veraapi:

# update

hello <- function() {
  print("Hello, world!")
}

predictLink <- function(link, cutoff=0.5){

  # Pre-processing a new link
  root_path <- getwd()
  model_path <- paste0(root_path, "/data/links_rf_model.RData")

  print(dir(paste0(root_path, "/data")))
  print(model_path)

  load(model_path)
  print(my_model)

  link_cleaned <- gsub("https://", "", link)
  link_cleaned <- gsub("/", " ", link_cleaned)
  link_cleaned <- gsub("-", " ", link_cleaned)

  # Step 1 - Create a corpus text
  corpus = tm::Corpus(tm::VectorSource(link_cleaned))

  ## Step 2 - Conversion to Lowercase
  corpus = tm::tm_map(corpus, tm::PlainTextDocument)
  corpus = tm::tm_map(corpus, tolower)

  #Step 3 - Removing Punctuation
  corpus = tm::tm_map(corpus, tm::removePunctuation)

  #Step 4 - Removing Stopwords and other words
  corpus = tm::tm_map(corpus, tm::removeWords, c("http", tm::stopwords("portuguese")))

  corpus = tm::tm_map(corpus, tm::removeNumbers)

  # Step 5 - Stemming: reducing the number of inflectional forms of words
  corpus = tm::tm_map(corpus, tm::stemDocument)

  # Step 6 - Create Document Term Matrix
  frequencies = tm::DocumentTermMatrix(corpus)
  #sparse = removeSparseTerms(frequencies, 0.995) #remove sparse terms
  tSparse_nonamed = as.data.frame(as.matrix(frequencies)) #convert into data frame

  tSparse <- tSparse_nonamed
  colnames(tSparse) = make.names(colnames(tSparse_nonamed)) #all the variable names R-friendly

  #present_vars <- my_model$finalModel$xNames %in% colnames(tSparse)
  #missing_vars <- !present_vars
  #missing_vars <- my_model$finalModel$xNames[missing_vars]

  missing_vars <- setdiff(my_model$finalModel$xNames, colnames(tSparse))
  extra_vars_matrix <- matrix(0, ncol = length(missing_vars), nrow = nrow(tSparse))
  colnames(extra_vars_matrix) <- missing_vars

  new_examples_matrix <- cbind(tSparse, extra_vars_matrix)

  # Predicting link

  predictions <- predict(my_model, new_examples_matrix, type = "prob")
  predictions_yes <- predictions[, 2]
  papers_predicted <- ifelse(predictions_yes >= cutoff, "True", "Fake")
  papers_predicted_df <- data.frame(Score = predictions_yes, Prediction = papers_predicted)
  papers_predicted_df

}