chess.R

library(tidyverse)
library(lubridate)

chess <- read_lines(
	"~/Downloads/all_with_filtered_anotations/all_with_filtered_anotations_since1998.txt",
	skip = 5
) %>%
	str_split(" ", 18, TRUE) %>%
	as_tibble %>%
	rename(
		id      = V1,
		d       = V2,
		result  = V3,
		welo    = V4,
		belo    = V5,
		len     = V6,
		date_c  = V7,
		resu_c  = V8,
		welo_c  = V9,
		belo_c  = V10,
		edate_c = V11,
		setup   = V12,
		fen     = V13,
		resu2_c = V14,
		oyrange = V15,
		bad_len = V16,
		hashes  = V17,
		game    = V18
	) %>%
	select(-hashes)

chess_clean <- chess %>%
	mutate(
		id      = id   %>% as.numeric,
		date_c  = case_when(date_c  == "date_false"    ~ FALSE, date_c  == "date_true"    ~ TRUE, TRUE ~ NA),
		resu_c  = case_when(resu_c  == "result_false"  ~ FALSE, resu_c  == "result_true"  ~ TRUE, TRUE ~ NA),
		welo_c  = case_when(welo_c  == "welo_false"    ~ FALSE, welo_c  == "welo_true"    ~ TRUE, TRUE ~ NA),
		belo_c  = case_when(belo_c  == "belo_false"    ~ FALSE, belo_c  == "belo_true"    ~ TRUE, TRUE ~ NA),
		edate_c = case_when(edate_c == "edate_false"   ~ FALSE, edate_c == "edate_true"   ~ TRUE, TRUE ~ NA),
		setup   = case_when(setup   == "setup_false"   ~ FALSE, setup   == "setup_true"   ~ TRUE, TRUE ~ NA),
		fen     = case_when(fen     == "fen_false"     ~ FALSE, fen     == "fen_true"     ~ TRUE, TRUE ~ NA),
		resu2_c = case_when(resu2_c == "result2_false" ~ FALSE, resu2_c == "result2_true" ~ TRUE, TRUE ~ NA),
		oyrange = case_when(oyrange == "oyrange_false" ~ FALSE, oyrange == "oyrange_true" ~ TRUE, TRUE ~ NA),
		bad_len = case_when(bad_len == "blen_false"    ~ FALSE, bad_len == "blen_true"    ~ TRUE, TRUE ~ NA)
	) %>%
	mutate(
		d      = if_else(date_c, as.character(NA), d),
		result = if_else(resu_c, as.character(NA), result),
		welo   = if_else(welo_c, as.character(NA) , welo),
		belo   = if_else(belo_c, as.character(NA), belo)
	) %>%
	select(-date_c, -resu_c, -welo_c, -belo_c) %>%
	mutate(
		d       = d    %>% ymd,
		welo    = welo %>% na_if("None") %>% as.numeric,
		belo    = belo %>% na_if("None") %>% as.numeric,
	)

chess_clean %>%
	filter(str_detect(game, "^W1\\."))