How can I loop through a C++ map of maps?
Old question but the remaining answers are outdated as of C++11 - you can use a ranged based for loop and simply do:
std::map<std::string, std::map<std::string, std::string>> mymap;
for(auto const &ent1 : mymap) {
// ent1.first is the first key
for(auto const &ent2 : ent1.second) {
// ent2.first is the second key
// ent2.second is the data
}
}
this should be much cleaner than the earlier versions, and avoids unnecessary copies.
Some favour replacing the comments with explicit definitions of reference variables (which get optimised away if unused):
for(auto const &ent1 : mymap) {
auto const &outer_key = ent1.first;
auto const &inner_map = ent1.second;
for(auto const &ent2 : inner_map) {
auto const &inner_key = ent2.first;
auto const &inner_value = ent2.second;
}
}
You can use an iterator.
typedef std::map<std::string, std::map<std::string, std::string>>::iterator it_type;
for(it_type iterator = m.begin(); iterator != m.end(); iterator++) {
// iterator->first = key
// iterator->second = value
// Repeat if you also want to iterate through the second map.
}
for(std::map<std::string, std::map<std::string, std::string> >::iterator outer_iter=map.begin(); outer_iter!=map.end(); ++outer_iter) {
for(std::map<std::string, std::string>::iterator inner_iter=outer_iter->second.begin(); inner_iter!=outer_iter->second.end(); ++inner_iter) {
std::cout << inner_iter->second << std::endl;
}
}
or nicer in C++0x:
for(auto outer_iter=map.begin(); outer_iter!=map.end(); ++outer_iter) {
for(auto inner_iter=outer_iter->second.begin(); inner_iter!=outer_iter->second.end(); ++inner_iter) {
std::cout << inner_iter->second << std::endl;
}
}
With C++17 (or later), you can use the "structured bindings" feature, which lets you define multiple variables, with different names, using a single tuple/pair. Example:
for (const auto& [name, description] : planet_descriptions) {
std::cout << "Planet " << name << ":\n" << description << "\n\n";
}
The original proposal (by luminaries Bjarne Stroustrup, Herb Sutter and Gabriel Dos Reis) is fun to read (and the suggested syntax is more intuitive IMHO); there's also the proposed wording for the standard which is boring to read but is closer to what will actually go in.