google_books_count pairs

pairs.R

library(foreach)
library(doSNOW)
library(dplyr)
library(httr)

google_books_count = function (search_terms) {
    require(httr)
    terms = str_c(search_terms, collapse = '+')
    query_url = str_c('https://www.googleapis.com/books/v1/volumes', 
                      '?', 
                      'q=', terms, 
                      '&project=lite&langRestrict=en&maxResults=1', 
                      '&key=', api_key)
    response = GET(query_url)
    suppressMessages(response_list <- parsed_content(response))
    log_f = log(response_list$totalItems)
    return(log_f)
}

load('pairs-nodist.Rdata')

## Register parallel cluster
cl = makeCluster(parallel::detectCores())
# registerDoParallel(cl)
registerDoSNOW(cl)
## Define progress bar
pb = txtProgressBar(max = nrow(pairs), style = 3)
progress = function(n) setTxtProgressBar(pb, n)
system.time(
    pairs <- foreach(pair = iter(pairs, by = 'row'), .combine = bind_rows, 
                     .packages = c('dplyr', 'stringr', 'xml2'), 
                     .options.snow = list(progress = progress)
    ) %dopar% {
        x = pair$Var1
        y = pair$Var2
        f_xy = google_books_count(c(x, y))
        # Sys.sleep(.5)
        tibble(Var1 = x, Var2 = y, f_xy)
    }
)

save(pairs, file = 'pairs-dist.Rdata')