Hello,
Trying to challenge myself with C++ STL.
I want to read in data from file and do some calculation, adapted from an exercise of a book (C++ for engineers and scientist, 3rd Ed, Gary Bronson).
infile is like:
ID Name Course Credit Grade
2333021 Bokow,R. NS201 3 A
2333021 Bokow,R. MG342 3 A
2333021 Bokow,R. FA302 1 A
2574063 Failin,D. MK106 3 C
2574063 Failin,D. MA208 3 B
2574063 Failin,D. CM201 3 C
2574063 Failin,D. CP101 2 B
2663628 Kingsley,M. QA140 3 A
2663628 Kingsley,M. CM245 3 B
2663628 Kingsley,M. EQ521 3 A
2663628 Kingsley,M. MK341 3 A
2663628 Kingsley,M. CP101 3 B
My idea is to read in record as map for which ID and Name is the key, and courses as value. As the number of courses are different for each student, create two structures, one for student ID and name; the other for each course; push all the courses of the same student into a vector of structure. So the final format is like
map < struct IDnName; vector < struct Course> record> Record
My coding flow is:
1) getline of each row; parse it into tokens ;
2) assign tokens into struct IDnName and struct Course ;
3) push the two structs into map , struct IDnName as key and struct Course as value ;
4) push struct Course into vector if IDnName already exists;
Challenges to me:
1) Is this the right algorithm to handle the situation? If not, which is the right way?
2) For learning purpose, with my coding flow to print the record of each student, I break down the job into two functions, a) key_print() b) val_print() ; but I am not sure the parameters should be passed by value or reference, which I always have problem with.
I've attached my code, for which there were too many errors to be listed here when I tried to compile. Hope someone could help me out. Thanks a lot!
Then comes to a new thing if two maps are used------they need share the key. While student_names only contains 3 unique entries(structure), student_courses contains 12 entries (structure) that are pushed into three vectors each for a student_names.
Do you mean to assign a new integer to both map?
The next step is to calculate the total credit of courses for each student; calculate the average grades for each student on his/her courses taken so far. So that I though use the string(or student ID) as key for retrieval and calculation.
The compiling error indicated the problem is the map iterator, which seems should be overloaded.
error: no match for 'operator=' in map_itr=mymap.std::map<_Key, _Tp, _Compare, _Alloc>::begin....
candidate is:
In file included from /usr/include/c++/4.7/map:60:0,
......
This challenge is too big for me at this moment. Thanks anyway!
If you wanted to bundle the data together and use it in a 'smart' way instead of updating individual variables, you could put it in a class with multiple members inside it. It still clearly belongs in multiple structures. How many times are you supposed to store someone's name and unique ID?
If you really wanted you could put the name and unique id in a pair<int, string>, which you should be able to plug into map < pair< int, string > , ... >
One of my original purposes of this practice is to experience more on vector and map, especially using a struct as key and vector for value in the map, so that I really did not mean to take the mximum resistance path . I like the slang "There are always more than one way to do it", and I tend to over-think many problems before I get to "take the minimum resistance path", but my always problem is how to find out the minimum resistance path.
Of course each unique student ID should be stored only once but not the student name, cases are students with same name but different ID number.
Thanks anyway and I will try using your method first then go back to the map<struct, vector<struct>> method, only for practice purpose to make a running code. Thanks a lot again!
To make your structure usable in a map, you should define an 'operator <' for it.
class asdf {
public:
bool operator <(const asdf &rhs) const {
// Return true if this class is less than RHS,
// false otherwise
}
};
How to do so can be fiendishly complex for complicated data, since you first need a coherent definition of what 'less' even means when comparing two student class records.
To make your structure usable in a map, you should define an 'operator <' for it. How to do so can be fiendishly complex for complicated data, since you first need a coherent definition of what 'less' even means when comparing two student class records.
Definitely I under-estimated the challenge.
Did not realize this only after another two hours of googling/reading. In my original post there were too many errors to be listed here when I tried to compile. which really discouraged me! (Better now.)
Thanks a lot!
---------- Post updated at 05:41 PM ---------- Previous update was at 02:26 PM ----------
Kind of got lost after discussions and searching.
Originally intended output is like:
---------------------
2333021 Bokow,R.
NS201 3 A
MG342 3 A
FA302 1 A
2574063 Failin,D.
MK106 3 C
MA208 3 B
CP101 2 B
2663628 Kingsley,M.
QA140 3 A
EQ521 3 A
MK341 3 A
CP101 3 B
So
1) I want student_ID be the key for the courses and names.
2) The records are read from a fstream which contains hundreds of thousands rows, imagine from a university record.
3) Also the student_ID may NOT be clustered together one row after another, so that read in record from file need to be checked for each entry before it is pushed into map.
Mainly technical problems:
One difference from your code is the the vector part which I thought is a structure containing course_abbreviation, credit and grade. Yours is a vector of string. Which is the part I did not catch with your code?
code:
#include <iostream>
#include <fstream>
#include <string>
#include <cstring>
#include <map>
#include <vector>
#include <cstdlib>
using namespace std;
struct grades {
string course_Abrrev;
int course_Credit;
char course_Grade;};
struct IDnName {
int student_ID;
string student_Name;};
int main(int argc, char *argv[])
{
vector <struct grades> Courses;
map < string, vector < grades > >marks;
map < string, string > names;
ifstream inFILE;
inFILE.open(argv[1]);
if (inFILE.fail()) {
cout << "ERROR: File failed to open!" << endl;
} else {
string line; //string for one line
struct IDnName oneName; // Allocate a place for one ID and name
struct grades oneCourse; // Allocate a place for a structure for a record of one course
while (inFILE.good()) {
char *sPtr; //string pointer for token
char *sArray = new char[line.length() + 1]; //Initialize pointer sArray for string--->char* conversion
getline(inFILE, line); //Read the whole line
strcpy(sArray, line.c_str());
//tokenize the line into array
sPtr = strtok(sArray, " ");
// oneName.student_ID = atoi(sPtr[0].c_str()); // did not work
oneName.student_ID = stoi(sPtr); //c++ method
while (sPtr != NULL) {
oneName.student_Name = sPtr;
names.insert(make_pair(oneName.student_ID, oneName.student_Name)); //LINE 54
sPtr = strtok(NULL, " ");
oneCourse.course_Abrrev = sPtr;
sPtr = strtok(NULL, " ");
// oneCourse.course_Credit = atoi(sPtr[0].c_str()); // did not work
oneCourse.course_Credit = stoi(sPtr);
sPtr = strtok(NULL, " ");
oneCourse.course_Grade = sPtr[0];
}
if (marks.find(oneName.student_ID) != marks.end()) { //LINE 65
marks.insert(make_pair(oneName.student_ID, vector<struct grades>())); //LINE 66
} else {
Courses.push_back(oneCourse);
}
}
}
// print part not ready yet
return 0;
}
Highlighted LINE 54, 65 & 66 are the parts throwing out lots of errors. Thanks!
That wasn't what I meant by using pairs, but I think you'd do map.insert( std::pair<type,type>(valuea, valueb) )
The reason your 'atoi' didn't work is because you're trying to call c_str on something that's already a c string! atoi(sptr); stoi is just an alias to atoi on systems that have it(unlike mine, which did not.) Use atoi.
That 'courses' structure doesn't look useful for anything, not sure why you have it, all the information is in 'marks'.
Finally -- I think you missed the point. Imagine how they make these map<>'s, it's going to be a list or array of some kind internally, right? So in memory somewhere there is a structure like this:
This isn't just wasted memory, it creates the opportunity for your program to store "impossible" states where element.key.studentid and element.value.studentid disagree. This seems even more likely to happen when keeping the same value around in different places as different types!
So:
Use studentid as an integer everywhere.
Never duplicate your key inside your data, only use it to fetch your data.
Thanks! Courses is a vector of struct, which seems needed as (map) value of marks. It is to hold the grades structures when more courses are pushed into the vector until eof. So the vector sizes are very likely different for each student. In the example input file, there are 3 students that have 3, 4 and 5 courses, respectively.
like:
Your program is a mess, had no error checking and crashed nonstop. That's why strtok is considered "evil" -- if you don't bother checking, bad things will happen. But a little checking makes it fine.
Nearly every line ended up commented out and replaced, I had to rewrite it instead.
#include <stdio.h>
#include <string.h>
#include <stdlib.h> // You need this for atoi, see 'man atoi'
#include <map>
#include <vector>
#include <string>
using namespace std;
struct grades {
string course_Abrrev;
int course_Credit;
char course_Grade;
};
int main(void)
{
char buf[4096];
map<int, string> names;
map<int, vector < grades > > marks;
// Read a line into 'buf'
while(fgets(buf, 4096, stdin) != NULL)
{
// Array of 10 pointers.
// Each will point to a position inside 'buf', or
// just point to "".
const char *token[10];
grades grade;
int n=0;
token[n]=strtok(buf, " \r\n\t");
while(n < 10)
{
n++;
token[n]=strtok(NULL, " \r\n\t");
if(token[n]==NULL) break;
}
// Set all NULL strings to "" instead.
while(n < 10) token[n++]="";
// There, now we have a nice array of tokens
// which are all guaranteed not to crash when used.
// The very worst any of them will be is blank...
n=atoi(token[0]); // Get student ID
grade.course_Abrrev=token[2];
grade.course_Credit=atoi(token[3]);
grade.course_Grade=*(token[3]);
names[n]=token[1]; // Store name
marks[n].push_back(grade); // store grade
}
for(map<int, string>::iterator i=names.begin();
i != names.end();
i++)
{
int id=(*i).first;
printf("%d %s\n", id, (*i).second.c_str());
for(vector<grades>::iterator n=marks[id].begin();
n != marks[id].end();
n ++ )
{
printf("\t%s %d %c\n",
(*n).course_Abrrev.c_str(),
(*n).course_Credit,
(*n).course_Grade);
}
}
}
Logically I was thinking in a totally different way! Definitely I missed many points that may be critical for programming. Let me try out your code first. Thanks a lot!
#include <stdio.h>
#include <string.h>
#include <stdlib.h> // You need this for atoi
#include <map>
#include <vector>
#include <string>
using namespace std;
struct course {
char abbrev[11];
int credit;
char grade;
};
int main(void)
{
char buf[4096];
map<int, string> names;
map<int, vector < course > > marks;
// Read a line into 'buf'
while(fgets(buf, 4096, stdin) != NULL)
{
course grade;
char name[21];
int id;
if(sscanf(buf, "%d %20s %10s %d %c",
&id,name,grade.abbrev,&grade.credit,&grade.grade) != 5)
{
fprintf(stderr, "Bad line %s\n", buf);
continue;
}
names[id]=name;
marks[id].push_back(grade);
}
for(map<int, string>::iterator i=names.begin();
i != names.end();
i++)
{
int id=(*i).first;
printf("%d %s\n", id, (*i).second.c_str());
for(vector<course>::iterator n=marks[id].begin();
n != marks[id].end();
n ++ )
{
printf("\t%s %d %c\n",
(*n).abbrev, (*n).credit, (*n).grade);
}
}
}
sscanf shortens it by 20 lines, you can just tell it "lines should like this" and it'll sort it into the arguments you asked for and tell you whether it could. It even protects against overflows if you tell it to (%20s, so it doesn't overrun a 21-char buffer).
This is what bugs me most about the 'scanf is dangerous, strtok is dangerous' camp. As opposed to what, hand-writing your own unchecked ad-hoc string parser? At least C has a parser.
It even protects against overflows if you tell it to (%20s, so it doesn't overrun a 21-char buffer).
Those are only concerns of advanced programmers. I can't predict the situations even you pointed it out.
BTW, the last version has a bug if there is a blank line in the input file....(No worry, I'll try to fix it myself.)
The biggest point for me is from the lines:
names[id] = name;
marks[id].push_back(grade);
I did not know this way to assign a map_pair so much simply.
A crashed program doesn't do the job whether you're beginner or advanced.
If you don't start thinking about it when you're a beginner, you never will.
No it doesn't:
$ cat student:
2333021 Bokow,R. NS201 3 A
2333021 Bokow,R. MG342 3 A
2333021 Bokow,R. FA302 1 A
2574063 Failin,D. MK106 3 C
2574063 Failin,D. MA208 3 B
2574063 Failin,D. CM201 3 C
2574063 Failin,D. CP101 2 B
2663628 Kingsley,M. QA140 3 A
2663628 Kingsley,M. CM245 3 B
2663628 Kingsley,M. EQ521 3 A
2663628 Kingsley,M. MK341 3 A
2663628 Kingsley,M. CP101 3 B
$ ./a.out
Bad line
2333021 Bokow,R.
NS201 3 A
MG342 3 A
FA302 1 A
2574063 Failin,D.
MK106 3 C
MA208 3 B
CM201 3 C
CP101 2 B
2663628 Kingsley,M.
QA140 3 A
CM245 3 B
EQ521 3 A
MK341 3 A
CP101 3 B
$
. You are really ideal for people like me and others who want to learn technology.