Why is that Trelliscope viewer showing empty panels, no images?

Question

I have the below data frame "p" on Pokemon.

library("dplyr")
library("trelliscopejs")

glimpse(p)
Rows: 801
Columns: 5
$ pokemon       <chr> "bulbasaur", "ivysaur", "venusaur~
$ type_1        <chr> "grass", "grass", "grass", "grass~
$ attack        <int> 49, 62, 82, 100, 52, 64, 84, 130,~
$ generation_id <chr> "1", "1", "1", NA, "1", "1", "1",~
$ url_image     <chr> "http://assets.pokemon.com/assets~

I am creating a panel column and then applying Trelliscope in order to access and view the pokemon images from the web (corresponding to the "url_image" of the dataframe). Though it is generating a viewer with 18 pages covering 801 panels as expected, all the panels are empty and I cannot see the images.

p <- p %>%  mutate(panel = img_panel(url_image), pokemon = cog(val = pokemon, default_label = TRUE))

trelliscope(p, name = "pokemon", ncol=6, nrow=3)

I have downloaded all the 801 images to a local folder "pokemon_local" and tried to view them in Trelliscope using the below code. Unfortunately, again though the Trelliscope viewer is getting generated, the panels are empty.

path <- file.path("D:/xyz/pokemon_local")
dir.create(path)

for (url in p$url_image)
  download.file(url, destfile=file.path(path,basename(url)), quiet=TRUE, mode="wb")

p$image <- basename(p$url_image)
p<- mutate(p, panel=img_panel_local(image))

trelliscope(p, name="pokemon", nrow=3, ncol=6, path=path)

Can someone help with an explanation and solution?

Answer 1

After several trial and error attempts, and based on the hint provided in @Waldi 's answer, I have discovered that I need to add "state" variable with a list of labels of "pokemon" and "pokdex" elements. If we omit either of them, it does not work. I don't know why? By the way, several other examples I tried with "gapminder" dataset (which is, unlike "pokemon dataset", not about images), we do not need to create "state" variable.

So if we change the trelliscope() function code in the question with the below code, both the options, will work.

1. accessing and viewing images from web url directly

   trelliscope(p, name="pokemon", nrow=3, ncol=6, state = list(labels = c("pokemon", "pokedex")))

2. accessing and viewing images stored on the local disk

   trelliscope(p, name="pokemon", nrow=3, ncol=6, state = list(labels = c("pokemon", "pokedex")), path=path)

Answer 2

Pokemon dataset is available online, which facilitates testing :

library(dplyr)
library(trelliscopejs)

pok <- read.csv("https://raw.githubusercontent.com/hafen/pokRdex/master/pokRdex_mod.csv") %>%
  mutate_at(vars(matches("_id$")), as.character)

pok %>%
  mutate(panel = img_panel(url_image)) %>%
  trelliscope("pokemon", nrow = 3, ncol = 6,
              state = list(labels = c("pokemon", "pokedex")))

Just go to Sort Tab in the sidebar and select a variable to sort on : the Pokemons appear ;-)

trelliscope Issues seem on hold since 2017, which might explain why this unexpected behaviour wasn't corrected.