anonymous_namespace{StringDictionary.cpp} Namespace Reference

Functions
int	checked_open (const char *path, const bool recover)
const uint64_t	round_up_p2 (const uint64_t num)
string_dict_hash_t	hash_string (const std::string_view &str)
template<class T >
void	throw_encoding_error (std::string_view str, std::string_view folder)
bool	is_like (const std::string &str, const std::string &pattern, const bool icase, const bool is_simple, const char escape)
bool	is_regexp_like (const std::string &str, const std::string &pattern, const char escape)

Variables
const int	SYSTEM_PAGE_SIZE = omnisci::get_page_size()

Function Documentation

int anonymous_namespace{StringDictionary.cpp}::checked_open	(	const char *	path,
		const bool	recover
	)

Definition at line 52 of file StringDictionary.cpp.

References logger::ERROR, LOG, and omnisci::open().

Referenced by StringDictionary::StringDictionary().

                                                       {
  auto fd = omnisci::open(path, O_RDWR | O_CREAT | (recover ? O_APPEND : O_TRUNC), 0644);
  if (fd > 0) {
    return fd;
  }
  auto err = std::string("Dictionary path ") + std::string(path) +
             std::string(" does not exist.");
  LOG(ERROR) << err;
  throw DictPayloadUnavailable(err);
}

Here is the call graph for this function:

Here is the caller graph for this function:

string_dict_hash_t anonymous_namespace{StringDictionary.cpp}::hash_string

(

const std::string_view &

str

)

Definition at line 81 of file StringDictionary.cpp.

References i.

Referenced by StringDictionary::getOrAddBulk(), StringDictionary::getOrAddImpl(), StringDictionary::getUnlocked(), StringDictionary::hashStrings(), StringDictionary::increaseHashTableCapacity(), StringDictionary::increaseHashTableCapacityFromStorageAndMemory(), and StringDictionary::StringDictionary().

                                                          {
  string_dict_hash_t str_hash = 1;
  // rely on fact that unsigned overflow is defined and wraps
  for (size_t i = 0; i < str.size(); ++i) {
    str_hash = str_hash * 997 + str[i];
  }
  return str_hash;
}

Here is the caller graph for this function:

bool anonymous_namespace{StringDictionary.cpp}::is_like	(	const std::string &	str,
		const std::string &	pattern,
		const bool	icase,
		const bool	is_simple,
		const char	escape
	)

Definition at line 615 of file StringDictionary.cpp.

References string_ilike(), string_ilike_simple(), string_like(), and string_like_simple().

Referenced by StringDictionaryProxy::getLike(), and StringDictionary::getLike().

                                {
  return icase
             ? (is_simple ? string_ilike_simple(
                                str.c_str(), str.size(), pattern.c_str(), pattern.size())
                          : string_ilike(str.c_str(),
                                         str.size(),
                                         pattern.c_str(),
                                         pattern.size(),
                                         escape))
             : (is_simple ? string_like_simple(
                                str.c_str(), str.size(), pattern.c_str(), pattern.size())
                          : string_like(str.c_str(),
                                        str.size(),
                                        pattern.c_str(),
                                        pattern.size(),
                                        escape));
}

Here is the call graph for this function:

Here is the caller graph for this function:

bool anonymous_namespace{StringDictionary.cpp}::is_regexp_like	(	const std::string &	str,
		const std::string &	pattern,
		const char	escape
	)

Definition at line 905 of file StringDictionary.cpp.

References regexp_like().

Referenced by StringDictionaryProxy::getRegexpLike(), and StringDictionary::getRegexpLike().

                                       {
  return regexp_like(str.c_str(), str.size(), pattern.c_str(), pattern.size(), escape);
}

Here is the call graph for this function:

Here is the caller graph for this function:

const uint64_t anonymous_namespace{StringDictionary.cpp}::round_up_p2

(

const uint64_t

num

)

Definition at line 63 of file StringDictionary.cpp.

References UINT32_MAX.

Referenced by StringDictionary::StringDictionary().

                                               {
  uint64_t in = num;
  in--;
  in |= in >> 1;
  in |= in >> 2;
  in |= in >> 4;
  in |= in >> 8;
  in |= in >> 16;
  in++;
  // TODO MAT deal with case where filesize has been increased but reality is
  // we are constrained to 2^31.
  // In that situation this calculation will wrap to zero
  if (in == 0 || (in > (UINT32_MAX))) {
    in = UINT32_MAX;
  }
  return in;
}

Here is the caller graph for this function:

template<class T >

void anonymous_namespace{StringDictionary.cpp}::throw_encoding_error	(	std::string_view	str,
		std::string_view	folder
	)

Definition at line 307 of file StringDictionary.cpp.

References logger::ERROR, LOG, and omnisci.dtypes::T.

                                                                     {
  std::ostringstream oss;
  oss << "The text encoded column stored at " << folder << ", has exceeded its limit of "
      << sizeof(T) * 8 << " bits (" << static_cast<size_t>(max_valid_int_value<T>() + 1)
      << " unique values)."
      << " There was an attempt to add the new string '" << str
      << "'. Table will need to be recreated with larger String Dictionary Capacity";
  LOG(ERROR) << oss.str();
  throw std::runtime_error(oss.str());
}

Variable Documentation

const int anonymous_namespace{StringDictionary.cpp}::SYSTEM_PAGE_SIZE = omnisci::get_page_size()

Definition at line 50 of file StringDictionary.cpp.

Referenced by StringDictionary::addMemoryCapacity(), and StringDictionary::addStorageCapacity().