using chromedriver for collecting corporate earnings

This commit is contained in:
2025-11-22 20:47:24 +01:00
parent fc56ae5d82
commit cd3f47d91f
3 changed files with 64 additions and 36 deletions

View File

@@ -1,47 +1,81 @@
// src/corporate/scraper.rs // src/corporate/scraper.rs
use super::types::{CompanyEvent, CompanyPrice}; use super::types::{CompanyEvent, CompanyPrice};
use reqwest::Client; use fantoccini::{Client, Locator};
use scraper::{Html, Selector}; use scraper::{Html, Selector};
use chrono::{NaiveDate, Datelike}; use chrono::{NaiveDate, Datelike};
use tokio::time::{sleep, Duration};
const USER_AGENT: &str = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36"; const USER_AGENT: &str = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36";
pub async fn fetch_earnings_history(ticker: &str) -> anyhow::Result<Vec<CompanyEvent>> { pub async fn dismiss_yahoo_consent(client: &Client) -> anyhow::Result<()> {
let client = Client::new(); let script = r#"
let url = format!("https://finance.yahoo.com/quote/{ticker}/history?filter=earnings"); (() => {
const agree = document.querySelector('button[name="agree"]');
if (agree) {
agree.click();
return true;
}
return false;
})()
"#;
let text = client for _ in 0..10 {
.get(&url) let done: bool = client.execute(script, vec![]).await?.as_bool().unwrap_or(false);
.header("User-Agent", USER_AGENT) if done {
.send() break;
.await? }
.text() sleep(Duration::from_millis(500)).await;
.await?; }
Ok(())
}
let document = Html::parse_document(&text); pub async fn fetch_earnings_history(client: &Client, ticker: &str) -> anyhow::Result<Vec<CompanyEvent>> {
let row_sel = Selector::parse(r#"table tbody tr"#).unwrap(); let url = format!("https://finance.yahoo.com/calendar/earnings?symbol={}", ticker);
client.goto(&url).await?;
dismiss_yahoo_consent(client).await?;
// Load all by clicking "Show More" if present
loop {
match client.find(Locator::XPath(r#"//button[contains(text(), 'Show More')]"#)).await {
Ok(btn) => {
btn.click().await?;
sleep(Duration::from_secs(2)).await;
}
Err(_) => break,
}
}
let html = client.source().await?;
let document = Html::parse_document(&html);
let row_sel = Selector::parse("table tbody tr").unwrap();
let mut events = Vec::new(); let mut events = Vec::new();
for row in document.select(&row_sel) { for row in document.select(&row_sel) {
let cols: Vec<String> = row.text().map(|s| s.trim().to_string()).collect(); let cols: Vec<String> = row.select(&Selector::parse("td").unwrap())
if cols.len() < 4 { continue; } .map(|td| td.text().collect::<Vec<_>>().join(" ").trim().to_string())
.collect();
if cols.len() < 6 { continue; }
let full_date = &cols[2];
let parts: Vec<&str> = full_date.split(" at ").collect();
let raw_date = parts[0].trim();
let time_str = if parts.len() > 1 { parts[1].trim() } else { "" };
let raw_date = cols[0].split(" - ").next().unwrap_or(&cols[0]);
let date = match parse_yahoo_date(raw_date) { let date = match parse_yahoo_date(raw_date) {
Ok(d) => d, Ok(d) => d,
Err(_) => continue, Err(_) => continue,
}; };
let eps_forecast = parse_float(&cols[1]); let eps_forecast = parse_float(&cols[3]);
let eps_actual = parse_float(&cols[2]); let eps_actual = if cols[4] == "-" { None } else { parse_float(&cols[4]) };
let surprise_pct = if let (Some(f), Some(a)) = (eps_forecast, eps_actual) { let surprise_pct = if let (Some(f), Some(a)) = (eps_forecast, eps_actual) {
if f.abs() > 0.001 { Some((a - f) / f.abs() * 100.0) } else { None } if f.abs() > 0.001 { Some((a - f) / f.abs() * 100.0) } else { None }
} else { None }; } else { None };
let time = if cols[0].contains("After") || cols[0].contains("AMC") { let time = if time_str.contains("PM") {
"AMC".to_string() "AMC".to_string()
} else if cols[0].contains("Before") || cols[0].contains("BMO") { } else if time_str.contains("AM") {
"BMO".to_string() "BMO".to_string()
} else { } else {
"".to_string() "".to_string()
@@ -51,7 +85,7 @@ pub async fn fetch_earnings_history(ticker: &str) -> anyhow::Result<Vec<CompanyE
ticker: ticker.to_string(), ticker: ticker.to_string(),
date: date.format("%Y-%m-%d").to_string(), date: date.format("%Y-%m-%d").to_string(),
time, time,
period: cols.get(3).cloned().unwrap_or_default(), period: "".to_string(), // No period info available, set to empty
eps_forecast, eps_forecast,
eps_actual, eps_actual,
revenue_forecast: None, revenue_forecast: None,
@@ -64,8 +98,7 @@ pub async fn fetch_earnings_history(ticker: &str) -> anyhow::Result<Vec<CompanyE
Ok(events) Ok(events)
} }
pub async fn fetch_price_history(ticker: &str, start: &str, end: &str) -> anyhow::Result<Vec<CompanyPrice>> { pub async fn fetch_price_history(client: &Client, ticker: &str, start: &str, end: &str) -> anyhow::Result<Vec<CompanyPrice>> {
let client = Client::new();
let start_ts = NaiveDate::parse_from_str(start, "%Y-%m-%d")? let start_ts = NaiveDate::parse_from_str(start, "%Y-%m-%d")?
.and_hms_opt(0, 0, 0).unwrap().and_utc() .and_hms_opt(0, 0, 0).unwrap().and_utc()
.timestamp(); .timestamp();
@@ -79,13 +112,8 @@ pub async fn fetch_price_history(ticker: &str, start: &str, end: &str) -> anyhow
"https://query1.finance.yahoo.com/v7/finance/download/{ticker}?period1={start_ts}&period2={end_ts}&interval=1d&events=history&includeAdjustedClose=true" "https://query1.finance.yahoo.com/v7/finance/download/{ticker}?period1={start_ts}&period2={end_ts}&interval=1d&events=history&includeAdjustedClose=true"
); );
let csv = client client.goto(&url).await?;
.get(&url) let csv = client.source().await?;
.header("User-Agent", USER_AGENT)
.send()
.await?
.text()
.await?;
let mut prices = Vec::new(); let mut prices = Vec::new();
for line in csv.lines().skip(1) { for line in csv.lines().skip(1) {
@@ -110,7 +138,7 @@ fn parse_float(s: &str) -> Option<f64> {
} }
fn parse_yahoo_date(s: &str) -> anyhow::Result<NaiveDate> { fn parse_yahoo_date(s: &str) -> anyhow::Result<NaiveDate> {
NaiveDate::parse_from_str(s, "%b %d, %Y") NaiveDate::parse_from_str(s, "%B %d, %Y")
.or_else(|_| NaiveDate::parse_from_str(s, "%B %d, %Y")) .or_else(|_| NaiveDate::parse_from_str(s, "%b %d, %Y"))
.map_err(|_| anyhow::anyhow!("Bad date: {s}")) .map_err(|_| anyhow::anyhow!("Bad date: {s}"))
} }

View File

@@ -6,7 +6,7 @@ use chrono::Local;
use std::collections::{HashMap, HashSet}; use std::collections::{HashMap, HashSet};
pub async fn run_full_update(tickers: Vec<String>, config: &Config) -> anyhow::Result<()> { pub async fn run_full_update(client: &fantoccini::Client, tickers: Vec<String>, config: &Config) -> anyhow::Result<()> {
println!("Updating {} tickers (prices from {})", tickers.len(), config.corporate_start_date); println!("Updating {} tickers (prices from {})", tickers.len(), config.corporate_start_date);
let today = chrono::Local::now().format("%Y-%m-%d").to_string(); let today = chrono::Local::now().format("%Y-%m-%d").to_string();
@@ -15,13 +15,13 @@ pub async fn run_full_update(tickers: Vec<String>, config: &Config) -> anyhow::R
print!("{:6} ", ticker); print!("{:6} ", ticker);
// Earnings // Earnings
if let Ok(events) = fetch_earnings_history(&ticker).await { if let Ok(events) = fetch_earnings_history(client, &ticker).await {
merge_and_save_events(&ticker, events.clone()).await?; merge_and_save_events(&ticker, events.clone()).await?;
println!("{} earnings", events.len()); println!("{} earnings", events.len());
} }
// Prices now using config.corporate_start_date // Prices now using config.corporate_start_date
if let Ok(prices) = fetch_price_history(&ticker, &config.corporate_start_date, &today).await { if let Ok(prices) = fetch_price_history(client, &ticker, &config.corporate_start_date, &today).await {
save_prices_for_ticker(&ticker, prices).await?; save_prices_for_ticker(&ticker, prices).await?;
} }

View File

@@ -40,7 +40,7 @@ async fn main() -> anyhow::Result<()> {
// === Corporate Earnings Update === // === Corporate Earnings Update ===
println!("\nUpdating Corporate Earnings"); println!("\nUpdating Corporate Earnings");
let tickers = config::get_tickers(); let tickers = config::get_tickers();
corporate::run_full_update(tickers, &config).await?; corporate::run_full_update(&client, tickers, &config).await?;
// === Cleanup === // === Cleanup ===
client.close().await?; client.close().await?;