Fast Slide Windows

slide_windows.R

slide_windows <- function(tbl, doc_var, window_size) {
  tbl %>%
    group_by(!!doc_var) %>%
    mutate(WordId = row_number() - 1,
           RowCount = n()) %>%
    ungroup() %>%
    crossing(InWindowIndex = 0:(window_size-1)) %>%
    filter((WordId - InWindowIndex) >= 0, # starting position of a window must be after the beginning of the document
           (WordId - InWindowIndex + window_size - 1) < RowCount # ending position of a window must be before the end of the document
     ) %>%
    mutate(window_id = WordId - InWindowIndex + 1)
}