Добавил:
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
Professional C++ [eng].pdf
Скачиваний:
284
Добавлен:
16.08.2013
Размер:
11.09 Mб
Скачать

Chapter 22

Algorithms and Function Objects Example:

Auditing Voter Registrations

Voter fraud can be a problem in the United States. People sometimes attempt to register and vote in two or more different counties. Additionally, convicted felons, who are ineligible to vote in some states, occasionally attempt to register and vote anyway. Using your newfound algorithm and function object skills, you could write a simple voter registration auditing function that checks the voter rolls for certain anomalies.

The Voter Registration Audit Problem Statement

The voter registration audit function should audit the information for a single state. Assume that voter registrations are stored by county in a map that maps county names to a list of voters. Your audit function should take this map and a list of convicted felons as parameters, and should remove all convicted felons from the lists of voters. Additionally, the function should find all voters who are registered in more than one county and should remove those names from all counties. For simplicity, assume that the list of voters is simply a list of string names. A real application would obviously require more data, such as address and party affiliation.

The auditVoterRolls() Function

This example takes a top-down approach, starting from the highest-level function and making calls to functions and functors that are not yet written. As the example progresses, the missing implementations will be filled in.

The top-level function, auditVoterRolls(), works in three steps:

1.Find all the duplicate names in all the registration lists by making a call to getDuplicates().

2.Combine the list of duplicates and the list of convicted felons, and remove duplicates in the combined list.

3.Remove from every voter list all the names found in the combined list of duplicates and convicted felons. The approach taken here is to use for_each() to process each list in the map, applying a user-defined functor RemoveNames to remove the offending names from each list.

Here’s the implementation of auditVoterRolls():

//

//auditVoterRolls

//Expects a map of string/list<string> pairs keyed on county names

//and containing lists of all the registered voters in those counties

//Removes from each list any name on the convictedFelons list and

//any name that is found on any other list

//

void auditVoterRolls(map<string, list<string> >& votersByCounty,

648

Mastering STL Algorithms and Function Objects

const list<string>& convictedFelons)

{

// Get all the duplicate names.

list<string> duplicates = getDuplicates(votersByCounty);

//Combine the duplicates and convicted felons--we want

//to remove names on both lists from all voter rolls. duplicates.insert(duplicates.end(), convictedFelons.begin(),

convictedFelons.end());

//If there were any duplicates, remove them.

//Use the list versions of sort and unique instead of the generic

//algorithms, because the list versions are more efficient. duplicates.sort();

duplicates.unique();

//Now remove all the names we need to remove.

for_each(votersByCounty.begin(), votersByCounty.end(), RemoveNames(duplicates));

}

The getDuplicates() Function

The getDuplicates() function must find any name that is on more than one voter registration list. There are several different approaches one could use to solve this problem. This implementation simply combines the lists from each county into one big list and sorts it. At that point, any duplicate names between the different lists will be next to each other in the big list. Now getDuplicates() can use the adjacent_find() algorithm on the big, sorted, list to find all consecutive duplicates. Here is the implementation:

//

//getDuplicates()

//Returns a list of all names that appear in more than one list in

//the map

//

//The implementation generates one large list of all the names from

//all the lists in the map, sorts it, then finds all duplicates

//in the sorted list with adjacent_find().

//

list<string> getDuplicates(const map<string, list<string> >& voters)

{

list<string> allNames, duplicates;

//Collect all the names from all the lists into one big list. map<string, list<string> >::const_iterator it;

for(it = voters.begin(); it != voters.end(); ++it) { allNames.insert(allNames.end(), it->second.begin(), it->second.end());

}

//Sort the list--use the list version, not the general algorithm,

//because the list version is faster.

allNames.sort();

649

Chapter 22

//

//Now that it’s sorted, all duplicate names will be next to each other.

//Use adjacent_find() to find instances of two or more identical names

//next to each other.

//

// Loop until adjacent_find returns the end iterator.

//

list<string>::iterator lit;

for (lit = allNames.begin(); lit != allNames.end(); ++lit) { lit = adjacent_find(lit, allNames.end());

if (lit == allNames.end()) { break;

}

duplicates.push_back(*lit);

}

//

//If someone was on more than two voter lists, he or she will

//show up more than once in the duplicates list. Sort the list

//and remove duplicates with unique.

//Use the list versions because they are faster than the generic versions.

duplicates.sort();

duplicates.unique();

return (duplicates);

}

The RemoveNames Functor

The auditVoterRolls() function uses the following line to remove all the offending (duplicate and felon) names from each list in the voter registration map:

for_each(votersByCounty.begin(), votersByCounty.end(), RemoveNames(duplicates));

The for_each() algorithm calls the RemoveNames functor on each string/list<string> pair in the map. The definition of the RemoveNames functor class looks like this:

//

//RemoveNames

//Functor class that takes a string/list<string> pair and removes

//any strings from the list that are found in a list of names

//(supplied in the constructor)

//

class RemoveNames : public unary_function<pair<const string, list<string> >, void>

{

public:

RemoveNames(const list<string>& names) : mNames(names) {} void operator() (pair<const string, list<string> >& val);

650

Mastering STL Algorithms and Function Objects

protected:

const list<string>& mNames;

};

Note that RemoveNames subclasses unary_function, a technique described earlier in this chapter. The constructor takes a reference to a list of names, which it stores for use in its function-call operator. Recall that the parameter to the functor callback is an element of the map, which is a string/list<string> pair. The function call operator’s job is to remove any names from the string list that are found in the mNames list. The implementation uses the remove_if() algorithm with a special predicate

NameInList.

//

//Function-call operator for RemoveNames functor.

//Uses remove_if() followed by erase to actually delete the names

//from the list

//Names are removed if they are in our list of mNames. Use the NameInList

//functor to check if the name is in the list.

//

void RemoveNames::operator() (pair<const string, list<string> >& val)

{

list<string>::iterator it = remove_if(val.second.begin(), val.second.end(), NameInList(mNames));

val.second.erase(it, val.second.end());

}

Remember that the remove() family of algorithms don’t really remove elements; they only move them to the end of the range. You must call erase() on the container to actually remove the elements.

The NameInList Functor

The RemoveNames functor calls remove_if() with a predicate functor of the NameInList class. The NameInList functor returns true if the string given to it as an argument is in the list mNames. The class definition looks like this:

//

//NameInList

//Functor to check if a string is in a list of strings (supplied

//at construction time).

//

class NameInList : public unary_function<string, bool>

{

public:

NameInList(const list<string>& names) : mNames(names) {} bool operator() (const string& val);

protected:

const list<string>& mNames;

};

651