spacr.qt.widgets.console_panel

ConsolePanel — merged pipeline console + AI chat panel.

One vertical scrolling area shows both pipeline stdout AND AI chat messages, separated by dark-gray “topic” bars (“Mask”, “Measure”, “spaCR AI”, …). Below the scroll sits an input row where the user can type at any time; a switch on the left decides whether the message goes to the AI or is ignored.

Public API

• begin_topic(label) — insert a dark-gray divider bar

  (used at the start of every pipeline run and every time we switch to/from the AI)

• append_stdout(text) — append pipeline output; if the last

  entry isn’t already a stdout block it starts a new one

• append_error(traceback) — same as stdout but red-tinted

• open_error_flow(tb, app) — inject the AI-explainer prompt for a

  traceback and stream the reply into a fresh spaCR-AI section

• clear() — wipe every entry

Streaming state

The panel owns the AI thread+worker itself so state stays coherent even as the user switches between pipeline apps.

Module Contents

class spacr.qt.widgets.console_panel.ConsolePanel(active_app_label: str = '', parent=None)

Bases: PySide6.QtWidgets.QWidget

Merged pipeline stdout + AI chat panel.

Owns the AI stream thread so provider switches and app changes do not orphan a running subprocess. See the module docstring for the full public surface.

Variables:

ai_stream_finished – emitted when an AI stream ends (ok or error) so the parent screen can flip its Cancel button back.

ai_stream_finished

set_active_app(label: str) → None

Set the label used in the next auto-inserted topic divider.

begin_topic(label: str) → None

Insert a divider bar labeled label (e.g. ‘Mask’). Callers can force a new section this way. AI content NEVER uses this — AI replies flow inline in the same stdout block.

append_stdout(text: str) → None

Append pipeline output. Opens a fresh stdout block (with a topic divider) at the very first stdout of a session; opens a divider-less block after a bubble breaks the flow.

append_error(tb: str) → None

Append a red-tinted error block, prefixed with an ERROR topic bar.

Parameters:

tb – traceback text; empty strings are ignored.

clear() → None

Wipe every entry (topic bars, stdout blocks, chat bubbles).

set_ai_active(on: bool) → None

Enable/disable AI routing for Enter-submits from the input.

set_ai_provider(provider_name: str | None) → None

Select the provider used for AI submissions, or None to unset.

cancel_ai() → None

Public — AppScreen calls this if the user cancels a stream.

is_ai_streaming() → bool

Return True while an AI response is being streamed.

shutdown() → None

Cancel any active stream and block until its QThread has exited. Must be called before the panel (or its parent window) is destroyed — otherwise Python drops the last reference to the running QThread and Qt aborts with: QThread: Destroyed while thread ‘’ is still running.

The cancel path kills the CLI subprocess directly so the stream reader unblocks immediately; we then wait for the worker’s run() to return and the QThread to quit normally.

closeEvent(event) → None

Ensure the AI thread is drained before Qt destroys the panel.

open_error_flow(traceback_text: str, active_app: str = '') → None

Send a traceback to the AI explainer and stream the reply inline.

Parameters:

• traceback_text – raw traceback captured from the pipeline.

• active_app – optional app label used in the framing prompt.