Parse text file for name value pair

grob115 · 12-04-2011, 10:24 AM

Hi, am looking for some sample code to parse a file. Here's an example config file I have.

Code:

NAME=FIRST_PERSON
AGE=11
SEX=M

NAME=SECOND_PERSON
AGE=23
SEX=M

NAME=THIRD_PERSON
AGE=33
SEX=F

The file can contain one or N number of these name-value pairs. The code needs to be able to do the following:
1) Read them all in.
2) Ensure each one has the variables NAME, AGE, and SEX all filled in.

Anyone can suggest a way to do this in C++?

Sergei Steshenko · 12-04-2011, 10:29 AM

Quote:

Originally Posted by grob115

...
Anyone can suggest a way to do this in C++?

Try to enter

C++ text parsing

into any web search engine. Start from plain old "C" 'index' function (probably a part of C++ standard library too) : http://linux.die.net/man/3/index .

SigTerm · 12-04-2011, 02:03 PM

Quote:

Originally Posted by grob115

Anyone can suggest a way to do this in C++?

There's more than one way to do it. So why don't you post what you already tried to do and explain where you encountered the problem?

jb_gpk · 12-04-2011, 06:35 PM

Quote:

Originally Posted by grob115

Anyone can suggest a way to do this in C++?

Why not awk?

jhwilliams · 12-04-2011, 08:47 PM

Quote:

Originally Posted by jb_gpk

Why not awk?

How do you do this in Awk? I ask not because I doubt it can be done; rather, I know it can, but am trying to learn awk. (Beyond '{ print $1 }'.)

jhwilliams · 12-05-2011, 02:02 AM

Tada, an awk script -- will complain if not all values are set.

Code:

#!/usr/bin/awk -f
BEGIN {
    RS=""
    FS="="
    OFS="\t"
}
($1 ~ /NAME/),($NR == 0) {
    if (NF != 6) {
        print "Badly formed input line"
        exit
    }

    print $2,$4,$6
}

grob115 · 12-05-2011, 06:44 AM

Great idea with the awk, though it needs to be done in C++ as I'm learning it by trying to do different things of which this is one of them. Am thinking of reading the stuff into a vector of struct, which consists of the expected variables per set. Is this a good approach?

grob115 · 12-05-2011, 10:43 AM

Hi, this is my attempt to do this, though am having a bunch of errors from the IDE in NetBeans as shown in the attachment.
Not sure why it's complaining about simple things like the if and while statements.
Also, not sure why it can't find the <iostream>, <ifstream>, or <ofstream> headers. I'm using remote build option btw and it is able to find the compilers, assemblers, and linkers on the remote Linux box.

Code:

#include <string>
#include <iostream>

using namespace std;

string name, age, sex;
string field, value;
string ifilename = "config.txt";
string lfilename = "out.log";
ifstream ifile(ifilename.c_str());
ofstream ofile(lfilename.c_str());
bool bname = false, bage = false, bsex = false;
int count = 0;

typedef struct {
  string name;
  int age;
  char sex;
} TypeStructConfig;

typedef std::vector<TypeStructConfig> VectorStructConfig;

VectorStructConfig structConfig;

if (count == 0) {
    std::cout << "h";
}

while (!ifile.eof()) {
    getline(ifile, field, '=');
    getline(ifile, value);

    switch (field) {
        case NAME:{
        structConfig(count).name = value;
        bname = true;
        break;
        }
        case AGE:{
        structConfig(count).age = value;
        bage = true;
        break;
        }
        case SEX:{
        structConfig(count).sex = value;
        bsex = true;
        break;
        }
        default:{
            ofile << field << " is an unknown configuration parameter.\n";
            exit;
        }
    }

    if (bname == false || bage == false || bsex == false) {
        ofile << "Not all required parameters are set.\n";
        exit;
    } else {
        count++;
    }
}

Nylex · 12-05-2011, 11:11 AM

I'm not sure about NetBeans, but to me it seems a few things are wrong:

Firstly, you're missing the line "#include <vector>". Also, C++ vectors are created empty, so you first need to add items to them before modifying them, e.g.

Code:

TypeStructConfig foo;
foo.name = "A N Other";
foo.age = 1;
foo.sex = 'M';
structConfig.push_back(foo);

Also, you use square brackets to access elements, so in lines like

Code:

structConfig(count).sex = value;

the round brackets are wrong. You can use the member function at() to access elements, instead of using array syntax (i.e. square brackets), e.g.

Code:

structConfig.at(count).sex = value;

I've only had a quick glance at the code, so there might be other things too.

jhwilliams · 12-05-2011, 11:12 AM

I don't really know C++ that well these days, I guess. But here's a C version:

Code:

#include <string.h>
#include <stdio.h>

#define LINE_MAX 100

int main() {
    char line[LINE_MAX];
    char *key;
    char *value;

    while (EOF != scanf("%s\n", line)) {
        key = strtok(line, "=");
        value = strtok(NULL, "=");
        printf("%s=%s\n", key, value);
    }
}

Cedrik · 12-05-2011, 11:31 AM

Quote:

Originally Posted by grob115

Hi, this is my attempt to do this, though am having a bunch of errors from the IDE in NetBeans as shown in the attachment.
Not sure why it's complaining about simple things like the if and while statements.

There is no main() function ?

Sergei Steshenko · 12-05-2011, 12:05 PM

Quote:

Originally Posted by jhwilliams

I don't really know C++ that well these days, I guess. But here's a C version:

Code:

#include <string.h>
#include <stdio.h>

#define LINE_MAX 100

int main() {
    char line[LINE_MAX];
    char *key;
    char *value;

    while (EOF != scanf("%s\n", line)) {
        key = strtok(line, "=");
        value = strtok(NULL, "=");
        printf("%s=%s\n", key, value);
    }
}

'scanf' is a bad idea because of possible buffer overflows.

SigTerm · 12-05-2011, 12:19 PM

Quote:

Originally Posted by grob115

Hi, this is my attempt to do this,

This is not a valid C++ program. There's no main() function, and code is mixed up with global variables, and there are undeclared identifiers (NAME/AGE/SEX).
Also you don't use "typedef struct{} StructName" in C++. Make sure it compiles, fix all the errors, and try again - because until you do this, there's nothing to talk about (unless you're trying to understand compiler error). If you have trouble writing a basic C++ program, start from beginning - try any C++ book or tutorial.

grob115 · 12-06-2011, 07:55 AM

Hi, reason there wasn't a "main" function was because I was thinking of putting this into one of the cpp files for the project, and use include this from the main source file.

Anyway, if I'm writing it as a stand alone program, it'd look like the following. However, am stuck on getting the vector of struct to work correctly. I'm stuck. Can someone please advise?

Code:

test.cpp:28: error: expected initializer before ‘<’ token
test.cpp:29: error: ‘TypeVectorStructConfig’ does not name a type
test.cpp: In function ‘int main(int, char**)’:
test.cpp:40: error: ‘VectorStructConfig’ was not declared in this scope
test.cpp:43: error: ‘VectorStructConfig’ was not declared in this scope
test.cpp:46: error: ‘VectorStructConfig’ was not declared in this scope

Code:

#include <cstdlib>
#include <string>
#include <iostream>
#include <fstream>

std::string name, age, sex;
std::string field, value;
std::string ifilename = "config.txt";
std::string lfilename = "out.log";
std::ifstream ifile(ifilename.c_str());
std::ofstream ofile(lfilename.c_str());
bool bname = false, bage = false, bsex = false;
int count = 0;

struct TypeStructConfig {
  std::string name;
  int age;
  char sex;
} ;

typedef std::vector<TypeStructConfig> TypeVectorStructConfig;
TypeVectorStructConfig VectorStructConfig;

/*
 * 
 */
int main(int argc, char** argv) { 

    while (getline(ifile, field, '=')) {        
        getline(ifile, value);
        
        if (field == "NAME") {
            VectorStructConfig[count].name = value;
            bname = true; 
        } else if (field == "AGE") {
            VectorStructConfig[count].age = value;
            bage = true;
        } else if (field == "SEX") {
            VectorStructConfig[count].sex = value;
            bsex = true;
        } else {
            ofile << field << " is an unknown configuration parameter.\n";
            return 1;
        }
    }

    if (bname == false || bage == false || bsex == false) {
        ofile << "Not all required parameters are set.\n";
        return 1;
    } else {
        count++;
    }
    
    return 0;
}

Nylex · 12-06-2011, 08:09 AM

For one thing, you're missing the "vector" header, as I pointed out in my post above. Also, you still aren't adding any elements to the vector. Again, see my post above (you need to use the push_back() member function).

The following line is wrong:

Code:

VectorStructConfig[count].age = value;

value is a string, but the age member of your struct is an int. You can get from a string to an int with a stringstream:

Code:

#include <iostream>
#include <sstream>
#include <string>
using namespace std;

int main()
{
  string s("123");
  stringstream ss;
  ss << s;
  int x;
  ss >> x;
  cout << x << endl;
  return 0;
}

Obviously, you need to be sure that what's being input is really a numeric value!

The line

Code:

VectorStructConfig[count].sex = value;

is wrong because value is a string, but the sex member is a char. You can either use string's at() member function or use array syntax to get the character:

Code:

string s("a");
cout << s.at(0) << endl; 
cout << s[0] << endl;

Edit: As SigTerm mentioned, you might want to try a C++ tutorial or book. There's a tutorial here.