Lat-lon coordinates from Google Takeout for Saved Places

# Extract lat-lon coordinates from Google Maps URL
# John L. Godlee (johngodlee@gmail.com)
# Last updated: 2026-04-26

# Packages
library(httr2)
library(googleway)

# Call 'pass' to get API key
api_key <- trimws(
  system2("pass", args = c("show", "google_geocoding_api_key"), 
          stdout = TRUE))

# Check if key retrieved successfully 
if (length(api_key) == 0 || api_key == "") {
  stop("Failed to retrieve API key from pass.")
}

# Set API key
set_key(api_key)

# Define function
geocode_url <- function(url) {
  
  # Expand URL 
  resp <- tryCatch({
    req <- request(url)
    req_perform(req)
  }, error = function(e) {
    message("Failed to resolve URL (Dead link/Timeout): ", url)
    return(NULL)
  })
  
  # Standardise failure output 
  fail_out <- data.frame(
    url = url, name = NA, address = NA, plus_code = NA, lat = NA, lng = NA)
  if (is.null(resp)) {
    return(fail_out)
  }

  # Extract URL from response
  full_url <- resp_url(resp)
  
  # Extract place name from URL
  regex_pattern <- "/place/([^/]+)/"
  match_info <- regexec(regex_pattern, full_url)
  place_match <- regmatches(full_url, match_info)

  # If place name matched
  if (length(place_match) > 0 && length(place_match[[1]]) >= 2) {
    
    raw_place_name <- place_match[[1]][2]
    place_name <- gsub("\\+", " ", raw_place_name)
    place_name <- URLdecode(place_name) 
    
    # Query Google Geocoding API
    api_result <- google_geocode(address = place_name)
    
    # Extract coordinates and other information
    if (api_result$status == "OK") {
      
      coords <- access_result(api_result, "coordinates") 
      address <- access_result(api_result, "address")
      
      plus_code <- NA
      if ("plus_code" %in% names(api_result$results)) {
        plus_code <- api_result$results$plus_code[1, "global_code"] 
      }
      
      out <- data.frame(
        url = url,
        name = place_name, 
        address = address, 
        plus_code = plus_code, 
        coords)[1, ]
      
      return(out)
      
    } else {
      message("API returned status '", api_result$status, "' for: ", place_name)
      return(fail_out)
    }
    
  } else {
    message("Could not extract place name from resolved URL: ", full_url)
    return(fail_out)
  }
}

# Import URLs
google_takeout <- read.csv("./Takeout/Saved/Edinburgh food.csv")
urls <- google_takeout$URL
urls <- urls[urls != "" & !is.na(urls)] # Also filter out NAs just in case

# Run function on URLs
result_list <- lapply(urls, geocode_url)
result_df <- do.call(rbind, result_list)

result_df