Search US Library of Congress images with R

search_loc_images.R

# Uses httr and jsonlite packages and optional DT to display results

library(httr)
library(jsonlite)

search_loc_images <- function(query, max_results = 10) {


  # Base URL
  base_url <- "https://www.loc.gov/photos/"

  # Query parameters
  params <- list(
    q = query,
    fo = "json",
    rights = "public",
    c = max_results
  )

  # Send GET request
  response <- GET(url = base_url, query = params)

  # Check status code
  if (status_code(response) == 200) {
    # Parse JSON content
    data <- content(response, as = "parsed", type = "application/json")

    # Check if 'results' exist and has length > 0
    if (!is.null(data$results) && length(data$results) > 0) {
      # Extract relevant info
      results_list <- lapply(data$results, function(item) {
        list(
          Title       = if (!is.null(item$title)) item$title else "N/A",
          Date        = if (!is.null(item$date)) item$date else "N/A",
          Description = if (!is.null(item$description) && length(item$description) > 0)
            item$description[[1]]
          else
            "N/A",
          `Image URL` = if (!is.null(item$image_url) && length(item$image_url) > 0)
            item$image_url[[1]]
          else
            "N/A",
          `LOC URL`   = if (!is.null(item$url)) item$url else "N/A"
        )
      })

      # Convert list of lists to a data frame
      df <- do.call(rbind.data.frame, results_list)
      return(df)
    } else {
      message("No results found.")
      return(NULL)
    }
  } else {
    message(paste("Request failed with status code", status_code(response)))
    return(NULL)
  }
}

# Example search:
results_df <- search_loc_images("New York City skyline", max_results = 10)



if (!is.null(results_df)) {
  print(results_df)
  datatable(results_df, escape = TRUE, options = list(pageLength = 10), filter = 'top')
} else {
  cat("No data returned.\n")
}

# If you want a nice searchable table
library(dplyr)
library(DT)

results_df_for_table <- results_df |>
  mutate(
    Thumbnail = paste0('<img src="', Image.URL, '" width="100" height="100">'),
    Thumbnail = paste0('<a href="', LOC.URL, '" target="_blank">', Thumbnail, '</a>'),
    Title = paste0('<a href="', LOC.URL, '" target="_blank">', Title, '</a>')
  ) |>
  select(Title, Thumbnail, Date, Description )

if (!is.null(results_df)) {
  datatable(results_df_for_table, escape = FALSE, options = list(pageLength = 10), filter = 'top')
} else {
  cat("No data returned.\n")
}