added corporate quarterly announcments for the last 4 years

src/corporate/scraper.rs

@@ -2,8 +2,10 @@
 use super::types::{CompanyEvent, CompanyPrice};
 use fantoccini::{Client, Locator};
 use scraper::{Html, Selector};
-use chrono::{NaiveDate, Datelike};
+use chrono::{NaiveDate};
 use tokio::time::{sleep, Duration};
+use yfinance_rs::{YfClient, Ticker, Range, Interval};
+use yfinance_rs::core::conversions::money_to_f64;
 
 const USER_AGENT: &str = "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36";
 
@@ -54,9 +56,9 @@ pub async fn fetch_earnings_history(client: &Client, ticker: &str) -> anyhow::Re
         let cols: Vec<String> = row.select(&Selector::parse("td").unwrap())
             .map(|td| td.text().collect::<Vec<_>>().join(" ").trim().to_string())
             .collect();
-        if cols.len() < 6 { continue; }
+        if cols.len() < 4 { continue; }
 
-        let full_date = &cols[2];
+        let full_date = &cols[0];
         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 { "" };
@@ -66,8 +68,8 @@ pub async fn fetch_earnings_history(client: &Client, ticker: &str) -> anyhow::Re
             Err(_) => continue,
         };
 
-        let eps_forecast = parse_float(&cols[3]);
-        let eps_actual = if cols[4] == "-" { None } else { parse_float(&cols[4]) };
+        let eps_forecast = parse_float(&cols[1]);
+        let eps_actual = if cols[2] == "-" { None } else { parse_float(&cols[2]) };
 
         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 }
@@ -85,7 +87,7 @@ pub async fn fetch_earnings_history(client: &Client, ticker: &str) -> anyhow::Re
             ticker: ticker.to_string(),
             date: date.format("%Y-%m-%d").to_string(),
             time,
-            period: "".to_string(),  // No period info available, set to empty
+            period: "".to_string(),
             eps_forecast,
             eps_actual,
             revenue_forecast: None,
@@ -98,38 +100,46 @@ pub async fn fetch_earnings_history(client: &Client, ticker: &str) -> anyhow::Re
     Ok(events)
 }
 
-pub async fn fetch_price_history(client: &Client, ticker: &str, start: &str, end: &str) -> anyhow::Result<Vec<CompanyPrice>> {
-    let start_ts = NaiveDate::parse_from_str(start, "%Y-%m-%d")?
-        .and_hms_opt(0, 0, 0).unwrap().and_utc()
-        .timestamp();
+pub async fn fetch_price_history(
+    ticker: &str,
+    start: &str,
+    end: &str,
+) -> anyhow::Result<Vec<CompanyPrice>> {
+    let client = YfClient::default();
+    let tk = Ticker::new(&client, ticker);
 
-    let end_ts = NaiveDate::parse_from_str(end, "%Y-%m-%d")?
-        .succ_opt().unwrap()
-        .and_hms_opt(0, 0, 0).unwrap().and_utc()
-        .timestamp();
+    // We request the maximum range – the library will automatically respect Yahoo's limits
+    let history = tk
+        .history(Some(Range::Max), Some(Interval::D1), true)
+        .await
+        .map_err(|e| anyhow::anyhow!("Yahoo Finance API error for {ticker}: {e:?}"))?;
 
-    let url = format!(
-        "https://query1.finance.yahoo.com/v7/finance/download/{ticker}?period1={start_ts}&period2={end_ts}&interval=1d&events=history&includeAdjustedClose=true"
-    );
+    let mut prices = Vec::with_capacity(history.len());
 
-    client.goto(&url).await?;
-    let csv = client.source().await?;
+    for candle in history {
+        let date_str = candle.ts.format("%Y-%m-%d").to_string();
+
+        // Filter by user-defined start / end
+        if date_str < (*start).to_string() || date_str > (*end).to_string() {
+            continue;
+        }
 
-    let mut prices = Vec::new();
-    for line in csv.lines().skip(1) {
-        let cols: Vec<&str> = line.split(',').collect();
-        if cols.len() < 7 { continue; }
         prices.push(CompanyPrice {
             ticker: ticker.to_string(),
-            date: cols[0].to_string(),
-            open: cols[1].parse()?,
-            high: cols[2].parse()?,
-            low: cols[3].parse()?,
-            close: cols[4].parse()?,
-            adj_close: cols[5].parse()?,
-            volume: cols[6].parse()?,
+            date: date_str,
+            open: money_to_f64(&candle.open),
+            high: money_to_f64(&candle.high),
+            low: money_to_f64(&candle.low),
+            // close_unadj is the raw (non-adjusted) close; close is the adjusted one
+            close: money_to_f64(&candle.close_unadj.unwrap_or(candle.close.clone())),
+            adj_close: money_to_f64(&candle.close),
+            volume: candle.volume.unwrap_or(0),
         });
     }
+
+    // Sort just in case (normally already sorted)
+    prices.sort_by_key(|p| p.date.clone());
+
     Ok(prices)
 }