Support interactive legends for match_expr()-based layers in add_legend() #197

Activity

1. walkerke commented on Apr 13, 2026

   

   walkerke

   on Apr 13, 2026

   Last edited by walkerke

   Owner

   More actions

   Thanks for the detailed example! Interactive legends do already work with match_expr(). The key is passing filter_column to tell the legend which column in the source data to filter on.

   Here's a minimal example:

     library(mapgl)
     library(sf)
   
     nc <- st_read(system.file("shape/nc.shp", package = "sf"))
   
     nc$region <- cut(
       nc$BIR74,
       breaks = c(0, 1000, 5000, 15000, Inf),
       labels = c("Low", "Medium", "High", "Very High")
     )
   
     colors <- c("#eff3ff", "#6baed6", "#2171b5", "#08306b")
   
     maplibre(bounds = nc) |>
       add_fill_layer(
         id = "counties",
         source = nc,
         fill_color = match_expr(
           column = "region",
           values = c("Low", "Medium", "High", "Very High"),
           stops = colors
         ),
         fill_opacity = 0.7,
         fill_outline_color = "white"
       ) |>
       add_legend(
         legend_title = "Births (1974)",
         values = c("Low", "Medium", "High", "Very High"),
         colors = colors,
         type = "categorical",
         layer_id = "counties",
         filter_column = "region",
         interactive = TRUE
       )

   

   As of right now, we don't have a classification helper for match expressions, so that argument won't apply in this case.
2. danielvartan commented on Apr 13, 2026

   

   danielvartan

   on Apr 13, 2026

   ContributorAuthor

   More actions

   Thank you for the example.

   Your example works when the source has the category field as a vector layer, but not when a static source is used.

   Unfortunately, this does not work in my case.

   In the example I used a precompiled PMTiles file based on commonly used tiles in geospatial research in Brazil, sourced from the geobr R package (see the PMTilesBR project). Using a standard, precompiled tile source is very useful because it allows you to build lightweight applications without the need to serve your own tiles on the web.

   In this case, the hosted file does not have the category field as a vector layer. The styling is done by assigning a value (in this case, a hex color) to each hex address, so adding a filter_column does not work: the classification is not part of the source tile.
3. walkerke commented on Apr 14, 2026

   

   walkerke

   on Apr 14, 2026

   Owner

   More actions

   ok I see what you're saying - right now the implementation requires that you have a pre-defined column to filter on. I'll think that through as it'll take a bit more infrastructure work to hook that up.

   👍React with 👍1Reacted by Daniel Vartanian