ClickHouse/src/Common/getResource.cpp

#include "getResource.h"
#include <dlfcn.h>
#include <string>
#include <boost/algorithm/string/replace.hpp>
#include <Common/SymbolIndex.h>


std::string_view getResource(std::string_view name)
{
    // Convert the resource file name into the form generated by `ld -r -b binary`.
    std::string name_replaced(name);
    std::replace(name_replaced.begin(), name_replaced.end(), '/', '_');
    std::replace(name_replaced.begin(), name_replaced.end(), '-', '_');
    std::replace(name_replaced.begin(), name_replaced.end(), '.', '_');
    boost::replace_all(name_replaced, "+", "_PLUS_");

#if defined USE_MUSL
    /// If static linking is used, we cannot use dlsym and have to parse ELF symbol table by ourself.
    return DB::SymbolIndex::instance()->getResource(name_replaced);

#else
    // In most `dlsym(3)` APIs, one passes the symbol name as it appears via
    // something like `nm` or `objdump -t`. For example, a symbol `_foo` would be
    // looked up with the string `"_foo"`.
    //
    // Apple's linker is confusingly different. The NOTES on the man page for
    // `dlsym(3)` claim that one looks up the symbol with "the name used in C
    // source code". In this example, that would mean using the string `"foo"`.
    // This apparently applies even in the case where the symbol did not originate
    // from C source, such as the embedded binary resource files used here. So
    // the symbol name must not have a leading `_` on Apple platforms. It's not
    // clear how this applies to other symbols, such as those which _have_ a leading
    // underscore in them by design, many leading underscores, etc.
#if defined OS_DARWIN
    std::string prefix = "binary_";
#else
    std::string prefix = "_binary_";
#endif
    std::string symbol_name_start = prefix + name_replaced + "_start";
    std::string symbol_name_end = prefix + name_replaced + "_end";

    const char * sym_start = reinterpret_cast<const char *>(dlsym(RTLD_DEFAULT, symbol_name_start.c_str()));
    const char * sym_end = reinterpret_cast<const char *>(dlsym(RTLD_DEFAULT, symbol_name_end.c_str()));

    if (sym_start && sym_end)
    {
        auto resource_size = static_cast<size_t>(std::distance(sym_start, sym_end));
        return { sym_start, resource_size };
    }
    return {};
#endif
}