feat: adds @cosmograph-org/cosmo GPU layout (#9)

* docs: removes solved todo

* chore: ignoring .vscode

* wip: implements simulation wrapper to include cosmo

* feat: introduces cosmo through unified wrapper interface

* docs: adds screenshot to readme

* fix: fixes the worst typo of my life

* docs: idk where this came from

* refactor: removes logging from editor

* refactor: editor pans instead of zooming when centering canvas

* refactor: rewrites layout effectors for d3

* refactor: giving up on centering viva for now

* refactor: lowers number of iterations for fr precomputation

* refactor: rewrites and reformats settings file

* feat: overhauls cosmo's size and coordinate system

* refactor: gets rid of flowbite-svelte range

* fix: allows vertical scrolling for sidebar

* fix: slider input was string

* wip: attempting to remove cosmo's labels

* feat: bumps alpha after updating layout parameter

* feat: introduces all of d3 and cosmo forces

* docs: removes img alt text

* build: bump dist

* build: usethis::use_package("tidyselect")

.gitignore

@@ -1,3 +1,5 @@
+.vscode
+
 # Temp
 inst/temp
 

DESCRIPTION

@@ -14,7 +14,8 @@ Imports:
     dplyr,
     igraph,
     jsonlite,
-    shiny
+    shiny,
+    tidyselect
 Suggests: 
     testthat (>= 3.0.0)
 Config/testthat/edition: 3

R/app.R

@@ -91,9 +91,19 @@ start_app <- function(graph, layout) {
 
   igraph::V(graph)$component <- flag_for_grouping
 
+  selected_node_cols <- graph |>
+    igraph::as_data_frame("vertices") |>
+    dplyr::select(
+      id = name,
+      initialX = x,
+      initialY = y,
+      component,
+      tidyselect::any_of(c("color", "size"))
+    )
+
   graph_json <- jsonlite::toJSON(list(
-    nodes = igraph::as_data_frame(graph, "vertices"),
-    links = igraph::as_data_frame(graph, "edges")
+    nodes = selected_node_cols,
+    links = igraph::as_data_frame(graph, "edges")[, 1:2]
   ))
 
   server <- function(input, output, session) {
@@ -155,7 +165,7 @@ precompute_layout <- function(graph, cols) {
 
   similarity_layout <- igraph::layout_with_fr(
     graph = similarity_graph,
-    niter = 1000
+    niter = 500
   ) * LAYOUT_SIZE_FACTOR
 
   # Centers layout around origin = [0, 0]

README.Rmd

@@ -19,18 +19,21 @@ knitr::opts_chunk$set(
 [![R-CMD-check](https://github.com/dalmolingroup/easylayout/actions/workflows/R-CMD-check.yaml/badge.svg)](https://github.com/dalmolingroup/easylayout/actions/workflows/R-CMD-check.yaml)
 <!-- badges: end -->
 
-easylayout is an R package that seamlessly bridges manipulation and
-visualization by leveraging the user's IDE itself (e.g., RStudio, VSCode). It
-is **not** yet another visualization library, but instead aims to interconnect
-existing libraries and streamline their usage into the R ecosystem. easylayout
-takes an igraph object and serializes it into a web application integrated with
-the IDE's interface through a Shiny server. The web application lays out the
-network by simulating attraction and repulsion forces. Simulation parameters
-can be adjusted in real-time. An editing mode allows moving and rotating nodes.
-The implementation aims for performance, so that even lower-end devices are
-able to work with relatively large networks. Once the user finishes tinkering
-the layout, it is sent back to the R session to be plotted through popular
-libraries like ggplot2 or even the base package itself.
+easylayout is an R package that leverages interactive force simulations within
+the IDE itself (e.g., RStudio, VSCode). It is **not** yet another visualization
+library, but instead aims to interconnect existing libraries and streamline
+their usage into the R ecosystem.
+
+![](https://github.com/user-attachments/assets/1b91cb11-77ef-47a5-b529-3805a9785a76)
+
+easylayout takes an igraph object and serializes it into a web application 
+integrated with the IDE's interface through a Shiny server. The web application
+lays out the network by simulating attraction and repulsion forces. Simulation
+parameters can be adjusted in real-time. An editing mode allows moving and
+rotating nodes. The implementation aims for performance, so that even lower-end
+devices are able to work with relatively large networks. Once the user finishes
+tinkering the layout, it is sent back to the R session to be plotted through
+popular libraries like ggplot2 or even the base package itself.
 
 ## Installation
 
@@ -57,7 +60,7 @@ igraph::V(g)$label <- NA
 igraph::V(g)$size <- sample(1:5, number_of_vertices, replace = TRUE)
 igraph::V(g)$color <- sample(rainbow(5), number_of_vertices, replace = TRUE)
 
-plot(g, layout = easylayout)
+plot(g, layout = easylayout, vertex.label = NA, margin = 0)
 ```
 
 ## Future work

README.md

@@ -8,19 +8,22 @@
 [![R-CMD-check](https://github.com/dalmolingroup/easylayout/actions/workflows/R-CMD-check.yaml/badge.svg)](https://github.com/dalmolingroup/easylayout/actions/workflows/R-CMD-check.yaml)
 <!-- badges: end -->
 
-easylayout is an R package that seamlessly bridges manipulation and
-visualization by leveraging the user’s IDE itself (e.g., RStudio,
-VSCode). It is **not** yet another visualization library, but instead
-aims to interconnect existing libraries and streamline their usage into
-the R ecosystem. easylayout takes an igraph object and serializes it
-into a web application integrated with the IDE’s interface through a
-Shiny server. The web application lays out the network by simulating
-attraction and repulsion forces. Simulation parameters can be adjusted
-in real-time. An editing mode allows moving and rotating nodes. The
-implementation aims for performance, so that even lower-end devices are
-able to work with relatively large networks. Once the user finishes
-tinkering the layout, it is sent back to the R session to be plotted
-through popular libraries like ggplot2 or even the base package itself.
+easylayout is an R package that leverages interactive force simulations
+within the IDE itself (e.g., RStudio, VSCode). It is **not** yet another
+visualization library, but instead aims to interconnect existing
+libraries and streamline their usage into the R ecosystem.
+
+![](https://github.com/user-attachments/assets/1b91cb11-77ef-47a5-b529-3805a9785a76)
+
+easylayout takes an igraph object and serializes it into a web
+application integrated with the IDE’s interface through a Shiny server.
+The web application lays out the network by simulating attraction and
+repulsion forces. Simulation parameters can be adjusted in real-time. An
+editing mode allows moving and rotating nodes. The implementation aims
+for performance, so that even lower-end devices are able to work with
+relatively large networks. Once the user finishes tinkering the layout,
+it is sent back to the R session to be plotted through popular libraries
+like ggplot2 or even the base package itself.
 
 ## Installation
 
@@ -48,7 +51,7 @@ igraph::V(g)$label <- NA
 igraph::V(g)$size <- sample(1:5, number_of_vertices, replace = TRUE)
 igraph::V(g)$color <- sample(rainbow(5), number_of_vertices, replace = TRUE)
 
-plot(g, layout = easylayout)
+plot(g, layout = easylayout, vertex.label = NA, margin = 0)
 ```
 
 ## Future work