refactor: enhance data source

src/data_source.rs

@@ -7,18 +7,27 @@ use serde_json::{Value, json};
 use std::fs;
 use std::path::Path;
 
+/// Represents a full item set for a champion, ready to be serialized to JSON.
 #[derive(Serialize, Deserialize)]
 struct ItemSet {
+    /// The title of the item set.
     title: String,
+    /// The type of the item set (e.g., "custom").
     #[serde(rename = "type")]
     type_: String,
+    /// The map this item set is for (e.g., "any").
     map: String,
+    /// The mode this item set is for (e.g., "any").
     mode: String,
+    /// Whether this item set has priority.
     priority: bool,
+    /// The sort rank of the item set.
     sortrank: u32,
+    /// The blocks of items in the set.
     blocks: Vec<Value>,
 }
 
+/// Represents a build (block) of items.
 #[derive(Serialize, Deserialize)]
 pub struct Build {
     #[serde(rename = "type")]
@@ -26,6 +35,7 @@ pub struct Build {
     pub items: Vec<Item>,
 }
 
+/// Represents a single item in a build.
 #[derive(Serialize, Deserialize)]
 pub struct Item {
     pub id: String,
@@ -39,26 +49,33 @@ pub struct Stat {
     pub patch: String,
 }
 
+/// Trait for a data source that can provide champion item set data.
 pub trait DataSource {
+    /// Returns the alias of the data source.
     fn get_alias(&self) -> &str;
 
+    /// Returns the timeout for the data source.
     fn get_timeout(&self) -> u64;
 
+    /// Returns a map of champion IDs to their possible positions.
     fn get_champs_with_positions(&self, champion: &Champion) -> IndexMap<u32, Vec<String>>;
 
-    fn make_item_set(&self, items: Vec<&str>, label: String) -> Value {
+    /// Creates a JSON value representing an item set block from a list of item IDs and a label.
+    fn make_item_set(&self, items: Vec<String>, label: String) -> Value {
         json!({
-            "items": items.iter().map(|x| json!({"id": x.to_string(), "count": 1})).collect::<Vec<Value>>(),
+            "items": items.iter().map(|x| json!({"id": x, "count": 1})).collect::<Vec<Value>>(),
             "type": label
         })
     }
 
+    /// Returns champion data with win percentage for the given positions.
     fn get_champ_data_with_win_pourcentage(
         &self,
         champ: &ChampInfo,
         positions: &[String],
     ) -> Vec<(String, Vec<Value>, Stat)>;
 
+    /// Writes item sets for the given champion and positions to the specified path.
     fn write_item_set(
         &self,
         champ: &ChampInfo,
@@ -73,19 +90,11 @@ pub trait DataSource {
         );
         let data = self.get_champ_data_with_win_pourcentage(champ, positions);
 
-        let mut missing_roles = vec![];
+        let missing_roles: Vec<_> = positions
-        for pos in positions {
+            .iter()
-            let mut ok = false;
+            .filter(|pos| !data.iter().any(|build| &build.0 == *pos))
-            for build in &data {
+            .cloned()
-                if build.0 == *pos {
+            .collect();
-                    ok = true;
-                    break;
-                }
-            }
-            if !ok {
-                missing_roles.push(pos.to_owned());
-            }
-        }
         if !missing_roles.is_empty() {
             error!(
                 "{}: Can't get data for {} at {}",

src/kb_data_source.rs

@@ -1,10 +1,10 @@
 use crate::ChampInfo;
 use crate::Champion as ChampionLoL;
-use crate::data_source::{Build, DataSource, Item, Stat};
+use crate::data_source::{DataSource, Item, Stat};
 use indexmap::IndexMap;
 use log::error;
 use serde_derive::Deserialize;
-use serde_json::{Value, json};
+use serde_json::Value;
 
 pub struct KBDataSource {
     client: ureq::Agent,
@@ -160,127 +160,74 @@ impl KBDataSource {
     }
 
     fn get_positions(position: Option<Position>) -> Vec<String> {
-        let mut positions = Vec::new();
+        let positions = vec!["TOP", "JUNGLE", "MID", "BOT", "SUPPORT"];
         if let Some(pos) = position {
-            if pos.top > 0 {
+            positions
-                positions.push("TOP".to_owned());
+                .into_iter()
-            }
+                .zip([pos.top, pos.jungle, pos.mid, pos.bot, pos.support])
-            if pos.jungle > 0 {
+                .filter_map(|(name, count)| {
-                positions.push("JUNGLE".to_owned());
+                    if count > 0 {
-            }
+                        Some(name.to_owned())
-            if pos.mid > 0 {
+                    } else {
-                positions.push("MID".to_owned());
+                        None
-            }
+                    }
-            if pos.bot > 0 {
+                })
-                positions.push("BOT".to_owned());
+                .collect()
-            }
+        } else {
-            if pos.support > 0 {
+            positions.into_iter().map(|s| s.to_owned()).collect()
-                positions.push("SUPPORT".to_owned());
-            }
         }
-
-        // TODO: find better solution, activate all positions for retrieve older builds
-        if positions.is_empty() {
-            positions.push("TOP".to_owned());
-            positions.push("JUNGLE".to_owned());
-            positions.push("MID".to_owned());
-            positions.push("BOT".to_owned());
-            positions.push("SUPPORT".to_owned());
-        }
-
-        positions
     }
 
     fn get_build(&self, build: &KBBuild) -> (String, Vec<Value>, Stat) {
         let mut starting_items: Vec<Item> = vec![];
         let mut blocks = vec![];
-        if build.str_item_sets[0].str_item0.item_id != 0 {
+        for i in 0..6 {
-            starting_items.push(Item {
+            let item_id = match i {
-                id: build.str_item_sets[0].str_item0.item_id.to_string(),
+                0 => build.str_item_sets[0].str_item0.item_id,
-                count: 1,
+                1 => build.str_item_sets[0].str_item1.item_id,
-            })
+                2 => build.str_item_sets[0].str_item2.item_id,
+                3 => build.str_item_sets[0].str_item3.item_id,
+                4 => build.str_item_sets[0].str_item4.item_id,
+                5 => build.str_item_sets[0].str_item5.item_id,
+                _ => 0,
+            };
+            if item_id != 0 {
+                starting_items.push(Item {
+                    id: item_id.to_string(),
+                    count: 1,
+                });
+            }
         }
-        if build.str_item_sets[0].str_item1.item_id != 0 {
+        blocks.push(self.make_item_set(
-            starting_items.push(Item {
+            starting_items.iter().map(|item| item.id.clone()).collect(),
-                id: build.str_item_sets[0].str_item1.item_id.to_string(),
+            format!(
-                count: 1,
-            })
-        }
-        if build.str_item_sets[0].str_item2.item_id != 0 {
-            starting_items.push(Item {
-                id: build.str_item_sets[0].str_item2.item_id.to_string(),
-                count: 1,
-            })
-        }
-        if build.str_item_sets[0].str_item3.item_id != 0 {
-            starting_items.push(Item {
-                id: build.str_item_sets[0].str_item3.item_id.to_string(),
-                count: 1,
-            })
-        }
-        if build.str_item_sets[0].str_item4.item_id != 0 {
-            starting_items.push(Item {
-                id: build.str_item_sets[0].str_item4.item_id.to_string(),
-                count: 1,
-            })
-        }
-        if build.str_item_sets[0].str_item5.item_id != 0 {
-            starting_items.push(Item {
-                id: build.str_item_sets[0].str_item5.item_id.to_string(),
-                count: 1,
-            })
-        }
-        blocks.push(json!(Build {
-            type_: format!(
                 "Early game items | skillOrder : {}",
                 build.skill_sets[0].skill_order
             ),
-            items: starting_items
+        ));
-        }));
         let mut final_items: Vec<Item> = vec![];
-        if build.item_sets[0].item0.item_id != 0 {
+        for item in [
-            final_items.push(Item {
+            &build.item_sets[0].item0,
-                id: build.item_sets[0].item0.item_id.to_string(),
+            &build.item_sets[0].item1,
-                count: 1,
+            &build.item_sets[0].item2,
-            })
+            &build.item_sets[0].item3,
+            &build.item_sets[0].item4,
+            &build.item_sets[0].item5,
+        ] {
+            if item.item_id != 0 {
+                final_items.push(Item {
+                    id: item.item_id.to_string(),
+                    count: 1,
+                });
+            }
         }
-        if build.item_sets[0].item1.item_id != 0 {
+        blocks.push(self.make_item_set(
-            final_items.push(Item {
+            final_items.iter().map(|item| item.id.clone()).collect(),
-                id: build.item_sets[0].item1.item_id.to_string(),
+            format!(
-                count: 1,
-            })
-        }
-        if build.item_sets[0].item2.item_id != 0 {
-            final_items.push(Item {
-                id: build.item_sets[0].item2.item_id.to_string(),
-                count: 1,
-            })
-        }
-        if build.item_sets[0].item3.item_id != 0 {
-            final_items.push(Item {
-                id: build.item_sets[0].item3.item_id.to_string(),
-                count: 1,
-            })
-        }
-        if build.item_sets[0].item4.item_id != 0 {
-            final_items.push(Item {
-                id: build.item_sets[0].item4.item_id.to_string(),
-                count: 1,
-            })
-        }
-        if build.item_sets[0].item5.item_id != 0 {
-            final_items.push(Item {
-                id: build.item_sets[0].item5.item_id.to_string(),
-                count: 1,
-            })
-        }
-        blocks.push(json!(Build {
-            type_: format!(
                 "Item order by time finished | Summoner : {}",
                 build.summoner.name
             ),
-            items: final_items
+        ));
-        }));
 
         (
             build.position.position_name.to_uppercase(),
@@ -325,12 +272,13 @@ impl DataSource for KBDataSource {
     ) -> Vec<(String, Vec<Value>, Stat)> {
         let mut champ_data = vec![];
         if let Some(token) = &self.token {
+            let url = format!(
+                "https://api.koreanbuilds.net/builds?chmpname={}&patchid=-2&position=COMPOSITE",
+                champ.name.replace(" ", "%20")
+            );
             let data: BuildResponse = match self
                 .client
-                .get(&format!(
+                .get(&url)
-                    "https://api.koreanbuilds.net/builds?chmpname={}&patchid=-2&position=COMPOSITE",
-                    champ.name
-                ))
                 .header("Accept", "application/json")
                 .header("Authorization", token.as_str())
                 .call()
@@ -343,7 +291,7 @@ impl DataSource for KBDataSource {
                     }
                 },
                 Err(x) => {
-                    error!("Call failed: {}", x);
+                    error!("Call failed for URL: {}, error: {}", url, x);
                     return vec![];
                 }
             };

src/main.rs

@@ -26,6 +26,7 @@ mod ms_data_source;
 use data_source::DataSource;
 use kb_data_source::KBDataSource;
 use ms_data_source::MSDataSource;
+use serde::{self, Deserialize, Deserializer};
 
 #[derive(Deserialize)]
 struct Realm {
@@ -41,7 +42,16 @@ pub struct Champion {
 pub struct ChampInfo {
     id: String,
     name: String,
-    key: String,
+    #[serde(deserialize_with = "from_string_to_u32")]
+    key: u32,
+}
+
+fn from_string_to_u32<'de, D>(deserializer: D) -> Result<u32, D::Error>
+where
+    D: Deserializer<'de>,
+{
+    let s = String::deserialize(deserializer)?;
+    s.parse::<u32>().map_err(serde::de::Error::custom)
 }
 
 const DEFAULT_LOL_CHAMPS_DIR: &str = "./champs";
@@ -56,40 +66,16 @@ const REG_KEY_WIN_64_UNINSTALL: &str =
 const REG_KEY_WIN_UNINSTALL: &str = r"SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall\";
 
 fn main() -> Result<(), Box<dyn std::error::Error>> {
-    let args: Vec<String> = env::args().collect();
+    setup_logging_from_args();
-    let mut level = LevelFilter::Info;
+    info!("CGG Item Sets v{}", env!("CARGO_PKG_VERSION"));
-    for s in &args {
-        if s.eq_ignore_ascii_case("-v") || s.eq_ignore_ascii_case("--verbose") {
-            level = LevelFilter::Debug;
-            break;
-        }
-    }
-    logsy::set_echo(true);
-    logsy::set_level(level);
-    info!("CGG Item Sets");
 
-    let lol_champs_dir: PathBuf = match lol_champ_dir() {
+    let lol_champs_dir = get_lol_champs_dir();
-        Ok(x) => x,
-        Err(_e) => PathBuf::from(DEFAULT_LOL_CHAMPS_DIR),
-    };
     info!("LoL Champs Folder: {}", lol_champs_dir.display());
 
-    let client: Agent = create_http_client();
+    let client = create_http_client();
-
+    let realm = fetch_realm(&client)?;
-    let realm: Realm = client
-        .get("https://ddragon.leagueoflegends.com/realms/euw.json")
-        .call()?
-        .body_mut()
-        .read_json()?;
     info!("LoL version: {}", realm.v);
-    let champion: Champion = client
+    let champion = fetch_champions(&client, &realm)?;
-        .get(&format!(
-            "https://ddragon.leagueoflegends.com/cdn/{}/data/en_US/champion.json",
-            realm.v
-        ))
-        .call()?
-        .body_mut()
-        .read_json()?;
     info!("LoL numbers of champs: {}", champion.data.len());
 
     let data_sources: Vec<Box<dyn DataSource + Sync + Send>> = vec![
@@ -108,13 +94,54 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
     Ok(())
 }
 
-fn get_champ_from_key(champs: &Champion, key: u32) -> Option<String> {
+fn setup_logging_from_args() {
-    for champ in champs.data.values() {
+    let args: Vec<String> = env::args().collect();
-        if key.to_string() == champ.key {
+    let mut level = LevelFilter::Info;
-            return Some(champ.id.to_owned());
+    for s in &args {
+        if s.eq_ignore_ascii_case("-v") || s.eq_ignore_ascii_case("--verbose") {
+            level = LevelFilter::Debug;
+            break;
         }
     }
-    None
+    logsy::set_echo(true);
+    logsy::set_level(level);
+}
+
+fn get_lol_champs_dir() -> PathBuf {
+    match lol_champ_dir() {
+        Ok(x) => x,
+        Err(e) => {
+            error!("Failed to detect LoL champs dir: {e}, using default.");
+            PathBuf::from(DEFAULT_LOL_CHAMPS_DIR)
+        }
+    }
+}
+
+fn fetch_realm(client: &Agent) -> Result<Realm, Box<dyn std::error::Error>> {
+    let url = "https://ddragon.leagueoflegends.com/realms/euw.json";
+    info!("Fetching realm info from {url}");
+    let realm: Realm = client.get(url).call()?.body_mut().read_json()?;
+    Ok(realm)
+}
+
+fn fetch_champions(client: &Agent, realm: &Realm) -> Result<Champion, Box<dyn std::error::Error>> {
+    let url = format!(
+        "https://ddragon.leagueoflegends.com/cdn/{}/data/en_US/champion.json",
+        realm.v
+    );
+    info!("Fetching champion data from {url}");
+    let champion: Champion = client.get(&url).call()?.body_mut().read_json()?;
+    Ok(champion)
+}
+
+fn get_champ_from_key(champs: &Champion, key: u32) -> Option<String> {
+    champs.data.values().find_map(|champ| {
+        if champ.key == key {
+            Some(champ.id.clone())
+        } else {
+            None
+        }
+    })
 }
 
 fn execute_data_source(

src/ms_data_source.rs

@@ -1,8 +1,9 @@
 use indexmap::IndexMap;
-use log::error;
+use log::{error, warn};
 use regex::Regex;
 use serde_derive::Deserialize;
 use serde_json::Value;
+use std::sync::LazyLock;
 
 use crate::ChampInfo;
 use crate::Champion;
@@ -19,25 +20,116 @@ struct MSChampion {
     search_terms: String,
 }
 
-fn get_champ_from_name(champs: &Champion, name: String) -> Option<u32> {
+// Compile regexes once for performance
-    for champ in champs.data.values() {
+static NUMBER_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"([0-9]+\.?[0-9]+)").unwrap());
+static ITEM_REGEX: LazyLock<Regex> = LazyLock::new(|| Regex::new(r"/item/([0-9]+)\.").unwrap());
+
+fn get_champ_from_name(champs: &Champion, name: &str) -> Option<u32> {
+    champs.data.values().find_map(|champ| {
         if name == champ.name || name == champ.id {
-            return champ.key.parse::<u32>().ok();
+            Some(champ.key)
+        } else {
+            None
+        }
+    })
+}
+
+fn find_next_number(rest: &str) -> f32 {
+    if let Some(cap) = NUMBER_REGEX.captures(rest) {
+        if let Some(matched) = cap.get(1) {
+            return matched.as_str().parse::<f32>().unwrap_or(0.0);
+        }
+    }
+    0.0
+}
+
+fn find_matching_div(html: &str, start_pos: usize) -> Option<(usize, usize)> {
+    let mut open_divs = 0;
+    let mut i = start_pos;
+    let len = html.len();
+
+    while i < len {
+        if html[i..].starts_with("<div") {
+            open_divs += 1;
+            i += 4;
+        } else if html[i..].starts_with("</div>") {
+            open_divs -= 1;
+            i += 6;
+            if open_divs == 0 {
+                return Some((start_pos, i));
+            }
+        } else {
+            i += 1;
         }
     }
     None
 }
 
-fn find_next_number(rest: &str) -> f32 {
+fn extract_items_from_section(page: &str, section_title: &str) -> Vec<String> {
-    let re = Regex::new(r"([0-9]+\.?[0-9]+)");
+    if let Some(h3_pos) = page.find(section_title) {
-    if let Ok(re) = re {
+        // Find the start of the div containing the h3
-        if let Some(cap) = re.captures(rest) {
+        let div_start = page[..h3_pos].rfind("<div").unwrap_or(0);
-            if let Some(matched) = cap.get(1) {
+        if let Some((div_start, div_end)) = find_matching_div(page, div_start) {
-                return matched.as_str().parse::<f32>().unwrap_or(0.0);
+            let div_html = &page[div_start..div_end];
-            }
+            return ITEM_REGEX
+                .captures_iter(div_html)
+                .map(|cap| cap[1].to_owned())
+                .collect();
+        } else {
+            warn!(
+                "Failed to find matching </div> for section '{}'",
+                section_title
+            );
         }
     }
-    0.0
+    Vec::new()
+}
+
+fn extract_skill_order_from_table(page: &str) -> String {
+    // Find the table containing the skill order (look for Q.png as anchor)
+    let table_start = page
+        .find("Q.png")
+        .and_then(|pos| page[..pos].rfind("<table"))
+        .unwrap_or(0);
+    let table_end = page[table_start..]
+        .find("</table>")
+        .map(|e| table_start + e + 8)
+        .unwrap_or(page.len());
+    let table_html = &page[table_start..table_end];
+
+    // Extract rows
+    let rows: Vec<&str> = table_html
+        .match_indices("<tr")
+        .map(|(i, _)| {
+            let end = table_html[i..]
+                .find("</tr>")
+                .map(|e| i + e + 5)
+                .unwrap_or(table_html.len());
+            &table_html[i..end]
+        })
+        .collect();
+
+    // Only process Q/W/E/R rows (skip header)
+    let skills = ["Q", "W", "E", "R"];
+    let mut order = [""; 18];
+    for (i, row) in rows.iter().skip(1).take(4).enumerate() {
+        let mut col = 0;
+        let mut pos = 0;
+        while let Some(td_start) = row[pos..].find("<td") {
+            let td_start = pos + td_start;
+            let td_end = row[td_start..]
+                .find("</td>")
+                .map(|e| td_start + e + 5)
+                .unwrap_or(row.len());
+            let td_html = &row[td_start..td_end];
+            if td_html.contains(&format!(">{}<", skills[i])) && col < 18 {
+                order[col] = skills[i];
+            }
+            col += 1;
+            pos = td_end;
+        }
+    }
+    order.join("")
 }
 
 impl MSDataSource {
@@ -67,11 +159,14 @@ impl DataSource for MSDataSource {
             .and_then(|mut resp| resp.body_mut().read_json())
         {
             Ok(champs) => champs,
-            Err(_) => return champs,
+            Err(e) => {
+                error!("Failed to fetch champions from MetaSRC: {}", e);
+                return champs;
+            }
         };
 
         for champ in champions {
-            if let Some(id) = get_champ_from_name(champion, champ.name.to_owned()) {
+            if let Some(id) = get_champ_from_name(champion, &champ.name) {
                 let allowed_roles = ["TOP", "ADC", "SUPPORT", "JUNGLE", "MID"];
                 let roles = champ
                     .search_terms
@@ -79,9 +174,11 @@ impl DataSource for MSDataSource {
                     .map(|s| s.to_uppercase())
                     .filter(|role| allowed_roles.contains(&role.as_str()))
                     .collect::<Vec<String>>();
-                champs.insert(id, roles);
+                if let Some(first_role) = roles.first() {
+                    champs.insert(id, vec![first_role.clone()]);
+                }
             } else {
-                error!("Could not find champ {} in champion data", champ.name);
+                warn!("Could not find champ '{}' in champion data", champ.name);
             }
         }
 
@@ -109,57 +206,48 @@ impl DataSource for MSDataSource {
         if let Ok(mut p) = rep {
             let page = match p.body_mut().read_to_string() {
                 Ok(s) => s,
-                Err(_) => return builds,
+                Err(e) => {
-            };
+                    warn!("Failed to read page for champ {}: {}", champ.id, e);
-
+                    return builds;
-            let mut pos = page.find("Patch ");
-            let patch = if let Some(p) = pos {
-                find_next_number(&page[p..]).to_string()
-            } else {
-                String::new()
-            };
-
-            pos = page.find("Win");
-            let win_rate: f32 = if let Some(p) = pos {
-                find_next_number(&page[p..])
-            } else {
-                0.0
-            };
-
-            pos = page.find("KDA:");
-            let kda: f32 = if let Some(p) = pos {
-                find_next_number(&page[p..])
-            } else {
-                0.0
-            };
-
-            pos = page.find("Games:");
-            let games: u32 = if let Some(p) = pos {
-                find_next_number(&page[p..]) as u32
-            } else {
-                0
-            };
-
-            let mut items = vec![];
-            let mut pos: Option<usize> = page.find("/item/");
-            while let Some(mut p) = pos {
-                p += 6;
-                if let Some(dot_pos) = page[p..].find('.') {
-                    let i = &page[p..p + dot_pos];
-                    items.push(i);
                 }
+            };
 
-                let next = page[p..].find("/item/");
+            // Extract patch, win rate, kda, games
-                if let Some(n) = next {
+            let patch = page
-                    pos = Some(p + n);
+                .find("Patch ")
-                } else {
+                .map(|p| find_next_number(&page[p..]).to_string())
-                    pos = None;
+                .unwrap_or_default();
-                }
+
-            }
+            let win_rate = page
+                .find("Win")
+                .map(|p| find_next_number(&page[p..]))
+                .unwrap_or(0.0);
+
+            let kda = page
+                .find("KDA:")
+                .map(|p| find_next_number(&page[p..]))
+                .unwrap_or(0.0);
+
+            let games = page
+                .find("Games:")
+                .map(|p| find_next_number(&page[p..]) as u32)
+                .unwrap_or(0);
+
+            let items = extract_items_from_section(&page, "Item Purchase Order");
+            let starting_items = extract_items_from_section(&page, "Starting Items");
 
             builds.push((
                 positions[0].to_owned(),
-                vec![self.make_item_set(items, "Set".to_owned())],
+                vec![
+                    self.make_item_set(
+                        starting_items,
+                        format!(
+                            "Starting Items | skillOrder: {}",
+                            extract_skill_order_from_table(&page)
+                        ),
+                    ),
+                    self.make_item_set(items, "Item Purchase Order".to_owned()),
+                ],
                 Stat {
                     win_rate,
                     games,
@@ -167,6 +255,11 @@ impl DataSource for MSDataSource {
                     patch,
                 },
             ));
+        } else {
+            warn!(
+                "Failed to fetch build page for champ {} at position {}",
+                champ.id, positions[0]
+            );
         }
 
         builds