R and phantom.js: Can't scrape main content block - r

Question

I'm trying to scrape the topline data points (Total GHG, GHG per capita, GHG per BTU) and download the charts from the following page, using R and phantom.js:
http://apps1.eere.energy.gov/sled/#/results/home?city=Omaha&abv=NE
This is my code:
url <- "http://apps1.eere.energy.gov/sled/#/results/home?city=Omaha&abv=NE"
writeLines(sprintf("var page = require('webpage').create();
page.open('%s', function () {
console.log(page.content); //page source
phantom.exit();
});", url), con="scrape.js")
system("phantomjs scrape.js > eere.html")
pg <- read_html("eere.html")
pg %>% html_nodes("CLASS") %>% html_text()
Where CLASS is the class that I haven't yet identified. Here's the html I'm getting for eere.html. It only contains header and footer content, and fails to grab the body of the page.
Any advice?

Answer 1

With PhantomJS via RSelenium, it would look like
library(RSelenium)
library(rvest)
# start remote driver and browser
remdr <- rsDriver(browser = 'phantomjs', verbose = FALSE)
remdr$client$navigate('http://apps1.eere.energy.gov/sled/#/results/home?city=Omaha&abv=NE')
Sys.sleep(2) # wait 2 secs for page to load if you're not running line by line
page <- remdr$client$getPageSource()
# parse HTML with rvest
page[[1]] %>% read_html() %>% html_nodes('h5') %>% html_text()
#> [1] "Total GHG: metric tons"
#> [2] "GHG per capita: metric tons/person"
#> [3] "GHG per BTU: metric tons/MMBTU"
# clean up
remdr$client$close()
remdr$server$stop()
#> [1] TRUE
RSelenium uses an OOP style that's uncommon in R, but workable. Consequently the docs are arranged in a similarly unusual manner, but they are actually thorough if you dig in.
Lastly, RSelenium is best avoided if you don't need it. It's an important and necessary tool in the R toolbox, but because of what it does it's inherently heavy and slow compared to the rest of R. Given the site offers good data download options and how to recreate the data in question, it's ultimately unnecessary here. For a couple pages it may be practical, but for more there's a point at which it's quicker to just rebuild the data.