Is there an easy way to convert a std::vector<std::string> to a std::vector<unsigned char> (and back again to a std::vector<std::string>, without having to manually convert each string and add a delimiter, such as a comma?
4 Answers
The short answer is: no.
The way vectors and strings are implemented are as independant, heap-allocated arrays. Therefore, you could transform vector<string> into char** (a jagged array of arrays of char), and vector<unsigned char> into char* (an array of char) when thinking about internals. This turns your problem into: is there any way to concatenate arrays without having to copy them?
No. No there is not.
Comments
std::vector<char> chars;
for (const std::string& s : strings)
{
for (char c : s)
{
chars.push_back(c);
}
chars.push_back(',');
}
It's a little more clumsy without the new for loop syntax, but you get the idea.
1 Comment
Kerrek SB
C++ has iterators! You can write
chars.insert(chars.end(), s.begin(), s.end());.Boost Serialization should let you stuff a data structure into a sequence of unsigned char and reconstitute it again.
Comments
The first question is why, and what are you trying to do? What does the
std::vector<std::string> represent, and what should the semantics of
the conversion be? If you just want to concatenate, then the simplest
solution would be something like:
std::vector<unsigned char> results;
for ( std::vector<std::string>::const_iterator iter = source.begin();
iter != source.end();
++ iter ) {
results.insert( results.end(), iter->begin(), iter->end() );
}
The implicit conversion of char to unsigned char will take care of
the reslt.
If you need to insert some sort of separator or terminator for each
string in the source, you can do that in the loop as well: for a
terminator, just append it (push_back) after the insert; for a
separator, I generally append it conditionally before the insert,
e.g.:
std::vector<unsigned char> results;
for ( std::vector<std::string>::const_iterator iter = source.begin();
iter != source.end();
++ iter ) {
if ( iter != source.begin() ) {
results.push_back( separator );
}
results.insert( results.end(), iter->begin(), iter->end() );
}
But the question is: why unsigned char? Presumably, because you are
formatting into a buffer for some specific protocol. Is some additional
formatting required? What is the format of a string in your protocol?
(Typically, it will be either length + data, or '\0' terminated.)
Does the protocol require some sort of alignment? (For XDR—one of
the most widely used protocols—, you'd need something like:
std::vector<unsigned char> results;
for ( std::vector<std::string>::const_iterator iter = source.begin();
iter != source.end();
++ iter ) {
size_t len = iter->size();
results.push_back( (len >> 24) & 0xFF );
results.push_back( (len >> 16) & 0xFF );
results.push_back( (len >> 8) & 0xFF );
results.push_back( (len ) & 0xFF );
results.insert( results.end(), iter->begin(), iter->end() );
while ( results.size() % 4 != 0 ) {
results.push_back( '\0' );
}
}
, for example.)
std::vector<unsigned char> char_vec(myString.begin(), myString.end()), right? I was wondering if there was an easy way to convert from a std::vector<std::string> to a std::vector<unsigned char>` in a similar way.std::vector<std::vector<unsigned char> >?