spacr.qt.widgets.qc_summary¶

One verdict for a project, assembled from verdicts that already exist.

Every check spaCR can make is already implemented somewhere – segmentation scorecards, leakage audits, the units stamp on every measurement row. What is missing is a place to see them together, so “is this run usable?” does not require opening five screens and remembering which of them was run last.

Nothing here computes a verdict. The rule comes from spacr.qt.prerun, which states it plainly: opening a plate’s masks costs seconds to minutes, and a screen that pays that on every visit is a screen nobody keeps. So each reader here does a directory listing, a stat and a parse, and reports what it found – including that it found nothing, which is emphatically not the same as finding nothing wrong.

Staleness follows the same rule. seg_qc.read_digest dates each scorecard against its mask stack and reports one written before its masks as out of date; this module carries that through to the card rather than re-deriving it, and applies the same idea to the other sources against their own inputs.

The vocabulary is borrowed, not reinvented. A segmentation flag’s explanation comes from spacr.seg_qc.FLAG_GUIDANCE, which is where those sentences were written and where they will be maintained.

Classes¶

`Dashboard`	Every card for one project, plus the verdict they add up to.
`QCCard`	One check, as it was found on disk.

Functions¶

`format_dashboard`(→ str)	The dashboard as plain text, for a log or a copy-paste.
`read_dashboard`(→ Dashboard)	Read every verdict already on disk for one project. Computes none.
`worst_verdict`(→ str)	The worst of several verdicts, by `VERDICT_ORDER`.

Module Contents¶

class spacr.qt.widgets.qc_summary.Dashboard[source]¶

Every card for one project, plus the verdict they add up to.

Variables:

root – the project folder the cards were read from.
verdict – the worst card verdict.
headline – one sentence naming the worst finding.
cards – the cards, in a fixed order so the screen does not reshuffle between refreshes.
checked_at – when the read happened.
blocks_run – constant False. Advisory by construction, and the field exists to say so in code rather than only in a docstring – the same posture spacr.seg_qc.QCDigest takes.

card(key: str) → QCCard | None[source]¶

The card with this key, or None.

Parameters:: key – a card identifier such as "segmentation", "leakage", "units", "plate" or "agreement".

property stale: bool[source]¶: Whether any card is older than the thing it describes.

class spacr.qt.widgets.qc_summary.QCCard[source]¶

One check, as it was found on disk.

Variables:

key – stable identifier – segmentation, leakage, units, plate, agreement.
title – what goes on the card.
verdict – one of VERDICT_ORDER.
headline – the one line worth reading.
detail – the sentences behind it, already in plain language. For segmentation these come from spacr.seg_qc.FLAG_GUIDANCE – the same words that screen uses, because two vocabularies for one flag is how a codebase ends up disagreeing with itself.
source – the file the verdict was read from, or "".
mtime – that file’s modification time, or 0.0.
stale – whether its inputs are newer than it is. A stale card is not downgraded to a worse verdict – it describes the previous run accurately, and pretending otherwise would hide which of the two is the problem.
how_to_produce – for a missing card, the one thing to run.

property display_verdict: str[source]¶: The verdict as it should be labelled, staleness included.

spacr.qt.widgets.qc_summary.format_dashboard(dashboard: Dashboard) → str[source]¶

The dashboard as plain text, for a log or a copy-paste.

Parameters:: dashboard – the dashboard to format: its verdict and headline first, then one line per card with that card’s details indented under it.

spacr.qt.widgets.qc_summary.read_dashboard(src: Any, *, segmentation_reader=None) → Dashboard[source]¶

Read every verdict already on disk for one project. Computes none.

Parameters:

src – project folder, plate folder, or a list of either.
segmentation_reader – substitute for spacr.seg_qc.read_digest(), for tests.

Returns:

a Dashboard. verdict == "missing" means no check has been run, which is not the same as "ok".

spacr.qt.widgets.qc_summary.worst_verdict(verdicts: Sequence[str]) → str[source]¶

The worst of several verdicts, by VERDICT_ORDER.

Parameters:: verdicts – verdict strings, each one of VERDICT_ORDER ("ok" best to "error" worst); unknown strings are ignored, and none known gives "missing".