coach_salaries.R

begin <- Sys.time()
library(jsonlite)
library(dplyr)
options(stringsAsFactors = FALSE)

coach <- data.frame()
for(a in 1:2989){
tmp <- jsonlite::fromJSON(paste0("https://sports.usatoday.com/ajaxservice/ncaa/salaries__coach__",a))

if(is.null(nrow(tmp$rows))){
salary <- data.frame()
coach <- rbind(coach,salary)
} else {
salary <- flatten(cbind(a,tmp$profile$coach_name,ifelse(!is.null(substr(tmp$profile$head_shot,22,26)),substr(tmp$profile$head_shot,22,26),NA),tmp$rows))
rownames(salary) <- 1:nrow(salary)+nrow(coach)
    
if(ncol(salary)!=16){
print(a)
print(salary$`tmp$profile$coach_name`[1])
salary <- as.data.frame(as.matrix(salary))
coach <- bind_rows(coach,salary)
    
} else {
print(a)
print(salary$`tmp$profile$coach_name`[1])
coach <- rbind(coach,salary)
	}
    }
}

strip_num <- function(x) {
x <- as.numeric(gsub("[^[:alnum:][:blank:]+?&/\\-]","",x))
return(x)
}

coach <- coach %>%
	select(1:4,6:7,9,11,13,15,17) %>%
	rename(coach_id=a,
    coach_name=`tmp$profile$coach_name`,
    sport=`ifelse(!is.null(substr(tmp$profile$head_shot, 22, 26)), substr(tmp$profile$head_shot, `) %>%
    mutate(total_pay.value=strip_num(total_pay.value),
	school_pay.value=strip_num(school_pay.value),
	other_pay.value=strip_num(other_pay.value),
    max_bonus.value=strip_num(max_bonus.value),
    last_year_bonus.value=strip_num(last_year_bonus.value),
    year.value=strip_num(year.value),
	coach_id=strip_num(coach_id),
   sport=ifelse(position.value %in% c('director','strength'),'athletics',sport))


write.csv(coach,"C:/Users/Benjamin/Desktop/Projects/coaching_salaries.csv")

	
end <- Sys.time()
print(end-begin)