Parsing doubles and words in a string - c++

Question

I was working on the following exercise from The C++ Programming Language:
Read a sequence of possibly whitespace-separated (name,value) pairs,
where the name is a single whitespace-separated word and the value is
an integer or a floating-point value. Compute and print the sum and
mean for each name and the sum and mean for all names.
For example, given:
hello world5.678popcorn 8.123 rock 123 hello world 8.761 popcorn 98 rock 1.9rock2.3
The output of my implementation is:
rock: Sum (127.2), Mean (42.4)
hello world: Sum (14.439), Mean (7.2195)
popcorn: Sum (106.123), Mean (53.0615)
My implementation:
#include <iostream>
#include <string>
#include <unordered_map>
std::unordered_map<std::string, double> pairs;
std::unordered_map<std::string, int> occurences;
void save(const std::string& name, const std::string& value);
void trim(std::string& s, const std::string& chars = " ");
int main() {
std::string line;
getline(std::cin, line);
std::string name;
std::string value;
bool name_saved;
for(char c : line) {
if(!name_saved && isdigit(c)) { // reached end of name
name_saved = true;
trim(name);
value += c;
} else if(!name_saved) { // add char to name
name += c;
} else if(name_saved) {
if(isdigit(c) || (c == '.' && (value.find_first_of(".") == std::string::npos))) { // add char to value
value += c;
} else { // reached end of value
trim(value);
save(name, value);
name = "";
value = "";
name_saved = false;
if(isalpha(c)) {
name += c;
}
}
}
}
if(value != "") {
save(name, value);
}
std::cout << "\n";
for(auto pair : pairs) {
std::cout << pair.first << ": " << "Sum (" << pair.second << "), Mean (" << (pair.second / occurences[pair.first]) << ")\n";
}
return 0;
}
void save(const std::string& name, const std::string& value) {
pairs[name] += std::stod(value);
occurences[name]++;
}
void trim(std::string& s, const std::string& chars) {
s.erase(0, s.find_first_not_of(chars));
s.erase(s.find_last_not_of(chars) + 1);
}
I was wondering what would a more efficient approach to this exercise be? I feel that my code is quite messy and I would like to get some input on what I could use to clean it up and make it more compact.

Answer 1

There are many many different soultions. It depends a little bit on your personal programming style and what you have learned already or not.
The above example cries for "regular expressions". You can read here in the Cpp- Reference about them. And especially the function std::regex_search will be your friend.
First to the regurlar expression. You are looking for a "text" with embedded spaces ("single whitespace-separated word "), followed by a int or float. This can be expressed easily with a "regex":
([a-zA-Z]+ ?[a-zA-Z]+) ?(\d+\.?\d*)
So, first we have 1 or more alpha-characters, then an optional space and then again 1 or more alpha-characters. This makes up the name.
For the value, we have 1 or more digits, maybe followed by a "." and maybe more digits.
You get a better understanding, if you paste the regex and the test string in some online regex-tester like this.
It will give you some more detailed explanation. Espcially for the brackets "()", which form groups.
An the groups can be extracted after a std::regex_search has found a match. Meaning: If std::regex_search will find a match for the given regex, it will return true and the resulting groups can be found in std::smatch, see here for a description.
And with all this, we can define a simple for loop to get all names and values from the test string.
for (std::string s{ test }; std::regex_search(s, sm, re); s = sm.suffix())
First we will initialize the "loop-run" variable, in this case a std::string and initialize it with the given test-string. Then we will search for a match in the test-string. If there was a match, then we will find the result in sm[1] and sm[2]. After we did all operations inside the loop body, we set the "loop-run" variable to the not yet matched rest of the test string. The suffix.
To goup, calculate and aggregate the values, we use a std::map. The key is the "name" and the value is a std::pair consisting of the count of names and the sum of the associated values. So, in the loop:
for (std::string s{ test }; std::regex_search(s, sm, re); s = sm.suffix()) {
// Count the occurences of a text
aggregator[sm[1]].first++;
// Sum up the values for a text
aggregator[sm[1]].second += std::stod(sm[2]);
}
we use aggregator[sm[1]] to create the name in or retrieve the name from map. In any case, we have then a reference to the current name entry, and we can increment the count and build the running sum.
This is a very simple 3 line approach, which does already nearly all the expected work.
The rest is simple calculation of expected sum and mean values and showing everything on the screen.
Please see the full code below:
#include <iostream>
#include <string>
#include <regex>
#include <vector>
#include <iterator>
#include <map>
// The regex for words with embedded space and floats/ints
const std::regex re{ R"(([a-zA-Z]+ ?[a-zA-Z]+) ?(\d+\.?\d*))" };
int main() {
// Definition Section --------------------------------------------------------------------------------
// The input test string
std::string test{"hello world5.678popcorn 8.123 rock 123 hello world 8.761 popcorn 98 rock 1.9rock2.3"};
// Here we will store the result. The text and the associated "count" and "sum"
std::map<std::string, std::pair<unsigned int, double>> aggregator{};
std::smatch sm;
// Find, store and calculate data -------------------------------------------------------------------
// Iterate though the string and get the text and the float value
for (std::string s{ test }; std::regex_search(s, sm, re); s = sm.suffix()) {
// Count the occurences of a text
aggregator[sm[1]].first++;
// Sum up the values for a text
aggregator[sm[1]].second += std::stod(sm[2]);
}
// Output data ----------------------------------------------------------------------------------------
// Since the task is to calculate also the overall results, we will do
unsigned int countOverall{};
double sumOverall{};
// Iterate over the "text" data and output sum and mean value per text and aggregate the overall values
for (const auto& [text, agg] : aggregator) {
// Output sum and mean per text
std::cout << "\n" << text << ": Sum (" << agg.second << "), Mean (" << agg.second / agg.first << ")";
// Aggregate overall values
countOverall += agg.first;
sumOverall += agg.second;
}
// Show overall result to the user.
std::cout << "\n\nSum overall: (" << sumOverall << "), Mean overall: (" << sumOverall / countOverall << ")\n\n";
return 0;
}
If this solution is "better" or not? Please decide yourself . . .