CS 698 – Lab 5

Read the data

count <- function(cause=NULL){
  homicides <- readLines("homicides.txt")
r <- regexpr("<dd>[Cc]ause(.*?)</dd>", homicides)
m <- regmatches(homicides, r)
all_cause<- gsub("<dd>[Cc]ause: |</dd>", "", m)
if(cause %in% NULL) stop("The cause should be non-NULL") 
if(cause %in% unique(all_cause)){
return(length(grep(cause,all_cause,ignore.case = T)))
}
else  stop("The cause does not exist") 
}

count("shooting")

## [1] 1003

agecount <- function(age=NULL){
  homicides <- readLines("homicides.txt")
r <- regexpr("<dd>(.*?)years old</dd>", homicides)
m <- regmatches(homicides, r)
all_age <- as.numeric(regmatches(m,regexpr("[0-9][0-9]",m)))
if(age %in% NULL) stop("The age should be non-NULL") 

return(sum(all_age==age))
}

agecount(21)

## [1] 60

# all_age<- sub(": | years old</dd>", "", m)