Report Styles • datagoodr

One DGF can be published in more than one shape. The style decides how much room a variable gets on the page; the design decides which blocks it gets at all (see vignette("report-aesthetics")). This article is about the first.

Every style opens with the same contents list, and every row of it glimpses the variable’s data — the first values as the stabilizer will write them, or the preview sample where no stabilizer runs. rail_contents(glimpse = ) sets how many characters it may use, or 0 leaves the data out.

The three styles

Style	`create_rg(style = )`	For
Rail (default)	`"rail"`	Reading a dataset for the first time. One variable per screen: a left rail with its type, label, name and description, beside its graphic, preview and stats.
Compact	`"compact"`	Documentation. The same blocks condensed — several variables per screen.
Vintage	`"vintage"`	The original three-column grid, kept so guides built with it still render.

Compare the same dataset in the first two: baseball and baseball, compact.

Switching

The style is chosen when the document is scaffolded, because it picks three things at once: the template, its stylesheet, and the figure sizes the graphics are drawn at.

library(datagoodr)

create_rg(dgf)                       # rail, the default
create_rg(dgf, style = "compact")    # same guide, documentation density
create_rg(dgf, style = "vintage")    # the original grid

Each writes a .qmd plus its own stylesheet next to it — datagoodr.css, datagoodr-compact.css, datagoodr-vintage.css — so a project can be restyled by editing its own copy without touching the package.

To publish a dataset in two styles at once, scaffold two documents and give them different names. Both read the same DGF, so there is still one source of truth:

create_rg(dgf, file = "research-guide.qmd")
create_rg(dgf, file = "research-guide-compact.qmd", style = "compact")

Changing the style of a guide you have already annotated means scaffolding the new one beside it rather than converting in place — update_rg() re-renders an existing document against a different DGF, but it does not restyle it.

What compact actually changes

Nothing is left out: the same blocks appear in the same order, for the same types. What changes is how much page each one takes.

PROPERTIES is transposed into a one-row strip at the top of the body (ROWS DISTINCT MISSING/NA) instead of a vertical table in the rail. In the rail that table is what set a variable’s height, so condensing anything else first would have shortened nothing.
Booleans are drawn as one divided bar (boolebar()) rather than a stack of one bar per category: the parts share a length, over a tenths scale, so they read as shares of a whole — which is what a boolean is.
The long tables and the preview are kept short — 8 factor levels instead of 50, 2 preview lines instead of 4. A truncated table says what it left out (+ 444 more levels) rather than stopping silently.
The numeric DISTRIBUTION SHAPE block goes, since skew and kurtosis are already drawn above the histogram.
Graphics are drawn smaller and the whitespace is roughly halved.

On the baseball guide that is a 35% shorter page.

Density is a knob, not a rewrite

Compact is the rail renderer at a higher density, so everything that customizes a rail guide works on it unchanged: a renderer you redefine in DG.R, a layout.* shadow, your own stylesheet. Its design is readable the same way as any other:

get_design(compact = TRUE)

A reporting function opts into the style’s density by declaring size (characters per line), max.lines (preview rows) or max.rows (table rows). The renderer fills them in from the active style, so the same layout row prints 8 levels in a compact guide and 50 in the default one — and your own renderer condenses with the rest if it takes the same arguments:

# in DG.R - a renderer that respects the guide's density
paste_levels_freq <- function(VNAME, LABEL = "FACTOR LEVELS", max.rows = 50) {
  # ... your own table, trimmed to max.rows ...
}

Mixing styles in one document

rail_sections() takes the style per call, so a document can draw the groups that deserve room at full size and the rest condensed — for example a guide whose numeric variables are the point and whose identifiers are just inventory:

groups <- rail_groups(dgf)
rail_sections(groups$Numeric,    "Numeric")                      # full size
rail_sections(groups$Identifier, "Identifier", style = "compact") # condensed

Use the stylesheet of the style that dominates the document; the two share all their structure, so either one renders both.