teste-vector

download.file("https://snap.stanford.edu/data/finefoods.txt.gz", "finefoods.txt.gz")


library(readr)
library(stringr)
library(purrr)
reviews <- read_lines("finefoods.txt.gz") 
reviews <- reviews[str_sub(reviews, 1, 12) == "review/text:"]
reviews <- str_sub(reviews, start = 14)

library(tokenizers)
library(forcats)

reviews <- tokenize_words(reviews)

vocab <- reviews %>% 
  unlist() %>% 
  table() %>%
  sort(decreasing = TRUE) %>%
  names() %>%
  head(49999)
  
reviews_int <- map(
  reviews,
  ~match(.x, vocab) %>% coalesce(0L)
)