more team match data:

2026-06-03 10:14:11 +02:00 · 2025-04-20 17:13:51 +02:00 · 2025-04-20 17:13:51 +02:00 · f3aadfe03d
commit f3aadfe03d
parent cc3ea78778
5 changed files with 5316 additions and 5145 deletions
--- a/src/lib.rs
+++ b/src/lib.rs
@ -5,7 +5,7 @@ pub mod parser;

 use crate::data::{
    GameMode, MapHistory, MatchInfo, MembershipHistory, Player, Seasons, Team, TeamRef,
-    TeamRosterData, TeamSeason, Transaction,
+    TeamRosterData, Transaction,
 };
 use crate::parser::{
    MapHistoryParser, MatchPageParser, Parser, PlayerDetailsParser, PlayerParser, SeasonsParser,
@ -18,6 +18,7 @@ use std::time::Duration;
 pub use steamid_ng::SteamID;
 use tokio::time::sleep;
 use tracing::warn;
+use ugc_scraper_types::TeamMatches;

 pub type Result<T, E = ScrapeError> = std::result::Result<T, E>;

@ -136,7 +137,7 @@ impl UgcClient {
    }

    /// Retrieve team match history
-    pub async fn team_matches(&self, id: u32) -> Result<Vec<TeamSeason>> {
+    pub async fn team_matches(&self, id: u32) -> Result<TeamMatches> {
        let body = self
            .request(format!(
                "https://www.ugcleague.com/team_page_matches.cfm?clan_id={}",
--- a/src/parser/team_matches.rs
+++ b/src/parser/team_matches.rs
@ -1,10 +1,13 @@
 use super::Parser;
-use crate::data::{MatchResult, TeamRef, TeamSeason, TeamSeasonMatch};
+use crate::data::{GameMode, MatchResult, TeamRef, TeamSeason, TeamSeasonMatch};
 use crate::parser::{match_id_from_link, select_text, team_id_from_link, ElementExt};
 use crate::{ParseError, Result};
 use scraper::{Html, Selector};
+use std::str::FromStr;
+use ugc_scraper_types::{Side, TeamMatches};

-const SELECTOR_SEASON_TITLE: &str =
+const SELECTOR_SEASON_TITLE: &str = ".container table.table.table-condensed.table-striped thead h4";
+const SELECTOR_SEASON_SEASON: &str =
    ".container table.table.table-condensed.table-striped thead h4 b";
 const SELECTOR_SEASON_MATCHES: &str =
    ".container table.table.table-condensed.table-striped tbody:nth-child(3n)";
@ -20,8 +23,12 @@ const SELECTOR_SEASON_POINTS: &str = "td:nth-child(9) small";
 const SELECTOR_SEASON_POINTS_OPPONENTS: &str = "td:nth-child(10) small";
 const SELECTOR_SEASON_MATCH_PAGE: &str = "td a[href^=\"matchpage\"]";

+const SELECTOR_TEAM_NAME: &str = r#"div.col-md-9 > h2 > b"#;
+const SELECTOR_TEAM_LINK: &str = r#"h2 > span.pull-right > a[href^="team_page.cfm"]"#;
+
 pub struct TeamMatchesParser {
    selector_title: Selector,
+    selector_season: Selector,
    selector_matches: Selector,
    selector_match: Selector,
    selector_division: Selector,
@ -34,6 +41,9 @@ pub struct TeamMatchesParser {
    selector_points: Selector,
    selector_points_opponent: Selector,
    selector_match_page: Selector,
+
+    selector_team_name: Selector,
+    selector_team_link: Selector,
 }

 impl Default for TeamMatchesParser {
@ -46,6 +56,7 @@ impl TeamMatchesParser {
    pub fn new() -> Self {
        TeamMatchesParser {
            selector_title: Selector::parse(SELECTOR_SEASON_TITLE).unwrap(),
+            selector_season: Selector::parse(SELECTOR_SEASON_SEASON).unwrap(),
            selector_matches: Selector::parse(SELECTOR_SEASON_MATCHES).unwrap(),
            selector_match: Selector::parse(SELECTOR_SEASON_MATCH).unwrap(),
            selector_division: Selector::parse(SELECTOR_SEASON_DIVISION).unwrap(),
@ -58,27 +69,44 @@ impl TeamMatchesParser {
            selector_points: Selector::parse(SELECTOR_SEASON_POINTS).unwrap(),
            selector_points_opponent: Selector::parse(SELECTOR_SEASON_POINTS_OPPONENTS).unwrap(),
            selector_match_page: Selector::parse(SELECTOR_SEASON_MATCH_PAGE).unwrap(),
+
+            selector_team_name: Selector::parse(SELECTOR_TEAM_NAME).unwrap(),
+            selector_team_link: Selector::parse(SELECTOR_TEAM_LINK).unwrap(),
        }
    }
 }

 impl Parser for TeamMatchesParser {
-    type Output = Vec<TeamSeason>;
+    type Output = TeamMatches;

    fn parse(&self, document: &str) -> Result<Self::Output> {
        let document = Html::parse_document(document);

-        document
+        let seasons = document
            .select(&self.selector_title)
+            .zip(document.select(&self.selector_season))
            .zip(document.select(&self.selector_matches))
-            .map(|(title, matches)| {
-                let title = title.first_text().ok_or(ParseError::EmptyText {
+            .map(|((title, season), matches)| {
+                let format = title.first_text().ok_or(ParseError::EmptyText {
                    selector: SELECTOR_SEASON_TITLE,
                    role: "season title",
                })?;
-                let season: u32 = title.trim_start_matches("Season ").parse().map_err(|_| {
+
+                let format = format
+                    .split(' ')
+                    .find_map(|part| GameMode::from_str(part).ok())
+                    .ok_or(ParseError::InvalidText {
+                        text: format.into(),
+                        role: "season format",
+                    })?;
+
+                let season = season.first_text().ok_or(ParseError::EmptyText {
+                    selector: SELECTOR_SEASON_SEASON,
+                    role: "season title",
+                })?;
+                let season: u32 = season.trim_start_matches("Season ").parse().map_err(|_| {
                    ParseError::InvalidText {
-                        text: title.to_string(),
+                        text: season.to_string(),
                        role: "season title",
                    }
                })?;
@ -167,10 +195,7 @@ impl Parser for TeamMatchesParser {

                        let opponent = opponent_link
                            .map(|link| {
-                                let name = link.first_text().ok_or(ParseError::EmptyText {
-                                    selector: SELECTOR_SEASON_OPPONENT,
-                                    role: "match opponent",
-                                })?;
+                                let name = link.first_text().unwrap_or_default();
                                let id = team_id_from_link(link.attr("href").unwrap_or_default())?;
                                Result::<_, ParseError>::Ok(TeamRef {
                                    name: name.to_string(),
@ -208,7 +233,12 @@ impl Parser for TeamMatchesParser {
                        Ok(TeamSeasonMatch {
                            week,
                            date: date.to_string(),
-                            side: side.to_string(),
+                            side: side.parse::<Side>().map_err(|error| {
+                                ParseError::InvalidText {
+                                    text: error.text,
+                                    role: "match side",
+                                }
+                            })?,
                            map: map.to_string(),
                            division: division.to_string(),
                            result,
@ -216,8 +246,34 @@ impl Parser for TeamMatchesParser {
                    })
                    .collect::<Result<_>>()?;

-                Ok(TeamSeason { season, matches })
+                Ok(TeamSeason {
+                    season,
+                    matches,
+                    format,
                })
-            .collect::<Result<Vec<_>>>()
+            })
+            .collect::<Result<Vec<_>>>()?;
+
+        let team_id = document
+            .select(&self.selector_team_link)
+            .next()
+            .and_then(|link| team_id_from_link(link.attr("href").unwrap_or_default()).ok())
+            .ok_or(ParseError::ElementNotFound {
+                selector: SELECTOR_TEAM_LINK,
+                role: "match team link",
+            })?;
+
+        let team_name = select_text(document.root_element(), &self.selector_team_name).ok_or(
+            ParseError::ElementNotFound {
+                selector: SELECTOR_TEAM_NAME,
+                role: "match team name",
+            },
+        )?;
+        let team = TeamRef {
+            id: team_id,
+            name: team_name.into(),
+        };
+
+        Ok(TeamMatches { team, seasons })
    }
 }
--- a/tests/snapshots/snapshot__parse_team_matches_html.snap
+++ b/tests/snapshots/snapshot__parse_team_matches_html.snap
@ -2,9 +2,15 @@
 source: tests/snapshot.rs
 expression: parsed
 ---
-[
+{
+  "team": {
+    "name": "Xenon",
+    "id": 7861
+  },
+  "seasons": [
    {
      "season": 41,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -113,6 +119,7 @@ expression: parsed
    },
    {
      "season": 40,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -270,6 +277,7 @@ expression: parsed
    },
    {
      "season": 39,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -437,6 +445,7 @@ expression: parsed
    },
    {
      "season": 38,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Silver",
@ -604,6 +613,7 @@ expression: parsed
    },
    {
      "season": 37,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Silver",
@ -800,6 +810,7 @@ expression: parsed
    },
    {
      "season": 36,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -948,6 +959,7 @@ expression: parsed
    },
    {
      "season": 35,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -1105,6 +1117,7 @@ expression: parsed
    },
    {
      "season": 34,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -1281,6 +1294,7 @@ expression: parsed
    },
    {
      "season": 32,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -1438,6 +1452,7 @@ expression: parsed
    },
    {
      "season": 31,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -1595,6 +1610,7 @@ expression: parsed
    },
    {
      "season": 30,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -1752,6 +1768,7 @@ expression: parsed
    },
    {
      "season": 29,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -1909,6 +1926,7 @@ expression: parsed
    },
    {
      "season": 28,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -2028,6 +2046,7 @@ expression: parsed
    },
    {
      "season": 27,
+      "format": "9v9",
      "matches": [
        {
          "division": "Premium EU",
@ -2195,6 +2214,7 @@ expression: parsed
    },
    {
      "season": 26,
+      "format": "9v9",
      "matches": [
        {
          "division": "Premium EU",
@ -2352,6 +2372,7 @@ expression: parsed
    },
    {
      "season": 25,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -2566,6 +2587,7 @@ expression: parsed
    },
    {
      "season": 24,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -2723,6 +2745,7 @@ expression: parsed
    },
    {
      "season": 23,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -2928,6 +2951,7 @@ expression: parsed
    },
    {
      "season": 22,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -3123,6 +3147,7 @@ expression: parsed
    },
    {
      "season": 21,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -3271,6 +3296,7 @@ expression: parsed
    },
    {
      "season": 20,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Platinum",
@ -3409,6 +3435,7 @@ expression: parsed
    },
    {
      "season": 19,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Gold",
@ -3614,6 +3641,7 @@ expression: parsed
    },
    {
      "season": 18,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Steel",
@ -3828,6 +3856,7 @@ expression: parsed
    },
    {
      "season": 17,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Silver",
@ -3976,6 +4005,7 @@ expression: parsed
    },
    {
      "season": 16,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Silver",
@ -4133,6 +4163,7 @@ expression: parsed
    },
    {
      "season": 15,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Silver",
@ -4290,6 +4321,7 @@ expression: parsed
    },
    {
      "season": 14,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Silver",
@ -4429,6 +4461,7 @@ expression: parsed
    },
    {
      "season": 13,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Silver",
@ -4624,6 +4657,7 @@ expression: parsed
    },
    {
      "season": 12,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Silver",
@ -4819,6 +4853,7 @@ expression: parsed
    },
    {
      "season": 11,
+      "format": "9v9",
      "matches": [
        {
          "division": "Euro Steel",
@ -4994,3 +5029,4 @@ expression: parsed
      ]
    }
  ]
+}
--- a/tests/snapshots/snapshot__parse_team_matches_old_html.snap
+++ b/tests/snapshots/snapshot__parse_team_matches_old_html.snap
@ -2,9 +2,15 @@
 source: tests/snapshot.rs
 expression: parsed
 ---
-[
+{
+  "team": {
+    "name": "ChaosTheory",
+    "id": 2157
+  },
+  "seasons": [
    {
      "season": 1,
+      "format": "8v8",
      "matches": [
        {
          "division": "Steel",
@ -145,3 +151,4 @@ expression: parsed
      ]
    }
  ]
+}
--- a/types/src/lib.rs
+++ b/types/src/lib.rs
@ -261,24 +261,96 @@ pub struct RosterHistory {
    pub role: MembershipRole,
 }

+#[derive(Debug, Clone)]
+#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
+pub struct TeamMatches {
+    pub team: TeamRef,
+    pub seasons: Vec<TeamSeason>,
+}
+
 #[derive(Debug, Clone)]
 #[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
 pub struct TeamSeason {
    pub season: u32,
+    pub format: GameMode,
    pub matches: Vec<TeamSeasonMatch>,
 }

+#[derive(Debug, Clone, PartialEq, Eq)]
+#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
+#[cfg_attr(feature = "serde", serde(rename_all = "snake_case"))]
+pub enum Side {
+    Home,
+    Visiting,
+}
+
+#[derive(Debug, Clone, Error)]
+#[error("Invalid side {text}")]
+pub struct InvalidSide {
+    pub text: String,
+}
+
+impl FromStr for Side {
+    type Err = InvalidSide;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        match s {
+            "home" => Ok(Side::Home),
+            "visiting" => Ok(Side::Visiting),
+            _ => Err(InvalidSide { text: s.into() }),
+        }
+    }
+}
+
 #[derive(Debug, Clone)]
 #[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
 pub struct TeamSeasonMatch {
    pub division: String,
    pub week: u8,
    pub date: String,
-    pub side: String,
+    pub side: Side,
    pub result: MatchResult,
    pub map: String,
 }

+impl TeamSeasonMatch {
+    pub fn match_info(&self, team: &TeamRef, format: GameMode) -> Option<MatchInfo> {
+        match &self.result {
+            MatchResult::Played {
+                opponent,
+                score,
+                score_opponent,
+                ..
+            }
+            | MatchResult::Pending {
+                opponent,
+                score,
+                score_opponent,
+                ..
+            } => {
+                let (team_home, team_away, score_home, score_away) = if self.side == Side::Home {
+                    (team.clone(), opponent.clone(), *score, *score_opponent)
+                } else {
+                    (opponent.clone(), team.clone(), *score_opponent, *score)
+                };
+                Some(MatchInfo {
+                    comment: None,
+                    comment_author: None,
+                    team_home,
+                    team_away,
+                    score_home,
+                    score_away,
+                    map: self.map.clone(),
+                    week: self.week,
+                    format,
+                    default_date: self.date.clone(),
+                })
+            }
+            _ => None,
+        }
+    }
+}
+
 #[derive(Debug, Clone)]
 #[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
 #[cfg_attr(feature = "serde", serde(rename_all = "snake_case", tag = "state"))]
@ -306,8 +378,7 @@ pub enum MatchResult {
 impl MatchResult {
    pub fn match_id(&self) -> Option<u32> {
        match self {
-            MatchResult::Played { id, .. } => Some(*id),
-            MatchResult::Pending { id, .. } => Some(*id),
+            MatchResult::Played { id, .. } | MatchResult::Pending { id, .. } => Some(*id),
            _ => None,
        }
    }