How to skip header row in csv using C++ - c++

Question

In my scenario,I need to create a parameters file using CSV .Every row means a config-data,the first field of the row treated as the header,using as a identifier. The format of CSV like below will be easy for me to parse:
1,field1,field2,field3,field4　// 1 indicated the TARGET that the other fields will be writted to.
1,field1,field2,field3,field4
2,field1,field2,field3,field4
2,field1,field2,field3,field4........
But it's not friendly to users.So,I define a csv file like below:
HeaderLine_Begin,1
field1,field2,field3,field4
field1,field2,field3,field4
HeaderLine_Begin,2
field1,field2,field3,field4
field1,field2,field3,field4
means,every row is the data will be writted to the target by the HeaderLine_Begin.I just separate the ID from the real data.
Then,I create a struct like this:
enum myenum
{
ON,OFF,NOCHANGE
};
struct Setting
{
int TargetID;
string field1;
string field2;
myenum field3;
myenum field4;
};
I knew how to write some code for reading csv line by line like below
filename +=".csv";
std::ifstream file(filename.c_str());
std::string line;
while ( file.good() )
{
getline ( file, line, '\n' ); // read a line until last
if(line.compare(0,1,"#") == 0) // ignore the comment line
continue;
ParseLine();// DONE.Parse the line if it's header row OR data row
}
file.close(); // close file
What I want to do is to create a list like vetor settings to keep the data.The flow should be,like,find the first headerID1,then find the next line.If the next line is dataline,treat it as the dataline belong to headerID1.If the next line is another headerID,loop again.
The problem is,there are no such std::getnextline(int lineIndex) for me to fetch the rows after I found the headerRow.

Answer 1

Your input loop should be more like:
int id = -1;
while (getline(file, line))
{
if (line.empty() || line[0] == '#')
continue;
if (starts_with_and_remove(line, "HeaderLine_Begin,"))
id = boost::lexical_cast<int>(line); // or id = atoi(line.c_str())
else
{
assert(id != -1);
...parse CSV, knowing "id" is in effect...
}
}
With:
bool stats_with_and_remove(std::string& lhs, const std::string& rhs)
{
if (lhs.compare(0, rhs.size(), lhs) == 0) // rhs.size() > lhs.size() IS safe
{
lhs.erase(0, rhs.size());
return true;
}
return false;
}

Answer 2

The simplest solution would be to use regular expressions:
std::string line;
int currentId = 0;
while ( std::getline( source, line ) ) {
trimCommentsAndWhiteSpace( line );
static std::regex const header( "HeaderLine_Begin,(\\d+)" );
std::smatch match;
if ( line.empty() ) {
// ignore
} else if ( std::regex_match( line, match, header ) ) {
std::istringstream s( match[ 1 ] );
s >> currentId;
} else {
// ...
}
}
I regularly use this strategy to parse .ini files, which pose
the same problem: section headers have a different syntax to
other things.
trimCommentsAndWhiteSpace can be as simple as:
void
trimCommentsAndWhiteSpace( std::string& line )
{
if ( !line.empty() && line[0] == '#' ) {
line = "";
}
}
It's fairly easy to expand it to handle end of line comments as
well, however, and it's usually a good policy (in contexts like
this) to trim leading and trailing whitespace---trailing
especially, since a human reader won't see it when looking at
the file.
Alternatively, of course, you could use a regular expression for
the lines you want to treet as comments ("\s*#.*"); this works
well with your current definition, but doesn't really extend
well for end of line comments, especially if you want to allow
# in quoted strings in your fields.
And one final comment: your loop is incorrect. You don't test
that getline succeeded before using its results, and
file.good() may return true even if there is nothing more to
read. (file.good() is one of those things that are there for
historical reasons; there's no case where it makes sense to use
it.)