So far I can read every line and print it out to the console:
void readFile(){
string line;
ifstream myfile("example1.pgm");
if (myfile.is_open()){
while (myfile.good()){
getline (myfile,line);
cout << line;
}
}
However a pgm file apparently will always have the following at the start before the data:
P2
# test.pgm
24 7
15
How can i adapt my code so that it checks that "P2" is present, ignores any comments (#), and stores the variables and subsequent pixel data?
I'm a bit lost and new to c++ so any help is appreicated.
Thanks
There are a lot of different ways to parse a file. For something like this, you could look at the answers on this site. Personally, I would go with a loop of getline() and test/parse every line (stored in the variable "line"), you can also use a stringstream since it is easier to use with multiple values :
Idea
First line : test that P2 (Portable graymap) is present, maybe with something like
if(line.compare("P2")) ...
Second line : do nothing, you can go on with the next getline()
Third line : store the size of the image; with a stringstream you could do this
int w,h;
ss >> w >> h;
Following lines : store the pixel data until you reach the end of the file
Resulting code
You can try this code and adapt it to your needs :
#include <iostream> // cout, cerr
#include <fstream> // ifstream
#include <sstream> // stringstream
using namespace std;
int main() {
int row = 0, col = 0, numrows = 0, numcols = 0;
ifstream infile("file.pgm");
stringstream ss;
string inputLine = "";
// First line : version
getline(infile,inputLine);
if(inputLine.compare("P2") != 0) cerr << "Version error" << endl;
else cout << "Version : " << inputLine << endl;
// Second line : comment
getline(infile,inputLine);
cout << "Comment : " << inputLine << endl;
// Continue with a stringstream
ss << infile.rdbuf();
// Third line : size
ss >> numcols >> numrows;
cout << numcols << " columns and " << numrows << " rows" << endl;
int array[numrows][numcols];
// Following lines : data
for(row = 0; row < numrows; ++row)
for (col = 0; col < numcols; ++col) ss >> array[row][col];
// Now print the array to see the result
for(row = 0; row < numrows; ++row) {
for(col = 0; col < numcols; ++col) {
cout << array[row][col] << " ";
}
cout << endl;
}
infile.close();
}
EDIT
Here is a good tutorial on how to use stringstreams.
A way of simplifying PNM (PBM/PGM/PPM) header processing is to build up a header string line-by-line until you have captured all of the requisite data. It doesn't take too much code to do this using only the standard C++ libraries...
#include <string>
#include <iostream>
#include <sstream>
#include <stdexcept>
...
std::string header, magic;
int width=0, height=0, maxsample=0, samples=0, bits=0, bytes=0;
do {
try { getline(is,magic); } catch ( const std::ios_base::failure & ) {}
if ( !magic.empty() && magic[0] != '#' ) header += magic+" ";
if ( !( std::stringstream(header+" 1") >> magic >> width >> height >> maxsample ).eof() ) break;
if ( ( (magic=="P1"||magic=="P4") && maxsample==1 ) || !is.good() ) break;
} while ( true );
samples = magic=="P1"?1:magic=="P2"?1:magic=="P3"?3:magic=="P4"?1:magic=="P5"?1:magic=="P6"?3:0;
bits = (magic=="P1"||magic=="P4")?1:maxsample<256?8:maxsample<256*256?16:0, bytes = (width*samples*bits+7)>>3;
if ( width<=0 || height<=0 || maxsample<=0 || samples<=0 || bits<=0 ) throw std::runtime_error("invalid PNM header");
This handles comments (if present) and the special-case of PBM (no 'maxsample') -- and it works regardless of whether or not exceptions are enabled on the input stream.
Once you've read the header, reading the image data is usually a simple matter since the format is defined to just be a sequential data dump (which may be either ASCII or binary depending on the 'magic' value). In the case of 16-bit binary-encoded samples, the format specification indicates that "The most significant byte is first" (big endian), so this case may require some platform-specific handling.
As written, this requires C++11 -- probably due to the way I'm using stringstream
as a temporary.
One caveat: In a pathological case, it is possible for this to waste a lot of time/RAM while attempting to read an invalid header -- since the getline
call isn't inherently bounded. There's a relatively simple solution (replace getline
with something more robust), but it requires a bit more code.
For production-quality applications, consider using libnetpbm.