stat.inkからSplatoon 3の戦歴データを取得する

getspl3log.Rmd

---
title: "stat.inkからSplatoon 3の戦歴データを取得する"
author: '@nozma'
date: "2022-11-29"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
library(rvest)
library(dplyr)
library(lubridate)
library(tidyr)
library(ggplot2)
library(jsonlite)
library(tidyjson)
library(stringr)
theme_set(theme_bw(base_family = "IPAexGothic"))
```

# 戦歴取得


```{r}
#| message=FALSE

# param ----
user_name <- "@nozma"
base_url <- glue::glue("https://stat.ink/{user_name}/spl3?page=")
max_page <- 7 # TODO:合計バトル数から自動計算

# init df ----
result_raw <- tibble()

# get data ----
for(i in seq_len(max_page)){
  read_html(paste0(base_url, i)) %>% 
    html_nodes("table") %>% 
    html_table() %>% 
    .[[1]] %>% 
    as_tibble(.name_repair = "unique") %>%
    # 軽い前処理
    filter(`...1` == "Detail") %>% # 日付行の除外
    select(!matches("^\\.")) %>%   # 空白カラムの除外
    # 切断ありフラグを追加(trのclassから取得)
    bind_cols( 
      read_html(paste0(base_url, i)) %>% 
        html_nodes("table") %>% 
        html_nodes("tr") %>% 
        html_attr("class") %>% 
        tibble(class = .) %>% 
        filter(grepl("battle-row", class)) %>% 
        transmute(disconnected = grepl("disconnected", class))
    ) %>% 
    bind_rows(result_raw) -> result_raw
  Sys.sleep(1) # 落ち着く
}
```

# API情報取得

ref. [Spl3 API: Battle － Post · fetus-hina/stat.ink Wiki](https://github.com/fetus-hina/stat.ink/wiki/Spl3-API:-Battle-%EF%BC%8D-Post)

```{r}
# ロビー
read_html("https://stat.ink/api/v3/lobby") %>% 
  html_element("p") %>% 
  html_text() %>% 
  fromJSON() %>% 
  unnest(name) -> lobby_info
# ルール
read_html("https://stat.ink/api/v3/rule") %>% 
  html_element("p") %>% 
  html_text() %>% 
  fromJSON() %>% 
  unnest(c(aliases, name, short_name), names_sep = "_") -> rule_info
# ステージ
read_html("https://stat.ink/api/v3/stage") %>% 
  html_element("p") %>% 
  html_text() %>% 
  fromJSON() %>% 
  unnest(c(aliases, name, short_name, release_at), names_sep = "_") -> stage_info
# ブキ
read_html("https://stat.ink/api/v3/weapon") %>% 
  html_element("p") %>% 
  html_text() %>% 
  gather_array() %>% 
  spread_all() %>% as_tibble() -> weapon_info
```

# 前処理

```{r}
result_raw %>% 
  rename_all(.funs = ~tolower(gsub(" ", "_", .))) %>%
  transmute(
    lobby, mode, stage,
    weapon, 
    sub_weapon, special, rank, judge, result,
    #`k/d`, ratio, rate, 
    #`kill_or_assist`, 
    specials = as.integer(specials), 
    inked = as.numeric(sub(",", "", inked)), 
    rank_in_team = as.integer(rank_in_team),
    elapsed = as.numeric(elapsed...22), # カラムの追加・削除があると動かなくなる可能性あり
    date_time = mdy_hm(date_time),
    disconnected,
    # 変数追加
    simple_result = if_else(grepl("Victory", result), "Victory", "Defeat"), # 勝ち負け
    knockout = judge == "KNOCKOUT",
    judge_friend = case_when( # 味方カウント
      knockout ~ NA_character_,
      simple_result == "Victory" ~ str_extract(judge, "^\\d{2}(\\.\\d)?"),
      simple_result == "Defeat" ~ case_when(
        nchar(judge) >= 4 ~ str_extract(judge, "^\\d{2}(\\.\\d)?"),
        nchar(judge) == 3 ~ str_extract(judge, "^\\d{1}"),
        nchar(judge) <= 2 ~ "0"
      ),
      TRUE ~ NA_character_
    ) %>% as.numeric(),
    judge_opponent = case_when( # 敵カウント
      knockout ~ NA_character_,
      simple_result == "Defeat" ~ str_extract(judge, "(\\d{2}(\\.\\d)?)%?$") %>% gsub("%", "", .),
      simple_result == "Victory" ~ case_when(
        nchar(judge) >= 4 ~ str_extract(judge, "(\\d{2}(\\.\\d)?)%?$") %>% gsub("%", "", .),
        nchar(judge) == 3 ~ str_extract(judge, "\\d{1}$"),
        nchar(judge) <= 2 ~ judge
      ),
      TRUE ~ NA_character_
    ) %>% as.numeric(),
    exempted = grepl("Exempted", result), # 免除フラグ
    kill = as.integer(stringr::str_extract(`k/d`, "^\\d{1,2}")), # キル
    assist = as.integer(kill_or_assist) - kill, # アシスト
    kill_assist = kill + assist, # キル+アシスト
    death = as.integer(stringr::str_extract(`k/d`, "\\s(\\d{1,2})\\s")) # デス
  ) %>% 
  # 回線落ち除外
  filter(judge != "") %>% 
  #日本語化 ----
  # ロビー
  left_join(
    lobby_info %>% select(en_US, ja_JP) %>% distinct(),
    by = c("lobby" = "en_US")
  ) %>% 
  select(!lobby) %>% rename(lobby = ja_JP) %>% 
  # ルール
  left_join(
    rule_info %>% select(name_en_US, short_name_ja_JP) %>% distinct(),
    by = c("mode" = "name_en_US")
  ) %>% 
  select(!mode) %>% rename(mode = short_name_ja_JP) %>% 
  # ステージ
  left_join(
    stage_info %>% select(name_en_US, short_name_ja_JP) %>% distinct(),
    by = c("stage" = "name_en_US")
  ) %>% 
  select(!stage) %>% rename(stage = short_name_ja_JP) %>% 
  # ブキ
  left_join(
    weapon_info %>% select(name.en_US, name.ja_JP) %>% distinct(),
    by = c("weapon" = "name.en_US")
  ) %>% 
  select(!weapon) %>% rename(weapon = name.ja_JP) %>% 
  # サブウェポン
  left_join(
    weapon_info %>% select(sub.name.en_US, sub.name.ja_JP) %>% distinct(),
    by = c("sub_weapon" = "sub.name.en_US")
  ) %>% 
  select(!sub_weapon) %>% rename(sub_weapon = sub.name.ja_JP) %>% 
  # スペシャル
  left_join(
    weapon_info %>% select(special.name.en_US, special.name.ja_JP) %>% distinct(),
    by = c("special" = "special.name.en_US")
  ) %>% 
  select(!special) %>% rename(special = special.name.ja_JP) %>% 
  arrange(desc(date_time)) %>% 
  # 不要な列の削除
  select(!c(judge, result)) -> result
```

出力サンプル: https://gist.github.com/nozma/e7fab790288b067b1b867572fa6e703a