added creating CompanyInfo mapping

This commit is contained in:
2025-12-04 13:33:32 +01:00
parent 95fd9ca141
commit ef2393ab70
13 changed files with 965 additions and 696 deletions

View File

@@ -3,69 +3,41 @@ mod economic;
mod corporate;
mod config;
mod util;
mod scraper;
use fantoccini::{ClientBuilder};
use serde_json::{Map, Value};
use tokio::signal;
use anyhow::Result;
use config::Config;
use scraper::webdriver::ChromeDriverPool;
use std::sync::Arc;
/// The entry point of the application.
///
/// This function loads the configuration, initializes a shared ChromeDriver pool,
/// and sequentially runs the full updates for corporate and economic data.
/// Sequential execution helps prevent resource exhaustion from concurrent
/// chromedriver instances and avoids spamming the target websites with too many requests.
///
/// # Errors
///
/// Returns an error if configuration loading fails, pool initialization fails,
/// or if either update function encounters an issue (e.g., network errors,
/// scraping failures, or chromedriver spawn failures like "program not found").
#[tokio::main]
async fn main() -> anyhow::Result<()> {
// === Ensure data directories exist ===
util::ensure_data_dirs().await?;
async fn main() -> Result<()> {
let config = Config::load().map_err(|err| {
println!("Failed to load Config .env: {}", err);
err
})?;
// === Load configuration ===
let config = config::Config::default();
// Initialize the shared ChromeDriver pool once
let pool_size = config.max_parallel_tasks;
let pool = Arc::new(ChromeDriverPool::new(pool_size).await?);
// === Start ChromeDriver ===
let mut child = std::process::Command::new("chromedriver-win64/chromedriver.exe")
.args(["--port=9515"]) // Level 3 = minimal logs
.spawn()?;
// Run economic update first, passing the shared pool
economic::run_full_update(&config, &pool).await?;
// Build capabilities to hide infobar + enable full rendering
let port = 9515;
let caps_value = serde_json::json!({
"goog:chromeOptions": {
"args": [
//"--headless",
"--disable-gpu",
"--disable-notifications",
"--disable-popup-blocking",
"--disable-blink-features=AutomationControlled"
],
"excludeSwitches": ["enable-automation"]
}
});
// Then run corporate update, passing the shared pool
corporate::run_full_update(&config, &pool).await?;
let caps_map: Map<String, Value> = caps_value.as_object()
.expect("Capabilities should be a JSON object")
.clone();
let mut client = ClientBuilder::native()
.capabilities(caps_map)
.connect(&format!("http://localhost:{}", port))
.await?;
// Graceful shutdown
let client_clone = client.clone();
tokio::spawn(async move {
signal::ctrl_c().await.unwrap();
client_clone.close().await.ok();
std::process::exit(0);
});
// === Economic Calendar Update ===
println!("Updating Economic Calendar (High Impact Only)");
economic::goto_and_prepare(&client).await?;
economic::run_full_update(&client, &config).await?;
// === Corporate Earnings Update ===
println!("\nUpdating Corporate Earnings");
corporate::run_full_update(&client, &config).await?;
// === Cleanup ===
client.close().await?;
child.kill()?;
println!("\nAll data updated successfully!");
Ok(())
}