// Program to calculate the probability of a given first and last name.

firstName = input["Enter first name(s): "]
lastName = input["Enter last name: "]

pop = input["Enter population [260 million]: "]
if pop == ""
   pop = 260 million
else
   pop = eval[pop]

println["\nProbability of name \"" + firstName + " " + lastName + "\": "]

firstFreq = frequency[firstName]
println["Frequency of " + firstName + " (in either sex) is: " + (firstFreq -> "percent")]

lastFreq = frequency[lastName, false]
println["Frequency of " + lastName + " is: " + (lastFreq -> "percent")]

if lastFreq == 0
   println["Probability of last name is too low for prediction."]
else
{
   together = firstFreq * lastFreq

   //println[together -> "percent"]

   println["\nThis name is held by 1 in " + ceil[1/together] + " people."]

   println["This name will occur with probability " + (probTogether[firstFreq, lastFreq, pop] -> "percent") + " in " + pop + " people."]

   println["This name will tend to occur at least once in every " + predictTogether[firstFreq, lastFreq] + " people."]

   println["There are probably " + 10 * round[together * pop, .1] + " people with this name in " + pop + " people."]

   // Functions used by above program

   // Probability of a single name with frequency freq being in a movie with
   // chars characters...
   probAlone[freq, chars] := 1 - (1-freq)^chars

   // Probability those two names appear together in n people
   probTogether[probA, probB, pop] := 1 - (1 - probA * probB)^pop

   // Predict the number of movies to get a 50% probability
   predictTogether[probA, probB] := ceil[- ln[2] / ln[1 - probA * probB]]
}

// Contact www.placesnamed.com and fetch probabilities.
frequency[names, first=true] :=
{
   nameList = split[%r/[\s,\/]+/, names]
   freq = 0
   for [name] nameList
   {
      if (name == "")
         next          // Ignore leading/trailing space
      
      [let1, let2] = name =~ %r/(.)(.)/  // Get first 2 letters
      url = "http://www.placesnamed.com/"+ let1 + "/" + let2 + "/" + name + ".asp"

      if (first)
      {
         pattern = regex[name + ".* popular ((?:fe)?male).*frequency is (.*?)%","i"]
         for [line] lines[url]
            if [sex, p] = line =~ pattern
            {
               println[name + " (" + sex + "): " + p + "%"]
               freq = freq + eval[p] percent
            }
      } else   // Last name
      {
         pattern = regex[name + ".* popular .*last.*frequency is (.*?)%","i"]

         for [line] lines[url]
            if [p] = line =~ pattern
               freq = freq + eval[p] percent
      }
   }

   if (first)
      return freq/2
   else
      return freq
}