hottub/src/providers/perverzija.rs

use std::vec;
use std::env;
use error_chain::error_chain;
use htmlentity::entity::{decode, ICodedDataTrait};
use reqwest::{Proxy};
use futures::future::join_all;

use crate::db;
use crate::providers::Provider;
use crate::util::cache::VideoCache;
use crate::util::flaresolverr::{FlareSolverrRequest, Flaresolverr};
use crate::util::time::parse_time_to_seconds;
use crate::videos::{self, VideoEmbed, VideoItem};
use crate::DbPool; // Make sure Provider trait is imported

error_chain! {
    foreign_links {
        Io(std::io::Error);
        HttpRequest(reqwest::Error);
    }
}

pub struct PerverzijaProvider {
    url: String,
}
impl PerverzijaProvider {
    pub fn new() -> Self {
        PerverzijaProvider {
            url: "https://tube.perverzija.com/".to_string(),
        }
    }
    async fn get(&self, cache:VideoCache, pool:DbPool, page: u8, featured: String) -> Result<Vec<VideoItem>> {
        //TODO
        // let mut url = Url::parse("https://example.net")?;
        // url.query_pairs_mut().append_pair("foo", "bar");
        // url.query_pairs_mut().append_pair("key", "dkhdsihdsaiufds");
        // url.query_pairs_mut().append_pair("hello", "world");
        // println!("{}", url.as_str());

        let mut prefix_uri = "".to_string();
        if featured == "featured" {
            prefix_uri = "featured-scenes/".to_string();
        }
        let mut url = format!("{}{}page/{}/", self.url, prefix_uri, page);
        if page == 1 {
            url = format!("{}{}", self.url, prefix_uri);
        }
        
        let old_items = match cache.get(&url) {
            Some((time, items)) => {
                if time.elapsed().unwrap_or_default().as_secs() < 60 * 60 {
                    println!("Cache hit for URL: {}", url);
                    return Ok(items.clone());
                }
                else{
                    items.clone()
                }
            }
            None => {
                vec![]
            }
        };

        
        let client = match env::var("BURP_URL").as_deref() {
            Ok(burp_url) => 
                reqwest::Client::builder()
                .user_agent("Mozilla/5.0 (iPhone; CPU iPhone OS 14_5 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) FxiOS/33.0 Mobile/15E148 Safari/605.1.15")
                .proxy(Proxy::https(burp_url).unwrap())
                .danger_accept_invalid_certs(true)
                .build()?,
            Err(_) => reqwest::Client::builder()
                .user_agent("Mozilla/5.0 (iPhone; CPU iPhone OS 14_5 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) FxiOS/33.0 Mobile/15E148 Safari/605.1.15")
                .danger_accept_invalid_certs(true)
                .build()?,
        };
        
        let response = client.get(url.clone()).send().await?;
        // print!("Response: {:?}\n", response);
        if response.status().is_success() {
            let text = response.text().await?;
            let video_items: Vec<VideoItem> = self.get_video_items_from_html(text.clone(), pool);
            if !video_items.is_empty() {
                cache.remove(&url);
                cache.insert(url.clone(), video_items.clone());
            } else{
                return Ok(old_items);
            }
            Ok(video_items)
        } else {
            let flare_url = env::var("FLARE_URL").expect("FLARE_URL not set");
            let flare = Flaresolverr::new(flare_url);
            let result = flare
                .solve(FlareSolverrRequest {
                    cmd: "request.get".to_string(),
                    url: url.clone(),
                    maxTimeout: 60000,
                })
                .await;
            let video_items = match result {
                Ok(res) => {
                    // println!("FlareSolverr response: {}", res);
                    self.get_video_items_from_html(res.solution.response, pool)
                }
                Err(e) => {
                    println!("Error solving FlareSolverr: {}", e);
                    return Err("Failed to solve FlareSolverr".into());
                }
            };
            if !video_items.is_empty() {
                cache.remove(&url);
                cache.insert(url.clone(), video_items.clone());
            } else {
                return Ok(old_items);
            }
            Ok(video_items)
        }
    }
    async fn query(&self, cache: VideoCache, pool:DbPool, page: u8, query: &str) -> Result<Vec<VideoItem>> {
        let search_string = query.replace(" ", "+");
        let mut url = format!(
            "{}page/{}/?s={}",
            self.url, page, search_string
        );
        if page == 1 {
            url = format!("{}?s={}", self.url, search_string);
        }

        // Check our Video Cache. If the result is younger than 1 hour, we return it.
        let old_items = match cache.get(&url) {
            Some((time, items)) => {
                if time.elapsed().unwrap_or_default().as_secs() < 60 * 60 {
                    return Ok(items.clone());
                }
                else{
                    let _ = cache.check().await;
                    return Ok(items.clone())
                }
            }
            None => {
                vec![]
            }
        };    
        let client = match env::var("BURP_URL").as_deref() {
            Ok(burp_url) => 
                reqwest::Client::builder()
                .user_agent("Mozilla/5.0 (iPhone; CPU iPhone OS 14_5 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) FxiOS/33.0 Mobile/15E148 Safari/605.1.15")
                .proxy(Proxy::https(burp_url).unwrap())
                .danger_accept_invalid_certs(true)
                .build()?,
            Err(_) => reqwest::Client::builder()
                .user_agent("Mozilla/5.0 (iPhone; CPU iPhone OS 14_5 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) FxiOS/33.0 Mobile/15E148 Safari/605.1.15")
                .danger_accept_invalid_certs(true)
                .build()?,
        };
        
        let response = client.get(url.clone()).send().await?;
        if response.status().is_success() {
            let text = response.text().await?;
            let video_items: Vec<VideoItem> = self.get_video_items_from_html_query(text.clone(), pool).await;
            if !video_items.is_empty() {
                cache.remove(&url);
                cache.insert(url.clone(), video_items.clone());
            } else{
                return Ok(old_items);
            }
            Ok(video_items)
        } else {
            let flare_url = env::var("FLARE_URL").expect("FLARE_URL not set");
            let flare = Flaresolverr::new(flare_url);
            let result = flare
                .solve(FlareSolverrRequest {
                    cmd: "request.get".to_string(),
                    url: url.clone(),
                    maxTimeout: 60000,
                })
                .await;
            let video_items = match result {
                Ok(res) => {
                    self.get_video_items_from_html_query(res.solution.response, pool).await
                }
                Err(e) => {
                    println!("Error solving FlareSolverr: {}", e);
                    return Err("Failed to solve FlareSolverr".into());
                }
            };
            if !video_items.is_empty() {
                cache.remove(&url);
                cache.insert(url.clone(), video_items.clone());
            } else{
                return Ok(old_items);
            }
            Ok(video_items)
        }
    }

    fn get_video_items_from_html(&self, html: String, pool: DbPool) -> Vec<VideoItem> {
        if html.is_empty() {
            println!("HTML is empty");
            return vec![];
        }
        let mut items: Vec<VideoItem> = Vec::new();
        let video_listing_content = html.split("video-listing-content").collect::<Vec<&str>>()[1];
        let raw_videos = video_listing_content
            .split("video-item post")
            .collect::<Vec<&str>>()[1..]
            .to_vec();
        for video_segment in &raw_videos {
            let vid = video_segment.split("\n").collect::<Vec<&str>>();
            if vid.len() > 20 {
                continue;
            }
            let mut title = vid[1].split(">").collect::<Vec<&str>>()[1]
                .split("<")
                .collect::<Vec<&str>>()[0]
                .to_string();
            // html decode
            title = decode(title.as_bytes()).to_string().unwrap_or(title);
            let url = vid[1].split("iframe src=&quot;").collect::<Vec<&str>>()[1]
                .split("&quot;")
                .collect::<Vec<&str>>()[0]
                .to_string().replace("index.php", "xs1.php");
            let id = url.split("data=").collect::<Vec<&str>>()[1]
                .split("&")
                .collect::<Vec<&str>>()[0]
                .to_string();
            let raw_duration = match vid.len() {
                10 => vid[6].split("time_dur\">").collect::<Vec<&str>>()[1]
                    .split("<")
                    .collect::<Vec<&str>>()[0]
                    .to_string(),
                _ => "00:00".to_string(),
            };
            let duration = parse_time_to_seconds(&raw_duration).unwrap_or(0) as u32;

            let thumb = match vid[4].contains("srcset=") {
                true => vid[4].split("sizes=").collect::<Vec<&str>>()[1]
                    .split("w, ")
                    .collect::<Vec<&str>>()
                    .last()
                    .unwrap()
                    .to_string()
                    .split(" ")
                    .collect::<Vec<&str>>()[0]
                    .to_string(),
                false => vid[4].split("src=\"").collect::<Vec<&str>>()[1]
                    .split("\"")
                    .collect::<Vec<&str>>()[0]
                    .to_string(),
            };
            let embed_html = vid[1].split("data-embed='").collect::<Vec<&str>>()[1]
                .split("'")
                .collect::<Vec<&str>>()[0]
                .to_string();
            let id_url = vid[1].split("data-url='").collect::<Vec<&str>>()[1]
                .split("'")
                .collect::<Vec<&str>>()[0]
                .to_string();
            
            let mut conn = pool.get().expect("couldn't get db connection from pool");
            let _ = db::insert_video(&mut conn, &id_url, &url);
            drop(conn);
            let referer_url = "https://xtremestream.xyz/".to_string();
            let embed = VideoEmbed::new(embed_html, url.clone());

            let mut tags: Vec<String> = Vec::new(); // Placeholder for tags, adjust as needed
            for tag in vid[0].split(" ").collect::<Vec<&str>>(){
                if tag.starts_with("tag-") {
                    let tag_name = tag.split("tag-").collect::<Vec<&str>>()[1]
                        .to_string();
                    if !tag_name.is_empty() {
                        tags.push(tag_name.replace("-", " ").to_string());
                    }
                }
            }
            let mut video_item = VideoItem::new(
                id,
                title,
                embed.source.clone(),
                "perverzija".to_string(),
                thumb,
                duration,
            ).tags(tags);
            // .embed(embed.clone());
            let mut format =
                videos::VideoFormat::new(url.clone(), "1080".to_string(), "m3u8".to_string());
            format.add_http_header("Referer".to_string(), referer_url.clone());
            if let Some(formats) = video_item.formats.as_mut() {
                formats.push(format);
            } else {
                video_item.formats = Some(vec![format]);
            }
            items.push(video_item);
        }

        return items;
    }

    async fn get_video_items_from_html_query(&self, html: String, pool:DbPool) -> Vec<VideoItem> {
        let raw_videos = html
            .split("video-item post")
            .collect::<Vec<&str>>()[1..]
            .to_vec();
        let futures = raw_videos.into_iter().map(|el| self.get_video_item(el, pool.clone()));
        let results: Vec<Result<VideoItem>> = join_all(futures).await;
        let items: Vec<VideoItem> = results
            .into_iter()
            .filter_map(Result::ok)
            .collect();
        return  items;
    }

    async fn get_video_item(&self, snippet: &str, pool: DbPool) -> Result<VideoItem> {
        let vid = snippet.split("\n").collect::<Vec<&str>>();
        if vid.len() > 30 {
            return Err("Unexpected video snippet length".into());
        }

        // for (index,line) in vid.iter().enumerate() {
        //     println!("Line {}: {}", index, line.to_string().trim());
        // }

        let mut title = vid[5].split(" title=\"").collect::<Vec<&str>>()[1]
            .split("\"")
            .collect::<Vec<&str>>()[0]
            .to_string();
        title = decode(title.as_bytes()).to_string().unwrap_or(title);

        let thumb = match vid[6].split(" src=\"").collect::<Vec<&str>>().len(){
            1=>{
                for (index,line) in vid.iter().enumerate() {
                    println!("Line {}: {}", index, line.to_string().trim());
                }
                return Err("Failed to parse thumbnail URL".into());
            }
            _ => vid[6].split(" src=\"").collect::<Vec<&str>>()[1]
                .split("\"")
                .collect::<Vec<&str>>()[0]
                .to_string(),
        };
        let duration = 0;

        let lookup_url = vid[5].split(" href=\"").collect::<Vec<&str>>()[1]
            .split("\"")
            .collect::<Vec<&str>>()[0]
            .to_string();
        let referer_url = "https://xtremestream.xyz/".to_string();

        let mut conn = pool.get().expect("couldn't get db connection from pool");
        let db_result = db::get_video(&mut conn,lookup_url.clone());
        match db_result {
            Ok(Some(url)) => {
                let mut id = url.split("data=").collect::<Vec<&str>>()[1]
                    .to_string();
                if id.contains("&"){
                    id = id.split("&").collect::<Vec<&str>>()[0].to_string()
                }
                let mut video_item = VideoItem::new(
                    id,
                    title,
                    url.clone(),
                    "perverzija".to_string(),
                    thumb,
                    duration,
                );
                let mut format =
                    videos::VideoFormat::new(url.clone(), "1080".to_string(), "m3u8".to_string());
                format.add_http_header("Referer".to_string(), referer_url.clone());
                if let Some(formats) = video_item.formats.as_mut() {
                    formats.push(format);
                } else {
                    video_item.formats = Some(vec![format]);
                }
                return Ok(video_item);
            }
            Ok(None) => (),
            Err(e) => {
                println!("Error fetching video from database: {}", e);
                // return Err(format!("Error fetching video from database: {}", e).into());
            }
        }
        drop(conn);

        let client = match env::var("BURP_URL").as_deref() {
            Ok(burp_url) => 
                reqwest::Client::builder()
                .user_agent("Mozilla/5.0 (iPhone; CPU iPhone OS 14_5 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) FxiOS/33.0 Mobile/15E148 Safari/605.1.15")
                .proxy(Proxy::https(burp_url).unwrap())
                .danger_accept_invalid_certs(true)
                .build()?,
            Err(_) => reqwest::Client::builder()
                .user_agent("Mozilla/5.0 (iPhone; CPU iPhone OS 14_5 like Mac OS X) AppleWebKit/605.1.15 (KHTML, like Gecko) FxiOS/33.0 Mobile/15E148 Safari/605.1.15")
                .danger_accept_invalid_certs(true)
                .build()?,
        };
        
        let response = client.get(lookup_url.clone()).send().await?;
        let text = match response.status().is_success(){
            true => response.text().await?,
            false => {
                return Err("Failed to fetch video details".into());
            }
        };


        let url = text.split("<iframe src=\"").collect::<Vec<&str>>()[1]
            .split("\"")
            .collect::<Vec<&str>>()[0]
            .to_string().replace("index.php","xs1.php");
        if url.starts_with("https://discord.com"){
            return Err("Discord link found, skipping video".into());
        }

        let mut conn = pool.get().expect("couldn't get db connection from pool");
        let _ = db::insert_video(&mut conn, &lookup_url, &url);
        drop(conn);
        if !url.contains("xtremestream.xyz"){
            return Err("Video URL does not contain xtremestream.xyz".into());
        }
        let mut id = url.split("data=").collect::<Vec<&str>>()[1]
            .to_string();
        if id.contains("&"){
            id = id.split("&").collect::<Vec<&str>>()[0].to_string()
        }
        // if !vid[6].contains(" src=\""){
        //     for (index,line) in vid.iter().enumerate() {
        //         println!("Line {}: {}", index, line.to_string().trim());
        //     }
        // }
        let mut tags: Vec<String> = Vec::new(); // Placeholder for tags, adjust as needed
        for tag in vid[0].split(" ").collect::<Vec<&str>>(){
            if tag.starts_with("tag-") {
                let tag_name = tag.split("tag-").collect::<Vec<&str>>()[1]
                    .to_string();
                if !tag_name.is_empty() {
                    tags.push(tag_name.replace("-", " ").to_string());
                }
            }
        }

        let mut video_item = VideoItem::new(
            id,
            title,
            url.clone(),
            "perverzija".to_string(),
            thumb,
            duration,
        )
        .tags(tags);
        // .embed(embed.clone());
        let mut format =
            videos::VideoFormat::new(url.clone(), "1080".to_string(), "m3u8".to_string());
        format.add_http_header("Referer".to_string(), referer_url.clone());
        if let Some(formats) = video_item.formats.as_mut() {
            formats.push(format);
        } else {
            video_item.formats = Some(vec![format]);
        }
        return Ok(video_item);
    }
}

impl Provider for PerverzijaProvider {
    async fn get_videos(
        &self,
        cache: VideoCache,
        pool: DbPool,
        _channel: String,
        sort: String,
        query: Option<String>,
        page: String,
        per_page: String,
        featured: String,
    ) -> Vec<VideoItem> {
        let _ = per_page;
        let _ = sort;
        let videos: std::result::Result<Vec<VideoItem>, Error> = match query {
            Some(q) => self.query(cache, pool, page.parse::<u8>().unwrap_or(1), &q).await,
            None => self.get(cache, pool, page.parse::<u8>().unwrap_or(1), featured).await,
        };
        match videos {
            Ok(v) => v,
            Err(e) => {
                println!("Error fetching videos: {}", e);
                vec![]
            }
        }
    }
}