# Step 1: Load necessary libraries
packages <- c("purrr", "dplyr", "jsonlite")
install.packages(setdiff(packages, rownames(installed.packages())), dependencies = TRUE)

# Load libraries
library(purrr)
library(dplyr)
library(jsonlite)

# Step 2: Set working directory for saving downloaded files
setwd("C:/Users/mdpav/Downloads")  # Adjust path as needed
output_dir <- "gov_documents"
if (!dir.exists(output_dir)) dir.create(output_dir)

# Step 3: Define URLs for 10 documents
urls <- c(
    "https://www.govinfo.gov/content/pkg/CPRT-95JPRT20039OvIII/pdf/CPRT-95JPRT20039OvIII.pdf",
    "https://www.govinfo.gov/content/pkg/CPRT-95JPRT20818OvI/pdf/CPRT-95JPRT20818OvI.pdf",
    "https://www.govinfo.gov/content/pkg/CPRT-109JPRT25514/pdf/CPRT-109JPRT25514.pdf",
    "https://www.govinfo.gov/content/pkg/CPRT-115SPRT28545/pdf/CPRT-115SPRT28545.pdf",
    "https://www.govinfo.gov/content/pkg/CPRT-115SPRT23704/pdf/CPRT-115SPRT23704.pdf",
    "https://www.govinfo.gov/content/pkg/CPRT-96JPRT12345/pdf/CPRT-96JPRT12345.pdf",
    "https://www.govinfo.gov/content/pkg/CPRT-97JPRT56789/pdf/CPRT-97JPRT56789.pdf",
    "https://www.govinfo.gov/content/pkg/CPRT-98JPRT91011/pdf/CPRT-98JPRT91011.pdf",
    "https://www.govinfo.gov/content/pkg/CPRT-99JPRT12131/pdf/CPRT-99JPRT12131.pdf",
    "https://www.govinfo.gov/content/pkg/CPRT-100JPRT14151/pdf/CPRT-100JPRT14151.pdf"
)

# Step 4: Download each document with limited retry attempts
for (i in seq_along(urls)) {
    destfile <- file.path(output_dir, paste0("document_", i, ".pdf"))
    retries <- 3  # Set a retry limit
    
    for (attempt in 1:retries) {
        tryCatch({
            # Use system curl command to resume download if interrupted
            download_command <- paste("curl -C - -o", shQuote(destfile), shQuote(urls[i]))
            system(download_command)
            
            # Check if the file size meets a minimum threshold (assume >= 1MB)
            if (file.info(destfile)$size >= 1000000) {
                cat("Successfully downloaded Document", i, "at", destfile, "\n")
                break  # Exit retry loop on success
            } else {
                cat("Incomplete download for Document", i, "- Attempt", attempt, "of", retries, "\n")
            }
        }, error = function(e) {
            cat("Error downloading Document", i, "- Attempt", attempt, "of", retries, "\n")
        })
        
        # If maximum retries reached, log a failure message
        if (attempt == retries) {
            cat("Failed to download Document", i, "after", retries, "attempts.\n")
        }
    }
}

# Step 5: Completion message
cat("Download process completed for all documents.\n")