Python and GUI workflows

Analysis workflows

ERLabPy supports three workflows. Two start from data and analysis code in a notebook. The third starts and remains in ImageTool Manager.

        flowchart LR
    subgraph notebook_workflows["Notebook workflows"]
        direction TB
        notebook["Notebook<br/><small>DataArray variables · analysis code</small>"]
        watched["Watch variable<br/><small>%watch analysis_map</small>"]
        show["Show current data<br/><small>%itool -m · show_in_manager()</small>"]
        manager["ImageTool Manager<br/><small>ImageTools · managed tools · figures</small>"]
        result["Derived result<br/><small>Provenance recorded</small>"]
        code["Copied Python code"]
        tool["Standalone tool<br/><small>dtool · ktool · goldtool · ftool</small>"]
        resume["Notebook analysis"]

        notebook --> watched
        watched <-->|"Synchronize variable and row"| manager
        notebook --> show --> manager
        manager -->|"Run managed operation"| result
        result -->|"Copy Full Code"| code --> resume
        notebook -->|"One-off analysis"| tool
        tool -->|"Close tool"| resume
    end

    subgraph manager_only_workflow["Manager-only workflow"]
        direction TB
        manager_only["ImageTool Manager<br/><small>Data · tools · figures</small>"]
        load["Load data<br/><small>Data Explorer · top-level row</small>"]
        analysis["Inspect and analyze<br/><small>ImageTool · ktool · dtool · ftool</small>"]
        managed_result["Derived results<br/><small>Nested rows · provenance</small>"]
        figure["Create figures<br/><small>Figure Composer</small>"]
        workspace["Save the session<br/><small>.itws workspace</small>"]

        manager_only --> load --> analysis --> managed_result
        managed_result --> figure
        managed_result --> workspace
    end

    notebook_workflows ~~~ manager_only_workflow

    class notebook,manager_only input
    class watched,show,manager,tool,load,analysis,figure focus
    class result,code,resume,managed_result,workspace output
    

Notebook and Manager

Use this workflow when Python code remains the main analysis record and Manager provides synchronized ImageTool windows, derived-result tracking, provenance, and Figure Composer. Use %watch when a notebook variable and a Manager row must stay synchronized. Use %itool -m or show_in_manager when Manager only needs the current data.

When a supported GUI operation creates new data, Manager shows the result below its source and records the operation in the Provenance tab. Use Copy Full Code to copy the corresponding Python statements and continue the analysis in the notebook.

Notebook and standalone tools

Standalone tools such as erlab.interactive.dtool(), erlab.interactive.ktool(), erlab.interactive.goldtool(), and erlab.interactive.ftool() run in the notebook process. With the default Qt event loop behavior, the current cell returns after the tool closes. This path is suitable for one-off analysis that does not need Manager state.

Manager-only workflow

Use this workflow when the analysis can remain in the GUI. Start ImageTool Manager and use Data Explorer to browse, preview, and load files. Loaded data appears as a top-level row. Open it in ImageTool, apply supported operations or open managed tools, and inspect derived results below their sources. Manager records the corresponding operations as provenance.

Create managed figures with Figure Composer. Save the session as an .itws workspace to restore the Manager windows, derived relationships, provenance, and figure recipes later.

Data and code transfer

Direction	Data or code
Files to ImageTool Manager	A loaded xarray object added as a top-level row
Notebook or script to ImageTool Manager	A shown or watched xarray object
GUI operation to Manager	A result DataArray grouped below the source that produced it
GUI operation to Python	Generated code that uses public Python APIs
Figure Composer to Python	Generated Matplotlib code
ImageTool Manager to workspace	Managed windows, state, row notes, derived relationships, and Figure Composer recipes

ImageTool reads the same dimensions and coordinates that Python reads. Manager rows and operation history do not change the dimensions, coordinates, or metadata of the DataArray.

Interactive tools help users select data ranges and parameter values. Generated code supports review, repeated analysis, and automation. Review generated code before adding it to a script or notebook.

The table in Operation map lists GUI actions and their corresponding Python operations. Use Python integration when you move data or code between Python and the GUI. GUI reference describes the applications and controls. Use Derived results and workspaces to update derived results, reuse recorded operations, or save a Manager workspace.