feat: batch stuff

src/main.rs

@@ -1,3 +1,4 @@
+use indicatif::{ProgressBar, ProgressStyle};
 use scraper::{Html, Selector};
 use regex::Regex;
 use std::fs::File;
@@ -5,9 +6,10 @@ use std::io::Read;
 use std::process::Command;
 use rusqlite::{Connection, Result};
 use std::time::Instant;
+use rayon::prelude::*;
 
 fn main() -> Result<(), Box<dyn std::error::Error>> {
-    let db = Connection::open("video_ids.db")?;
+    let mut db = Connection::open("video_ids.db")?;
 
     db.execute(
         "CREATE TABLE IF NOT EXISTS video_ids (
@@ -17,12 +19,60 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
         [],
     )?;
 
-    get_ids_history(db)?;
+    get_ids_history(&mut db)?;
+
+    let mut stmt = db.prepare("SELECT id FROM video_ids WHERE duration IS NULL OR duration = ''")?;
+    let ids_to_fetch: Vec<String> = stmt.query_map([], |row| row.get(0))?
+                                        .collect::<Result<Vec<String>, _>>()?;
+    stmt.finalize()?;
+    
+    let total_ids = ids_to_fetch.len();
+    if total_ids == 0 {
+        println!("No IDs to fetch.");
+        return Ok(());
+    }
+
+    let progress_bar = ProgressBar::new(total_ids as u64);
+    progress_bar.set_style(ProgressStyle::default_bar()
+        .template("{msg} [{elapsed_precise}] {wide_bar} {pos}/{len} ({eta})")?
+        .progress_chars("##-"));
+    progress_bar.set_message("Fetching video durations...");
+
+    let start_time = Instant::now();
+    let batch_size = 30;
+    let results: Vec<(String, String)> = ids_to_fetch
+        .par_chunks(batch_size)
+        .flat_map(|chunk| {
+            chunk.iter().map(|id| {
+                let duration = get_video_duration(id).unwrap_or_else(|_| "Error".to_string());
+                progress_bar.inc(1);
+                (id.clone(), duration)
+            }).collect::<Vec<_>>()
+        })
+        .collect();
+
+    let elapsed_time = start_time.elapsed();
+    println!("Fetching video durations took: {:?}", elapsed_time);
+
+    let update_start = Instant::now();
+    let tx = db.transaction()?;
+    {
+        let mut stmt = tx.prepare_cached("UPDATE video_ids SET duration = ? WHERE id = ?")?;
+        for (id, duration) in results {
+            stmt.execute(rusqlite::params![duration, id])?;
+            progress_bar.inc(1);
+        }
+    }
+    tx.commit()?;
+    progress_bar.finish_with_message("Done!");
+    let update_duration = update_start.elapsed();
+    println!("Updating database took: {:?}", update_duration);
+    println!("Total time taken: {:?}", elapsed_time + update_duration);
 
     Ok(())
 }
 
-fn get_ids_history(mut db: Connection) -> Result<(), Box<dyn std::error::Error>> {
+fn get_ids_history(db: &mut Connection) -> Result<(), Box<dyn std::error::Error>> {
     let read_start = Instant::now();
 
     let mut id_db: Vec<&str> = Vec::new();
@@ -55,7 +105,7 @@ fn get_ids_history(mut db: Connection) -> Result<(), Box<dyn std::error::Error>>
 
     // database inserting
     let insert_start = Instant::now();
-    
+
     let tx = db.transaction()?;
     {
         let mut stmt = tx.prepare_cached("INSERT OR IGNORE INTO video_ids (id) VALUES (?)")?;
@@ -75,20 +125,32 @@ fn get_ids_history(mut db: Connection) -> Result<(), Box<dyn std::error::Error>>
 fn get_video_duration(id: &str) -> Result<String, Box<dyn std::error::Error>> {
     let url = format!("https://www.youtube.com/watch?v={}", id);
     
-    let output = Command::new("yt-dlp")
-        .args(&[
-            "--get-duration",
-            url.as_str(),
-        ])
-        .output()?;
+    // Add timeout and retry logic
+    let max_retries = 3;
+    let mut retries = 0;
     
-    if output.status.success() {
-        let stdout = String::from_utf8_lossy(&output.stdout);
-        let video_url = stdout.lines().next().unwrap_or("");
-        return Ok(video_url.to_string());
-    } else {
-        let stderr = String::from_utf8_lossy(&output.stderr);
-        eprintln!("Error: {}", stderr);
-        return Err("Failed to get video link".into());
+    while retries < max_retries {
+        let output = Command::new("yt-dlp")
+            .args(&[
+                "--get-duration",
+                "--no-warnings",
+                "--socket-timeout", "10",
+                url.as_str(),
+            ])
+            .output()?;
+        
+        if output.status.success() {
+            let stdout = String::from_utf8_lossy(&output.stdout);
+            let duration = stdout.trim();
+            
+            if !duration.is_empty() {
+                return Ok(duration.to_string());
+            }
+        }
+        
+        retries += 1;
+        std::thread::sleep(std::time::Duration::from_millis(500 * retries));
     }
-}
+    
+    Err(format!("Failed to get duration for video ID: {}", id).into())
+}