Question

In the MongoDB shell, if I type someDoc._id, Mongo replies with something like ObjectId(4f6b83af44c75956279e7777). How is that string generated from the ObjectId bytes?

Links to the javascript source for this are welcome, as are links to the source for other drivers.

Answer 1

Looks like that comes from BSONElement::toString:

case jstOID:
    s << "ObjectId('";
    s << __oid() << "')";
    break;

And __oid() gives you a mongo::OID and the << operator calls str:

inline StringBuilder& operator<< (StringBuilder& s, const OID& o) { return (s << o.str()); }

and that just calls toHexLower on the raw bytes:

string str() const { return toHexLower(data, 12); }

Finally, toHexLower does this:

inline string toHexLower(const void* inRaw, int len) {
    static const char hexchars[] = "0123456789abcdef";

    StringBuilder out;
    const char* in = reinterpret_cast<const char*>(inRaw);
    for (int i=0; i<len; ++i) {
        char c = in[i];
        char hi = hexchars[(c & 0xF0) >> 4];
        char lo = hexchars[(c & 0x0F)];

        out << hi << lo;
    }

    return out.str();
}

The stringification isn’t done in JavaScript, it is done in C++.

The above uses the 2.0 branch of MongoDB but the (current) master is the same (or nearly so).