---
title: "Getting started with rgexf"
author: "George G. Vega Yon"
date: "August 11, 2021"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Getting started with rgexf}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r setup, echo=FALSE}
knitr::opts_chunk$set(fig.width = 7)
```

# Introduction

The `rgexf` package provides a way to interact with [GEXF files](https://gexf.net/).
The GEXF standard was developed by the [Gephi](https://gephi.org/)--"Like Photoshop for graphs"--core team, 
and can be used to save static and dynamic networks.

With the `rgexf` package, users can create `gexf` (R) objects from scratch,
import GEXF files, coerce `gexf` objects into `igraph` objects, and
visualize graphs using the [`gexf-js` JavaScript library](https://github.com/raphv/gexf-js).
In this vignette, we will illustrate how can we (a) import a GEXF file into R
and visualize it with `igraph,` and (b) create a `gexf` object from scratch.

# Reading GEXF files

The `rgexf` package comes with a network from Les Misérables, which is
featured in Gephi. To read GEXF files, we can use the `read.gexf` function:

```{r read-gexf}
# Loading rgexf
library(rgexf)

# Accessing the path of the file
fn    <- system.file(
  "gexf-graphs/lesmiserables.gexf", package = "rgexf"
  )
lesmi <- read.gexf(fn)

# Taking a look at the first handful of nodes and edges
head(lesmi)
```

Moreover, we can directly plot the graph using the `plot.gexf` method--through
the `gexf-js` library--or coercing it into an `igraph` object and use
igraph's plotting engine:

```{r igraph-plot}
lesmi_ig <- gexf.to.igraph(lesmi)
lesmi_ig

op <- par(mai = rep(0, 4)) # For extra space
plot(lesmi_ig)
par(op)
```

We can also go back:

```{r go-back}
head(igraph.to.gexf(lesmi_ig))
```


Using the `plot.gexf` method--which uses the `gexf-js` JavaScript library--results in a Web visualization of the graph, like this:

```r
plot_gexfjs(g)
```

```{r gexf-js, echo = FALSE, fig.alt = "Les Misérables network rendered with gexf-js"}
knitr::include_graphics("lesmiserables.png")
```

Starting with version 0.17.0, the default plot method uses `sigmajs` instead of `gexf-js`. Nonetheless, `gexf-js` is still available and can be embedded in Quarto/R Markdown documents: the last code chunk in this vignette shows how to do it.

# Creating GEXF files

This example was extracted directly from the demo "gexfrandom." Here
we create three networks with the same set of vertices and layout,
we color them and finalize plotting it with igraph.

```{r}
# Random graph demo
set.seed(11)
```

Creating the vertices:

```{r}
# Vertex
n <- 30
prb <- .3
vertex1 <- data.frame(id = 1:n, label = 1:n)
vertex2 <- data.frame(id = (n + 1):(2 * n), label = (n + 1):(2 * n))
vertex3 <- data.frame(
  id = (2 * n + 1):(3 * n), label = (2 * n + 1):(3 * n)
  )
```

Building edges:

```{r}
# Edges
edges1 <- combn(vertex1$label, 2)
edges1 <- edges1[, which(runif(ncol(edges1)) > (1 - prb))]
edges1 <- data.frame(source = edges1[1, ], target = edges1[2, ])

edges2 <- combn(vertex2$label, 2)
edges2 <- edges2[, which(runif(ncol(edges2)) > (1 - prb))]
edges2 <- data.frame(source = edges2[1, ], target = edges2[2, ])

edges3 <- combn(vertex3$label, 2)
edges3 <- edges3[, which(runif(ncol(edges3)) > (1 - prb))]
edges3 <- data.frame(source = edges3[1, ], target = edges3[2, ])
```

We can and add visual attributes:

```{r}
# Visual attributes
size <- runif(n, max = 100)
color <- terrain.colors(n)
color <- color[order(runif(n))][1:n]
color <- cbind(t(col2rgb(color)), 1)

color2 <- heat.colors(n)
color2 <- color2[order(runif(n))][1:n]
color2 <- cbind(t(col2rgb(color2)), 1)

color3 <- topo.colors(n)
color3 <- color3[order(runif(n))][1:n]
color3 <- cbind(t(col2rgb(color3)), 1)
```

Generating a layout:

```{r}
# Nice layout
pos <- matrix(0, nrow = n, ncol = 3)

for (i in 2:n) {
  pos[i, 1] <- pos[i - 1, 1] + cos(2 * pi * (i * 1.7 - 1) / n)
  pos[i, 2] <- pos[i - 1, 2] + sin(2 * pi * (i - 1) / n)
}

pos <- pos / (max(pos) - min(pos))
pos2 <- pos
pos2[, 1] <- pos2[, 1] + max(pos2[, 1]) - min(pos[, 1])
pos3 <- pos
pos3[, 1] <- pos3[, 1] + max(pos2[, 1]) - min(pos[, 1])
```

And finally, we can build the `gexf` object:

```{r}
graph <- gexf(
  rbind(vertex1, vertex2, vertex3),
  rbind(edges1, edges2, edges3),
  nodesVizAtt = list(
    size = c(size, size, size),
    color = rbind(color, color2, color3),
    position = rbind(pos, pos2, pos3)
  )
)

# Taking a quick look
head(graph)
```

As before, we can either directly call the plot function
for `gexf` objects, or coerce it into an `igraph` object:

```{r igraph-plot-2}
# plot(graph) 
op <- par(mai = rep(0, 4)) # For extra space
plot(gexf.to.igraph(graph))
par(op)
```

## Embedding `gexf-js` in quarto/rmarkdown documents

To embed `gexf-js` in quarto/rmarkdown documents, we can use the `gexfjs` function. This function takes a `gexf` object and returns an HTML widget that can be rendered in the document:

```{r}
gexfjs(graph)
```

Different from the `plot_gexfjs` function, the `gexfjs` inserts the visualization in the document, while `plot_gexfjs` opens it in a new browser window.