figurecapture.h File Reference

Manifest-driven screenshot capture for the user manual's figures. More...

#include <QElapsedTimer>
#include <QJsonObject>
#include <QList>
#include <QObject>
#include <QSize>
#include <QString>

Include dependency graph for figurecapture.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
struct	openswmmvis::ui::FigureSpec
	One row of the manifest: a figure and how to reach it. More...
struct	openswmmvis::ui::FigureResult
	Result of attempting one row; serialised into run.json. More...
class	openswmmvis::ui::FigureCapture
	Drives a manifest of figures to PNG, then quits the application. More...

Namespaces
namespace	openswmmvis
namespace	openswmmvis::ui

Enumerations
enum class	openswmmvis::ui::FigureLane {   openswmmvis::ui::Offscreen ,   openswmmvis::ui::Live ,   openswmmvis::ui::Human }
	Which platform a figure can be captured on. More...

Detailed Description

Manifest-driven screenshot capture for the user manual's figures.

Author

Caleb Buahin caleb.nosp@m..bua.nosp@m.hin@g.nosp@m.mail.nosp@m..com

Date

2026

License\n GPL-3.0-or-later

The manual (docs/manual/) declares every screenshot as a \figtodo placeholder and turns it into \fig once the PNG exists in docs/manual/images/. This class produces those PNGs without a human at the keyboard: SWMMVIS_CAPTURE_MANIFEST=<abs .json> names a list of figures, each describing the UI state to reach and the widget to grab.

Generalises the three one-off SWMMVIS_SNAPSHOT_* dev hooks in swmmvis.cpp (which stay, being documented in appendices/a04_performance.md) into one table so a whole chapter is captured per launch — 326 one-shot launches would be ~45 min of pure startup, grouped runs are ~2 min.

Two run modes, one manifest. QT_QPA_PLATFORM=offscreen handles widget-only figures (dialogs, docks, panels) with no display; anything containing scene-graph content (2D mesh, 2D results) reads back BLANK offscreen and must run on the live cocoa platform. Rows are not classified by hand: every grab is scored and a blank one is reported so it can be re-shot on the live lane.

Deliberately public-API only (ActionRegistry, ThemeManager, QWidget) so it needs no access to SWMMVis internals. Models are supplied by the launcher through SWMMVIS_OPEN_ON_STARTUP, one launch per model.