Add geckodriver url env var

scraper/lib.rs

@@ -8,15 +8,26 @@ use fantoccini::{Client, ClientBuilder, Locator};
 use bin_stuff::{Bin, BinDates};
 use log::{error, info};
 
-pub async fn get_stuff(postcode: &str, address: &str) -> Result<Vec<BinDates>, Box<dyn Error>> {
+pub async fn get_stuff(
+    postcode: &str,
+    address: &str,
+    driver_url: Option<String>,
+) -> Result<Vec<BinDates>, Box<dyn Error>> {
     let mut capabilities = Capabilities::new();
     let options = serde_json::json!({ "args": ["--headless"] });
     capabilities.insert("moz:firefoxOptions".to_string(), options);
 
+    info!("Attempting to connect to webdriver client");
+    let default_driver_url = "http://127.0.0.1:4444".to_string();
+    if driver_url.is_none() {
+        info!(
+            "No driver_url provided. Defaulting to {}",
+            default_driver_url
+        );
+    }
     let client = match ClientBuilder::native()
         .capabilities(capabilities)
-        // TODO: Geckodriver URL as env var
-        .connect("http://127.0.0.1:4444")
+        .connect(&driver_url.unwrap_or("http://127.0.0.1:4444".to_string()))
         .await
     {
         Ok(client) => client,
@@ -25,6 +36,7 @@ pub async fn get_stuff(postcode: &str, address: &str) -> Result<Vec<BinDates>, B
             return Err(Box::new(e));
         }
     };
+    info!("Got webdriver client");
 
     // NOTE: Some of the fields get different IDs when submitting each step it seems
     let max_attempts = 3;

server/src/email_sender.rs

@@ -8,6 +8,7 @@ pub async fn do_the_stuff(
     users: &[User],
     aws_client: &aws_sdk_sesv2::Client,
     from_email_address: &str,
+    geckodriver_url: String,
 ) -> Result<(), Box<dyn Error>> {
     for person in users {
         println!("Found {:?}", person);
@@ -20,7 +21,12 @@ pub async fn do_the_stuff(
         // Need to scrape the page from an actual browser. Tried curl/reqwest requests, but submitting
         // the post request would redirect back to the first page. Some sort of request token missing
         // when we do this or something (first step of form submission changes the url).
-        let bins = scraper::get_stuff(&person.postcode, &person.address).await?;
+        let bins = scraper::get_stuff(
+            &person.postcode,
+            &person.address,
+            Some(geckodriver_url.clone()),
+        )
+        .await?;
 
         let today = chrono::Utc::now().date_naive();
         let next_collection_date = next_collection_date_from(today);

server/src/main.rs

@@ -46,6 +46,7 @@ struct AppState {
     pool: SqlitePool,
     aws_client: Client,
     from_email_address: String,
+    geckodriver_url: String,
 }
 
 #[tokio::main]
@@ -61,6 +62,17 @@ async fn main() -> Result<(), Box<dyn Error>> {
     let _aws_secret_access_key =
         env::var("AWS_SECRET_ACCESS_KEY").expect("AWS_SECRET_ACCESS_KEY must be specified");
     let db_url = env::var("DATABASE_URL").expect("DATABASE_URL must be specified");
+    let geckodriver_url_default = "http://127.0.0.1:4444".to_string();
+    let geckodriver_url = match env::var("GECKODRIVER_URL") {
+        Ok(url) => url,
+        Err(_) => {
+            info!(
+                "GECKODRIVER_URL was not specified. Defaulting to {}",
+                geckodriver_url_default
+            );
+            geckodriver_url_default
+        }
+    };
     let region_provider = RegionProviderChain::default_provider().or_else("eu-west-1");
     let config = aws_config::from_env().region(region_provider).load().await;
     let aws_client = Client::new(&config);
@@ -81,6 +93,7 @@ async fn main() -> Result<(), Box<dyn Error>> {
         pool,
         aws_client,
         from_email_address,
+        geckodriver_url,
     };
     let scheduler_app_state = app_state.clone();
 
@@ -124,6 +137,7 @@ async fn scrape_and_email_stuff(app_state: AppState) {
         &people_to_notify,
         &app_state.aws_client,
         &app_state.from_email_address,
+        app_state.geckodriver_url,
     )
     .await
     {