hottub/src/providers/perverzija.rs

use std::vec;
use std::env;
use error_chain::error_chain;
use htmlentity::entity::{decode, ICodedDataTrait};
use futures::future::join_all;
use serde::Serialize;
use wreq::Client;
use wreq_util::Emulation;
use serde::Deserialize;
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::ServerOptions;
use crate::videos::{self, VideoEmbed, VideoItem};
use crate::DbPool;


error_chain! {
    foreign_links {
        Io(std::io::Error);
        HttpRequest(wreq::Error);
        JsonError(serde_json::Error);
    }
}


#[derive(Debug, Deserialize, Serialize)]
struct PerverzijaDbEntry {
    url_string: String,
    tags_strings: Vec<String>,
}

#[derive(Debug, Clone)]
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>> {

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

        
        let client = Client::builder()
        .emulation(Emulation::Firefox136)
        .build()?;
        
        let response = client.get(url_str.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_str);
                cache.insert(url_str.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_str.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_str);
                cache.insert(url_str.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 mut query_parse = true;
        let search_string = query.replace(" ", "+");
        let mut url_str = format!(
            "{}page/{}/?s={}",
            self.url, page, search_string
        );
        if page == 1 {
            url_str = format!("{}?s={}", self.url, search_string);
        }

        if query.starts_with("@studio:") {
            let studio_name = query.replace("@studio:", "");
            url_str = format!("{}studio/{}/page/{}/", self.url, studio_name, page);
            query_parse = false;
        } else if query.starts_with("@stars:") {
            let stars_name = query.replace("@stars:", "");
            url_str = format!("{}stars/{}/page/{}/", self.url, stars_name, page);
            query_parse = false;
        }
        url_str = url_str.replace("page/1/", "");
        // Check our Video Cache. If the result is younger than 1 hour, we return it.
        let old_items = match cache.get(&url_str) {
            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 = Client::builder()
        .emulation(Emulation::Firefox136)
        .build()?;
        
        let response = client.get(url_str.clone()).send().await?;
        if response.status().is_success() {
            let text = response.text().await?;
            let video_items: Vec<VideoItem> = match query_parse{
                true => {self.get_video_items_from_html_query(text.clone(), pool).await},
                false => {self.get_video_items_from_html(text.clone(), pool)}
            };
            if !video_items.is_empty() {
                cache.remove(&url_str);
                cache.insert(url_str.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_str.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_str);
                cache.insert(url_str.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;
            }
            // for (index, line) in vid.iter().enumerate() {
            //     println!("Line {}: {}", index, line.to_string().trim());
            // }
            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);
            if !vid[1].contains("iframe src=&quot;") {
                continue;
            }
            let url_str = vid[1].split("iframe src=&quot;").collect::<Vec<&str>>()[1]
                .split("&quot;")
                .collect::<Vec<&str>>()[0]
                .to_string().replace("index.php", "xs1.php");
            if url_str.starts_with("https://streamtape.com/"){
                continue; // Skip Streamtape links
            }
            let id = url_str.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;

            if !vid[4].contains("srcset=") && vid[4].split("src=\"").collect::<Vec<&str>>().len() == 1{
                for (index, line) in vid.iter().enumerate(){
                    println!("Line {}: {}\n\n", index, line);
                }
            }

            let mut thumb = "".to_string();
            for v in vid.clone(){
                let line = v.trim();
                if line.starts_with("<img "){
                    thumb = line.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_str);
            drop(conn);
            let referer_url = "https://xtremestream.xyz/".to_string();
            let embed = VideoEmbed::new(embed_html, url_str.clone());

            let mut tags: Vec<String> = Vec::new(); // Placeholder for tags, adjust as needed

            let studios_parts = vid[7].split("a href=\"").collect::<Vec<&str>>();
            for studio in studios_parts.iter().skip(1) {
                if studio.starts_with("https://tube.perverzija.com/studio/"){
                    tags.push(
                        studio.split("/\"").collect::<Vec<&str>>()[0]
                        .replace("https://tube.perverzija.com/studio/", "@studio:")
                            .to_string(),
                    );
                } 
            }

            for tag in vid[0].split(" ").collect::<Vec<&str>>(){
                if tag.starts_with("stars-") {
                    let tag_name = tag.split("stars-").collect::<Vec<&str>>()[1].split("\"").collect::<Vec<&str>>()[0]
                        .to_string();
                    if !tag_name.is_empty() {
                        tags.push(format!("@stars:{}", tag_name));
                    }
                }
            }

            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_str.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());
        }

        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(entry)) => {
                if entry.starts_with("{"){    // replace old urls with new json objects
                    let entry = serde_json::from_str::<PerverzijaDbEntry>(entry.as_str())?;
                    let url_str = entry.url_string;
                    let tags = entry.tags_strings;
                    if url_str.starts_with("!"){
                        return Err("Video was removed".into());
                    }
                    let mut id = url_str.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_str.clone(),
                        "perverzija".to_string(),
                        thumb,
                        duration,
                    )
                    .tags(tags)
                    ;
                    let mut format =
                        videos::VideoFormat::new(url_str.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)
                }
                else{
                    let _ = db::delete_video(&mut conn,lookup_url.clone());
                };
            }
            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 = Client::builder()
        .emulation(Emulation::Firefox136)
        .build()?;
        
        let response = client.get(lookup_url.clone()).send().await?;
        let text = match response.status().is_success(){
            true => response.text().await?,
            false => {
                println!("Failed to fetch video details");
                return Err("Failed to fetch video details".into());
            }
        };


        let mut url_str = text.split("<iframe src=\"").collect::<Vec<&str>>()[1]
            .split("\"")
            .collect::<Vec<&str>>()[0]
            .to_string().replace("index.php","xs1.php");
        if !url_str.contains("xtremestream.xyz"){
            url_str = "!".to_string()
        }

        let mut tags: Vec<String> = Vec::new(); // Placeholder for tags, adjust as needed

        let studios_parts = text.split("<strong>Studio: </strong>").collect::<Vec<&str>>()[1].split("</div>").collect::<Vec<&str>>()[0].split("<a href=\"").collect::<Vec<&str>>();
        for studio in studios_parts.iter().skip(1) {
            if studio.starts_with("https://tube.perverzija.com/studio/"){
                tags.push(
                    studio.split("/\"").collect::<Vec<&str>>()[0]
                    .replace("https://tube.perverzija.com/studio/", "@studio:")
                        .to_string(),
                );
            } 
        }
        if text.contains("<strong>Stars: </strong>"){
            let stars_parts: Vec<&str> = text.split("<strong>Stars: </strong>").collect::<Vec<&str>>()[1].split("</div>").collect::<Vec<&str>>()[0].split("<a href=\"").collect::<Vec<&str>>();
            for star in stars_parts.iter().skip(1) {
                if star.starts_with("https://tube.perverzija.com/stars/"){
                    tags.push(
                        star.split("/\"").collect::<Vec<&str>>()[0]
                        .replace("https://tube.perverzija.com/stars/", "@stars:")
                            .to_string(),
                    );
                } 
            }
    }

        let tags_parts: Vec<&str> = text.split("<strong>Tags: </strong>").collect::<Vec<&str>>()[1].split("</div>").collect::<Vec<&str>>()[0].split("<a href=\"").collect::<Vec<&str>>();
        for star in tags_parts.iter().skip(1) {
            if star.starts_with("https://tube.perverzija.com/stars/"){
                tags.push(
                    star.split("/\"").collect::<Vec<&str>>()[0]
                    .replace("https://tube.perverzija.com/stars/", "@stars:")
                        .to_string(),
                );
            } 
        }

        let perverzija_db_entry = PerverzijaDbEntry {
            url_string: url_str.clone(),
            tags_strings: tags.clone(),
        };
        let mut conn = pool.get().expect("couldn't get db connection from pool");
        let insert_result = db::insert_video(&mut conn, &lookup_url, &serde_json::to_string(&perverzija_db_entry)?);
        match insert_result{
            Ok(_) => (),
            Err(e) => {println!("{:?}", e); }
        }
        drop(conn);
        if !url_str.contains("xtremestream.xyz"){
            return Err("Video URL does not contain xtremestream.xyz".into());
        }
        let mut id = url_str.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());
        //     }
        // }
        // for (index, line) in vid.iter().enumerate() {
        //     println!("Line {}: {}", index, line.to_string().trim());
        // }
        

        

        let mut video_item = VideoItem::new(
            id,
            title,
            url_str.clone(),
            "perverzija".to_string(),
            thumb,
            duration,
        )
        .tags(tags);
        // .embed(embed.clone());
        let mut format =
            videos::VideoFormat::new(url_str.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,
        sort: String,
        query: Option<String>,
        page: String,
        per_page: String,
        options: ServerOptions,
    ) -> 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), options.featured.unwrap()).await,
        };
        match videos {
            Ok(v) => v,
            Err(e) => {
                println!("Error fetching videos: {}", e);
                vec![]
            }
        }
    }
}