In giocomai/edjnetquotefinder: EDJNet's QuoteFinder

knitr::opts_chunk$set(echo = FALSE)

library("dplyr")
library("reshape2")
library("stringr")
library("tidytext")
library("rtweet")
library("wordcloud")
library("RColorBrewer")
library("webshot")
library("htmlwidgets")
library("wordcloud2")
library("purrr")

library("extrafont")
extrafont::loadfonts()
# library("devtools")
# devtools::install_github("mkearney/rtweet")
# library("rtweet")

dir.create(path = "img",showWarnings = FALSE)


bluesFunc <- colorRampPalette(colors = c("#4292C6", "#08306B"))
blues <- bluesFunc(5)

twitter_token <- readRDS(file = "twitter_token.rds")

dataFolder <- "/home/qf/qf_data/tweets_processed/"

dataset <- readRDS(file = file.path(dataFolder, "dataset.rds"))
hashtags <- readRDS(file = file.path(dataFolder, "tweets_hashtags_list.rds"))
trendingHashtags <- readRDS(file = file.path(dataFolder, "tweets_trending_hashtags_list.rds"))

Pick a tweet type randomly

RandomTweetType <-
  tibble::tribble(~id, ~TweetType,
                  1, "sentiment wordcloud first trending english hashtag",
                  2, "sentiment wordcloud second trending english hashtag",
                  3, "sentiment wordcloud third trending english hashtag",
                  4, "sentiment wordcloud all tweets en",
                  5, "wc2 all english 1 month",
                  6, "wc2 all german 1 month",
                  7, "wc2 all italian 1 month",
                  8, "wc2 all french 1 month",
                  9, "wc2 all polish 1 month",
                  10, "wc2 all spanish 1 month",
                  11, "random top english hashtag",
                  12, "random top english hashtag",
                  13, "random top italian hashtag",
                  14, "random top german hashtag",
                  15, "random top french hashtag",

  )

selected <- sample(x = 1:nrow(RandomTweetType), size = 1)

RandomTweetType %>% dplyr::slice(selected)

if (selected == 1) {

filename <- file.path("img", paste(Sys.time(), "sentiment wordcloud.png"))

png(filename = filename, width=1024,height=1024, res = 150)
par(mar = rep(0, 4))
dataset %>% 
  filter(date>Sys.Date()-31) %>% 
  filter(lang=="en") %>% 
  filter(purrr::map_lgl(.x = hashtags, .f = function (x) is.element(el = as.character(stringr::str_remove(string = trendingHashtags$en[1], pattern = "#")), set = x))) %>%
  select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  anti_join(stop_words, by = "word") %>% 
  inner_join(get_sentiments("bing"), by = "word") %>%
  count(word, sentiment, sort = TRUE) %>%
  acast(word ~ sentiment, value.var = "n", fill = 0) %>%
  comparison.cloud(colors = c("#F8766D", "#00BFC4"),
                   max.words = 100, scale = c(2.5, 1),
                   family = "Roboto", font = 1)
dev.off()

# Post tweet
post_tweet(status = paste0("What did MEPs have to say good or bad about", paste0(" #", trendingHashtags$en[1]), "? Here's a sentiment analysis wordcloud based on tweets posted in the last month. Explore the tweets at https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)


}
dataset

if (selected == 2) {
filename <- file.path("img", paste(Sys.time(), "sentiment wordcloud.png"))

png(filename = filename, width=1024,height=1024, res = 150)
par(mar = rep(0, 4))
dataset %>% 
  filter(date>Sys.Date()-31) %>% 
  filter(lang=="en") %>% 
  filter(purrr::map_lgl(.x = hashtags, .f = function (x) is.element(el = as.character(stringr::str_remove(string = trendingHashtags$en[2], pattern = "#")), set = x))) %>%
  select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  anti_join(stop_words, by = "word") %>% 
  inner_join(get_sentiments("bing"), by = "word") %>%
  count(word, sentiment, sort = TRUE) %>%
  acast(word ~ sentiment, value.var = "n", fill = 0) %>%
  comparison.cloud(colors = c("#F8766D", "#00BFC4"),
                   max.words = 100, scale = c(2.5, 1),
                   family = "Roboto", font = 1)
dev.off()

# Post tweet
post_tweet(status = paste0("What did MEPs have to say good or bad abou t", trendingHashtags$en[2], "? Here's a sentiment analysis wordcloud based on tweets posted in the last month. Explore the tweets at https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)
}

if (selected == 3) {
filename <- file.path("img", paste(Sys.time(), "sentiment wordcloud.png"))

png(filename = filename, width=1024,height=1024, res = 150)
par(mar = rep(0, 4))
dataset %>% 
  filter(date>Sys.Date()-31) %>% 
  filter(lang=="en") %>% 
  filter(purrr::map_lgl(.x = hashtags, .f = function (x) is.element(el = as.character(stringr::str_remove(string = trendingHashtags$en[3], pattern = "#")), set = x))) %>%
  select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  anti_join(stop_words, by = "word") %>% 
  inner_join(get_sentiments("bing"), by = "word") %>%
  count(word, sentiment, sort = TRUE) %>%
  acast(word ~ sentiment, value.var = "n", fill = 0) %>%
  comparison.cloud(colors = c("#F8766D", "#00BFC4"),
                   max.words = 100, scale = c(2.5, 1),
                   family = "Roboto", font = 1)
dev.off()

# Post tweet
post_tweet(status = paste0("What did MEPs have to say good or bad about", trendingHashtags$en[3], "? Here's a sentiment analysis wordcloud based on tweets posted in the last month. Explore the tweets at https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)
}

if (selected == 4) {
filename <- file.path("img", paste(Sys.time(), "sentiment wordcloud.png"))

png(filename = filename, width=1024,height=1024, res = 150)
par(mar = rep(0, 4))
dataset %>% 
  filter(date>Sys.Date()-31) %>% 
  filter(lang=="en") %>%
  select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  anti_join(stop_words, by = "word") %>% 
  inner_join(get_sentiments("bing"), by = "word") %>%
  count(word, sentiment, sort = TRUE) %>%
  acast(word ~ sentiment, value.var = "n", fill = 0) %>%
  comparison.cloud(colors = c("#F8766D", "#00BFC4"),
                   max.words = 100, scale = c(2.5, 1),
                   family = "Roboto", font = 1)
dev.off()



post_tweet(status = paste0("What are the most frequent positive and negative keywords used in their tweets by MEPs in the last month? Explore the tweets at https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)

}

if (selected == 5) {
dataset <- dataset %>% 
  filter(lang=="en") %>% 
    select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  # remove stopwords, if list for the relevant language is available, otherwise do nothing
  when(is.element(el = "en", set = stopwords::stopwords_getlanguages(source = "stopwords-iso")) ~
         anti_join(., tibble::tibble(word = stopwords::stopwords(language = "en", source = "stopwords-iso")), by = "word"),
       ~ .) %>% 
count(word, sort = TRUE)  
# customise output color, gradients of blue by frequency
colour <- dataset %>% 
  mutate(colour = case_when(
    n <= quantile(n)[1] ~ blues[1],
    n > quantile(n)[1]& n<=quantile(n)[2] ~ blues[2],
    n > quantile(n)[2]& n<=quantile(n)[3] ~ blues[3],
    n > quantile(n)[3]& n<=quantile(n)[4] ~ blues[4],
    n > quantile(n)[4]& n<=quantile(n)[5] ~ blues[4]
  )) %>% pull(colour)
#sizeVar <- as.numeric(quantile(dataset$n)[5]/quantile(dataset$n)[1]/nrow(dataset)*5)
# graph <- dataset %>% 
#   wordcloud2(size = sizeVar, color = colour)
graph <- dataset %>% wordcloud2(color = colour, size = 1.2, shape = "star")
filename <- file.path("img", paste(Sys.time(), "star_wordcloud2_1month EN.png", sep = "-"))
saveWidget(graph,"tmp.html",selfcontained = FALSE)
webshot("tmp.html", filename, delay = 20, vwidth = 2048, vheight = 1512)

post_tweet(status = paste0("One months of MEPs on Twitter, in a star. Explore the tweets also in other languages at https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)

}

if (selected == 6) {
dataset <- dataset %>% 
  filter(lang=="de") %>% 
    select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  # remove stopwords, if list for the relevant language is available, otherwise do nothing
  when(is.element(el = "de", set = stopwords::stopwords_getlanguages(source = "stopwords-iso")) ~
         anti_join(., tibble::tibble(word = stopwords::stopwords(language = "de", source = "stopwords-iso")), by = "word"),
       ~ .) %>% 
count(word, sort = TRUE)  
# customise output color, gradients of blue by frequency
colour <- dataset %>% 
  mutate(colour = case_when(
    n <= quantile(n)[1] ~ blues[1],
    n > quantile(n)[1]& n<=quantile(n)[2] ~ blues[2],
    n > quantile(n)[2]& n<=quantile(n)[3] ~ blues[3],
    n > quantile(n)[3]& n<=quantile(n)[4] ~ blues[4],
    n > quantile(n)[4]& n<=quantile(n)[5] ~ blues[4]
  )) %>% pull(colour)
#sizeVar <- as.numeric(quantile(dataset$n)[5]/quantile(dataset$n)[1]/nrow(dataset)*5)
# graph <- dataset %>% 
#   wordcloud2(size = sizeVar, color = colour)
graph <- dataset %>% wordcloud2(color = colour, size = 1.2, shape = "star")
filename <- file.path("img", paste(Sys.time(), "star_wordcloud2_1month DE.png", sep = "-"))
saveWidget(graph,"tmp.html",selfcontained = FALSE)
webshot("tmp.html", filename, delay = 20, vwidth = 2048, vheight = 1512)

post_tweet(status = paste0("One month of MEPs posting in German on Twitter, in a star. Explore the tweets also in other languages at https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)

}

if (selected == 7) {
dataset <- dataset %>% 
  filter(lang=="it") %>% 
    select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  # remove stopwords, if list for the relevant language is available, otherwise do nothing
  when(is.element(el = "it", set = stopwords::stopwords_getlanguages(source = "stopwords-iso")) ~
         anti_join(., tibble::tibble(word = stopwords::stopwords(language = "it", source = "stopwords-iso")), by = "word"),
       ~ .) %>% 
count(word, sort = TRUE)  
# customise output color, gradients of blue by frequency
colour <- dataset %>% 
  mutate(colour = case_when(
    n <= quantile(n)[1] ~ blues[1],
    n > quantile(n)[1]& n<=quantile(n)[2] ~ blues[2],
    n > quantile(n)[2]& n<=quantile(n)[3] ~ blues[3],
    n > quantile(n)[3]& n<=quantile(n)[4] ~ blues[4],
    n > quantile(n)[4]& n<=quantile(n)[5] ~ blues[4]
  )) %>% pull(colour)
#sizeVar <- as.numeric(quantile(dataset$n)[5]/quantile(dataset$n)[1]/nrow(dataset)*5)
# graph <- dataset %>% 
#   wordcloud2(size = sizeVar, color = colour)
graph <- dataset %>% wordcloud2(color = colour, size = 1.1, shape = "star")
filename <- file.path("img", paste(Sys.time(), "star_wordcloud2_1month IT.png", sep = "-"))
saveWidget(graph,"tmp.html",selfcontained = FALSE)
webshot("tmp.html", filename, delay = 20, vwidth = 2048, vheight = 1512)

post_tweet(status = paste0("What did MEPs tweet in Italian in the last month? Here's a summary in a star. Find out more: https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)

}

if (selected == 8) {
dataset <- dataset %>% 
  filter(lang=="fr") %>% 
    select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  # remove stopwords, if list for the relevant language is available, otherwise do nothing
  when(is.element(el = "fr", set = stopwords::stopwords_getlanguages(source = "stopwords-iso")) ~
         anti_join(., tibble::tibble(word = stopwords::stopwords(language = "fr", source = "stopwords-iso")), by = "word"),
       ~ .) %>% 
count(word, sort = TRUE)  
# customise output color, gradients of blue by frequency
colour <- dataset %>% 
  mutate(colour = case_when(
    n <= quantile(n)[1] ~ blues[1],
    n > quantile(n)[1]& n<=quantile(n)[2] ~ blues[2],
    n > quantile(n)[2]& n<=quantile(n)[3] ~ blues[3],
    n > quantile(n)[3]& n<=quantile(n)[4] ~ blues[4],
    n > quantile(n)[4]& n<=quantile(n)[5] ~ blues[4]
  )) %>% pull(colour)
#sizeVar <- as.numeric(quantile(dataset$n)[5]/quantile(dataset$n)[1]/nrow(dataset)*5)
# graph <- dataset %>% 
#   wordcloud2(size = sizeVar, color = colour)
graph <- dataset %>% wordcloud2(color = colour, size = 1.2, shape = "star")
filename <- file.path("img", paste(Sys.time(), "star_wordcloud2_1month FR.png", sep = "-"))
saveWidget(graph,"tmp.html",selfcontained = FALSE)
webshot("tmp.html", filename, delay = 20, vwidth = 2048, vheight = 1512)

post_tweet(status = paste0("What did MEPs tweet in French in the last month? Here's a summary in a star. Find out more: https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)

}

if (selected == 9) {
dataset <- dataset %>% 
  filter(lang=="pl") %>% 
    select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  # remove stopwords, if list for the relevant language is available, otherwise do nothing
  when(is.element(el = "pl", set = stopwords::stopwords_getlanguages(source = "stopwords-iso")) ~
         anti_join(., tibble::tibble(word = stopwords::stopwords(language = "pl", source = "stopwords-iso")), by = "word"),
       ~ .) %>% 
count(word, sort = TRUE)  
# customise output color, gradients of blue by frequency
colour <- dataset %>% 
  mutate(colour = case_when(
    n <= quantile(n)[1] ~ blues[1],
    n > quantile(n)[1]& n<=quantile(n)[2] ~ blues[2],
    n > quantile(n)[2]& n<=quantile(n)[3] ~ blues[3],
    n > quantile(n)[3]& n<=quantile(n)[4] ~ blues[4],
    n > quantile(n)[4]& n<=quantile(n)[5] ~ blues[4]
  )) %>% pull(colour)
#sizeVar <- as.numeric(quantile(dataset$n)[5]/quantile(dataset$n)[1]/nrow(dataset)*5)
# graph <- dataset %>% 
#   wordcloud2(size = sizeVar, color = colour)
graph <- dataset %>% wordcloud2(color = colour, size = 1.2, shape = "star")
filename <- file.path("img", paste(Sys.time(), "star_wordcloud2_1month PL.png", sep = "-"))
saveWidget(graph,"tmp.html",selfcontained = FALSE)
webshot("tmp.html", filename, delay = 20, vwidth = 2048, vheight = 1512)

post_tweet(status = paste0("What did MEPs tweet in Polish in the last month? Here's a summary in a star. Find out more: https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)

}

if (selected == 10) {
dataset <- dataset %>% 
  filter(lang=="es") %>% 
    select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  # remove stopwords, if list for the relevant language is available, otherwise do nothing
  when(is.element(el = "es", set = stopwords::stopwords_getlanguages(source = "stopwords-iso")) ~
         anti_join(., tibble::tibble(word = stopwords::stopwords(language = "es", source = "stopwords-iso")), by = "word"),
       ~ .) %>% 
count(word, sort = TRUE)  
# customise output color, gradients of blue by frequency
colour <- dataset %>% 
  mutate(colour = case_when(
    n <= quantile(n)[1] ~ blues[1],
    n > quantile(n)[1]& n<=quantile(n)[2] ~ blues[2],
    n > quantile(n)[2]& n<=quantile(n)[3] ~ blues[3],
    n > quantile(n)[3]& n<=quantile(n)[4] ~ blues[4],
    n > quantile(n)[4]& n<=quantile(n)[5] ~ blues[4]
  )) %>% pull(colour)
#sizeVar <- as.numeric(quantile(dataset$n)[5]/quantile(dataset$n)[1]/nrow(dataset)*5)
# graph <- dataset %>% 
#   wordcloud2(size = sizeVar, color = colour)
graph <- dataset %>% wordcloud2(color = colour, size = 1.2, shape = "star")
filename <- file.path("img", paste(Sys.time(), "star_wordcloud2_1month ES.png", sep = "-"))
saveWidget(graph,"tmp.html",selfcontained = FALSE)
webshot("tmp.html", filename, delay = 20, vwidth = 2048, vheight = 1512)

post_tweet(status = paste0("What did MEPs tweet in Spanish in the last month? Here's a summary in a star. Find out more: https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)

}

if (selected == 11|selected == 12) {
randomTopHashtag <- unlist(sample(head(trendingHashtags$en, 3), 1))
NrOfTweets <- str_extract(string = randomTopHashtag, pattern = "([[:digit:]]+)")
hashtag <- str_extract(string = randomTopHashtag, pattern = "([[:alnum:]]+)")
dataset <- dataset %>% 
  filter(lang=="en") %>% 
  filter(purrr::map_lgl(.x = hashtags, .f = function (x) is.element(el = as.character(hashtag), set = x))) %>%
    select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  # remove stopwords, if list for the relevant language is available, otherwise do nothing
  when(is.element(el = "en", set = stopwords::stopwords_getlanguages(source = "stopwords-iso")) ~
         anti_join(., data_frame(word = stopwords::stopwords(language = "en", source = "stopwords-iso")), by = "word"),
       ~ .) %>% 
count(word, sort = TRUE)  
# customise output color, gradients of blue by frequency
colour <- dataset %>% 
  mutate(colour = case_when(
    n <= quantile(n)[1] ~ blues[1],
    n > quantile(n)[1]& n<=quantile(n)[2] ~ blues[2],
    n > quantile(n)[2]& n<=quantile(n)[3] ~ blues[3],
    n > quantile(n)[3]& n<=quantile(n)[4] ~ blues[4],
    n > quantile(n)[4]& n<=quantile(n)[5] ~ blues[4]
  )) %>% pull(colour)
sizeVar <- as.numeric(quantile(dataset$n)[5]/quantile(dataset$n)[1]/nrow(dataset)*44)

sizeVar <- min(1.8, sizeVar)

graph <- dataset %>%
  wordcloud2(size = sizeVar, color = colour)
graph <- dataset %>% wordcloud2(color = colour, size = sizeVar)
filename <- file.path("img", paste(Sys.time(), "wordcloud2 1month hashtag", hashtag, "EN.png", sep = "-"))
saveWidget(graph,"tmp.html",selfcontained = FALSE)
webshot("tmp.html", filename, delay = 20, vwidth = 2048, vheight = 1512)
randomTopHashtag

# Post tweet
post_tweet(status = paste0("Here's one of the trending hashtags among MEPs tweeting in English this week: #", hashtag, ". Find out more at https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)

}

if (selected == 13) {
randomTopHashtag <- unlist(sample(head(trendingHashtags$it, 3), 1))
NrOfTweets <- str_extract(string = randomTopHashtag, pattern = "([[:digit:]]+)")
hashtag <- str_extract(string = randomTopHashtag, pattern = "([[:alnum:]]+)")
dataset <- dataset %>% 
  filter(lang=="it") %>% 
  filter(purrr::map_lgl(.x = hashtags, .f = function (x) is.element(el = as.character(hashtag), set = x))) %>%
    select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  # remove stopwords, if list for the relevant language is available, otherwise do nothing
  when(is.element(el = "it", set = stopwords::stopwords_getlanguages(source = "stopwords-iso")) ~
         anti_join(., data_frame(word = stopwords::stopwords(language = "it", source = "stopwords-iso")), by = "word"),
       ~ .) %>% 
count(word, sort = TRUE)  
# customise output color, gradients of blue by frequency
colour <- dataset %>% 
  mutate(colour = case_when(
    n <= quantile(n)[1] ~ blues[1],
    n > quantile(n)[1]& n<=quantile(n)[2] ~ blues[2],
    n > quantile(n)[2]& n<=quantile(n)[3] ~ blues[3],
    n > quantile(n)[3]& n<=quantile(n)[4] ~ blues[4],
    n > quantile(n)[4]& n<=quantile(n)[5] ~ blues[4]
  )) %>% pull(colour)
sizeVar <- as.numeric(quantile(dataset$n)[5]/quantile(dataset$n)[1]/nrow(dataset)*44)

sizeVar <- min(1.8, sizeVar)

graph <- dataset %>%
  wordcloud2(size = sizeVar, color = colour)
graph <- dataset %>% wordcloud2(color = colour, size = sizeVar)
filename <- file.path("img", paste(Sys.time(), "wordcloud2 1month hashtag", hashtag, "it.png", sep = "-"))
saveWidget(graph,"tmp.html",selfcontained = FALSE)
webshot("tmp.html", filename, delay = 20, vwidth = 2048, vheight = 1512)
randomTopHashtag

# Post tweet
post_tweet(status = paste0("Here's one of the trending hashtags among MEPs tweeting in Italian this week: #", hashtag, ". Find out more at https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)

}

if (selected == 14) {
randomTopHashtag <- unlist(sample(head(trendingHashtags$de, 3), 1))
NrOfTweets <- str_extract(string = randomTopHashtag, pattern = "([[:digit:]]+)")
hashtag <- str_extract(string = randomTopHashtag, pattern = "([[:alnum:]]+)")
dataset <- dataset %>% 
  filter(lang=="de") %>% 
  filter(purrr::map_lgl(.x = hashtags, .f = function (x) is.element(el = as.character(hashtag), set = x))) %>%
    select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  # remove stopwords, if list for the relevant language is available, otherwise do nothing
  when(is.element(el = "de", set = stopwords::stopwords_getlanguages(source = "stopwords-iso")) ~
         anti_join(., data_frame(word = stopwords::stopwords(language = "de", source = "stopwords-iso")), by = "word"),
       ~ .) %>% 
count(word, sort = TRUE)  
# customise output color, gradients of blue by frequency
colour <- dataset %>% 
  mutate(colour = case_when(
    n <= quantile(n)[1] ~ blues[1],
    n > quantile(n)[1]& n<=quantile(n)[2] ~ blues[2],
    n > quantile(n)[2]& n<=quantile(n)[3] ~ blues[3],
    n > quantile(n)[3]& n<=quantile(n)[4] ~ blues[4],
    n > quantile(n)[4]& n<=quantile(n)[5] ~ blues[4]
  )) %>% pull(colour)
sizeVar <- as.numeric(quantile(dataset$n)[5]/quantile(dataset$n)[1]/nrow(dataset)*44)

sizeVar <- min(1.8, sizeVar)

graph <- dataset %>%
  wordcloud2(size = sizeVar, color = colour)
graph <- dataset %>% wordcloud2(color = colour, size = sizeVar)
filename <- file.path("img", paste(Sys.time(), "wordcloud2 1month hashtag", hashtag, "de.png", sep = "-"))
saveWidget(graph,"tmp.html",selfcontained = FALSE)
webshot("tmp.html", filename, delay = 20, vwidth = 2048, vheight = 1512)
randomTopHashtag

# Post tweet
post_tweet(status = paste0("Here's one of the trending hashtags among MEPs tweeting in German this week: #", hashtag, ". Find out more at https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)

}

if (selected == 15) {
randomTopHashtag <- unlist(sample(head(trendingHashtags$fr, 3), 1))
NrOfTweets <- str_extract(string = randomTopHashtag, pattern = "([[:digit:]]+)")
hashtag <- str_extract(string = randomTopHashtag, pattern = "([[:alnum:]]+)")
dataset <- dataset %>% 
  filter(lang=="fr") %>% 
  filter(purrr::map_lgl(.x = hashtags, .f = function (x) is.element(el = as.character(hashtag), set = x))) %>%
    select(clean_text) %>% 
  unnest_tokens(input = clean_text, output = word) %>% 
  # remove stopwords, if list for the relevant language is available, otherwise do nothing
  when(is.element(el = "fr", set = stopwords::stopwords_getlanguages(source = "stopwords-iso")) ~
         anti_join(., data_frame(word = stopwords::stopwords(language = "fr", source = "stopwords-iso")), by = "word"),
       ~ .) %>% 
count(word, sort = TRUE)  
# customise output color, gradients of blue by frequency
colour <- dataset %>% 
  mutate(colour = case_when(
    n <= quantile(n)[1] ~ blues[1],
    n > quantile(n)[1]& n<=quantile(n)[2] ~ blues[2],
    n > quantile(n)[2]& n<=quantile(n)[3] ~ blues[3],
    n > quantile(n)[3]& n<=quantile(n)[4] ~ blues[4],
    n > quantile(n)[4]& n<=quantile(n)[5] ~ blues[4]
  )) %>% pull(colour)
sizeVar <- as.numeric(quantile(dataset$n)[5]/quantile(dataset$n)[1]/nrow(dataset)*44)

sizeVar <- min(1.8, sizeVar)

graph <- dataset %>%
  wordcloud2(size = sizeVar, color = colour)
graph <- dataset %>% wordcloud2(color = colour, size = sizeVar)
filename <- file.path("img", paste(Sys.time(), "wordcloud2 1month hashtag", hashtag, "fr.png", sep = "-"))
saveWidget(graph,"tmp.html",selfcontained = FALSE)
webshot("tmp.html", filename, delay = 20, vwidth = 2048, vheight = 1512)
randomTopHashtag

# Post tweet
post_tweet(status = paste0("Here's one of the trending hashtags among MEPs tweeting in French this week: #", hashtag, ". Find out more at https://applications.europeandatajournalism.eu/quotefinder/"), media = filename, token = twitter_token)

}