
  - Missing newlines before headings
29
25
  - Missing spaces after heading markers
@@ -7,6 +7,8 @@ import time
7
7
  from dataclasses import dataclass, field
8
8
  from typing import Any
9
9
 
10
+ from soothe_cli.runtime.token_usage import TokenUsage
11
+
10
12
  SpinnerStatus = str | None
11
13
  """Spinner line label, or `None` to hide.
12
14
 
@@ -17,68 +19,46 @@ Common single-word values include ``Thinking``, ``Planning``, ``Executing``,
17
19
 
18
20
  @dataclass
19
21
  class ModelStats:
20
- """Token stats for a single model within a session.
21
-
22
- Attributes:
23
- request_count: Number of LLM API requests made to this model.
24
- input_tokens: Cumulative input tokens sent to this model.
25
- output_tokens: Cumulative output tokens received from this model.
26
- """
22
+ """Token stats for a single model within a session."""
27
23
 
28
24
  request_count: int = 0
29
25
  input_tokens: int = 0
30
26
  output_tokens: int = 0
27
+ total_tokens: int = 0
28
+ cache_read_tokens: int = 0
29
+ cache_creation_tokens: int = 0
31
30
 
32
31
 
33
32
  @dataclass
34
33
  class SessionStats:
35
- """Stats accumulated over a single agent turn (or full session).
36
-
37
- Attributes:
38
- request_count: Total LLM API requests made (each chunk with
39
- usage_metadata counts as one completed request).
40
- input_tokens: Cumulative input tokens across all LLM requests.
41
- output_tokens: Cumulative output tokens across all LLM requests.
42
- wall_time_seconds: Wall-clock duration from stream start to end.
43
- per_model: Per-model breakdown keyed by model name.
44
- Populated only when `record_request` receives a non-empty
45
- `model_name`. Empty dict means no named-model requests were
46
- recorded; `print_usage_table` omits the model table in that case and
47
- shows only the wall-time line (if applicable).
48
- """
34
+ """Stats accumulated over a single agent turn (or full session)."""
49
35
 
50
36
  request_count: int = 0
51
37
  input_tokens: int = 0
52
38
  output_tokens: int = 0
39
+ total_tokens: int = 0
40
+ cache_read_tokens: int = 0
41
+ cache_creation_tokens: int = 0
53
42
  wall_time_seconds: float = 0.0
54
43
  per_model: dict[str, ModelStats] = field(default_factory=dict)
55
44
  event_stats: TurnEventStats | None = None
56
45
 
57
- def record_request(
58
- self,
59
- model_name: str,
60
- input_toks: int,
61
- output_toks: int,
62
- ) -> None:
63
- """Accumulate token counts for one completed LLM request.
64
-
65
- Updates both the session totals and the per-model breakdown.
66
-
67
- Args:
68
- model_name: The model that served this request (used as the
69
- per-model key). Pass an empty string to skip the per-model
70
- breakdown for this request.
71
- input_toks: Input tokens for this request.
72
- output_toks: Output tokens for this request.
73
- """
46
+ def record_request(self, model_name: str, usage: TokenUsage) -> None:
47
+ """Accumulate one LLM request's full token breakdown."""
74
48
  self.request_count += 1
75
- self.input_tokens += input_toks
76
- self.output_tokens += output_toks
49
+ self.input_tokens += usage.input_tokens
50
+ self.output_tokens += usage.output_tokens
51
+ self.total_tokens += usage.total_tokens
52
+ self.cache_read_tokens += usage.cache_read_tokens
53
+ self.cache_creation_tokens += usage.cache_creation_tokens
77
54
  if model_name:
78
55
  entry = self.per_model.setdefault(model_name, ModelStats())
79
56
  entry.request_count += 1
80
- entry.input_tokens += input_toks
81
- entry.output_tokens += output_toks
57
+ entry.input_tokens += usage.input_tokens
58
+ entry.output_tokens += usage.output_tokens
59
+ entry.total_tokens += usage.total_tokens
60
+ entry.cache_read_tokens += usage.cache_read_tokens
61
+ entry.cache_creation_tokens += usage.cache_creation_tokens
82
62
 
83
63
  def merge(self, other: SessionStats) -> None:
84
64
  """Merge another `SessionStats` into this one (mutates *self*).
@@ -91,12 +71,18 @@ class SessionStats:
91
71
  self.request_count += other.request_count
92
72
  self.input_tokens += other.input_tokens
93
73
  self.output_tokens += other.output_tokens
74
+ self.total_tokens += other.total_tokens
75
+ self.cache_read_tokens += other.cache_read_tokens
76
+ self.cache_creation_tokens += other.cache_creation_tokens
94
77
  self.wall_time_seconds += other.wall_time_seconds
95
78
  for model, ms in other.per_model.items():
96
79
  entry = self.per_model.setdefault(model, ModelStats())
97
80
  entry.request_count += ms.request_count
98
81
  entry.input_tokens += ms.input_tokens
99
82
  entry.output_tokens += ms.output_tokens
83
+ entry.total_tokens += ms.total_tokens
84
+ entry.cache_read_tokens += ms.cache_read_tokens
85
+ entry.cache_creation_tokens += ms.cache_creation_tokens
100
86
  if other.event_stats is not None:
101
87
  if self.event_stats is None:
102
88
  self.event_stats = TurnEventStats()
@@ -18,6 +18,8 @@ class TokenEventTrace:
18
18
  plan_phase_missing_total_field: int = 0
19
19
  step_completed_events: int = 0
20
20
  step_completed_with_total_field: int = 0
21
+ completed_events: int = 0
22
+ completed_with_total_field: int = 0
21
23
  authoritative_applied: int = 0
22
24
  display_refreshes: int = 0
23
25
  anomalies: list[str] = field(default_factory=list)
@@ -30,6 +32,8 @@ class TokenEventTrace:
30
32
  self.plan_phase_missing_total_field = 0
31
33
  self.step_completed_events = 0
32
34
  self.step_completed_with_total_field = 0
35
+ self.completed_events = 0
36
+ self.completed_with_total_field = 0
33
37
  self.authoritative_applied = 0
34
38
  self.display_refreshes = 0
35
39
  self.anomalies.clear()
@@ -78,6 +82,17 @@ class TokenEventTrace:
78
82
  has_total_field,
79
83
  )
80
84
 
85
+ def note_completed(self, *, total_tokens_used: int | None, has_total_field: bool) -> None:
86
+ """Record the turn-end authoritative total from `strange_loop.completed`."""
87
+ self.completed_events += 1
88
+ if has_total_field:
89
+ self.completed_with_total_field += 1
90
+ logger.debug(
91
+ "[token-events] completed total=%s has_total_field=%s",
92
+ total_tokens_used if has_total_field else "—",
93
+ has_total_field,
94
+ )
95
+
81
96
  def note_authoritative_merge(
82
97
  self,
83
98
  *,
@@ -123,7 +138,9 @@ class TokenEventTrace:
123
138
  """Emit one summary line per turn; flags missing events."""
124
139
  has_stream = self.stream_chunks > 0
125
140
  has_backend_total = (
126
- self.plan_phase_with_total_field > 0 or self.step_completed_with_total_field > 0
141
+ self.plan_phase_with_total_field > 0
142
+ or self.step_completed_with_total_field > 0
143
+ or self.completed_with_total_field > 0
127
144
  )
128
145
  has_any_usage = display_total > 0 or turn_input > 0 or turn_output > 0
129
146
 
@@ -163,4 +180,4 @@ class TokenEventTrace:
163
180
  )
164
181
 
165
182
 
166
- __all__ = ["TokenEventTrace", "logger"]
183
+ __all__ = ["TokenEventTrace"]
@@ -3,11 +3,13 @@
3
3
  from __future__ import annotations
4
4
 
5
5
  import logging
6
+ from dataclasses import dataclass
6
7
  from typing import Any
7
8
 
8
9
  logger = logging.getLogger(__name__)
9
10
 
10
11
  __all__ = [
12
+ "TokenUsage",
11
13
  "coerce_total_tokens_used",
12
14
  "extract_stream_message_token_usage",
13
15
  "fetch_conversation_token_count",
@@ -23,41 +25,75 @@ def coerce_total_tokens_used(value: Any) -> int:
23
25
  return 0
24
26
 
25
27
 
26
- def _coerce_int(value: Any) -> int:
27
- return coerce_total_tokens_used(value)
28
+ @dataclass(frozen=True, slots=True)
29
+ class TokenUsage:
30
+ """Full per-call token breakdown (cache stays a subset of input)."""
28
31
 
32
+ input_tokens: int = 0
33
+ output_tokens: int = 0
34
+ cache_read_tokens: int = 0
35
+ cache_creation_tokens: int = 0
36
+ total_tokens: int = 0
29
37
 
30
- def _usage_from_mapping(usage: dict[str, Any]) -> tuple[int, int, int]:
31
- """Return `(input_tokens, output_tokens, total_tokens)` from a usage dict."""
32
- input_toks = _coerce_int(usage.get("input_tokens") or usage.get("prompt_tokens"))
33
- output_toks = _coerce_int(usage.get("output_tokens") or usage.get("completion_tokens"))
34
- total_toks = _coerce_int(usage.get("total_tokens"))
35
- if not total_toks and (input_toks or output_toks):
36
- total_toks = input_toks + output_toks
37
- return input_toks, output_toks, total_toks
38
+ @classmethod
39
+ def from_mapping(cls, usage: dict[str, Any]) -> TokenUsage:
40
+ """Build a breakdown from a LangChain/OpenAI usage mapping."""
41
+ input_toks = coerce_total_tokens_used(
42
+ usage.get("input_tokens") or usage.get("prompt_tokens")
43
+ )
44
+ output_toks = coerce_total_tokens_used(
45
+ usage.get("output_tokens") or usage.get("completion_tokens")
46
+ )
47
+ total_toks = coerce_total_tokens_used(usage.get("total_tokens")) or input_toks + output_toks
48
+
49
+ cache_read = 0
50
+ cache_creation = 0
51
+ details = usage.get("input_token_details")
52
+ if isinstance(details, dict):
53
+ cache_read = coerce_total_tokens_used(details.get("cache_read"))
54
+ cache_creation = coerce_total_tokens_used(details.get("cache_creation"))
55
+ prompt_details = usage.get("prompt_tokens_details")
56
+ if isinstance(prompt_details, dict):
57
+ cache_read = cache_read or coerce_total_tokens_used(prompt_details.get("cached_tokens"))
58
+ cache_read = cache_read or coerce_total_tokens_used(usage.get("cache_read_input_tokens"))
59
+ cache_creation = cache_creation or coerce_total_tokens_used(
60
+ usage.get("cache_creation_input_tokens")
61
+ )
62
+
63
+ return cls(input_toks, output_toks, cache_read, cache_creation, total_toks)
64
+
65
+ def has_any(self) -> bool:
66
+ """True when any token count is present."""
67
+ return bool(
68
+ self.input_tokens
69
+ or self.output_tokens
70
+ or self.total_tokens
71
+ or self.cache_read_tokens
72
+ or self.cache_creation_tokens
73
+ )
38
74
 
39
75
 
40
- def extract_stream_message_token_usage(message: Any) -> tuple[int, int, int]:
41
- """Extract token usage from a streamed LangChain or wire-format message.
76
+ def extract_stream_message_token_usage(message: Any) -> TokenUsage:
77
+ """Extract a full token breakdown from a streamed LangChain or wire message.
42
78
 
43
79
  Providers vary:
44
- - LangChain `usage_metadata` (`input_tokens` / `output_tokens`)
80
+ - LangChain `usage_metadata` (`input_tokens` / `output_tokens` + cache details)
45
81
  - OpenAI-style `response_metadata.token_usage` (`prompt_tokens` / `completion_tokens`)
46
82
  - Flat wire dicts with either shape
47
83
  """
48
84
  if message is None:
49
- return 0, 0, 0
85
+ return TokenUsage()
50
86
 
51
87
  usage = getattr(message, "usage_metadata", None)
52
88
  if isinstance(usage, dict) and usage:
53
- return _usage_from_mapping(usage)
89
+ return TokenUsage.from_mapping(usage)
54
90
 
55
91
  response_metadata = getattr(message, "response_metadata", None)
56
92
  if isinstance(response_metadata, dict):
57
93
  for key in ("token_usage", "usage"):
58
94
  nested = response_metadata.get(key)
59
95
  if isinstance(nested, dict) and nested:
60
- return _usage_from_mapping(nested)
96
+ return TokenUsage.from_mapping(nested)
61
97
 
62
98
  if isinstance(message, dict):
63
99
  from soothe_cli.runtime.wire.message_text import wire_message_body
@@ -71,21 +107,20 @@ def extract_stream_message_token_usage(message: Any) -> tuple[int, int, int]:
71
107
  continue
72
108
  direct_usage = body.get("usage_metadata")
73
109
  if isinstance(direct_usage, dict) and direct_usage:
74
- return _usage_from_mapping(direct_usage)
110
+ return TokenUsage.from_mapping(direct_usage)
75
111
  response = body.get("response_metadata")
76
112
  if isinstance(response, dict):
77
113
  for key in ("token_usage", "usage"):
78
114
  nested_usage = response.get(key)
79
115
  if isinstance(nested_usage, dict) and nested_usage:
80
- return _usage_from_mapping(nested_usage)
116
+ return TokenUsage.from_mapping(nested_usage)
81
117
 
82
- return 0, 0, 0
118
+ return TokenUsage()
83
119
 
84
120
 
85
121
  def message_has_token_usage_metadata(message: Any) -> bool:
86
122
  """Return True when a stream message carries provider token usage."""
87
- input_toks, output_toks, total_toks = extract_stream_message_token_usage(message)
88
- return bool(input_toks or output_toks or total_toks)
123
+ return extract_stream_message_token_usage(message).has_any()
89
124
 
90
125
 
91
126
  async def fetch_conversation_token_count(daemon_session: Any, loop_id: str | None) -> int | None:
@@ -15,6 +15,7 @@ from soothe_client.appkit.turn import (
15
15
  from soothe_client.core.events import (
16
16
  STRANGE_LOOP_COMPLETED,
17
17
  STRANGE_LOOP_PLAN_DECISION,
18
+ STRANGE_LOOP_PLAN_PHASE,
18
19
  STRANGE_LOOP_STARTED,
19
20
  STRANGE_LOOP_STEP_COMPLETED,
20
21
  STRANGE_LOOP_STEP_QUEUED,
@@ -27,6 +28,7 @@ from soothe_client.ux.stream_tool_wire import STREAM_TOOL_CALL_UPDATE, TOOL_CALL
27
28
 
28
29
  from soothe_cli.runtime.presentation.engine import PresentationEngine
29
30
  from soothe_cli.runtime.state.session_stats import TurnEventStats
31
+ from soothe_cli.runtime.token_usage import message_has_token_usage_metadata
30
32
  from soothe_cli.runtime.wire.chunk_filter import (
31
33
  message_has_tool_invocation_metadata,
32
34
  updates_chunk_is_noop,
@@ -43,6 +45,7 @@ _MAIN_LOOP_CUSTOM_TYPES = frozenset(
43
45
  STRANGE_LOOP_STARTED,
44
46
  STRANGE_LOOP_COMPLETED,
45
47
  STRANGE_LOOP_PLAN_DECISION,
48
+ STRANGE_LOOP_PLAN_PHASE,
46
49
  STRANGE_LOOP_STEP_STARTED,
47
50
  STRANGE_LOOP_STEP_QUEUED,
48
51
  STRANGE_LOOP_STEP_COMPLETED,
@@ -83,6 +86,8 @@ class TurnPrepareState:
83
86
  def _message_priority(message: Any, *, is_summarization: bool) -> int:
84
87
  if isinstance(message, ToolMessage):
85
88
  return PRIORITY_NORMAL
89
+ if message_has_token_usage_metadata(message):
90
+ return PRIORITY_HIGH
86
91
  if message_has_tool_invocation_metadata(message):
87
92
  return PRIORITY_NORMAL
88
93
  if is_summarization:
@@ -281,13 +281,13 @@ class SootheApp(
281
281
  """Loop id that ``_loop_*_tokens`` counters belong to."""
282
282
 
283
283
  self._loop_baseline_tokens: int = 0
284
- """Persisted loop usage total loaded from checkpoint (resume baseline)."""
284
+ """Cumulative usage from completed goals (resume baseline)."""
285
285
 
286
- self._loop_input_tokens: int = 0
287
- """Input tokens for the in-flight goal/turn (stream or backend)."""
286
+ self._loop_goal_run_tokens: int = 0
287
+ """Authoritative per-goal cumulative total from backend lifecycle events."""
288
288
 
289
- self._loop_output_tokens: int = 0
290
- """Output tokens for the in-flight goal/turn (stream or backend)."""
289
+ self._loop_turn_committed: bool = False
290
+ """True once the active goal's usage has been folded into the baseline."""
291
291
 
292
292
  self._tokens_approximate: bool = False
293
293
  """Whether the cached token count is stale (interrupted generation)."""
@@ -263,6 +263,7 @@ class _ModelMixin:
263
263
  from soothe_cli.settings import settings
264
264
  from soothe_cli.tui.widgets.context_data import load_token_usage_snapshot
265
265
 
266
+ inflight = self._inflight_turn_stats
266
267
  return await load_token_usage_snapshot(
267
268
  context_tokens=self._loop_token_total(),
268
269
  approximate=self._tokens_approximate,
@@ -270,8 +271,10 @@ class _ModelMixin:
270
271
  daemon_session=self._daemon_session,
271
272
  model_name=settings.model_name,
272
273
  context_limit=settings.model_context_limit,
273
- input_tokens=self._loop_baseline_tokens + self._loop_input_tokens,
274
- output_tokens=self._loop_output_tokens,
274
+ input_tokens=inflight.input_tokens if inflight else 0,
275
+ output_tokens=inflight.output_tokens if inflight else 0,
276
+ cache_read_tokens=inflight.cache_read_tokens if inflight else 0,
277
+ cache_creation_tokens=inflight.cache_creation_tokens if inflight else 0,
275
278
  )
276
279
 
277
280
  async def _show_context_viewer(self) -> None:
@@ -221,7 +221,6 @@ class _StartupMixin:
221
221
  self._ui_adapter._get_loop_token_breakdown = self._loop_token_breakdown
222
222
  self._ui_adapter._on_begin_loop_turn_tokens = self._begin_loop_turn_tokens
223
223
  self._ui_adapter._apply_authoritative_loop_tokens = self._apply_authoritative_loop_tokens
224
- self._ui_adapter._on_tokens_hide = self._hide_tokens
225
224
  self._dump_tui_debug_snapshot()
226
225
 
227
226
  # Seed static slash commands now that the first frame has rendered.
@@ -59,24 +59,24 @@ class _UIMixin:
59
59
  self._status_bar.set_tokens(count, approximate=approximate)
60
60
 
61
61
  def _loop_token_total(self) -> int:
62
- """Return persisted loop usage plus the active goal/turn portion."""
63
- return self._loop_baseline_tokens + self._loop_input_tokens + self._loop_output_tokens
62
+ """Return the current cumulative loop usage (display total)."""
63
+ return self._display_token_total()
64
64
 
65
65
  def _current_turn_token_total(self) -> int:
66
- """Best-effort tokens for the in-flight goal/turn (stream + backend)."""
67
- goal_total = self._loop_input_tokens + self._loop_output_tokens
66
+ """Best-effort tokens for the in-flight goal (backend vs stream)."""
67
+ goal_total = self._loop_goal_run_tokens
68
68
  inflight = getattr(self, "_inflight_turn_stats", None)
69
- if inflight is None:
70
- return goal_total
71
- stream_total = inflight.input_tokens + inflight.output_tokens
69
+ stream_total = inflight.input_tokens + inflight.output_tokens if inflight is not None else 0
72
70
  return max(goal_total, stream_total)
73
71
 
74
72
  def _display_token_total(self) -> int:
75
- """Return loop usage including the in-flight turn when a worker is active."""
73
+ """Return completed-goal baseline plus the active goal's usage."""
74
+ if self._loop_turn_committed:
75
+ return self._loop_baseline_tokens
76
76
  return self._loop_baseline_tokens + self._current_turn_token_total()
77
77
 
78
78
  def _refresh_token_displays(self, *, approximate: bool | None = None) -> None:
79
- """Push the current token total to the thinking row or status bar."""
79
+ """Push the live token total to the thinking row and status bar."""
80
80
  if approximate is not None:
81
81
  self._tokens_approximate = self._tokens_approximate or approximate
82
82
  total = self._display_token_total()
@@ -85,14 +85,11 @@ class _UIMixin:
85
85
  status_bar = getattr(self, "_status_bar", None)
86
86
  if loading_widget is not None:
87
87
  loading_widget.set_token_usage(total, approximate=approximate_flag)
88
- if status_bar:
89
- status_bar.hide_tokens()
90
88
  target = "thinking_row"
91
- elif status_bar:
92
- self._update_tokens(total, approximate=approximate_flag)
93
- target = "status_bar"
94
89
  else:
95
- target = "none"
90
+ target = "status_bar"
91
+ if status_bar:
92
+ self._update_tokens(total, approximate=approximate_flag)
96
93
  trace = self._token_event_trace()
97
94
  if trace is not None:
98
95
  trace.note_display_refresh(display_total=total, target=target)
@@ -100,7 +97,7 @@ class _UIMixin:
100
97
  def _loop_token_breakdown(self) -> tuple[int, int, int]:
101
98
  """Return `(baseline, goal_run, display_total)` for debug tracing."""
102
99
  baseline = self._loop_baseline_tokens
103
- goal_run = self._loop_input_tokens + self._loop_output_tokens
100
+ goal_run = self._loop_goal_run_tokens
104
101
  return baseline, goal_run, self._display_token_total()
105
102
 
106
103
  def _reset_loop_token_usage(
@@ -113,8 +110,8 @@ class _UIMixin:
113
110
  """Clear per-loop token counters (loop switch, /clear, or scope change)."""
114
111
  self._loop_token_scope_id = str(loop_id or "").strip() or None
115
112
  self._loop_baseline_tokens = max(0, baseline)
116
- self._loop_input_tokens = 0
117
- self._loop_output_tokens = 0
113
+ self._loop_goal_run_tokens = 0
114
+ self._loop_turn_committed = False
118
115
  self._refresh_token_displays(approximate=approximate)
119
116
 
120
117
  def _ensure_loop_token_scope(self, loop_id: str | None) -> None:
@@ -134,10 +131,10 @@ class _UIMixin:
134
131
  self._reset_loop_token_usage(loop_id, baseline=baseline, approximate=approximate)
135
132
 
136
133
  def _begin_loop_turn_tokens(self) -> None:
137
- """Reset per-turn goal counters; baseline from prior turns is retained."""
134
+ """Reset per-goal counters; baseline from prior goals is retained."""
138
135
  self._ensure_loop_token_scope(self._lc_loop_id)
139
- self._loop_input_tokens = 0
140
- self._loop_output_tokens = 0
136
+ self._loop_goal_run_tokens = 0
137
+ self._loop_turn_committed = False
141
138
 
142
139
  def _record_loop_turn_tokens(
143
140
  self,
@@ -146,16 +143,15 @@ class _UIMixin:
146
143
  *,
147
144
  approximate: bool = False,
148
145
  ) -> None:
149
- """Fold one turn's usage into the loop baseline and refresh the bar."""
146
+ """Fold one goal's usage into the loop baseline and refresh the bar."""
150
147
  self._ensure_loop_token_scope(self._lc_loop_id)
151
148
  stream_total = max(0, input_tokens) + max(0, output_tokens)
152
- goal_total = self._loop_input_tokens + self._loop_output_tokens
153
- turn_total = max(stream_total, goal_total)
149
+ turn_total = max(stream_total, self._loop_goal_run_tokens)
154
150
  if turn_total <= 0:
155
151
  return
156
152
  self._loop_baseline_tokens += turn_total
157
- self._loop_input_tokens = 0
158
- self._loop_output_tokens = 0
153
+ self._loop_goal_run_tokens = 0
154
+ self._loop_turn_committed = True
159
155
  if approximate:
160
156
  self._tokens_approximate = True
161
157
  self._refresh_token_displays()
@@ -170,34 +166,26 @@ class _UIMixin:
170
166
  *,
171
167
  source: str = "backend",
172
168
  ) -> None:
173
- """Merge backend goal-run token totals into the visible loop counter."""
169
+ """Merge a backend per-goal cumulative total into the live counter."""
174
170
  from soothe_cli.runtime.token_usage import coerce_total_tokens_used
175
171
 
176
172
  goal_total = coerce_total_tokens_used(goal_run_tokens)
177
173
  self._ensure_loop_token_scope(self._lc_loop_id)
178
- previous_goal = self._loop_input_tokens + self._loop_output_tokens
179
- applied = goal_total >= previous_goal
180
- if applied and goal_total > 0:
181
- self._loop_input_tokens = 0
182
- self._loop_output_tokens = goal_total
183
- self._refresh_token_displays()
184
- else:
185
- self._refresh_token_displays()
174
+ previous_goal = self._loop_goal_run_tokens
175
+ applied = goal_total > previous_goal
176
+ if applied:
177
+ self._loop_goal_run_tokens = goal_total
178
+ self._refresh_token_displays()
186
179
  trace = self._token_event_trace()
187
180
  if trace is not None:
188
181
  trace.note_authoritative_merge(
189
182
  source=source,
190
183
  goal_run_total=goal_total,
191
184
  previous_goal_run=previous_goal,
192
- applied=applied and goal_total > 0,
185
+ applied=applied,
193
186
  display_total=self._display_token_total(),
194
187
  )
195
188
 
196
- def _hide_tokens(self) -> None:
197
- """Hide the status-bar token display while the thinking row is active."""
198
- if self._status_bar:
199
- self._status_bar.hide_tokens()
200
-
201
189
  def _check_hydration_needed(self) -> None:
202
190
  """Check if we need to hydrate messages from the store.
203
191