steam_user/services/

match_history.rs

//! Match history service for CS:GO/CS2 match data retrieval.

use std::{collections::HashMap, sync::OnceLock};

use chrono::{DateTime, Duration, NaiveDateTime, Utc};
use scraper::{Html, Selector};
use sha2::{Digest, Sha256};

use crate::{
    client::SteamUser,
    endpoint::steam_endpoint,
    error::SteamUserError,
    types::match_history::{Match, MatchHistoryResponse, MatchHistoryType, MatchPlayer, Team},
    utils::avatar::get_avatar_hash_from_url,
};

/// CS:GO/CS2 App ID.
const CSGO_APP_ID: u32 = 730;

static SEL_SCOREBOARD_TABLE: OnceLock<Selector> = OnceLock::new();
fn sel_scoreboard_table() -> &'static Selector {
    SEL_SCOREBOARD_TABLE.get_or_init(|| Selector::parse("table.csgo_scoreboard_inner_right").expect("valid CSS selector"))
}

static SEL_TR: OnceLock<Selector> = OnceLock::new();
fn sel_tr() -> &'static Selector {
    SEL_TR.get_or_init(|| Selector::parse("tr").expect("valid CSS selector"))
}

static SEL_TH: OnceLock<Selector> = OnceLock::new();
fn sel_th() -> &'static Selector {
    SEL_TH.get_or_init(|| Selector::parse("th").expect("valid CSS selector"))
}

static SEL_TD: OnceLock<Selector> = OnceLock::new();
fn sel_td() -> &'static Selector {
    SEL_TD.get_or_init(|| Selector::parse("td").expect("valid CSS selector"))
}

static SEL_LEFT_TABLE_TD: OnceLock<Selector> = OnceLock::new();
fn sel_left_table_td() -> &'static Selector {
    SEL_LEFT_TABLE_TD.get_or_init(|| Selector::parse("table.csgo_scoreboard_inner_left tr > td").expect("valid CSS selector"))
}

static SEL_ANCHOR: OnceLock<Selector> = OnceLock::new();
fn sel_anchor() -> &'static Selector {
    SEL_ANCHOR.get_or_init(|| Selector::parse("a").expect("valid CSS selector"))
}

static SEL_LINK_TITLE: OnceLock<Selector> = OnceLock::new();
fn sel_link_title() -> &'static Selector {
    SEL_LINK_TITLE.get_or_init(|| Selector::parse("a.linkTitle").expect("valid CSS selector"))
}

static SEL_PLAYER_AVATAR_IMG: OnceLock<Selector> = OnceLock::new();
fn sel_player_avatar_img() -> &'static Selector {
    SEL_PLAYER_AVATAR_IMG.get_or_init(|| Selector::parse(".playerAvatar a > img[src]").expect("valid CSS selector"))
}

impl SteamUser {
    /// Retrieves the CS:GO/CS2 match history for a specific mode and tab.
    ///
    /// # Arguments
    ///
    /// * `match_type` - The match history type to fetch.
    /// * `token` - Optional continuation token for pagination.
    ///
    /// # Returns
    ///
    /// Returns a [`MatchHistoryResponse`] containing the parsed matches and the
    /// next continuation token.
    #[steam_endpoint(GET, host = Community, path = "/profiles/{steam_id}/gcpd/730/", kind = Read)]
    pub async fn get_match_history(&self, match_type: MatchHistoryType, token: Option<&str>) -> Result<MatchHistoryResponse, SteamUserError> {
        match token {
            Some(t) if !t.is_empty() => self.get_paginated_match_history(match_type, t).await,
            _ => self.get_initial_match_history(match_type).await,
        }
    }

    /// Fetches the initial match history page (HTML response).
    #[steam_endpoint(GET, host = Community, path = "/profiles/{steam_id}/gcpd/{app_id}/", kind = Read)]
    #[tracing::instrument(skip(self))]
    async fn get_initial_match_history(&self, match_type: MatchHistoryType) -> Result<MatchHistoryResponse, SteamUserError> {
        let steam_id = self.session.steam_id.ok_or(SteamUserError::NotLoggedIn)?;

        // Initial request does not use ajax=1 according to user requirement
        let response = self.get_path(format!("/profiles/{}/gcpd/{}/", steam_id.steam_id64(), CSGO_APP_ID)).query(&[("tab", match_type.as_str())]).send().await?;
        self.check_response(&response)?;

        let text = response.text().await?;

        // Extract continue token + text first (cheap string finds), then hand
        // the body off to a blocking task for the heavy HTML parse + selector
        // traversal so the async runtime stays responsive.
        let continue_token = extract_between(&text, "var g_sGcContinueToken =", ";").map(|s| s.trim().trim_matches('\'').to_string()).unwrap_or_default();
        let continue_text = extract_between(&text, "load_more_button_continue_text\" class=\"returnLink\">", "</div>").map(|s| s.to_string()).unwrap_or_default();

        let matches = tokio::task::spawn_blocking(move || parse_match_history(&text, match_type)).await.map_err(|e| SteamUserError::Other(format!("match-history parse task failed: {e}")))?;

        Ok(MatchHistoryResponse { continue_token, continue_text, matches })
    }

    /// Fetches subsequent match history pages using AJAX (JSON response).
    #[steam_endpoint(GET, host = Community, path = "/profiles/{steam_id}/gcpd/{app_id}", kind = Read)]
    #[tracing::instrument(skip(self))]
    async fn get_paginated_match_history(&self, match_type: MatchHistoryType, token: &str) -> Result<MatchHistoryResponse, SteamUserError> {
        let steam_id = self.session.steam_id.ok_or(SteamUserError::NotLoggedIn)?;

        // AJAX request uses ajax=1 and continue_token.
        // sessionid is automatically added by self.get()
        let params = vec![("ajax", "1"), ("tab", match_type.as_str()), ("continue_token", token)];

        let response = self.get_path(format!("/profiles/{}/gcpd/{}", steam_id.steam_id64(), CSGO_APP_ID)).query(&params).send().await?;
        self.check_response(&response)?;

        let json = response.json::<serde_json::Value>().await?;

        let continue_token = json.get("continue_token").and_then(|v| v.as_str()).unwrap_or("").to_string();
        let continue_text = json.get("continue_text").and_then(|v| v.as_str()).unwrap_or("").to_string();
        let html = json.get("html").and_then(|v| v.as_str()).unwrap_or("").to_string();

        // Heavy HTML parse runs on the blocking pool so the async runtime
        // stays responsive on large match-history responses.
        let matches = tokio::task::spawn_blocking(move || parse_match_history(&html, match_type)).await.map_err(|e| SteamUserError::Other(format!("match-history parse task failed: {e}")))?;

        Ok(MatchHistoryResponse { continue_token, continue_text, matches })
    }
}

/// Helper to extract a substring between two delimiters.
fn extract_between<'a>(text: &'a str, start: &str, end: &str) -> Option<&'a str> {
    let start_idx = text.find(start)? + start.len();
    let rest = &text[start_idx..];
    let end_idx = rest.find(end)?;
    Some(&rest[..end_idx])
}

/// Parses match history HTML and returns a list of Match structs.
fn parse_match_history(html: &str, match_type: MatchHistoryType) -> Vec<Match> {
    if html.is_empty() {
        return Vec::new();
    }

    let document = Html::parse_document(html);
    let mut matches = Vec::new();

    // Select each match scoreboard table
    for table in document.select(sel_scoreboard_table()) {
        if let Some(m) = parse_single_match(&document, &table, match_type) {
            matches.push(m);
        }
    }

    matches
}

/// Parses a single match from its scoreboard table.
#[tracing::instrument(skip(_document, table))]
fn parse_single_match(_document: &Html, table: &scraper::ElementRef, match_type: MatchHistoryType) -> Option<Match> {
    // Local variables to collect match info (previously in MatchInfo struct)
    let mut map: Option<String> = None;
    let mut time: Option<DateTime<Utc>> = None;
    let mut timestamp: Option<i64> = None;
    let mut wait_time: Option<Duration> = None;
    let mut duration: Option<Duration> = None;
    let mut gotv_replay: Option<String> = None;
    let mut viewers: Option<i32> = None;
    let mut ranked = false;

    // Find parent tr element to get match info
    // We need to traverse up to find the match info table on the left side
    // This is a bit tricky with scraper, so we'll parse from the document

    // Get the table's HTML to find context
    let _table_html = table.html();

    // Try to find corresponding match info by looking at sibling elements
    // For now, we'll parse the player table and extract what we can

    let rows: Vec<_> = table.select(sel_tr()).collect();

    if rows.is_empty() {
        return None;
    }

    // First row should be headers
    let headers: Vec<String> = rows.first()?.select(sel_th()).map(|h| h.text().collect::<String>().trim().to_string()).collect();

    if headers.is_empty() {
        return None;
    }

    // Parse player rows
    let mut history_table: Vec<HashMap<String, String>> = Vec::new();
    let mut player_name_data: Vec<PlayerNameData> = Vec::new();

    for (i, row) in rows.iter().enumerate() {
        if i == 0 {
            continue; // Skip header row
        }

        let cells: Vec<_> = row.select(sel_td()).collect();
        let mut row_data = HashMap::new();

        for (j, cell) in cells.iter().enumerate() {
            if j >= headers.len() {
                continue;
            }
            let header = &headers[j];

            // Check if this is the player name column
            if cell.value().attr("class").unwrap_or("").contains("inner_name") {
                let name_data = parse_player_name_cell(cell);
                player_name_data.push(name_data.clone());
                row_data.insert(header.clone(), name_data.name);
            } else {
                row_data.insert(header.clone(), cell.text().collect::<String>().trim().to_string());
            }
        }

        history_table.push(row_data);
    }

    if history_table.is_empty() {
        return None;
    }

    // Find scoreboard row (middle row with score like "16 : 14")
    let scoreboard_index = history_table.len() / 2;
    if scoreboard_index < 1 || scoreboard_index >= history_table.len() {
        return None;
    }

    // Parse score from the scoreboard row
    let score_str = history_table.get(scoreboard_index)?.get("Player Name").cloned().unwrap_or_default();

    let scores: Vec<i32> = score_str.split(':').map(|s| s.trim().parse().unwrap_or(0)).collect();

    let ct_score = scores.first().copied().unwrap_or(0);
    let t_score = scores.get(1).copied().unwrap_or(0);

    // Parse players
    let mut players = Vec::new();
    let mut name_data_iter = player_name_data.iter();

    for (i, row) in history_table.iter().enumerate() {
        // Log HTML of the corresponding row (i+1 because rows[0] is the header)
        if let Some(html_row) = rows.get(i + 1) {
            tracing::trace!(row = i, html = %html_row.html(), "match history row");
        }

        if i == scoreboard_index {
            // Skip the scoreboard row - it doesn't have inner_name class,
            // so no entry was added to player_name_data for it
            continue;
        }

        let team = if i < scoreboard_index { Team::Ct } else { Team::T };
        let name_data = name_data_iter.next();

        let mut player = MatchPlayer { team, ..Default::default() };

        if let Some(nd) = name_data {
            player.name = nd.name.clone();
            player.link = nd.link.clone();
            player.miniprofile = nd.miniprofile;
            player.avatar_hash = nd.avatar_hash.clone();
            player.custom_url = nd.custom_url.clone();

            // Convert miniprofile (account ID) to steam ID 64
            if let Some(mp) = nd.miniprofile {
                player.steam_id = Some(steamid::SteamID::from_individual_account_id(mp));
            }
        }

        // Parse numeric stats
        player.ping = row.get("Ping").and_then(|s| s.parse().ok());
        player.kills = row.get("K").and_then(|s| s.parse().ok());
        player.assists = row.get("A").and_then(|s| s.parse().ok());
        player.deaths = row.get("D").and_then(|s| s.parse().ok());
        player.score = row.get("Score").and_then(|s| s.parse().ok());

        // Parse MVP stars
        if let Some(mvp_str) = row.get("★") {
            player.mvp = parse_mvp(mvp_str);
        }

        // Parse HSP (headshot percentage)
        if let Some(hsp_str) = row.get("HSP") {
            player.hsp = hsp_str.trim().trim_end_matches('%').parse().ok();
        }

        // Validate player validity based on miniprofile (Steam Account ID) presence
        // This is much more reliable than checking for empty names or stats
        let has_valid_id = if let Some(nd) = &name_data { nd.miniprofile.is_some() } else { false };

        if !has_valid_id {
            // Only log strict warning if it looks like a ghost row that might be confusing
            if !player.name.is_empty() {
                tracing::warn!(player_name = %player.name, "skipping player row: no miniprofile (likely invalid/bot)");
            }
            continue;
        }

        players.push(player);
    }

    // Try to parse match info from the left-side table
    // This requires finding the parent structure
    // Try to parse match info from the left-side table
    // Traverse up to find the common parent (tr) that holds both left and right
    // tables potential structure: tr -> td -> table.csgo_scoreboard_inner_right
    // (current 'table') we want: tr -> td -> table.csgo_scoreboard_inner_left

    // Note: table.parent() returns NodeRef, so we need to wrap it back into
    // ElementRef to use select
    let container_row_node = table.parent().and_then(|td| td.parent());

    if let Some(row_node) = container_row_node {
        if let Some(row) = scraper::ElementRef::wrap(row_node) {
            // Scope selection to this specific row
            let left_tds = row.select(sel_left_table_td());

            for td in left_tds {
                let text = td.text().collect::<String>().trim().to_string();

                if text.starts_with("Competitive ") {
                    map = Some(text.strip_prefix("Competitive ").unwrap_or(&text).to_string());
                } else if text.starts_with("Premier ") {
                    map = Some(text.strip_prefix("Premier ").unwrap_or(&text).to_string());
                } else if text.ends_with(" GMT") {
                    // Parse timestamp
                    if let Some(dt) = parse_match_timestamp(&text) {
                        time = Some(dt);
                        timestamp = Some(dt.timestamp_millis());
                    }
                } else if text.starts_with("Wait Time: ") {
                    wait_time = text.strip_prefix("Wait Time: ").and_then(|s| parse_duration(s.trim()));
                } else if text.starts_with("Match Duration: ") {
                    duration = text.strip_prefix("Match Duration: ").and_then(|s| parse_duration(s.trim()));
                } else if text == "Download GOTV Replay" || text == "Download Replay" || text == "Tải bản phát lại" {
                    // Find the link
                    if let Some(a) = td.select(sel_anchor()).next() {
                        gotv_replay = a.value().attr("href").map(|s| s.to_string());
                    }
                } else if text.starts_with("Viewers: ") {
                    viewers = text.strip_prefix("Viewers: ").and_then(|s| s.parse().ok());
                } else if text.starts_with("Ranked: Yes") {
                    ranked = true;
                }
            }
        }
    }

    // Generate match hash
    let match_hash = generate_match_hash(map.as_deref(), time, duration, &players);

    Some(Match {
        match_hash,
        map,
        time,
        timestamp,
        wait_time,
        duration,
        gotv_replay,
        viewers,
        ranked,
        players,
        scoreboard: [ct_score, t_score],
        match_type,
    })
}

/// Player name data extracted from the name cell.
#[derive(Debug, Clone, Default)]
struct PlayerNameData {
    name: String,
    link: Option<String>,
    miniprofile: Option<u32>,
    avatar_hash: Option<String>,
    custom_url: Option<String>,
}

/// Parses the player name cell to extract name, link, miniprofile, and avatar.
fn parse_player_name_cell(cell: &scraper::ElementRef) -> PlayerNameData {
    let mut data = PlayerNameData::default();

    // Find the link - this is where the actual player name is stored
    if let Some(a) = cell.select(sel_link_title()).next() {
        // Get the name directly from the anchor element (most reliable)
        data.name = a.text().collect::<String>().trim().to_string();
        data.link = a.value().attr("href").map(|s| s.to_string());
        data.miniprofile = a.value().attr("data-miniprofile").and_then(|s| s.parse().ok());

        if data.miniprofile.is_none() {
            tracing::warn!(player_name = %data.name, "failed to parse miniprofile for player");
            if let Some(mp_str) = a.value().attr("data-miniprofile") {
                tracing::warn!(data_miniprofile = %mp_str, "data-miniprofile attribute present but unparseable");
            } else {
                tracing::warn!("data-miniprofile attribute missing");
            }
        }

        // Extract custom URL from link
        if let Some(ref link) = data.link {
            data.custom_url = get_custom_url_from_profile_url(link);
        }
    } else {
        tracing::warn!(player_name = %data.name, "no profile link found for player");
    }

    // Find avatar
    if let Some(img) = cell.select(sel_player_avatar_img()).next() {
        if let Some(src) = img.value().attr("src") {
            data.avatar_hash = get_avatar_hash_from_url(src);
        }
    }

    data
}

// get_avatar_hash_from_url is imported from crate::utils::avatar

/// Extracts custom URL from profile URL.
fn get_custom_url_from_profile_url(url: &str) -> Option<String> {
    if url.contains("/id/") {
        url.split("/id/").nth(1).map(|s| s.trim_end_matches('/').to_string())
    } else {
        None
    }
}

/// Parses match timestamp string to DateTime<Utc>.
fn parse_match_timestamp(time_str: &str) -> Option<DateTime<Utc>> {
    // Format: "2024-01-01 12:00:00 GMT"
    let clean = time_str.replace(" GMT", "").trim().to_string();
    NaiveDateTime::parse_from_str(&clean, "%Y-%m-%d %H:%M:%S").ok().map(|ndt| DateTime::from_naive_utc_and_offset(ndt, Utc))
}

/// Parses a duration string like "MM:SS" or "HH:MM:SS" into chrono::Duration.
fn parse_duration(s: &str) -> Option<Duration> {
    let parts: Vec<&str> = s.split(':').collect();
    match parts.len() {
        2 => {
            // MM:SS
            let mins: i64 = parts[0].parse().ok()?;
            let secs: i64 = parts[1].parse().ok()?;
            Some(Duration::minutes(mins) + Duration::seconds(secs))
        }
        3 => {
            // HH:MM:SS
            let hours: i64 = parts[0].parse().ok()?;
            let mins: i64 = parts[1].parse().ok()?;
            let secs: i64 = parts[2].parse().ok()?;
            Some(Duration::hours(hours) + Duration::minutes(mins) + Duration::seconds(secs))
        }
        _ => None,
    }
}

/// Parses MVP string to count.
fn parse_mvp(mvp_str: &str) -> Option<i32> {
    let trimmed = mvp_str.trim();
    if trimmed.is_empty() {
        Some(0)
    } else if trimmed == "★" {
        Some(1)
    } else if trimmed.starts_with("★") {
        trimmed.replace("★", "").trim().parse().ok()
    } else {
        Some(0)
    }
}

/// Generates a unique match hash from match info.
///
/// Player miniprofiles (sorted) are included so that matches with partial
/// parse failures (all-None metadata) still produce distinct hashes instead
/// of all colliding on `sha256("")`.
fn generate_match_hash(map: Option<&str>, time: Option<DateTime<Utc>>, duration: Option<Duration>, players: &[MatchPlayer]) -> String {
    let mut miniprofiles: Vec<u32> = players.iter().filter_map(|p| p.miniprofile).collect();
    miniprofiles.sort_unstable();
    let players_part: String = miniprofiles.iter().map(|id| id.to_string()).collect::<Vec<_>>().join(",");

    let id_parts = format!("{}{}{}{}", map.unwrap_or(""), time.map(|t| t.to_rfc3339()).unwrap_or_default(), duration.map(|d| d.num_seconds().to_string()).unwrap_or_default(), players_part,);

    // Remove non-alphanumeric characters and commas are already alphanumeric-safe,
    // but we keep the filter for the other fields.
    let clean: String = id_parts.chars().filter(|c| c.is_alphanumeric() || *c == ',').collect();

    let mut hasher = Sha256::new();
    hasher.update(clean.as_bytes());
    hex::encode(hasher.finalize())
}