bino graph

bino graph prints a dependency graph of your bundle. Use it to understand how datasources, datasets, layouts, and artefacts are connected.

Usage

bino graph [flags]

Common flags:

• --work-dir - report bundle directory.
• --artefact - focus on specific artefact names (can be repeated).
• --exclude-artefact - exclude certain artefact names (can be repeated).
• --view - output format: tree (default) or flat.

Examples

View the full dependency tree for all reports:

bino graph --view tree

Focus on a single artefact:

bino graph --artefact monthly_sales --view tree

You can copy graph output into your docs and accompany it with a der diagram image for easier reading.

Column references

Every component in the graph lists the dataset columns it reads. Check it before you rename or remove a column.

Each dataset gets one entry:

columns[<dataset>]=<role>:<column>,<column> <role>:<column>,...

<dataset> is the name as written in the component, $name for a DataSource. A component that reads several datasets gets one entry per dataset.

The roles, in output order:

• scenario - scenario columns from scenarios, and from columnthereof in a Table.
• variance - the scenario columns a variance is computed from. dac1_pp1_pos reads ac1 and pp1.
• measure - value columns of ChartScatter, ChartBubble and ChartBullet, from fields such as x, y or actual.
• group - columns the rows are grouped by, such as rowGroup and rowGroupIndex for a grouped Table, or category for a chart level.
• order - columns from order.
• filter - columns named in filter.
• attribute - columns in Table attributes expressions, such as pl1 in max(pl1).
• template - columns in Text values like \${data.kpi[0].ac1}.
• implicit - columns the component always reads. A Table reads category, categoryIndex and operation, a ChartTime reads date and operation.

The column auto means the engine picks the column from the data at render time. You see it when a field such as scenarios or level is empty or set to auto.

inherited-page and inherited-closest are resolved from the title fields (titleScenarios, titleVariances, titleOrder) of the page and card around the component. A value that cannot be resolved, such as inherited-page without a page title or a ${PARAM} placeholder, is shown as written.

In the tree view each entry is an extra line under the component:

sales-dashboard [ReportArtefact] hash=0904a64ef07d, file=report.yaml
└── table-page [LayoutPage] hash=fe35c694dc2c, file=pages.yaml
    └── Table table-page#0 [Component] hash=c7c45b5d3de5, kind=Table, dataset=financial_summary (DataSet), file=pages.yaml
        │   columns[financial_summary]=scenario:ac1,pp1,pl1 variance:ac1,pp1,pl1 group:rowGroup,rowGroupIndex attribute:pl1 implicit:category,categoryIndex,operation
        └── financial_summary [DataSet] hash=54beb5c30327, file=datasets.yaml
            └── kpi_table_data [DataSource] hash=18d390f4437d, type=inline, file=data.yaml

In the flat view the entries are added to the DETAILS cell:

KIND       NAME                HASH          DEPENDS ON                 DETAILS
Component  Table table-page#0  c7c45b5d3de5  DataSet:financial_summary  hash=c7c45b5d3de5, kind=Table, dataset=financial_summary (DataSet), file=pages.yaml, columns[financial_summary]=scenario:ac1,pp1,pl1 variance:ac1,pp1,pl1 group:rowGroup,rowGroupIndex attribute:pl1 implicit:category,categoryIndex,operation

bino build writes the flat view to a .bngraph file next to each PDF (for example monthly_sales.pdf.bngraph), so the file lists the columns too.

Limits

• The list is built from the manifests only. bino does not run any SQL for it.
• The rules follow template engine v1.0.0-next.27. A newer engine or a new spec field is not detected.
• auto is not resolved to a column. Dataset defaults (_spec_* columns) are not read, so a field set only by a dataset default shows as auto or not at all.
• Fields the engine does not use to read data print nothing, for example Table barColumns and ChartTime stack and dateInterval.
• A component bound to several datasets lists the same columns for each of them, even though the engine reads only one.
• Inline datasets appear under their generated name _inline_dataset_<hash>.
• filter and Text values are read by simple pattern matching, so an unusual expression can give a wrong list. Text placeholders are only seen when written as \${...}.
• The engine finds the inherited page or card once per component, from its first inherited field. bino resolves each field on its own, so the list can differ when one component mixes inherited-page and inherited-closest inside a card.
• Tree nodes that use ref list no columns. Components on a page that an artefact uses with params show the ${PARAM} placeholders as written.