Daily downloads trend for `R` packages published on `CRAN`

cran-downloads-trend.R

# A function to quickly grasp the `CRAN` daily downloads.
plot_cran_downloads_trend <- function(package, x_axis_frequency = "3 months") {
    # Packages required.
    packages <- c("cranlogs", "mgcv", "curl", "ggplot2", "scales")
    
    # Load required packages.
    sapply(packages, function(pkg) {
        # Require the package.
        status <- suppressPackageStartupMessages(require(pkg, character.only = TRUE))

        # If the package is not installed, install it.
        if (!status) {
            # Install the package.
            install.packages(pkg, dependencies = TRUE)

            # Load the package.
            suppressPackageStartupMessages(require(pkg, character.only = TRUE))
        }
    })
    
    # Specify the `CRAN` archive page.
    base_cran_archive_page <- "https://cran.r-project.org/src/contrib/Archive/"

    # Specify the `CRAN` package page.
    base_cran_package_page <- "https://CRAN.R-project.org/package="

    # Create the archive page.
    cran_archive_page <- paste0(base_cran_archive_page, package, "/")

    # Create the package page.
    cran_package_page <- paste0(base_cran_package_page, package)

    # Open a connect to the `CRAN` archive page for the current package.
    connection <- curl(cran_archive_page)

    # Try to get the version from the archive or package page.
    page_text <- tryCatch(
        expr = {
            # Extract archive page text.
            readLines(connection)
        },
        error = function(e) {
            # Close the connection.
            close(connection)

            # Use the package page in case of `404` `HTTP` status.
            connection <- curl(cran_package_page)

            # Extract package page text.
            readLines(connection)
        },

        # Close the connection.
        finally = {
            # Close.
            close(connection)
        }
    )

    # Extract the first date as the date the package was published on `CRAN`.
    publish_date <- as.Date(
        # Extract the date.
        regmatches(page_text, regexpr("\\d{4}-\\d{2}-\\d{2}", page_text))[1]
    )

    # Recall what day was yesterday (i.e., usually trackers lag by one day).
    yesterday <- as.Date(format(Sys.time(), "%Y-%m-%d")) - 1

    # Assume the package has been always on `CRAN` since first published.
    number_days <- as.numeric(yesterday - publish_date)

    # Fetch the daily downloads.
    downloads <- cran_downloads(package, from = publish_date, to = yesterday)

    # Compute mean daily downloads.
    mean_daily_downloads <- mean(downloads$count)

    # Estimate trend.
    model <- gam(count ~ s(as.numeric(date), bs = "cr"), data = downloads)

    # Plot.
    plot <- ggplot(downloads, aes(x = date, y = count)) +
        # Daily downloads.
        geom_point(
            color = "gray",
            shape = 19,
            alpha = 0.8,
            size = 2,
            stroke = NA
        ) +
        
        # The download trend.
        geom_line(
            aes(
                y = fitted(model)
            ),
            color = "darkred",
            linewidth = 1
        ) +
            
        # The mean daily downloads.
        geom_hline(
            yintercept = mean_daily_downloads,
            color = "black",
            alpha = 0.8,
            linetype = "dashed"
        ) +
            
        # The download dates.
        scale_x_date(
            date_breaks = x_axis_frequency,
            date_labels = "%b %Y"
        ) +
        
        # The actual download values.
        scale_y_continuous(
            labels = comma,
            breaks = seq(0, max(downloads$count), by = mean_daily_downloads)
        ) +
            
        # The plot text.
        labs(
            title = paste0("Daily downloads of '", package, "' package on CRAN"),
            subtitle = paste0(
                "First published on ", publish_date, " (", scales::comma(number_days), " days ago) | ",
                "Total downloads: ", scales::comma(sum(downloads$count)), " | ",
                "Average daily downloads: ", scales::comma(round(mean_daily_downloads)), " | ",
                format(Sys.time(), "Plotted on %Y-%m-%d at %H:%M:%S")
            ),
            x = "Date",
            y = "Number of Downloads"
        ) +
            
        # Theme options
        theme_bw() +
        theme(
            axis.text.x = element_text(angle = 45, hjust = 1),
        )
    
    # Plot the plot.
    plot(plot)

    # Return the plot silently.
    invisible(plot)
}

# Plot the daily downloads of the `ggplot2` package.
plot_cran_downloads_trend("ggplot2", x_axis_frequency = "1 year")

# Plot the daily downloads of the `Rcpp` package.
plot_cran_downloads_trend("Rcpp", x_axis_frequency = "1 year")

# Plot the daily downloads of the `lme4` package.
plot_cran_downloads_trend("lme4", x_axis_frequency = "1 year")

# Plot the daily downloads of the `lavaan` package.
plot_cran_downloads_trend("lavaan", x_axis_frequency = "5 months")

Example of a resulting plot (i.e., for lavaan in this case):