Boost C++ Libraries of the most highly regarded and expertly designed C++ library projects in the world. Herb Sutter and Andrei Alexandrescu, C++ Coding Standards


Class template tz_db_base

boost::date_time::tz_db_base — Creates a database of time_zones from csv datafile.


// In header: <boost/date_time/tz_db_base.hpp>

template<typename time_zone_type, typename rule_type> 
class tz_db_base {
  // types
  typedef char                                                           char_type;             
  typedef time_zone_type::base_type                                      time_zone_base_type;   
  typedef time_zone_type::time_duration_type                             time_duration_type;    
  typedef time_zone_names_base< char_type >                              time_zone_names;       
  typedef boost::date_time::dst_adjustment_offsets< time_duration_type > dst_adjustment_offsets;
  typedef std::basic_string< char_type >                                 string_type;           

  // construct/copy/destruct

  // public member functions
  void load_from_stream(std::istream &);
  void load_from_file(const std::string &);
  bool add_record(const string_type &, 
                  boost::shared_ptr< time_zone_base_type >);
  boost::shared_ptr< time_zone_base_type > 
  time_zone_from_region(const string_type &) const;
  std::vector< std::string > region_list() const;

  // private member functions
  rule_type * parse_rules(const string_type &, const string_type &) const;
  week_num get_week_num(int) const;
  void split_rule_spec(int &, int &, int &, string_type) const;
  bool parse_string(string_type &);


The csv file containing the zone_specs used by the tz_db_base is intended to be customized by the library user. When customizing this file (or creating your own) the file must follow a specific format.

This first line is expected to contain column headings and is therefore not processed by the tz_db_base.

Each record (line) must have eleven fields. Some of those fields can be empty. Every field (even empty ones) must be enclosed in double-quotes. Ex:

"America/Phoenix" <- string enclosed in quotes
""                <- empty field

Some fields represent a length of time. The format of these fields must be:

"{+|-}hh:mm[:ss]" <- length-of-time format

Where the plus or minus is mandatory and the seconds are optional.

Since some time zones do not use daylight savings it is not always necessary for every field in a zone_spec to contain a value. All zone_specs must have at least ID and GMT offset. Zones that use daylight savings must have all fields filled except: STD ABBR, STD NAME, DST NAME. You should take note that DST ABBR is mandatory for zones that use daylight savings (see field descriptions for further details).

******* Fields and their description/details *********

ID: Contains the identifying string for the zone_spec. Any string will do as long as it's unique. No two ID's can be the same.

STD ABBR: STD NAME: DST ABBR: DST NAME: These four are all the names and abbreviations used by the time zone being described. While any string will do in these fields, care should be taken. These fields hold the strings that will be used in the output of many of the local_time classes. Ex:

time_zone nyc = tz_db.time_zone_from_region("America/New_York");
local_time ny_time(date(2004, Aug, 30), IS_DST, nyc);
cout << ny_time.to_long_string() << endl;
// 2004-Aug-30 00:00:00 Eastern Daylight Time
cout << ny_time.to_short_string() << endl;
// 2004-Aug-30 00:00:00 EDT

NOTE: The exact format/function names may vary - see local_time documentation for further details.

GMT offset: This is the number of hours added to utc to get the local time before any daylight savings adjustments are made. Some examples are: America/New_York offset -5 hours, & Africa/Cairo offset +2 hours. The format must follow the length-of-time format described above.

DST adjustment: The amount of time added to gmt_offset when daylight savings is in effect. The format must follow the length-of-time format described above.

DST Start Date rule: This is a specially formatted string that describes the day of year in which the transition take place. It holds three fields of it's own, separated by semicolons. The first field indicates the "nth" weekday of the month. The possible values are: 1 (first), 2 (second), 3 (third), 4 (fourth), 5 (fifth), and -1 (last). The second field indicates the day-of-week from 0-6 (Sun=0). The third field indicates the month from 1-12 (Jan=1).

Examples are: "-1;5;9"="Last Friday of September", "2;1;3"="Second Monday of March"

Start time: Start time is the number of hours past midnight, on the day of the start transition, the transition takes place. More simply put, the time of day the transition is made (in 24 hours format). The format must follow the length-of-time format described above with the exception that it must always be positive.

DST End date rule: See DST Start date rule. The difference here is this is the day daylight savings ends (transition to STD).

End time: Same as Start time.

tz_db_base public construct/copy/destruct

  1. tz_db_base();
    Constructs an empty database.

tz_db_base public member functions

  1. void load_from_stream(std::istream & in);
    Process csv data file, may throw exceptions.

    May throw bad_field_count exceptions

  2. void load_from_file(const std::string & pathspec);
    Process csv data file, may throw exceptions.

    May throw data_not_accessible, or bad_field_count exceptions

  3. bool add_record(const string_type & region, 
                    boost::shared_ptr< time_zone_base_type > tz);
    returns true if record successfully added to map

    Takes a region name in the form of "America/Phoenix", and a time_zone object for that region. The id string must be a unique name that does not already exist in the database.

  4. boost::shared_ptr< time_zone_base_type > 
    time_zone_from_region(const string_type & region) const;
    Returns a time_zone object built from the specs for the given region.

    Returns a time_zone object built from the specs for the given region. If region does not exist a local_time::record_not_found exception will be thrown

  5. std::vector< std::string > region_list() const;
    Returns a vector of strings holding the time zone regions in the database.

tz_db_base private member functions

  1. rule_type * parse_rules(const string_type & sr, const string_type & er) const;
    parses rule specs for transition day rules
  2. week_num get_week_num(int nth) const;
    helper function for parse_rules()
  3. void split_rule_spec(int & nth, int & d, int & m, string_type rule) const;
    splits the [start|end]_date_rule string into 3 ints
  4. bool parse_string(string_type & s);
    Take a line from the csv, turn it into a time_zone_type.

    Take a line from the csv, turn it into a time_zone_type, and add it to the map. Zone_specs in csv file are expected to have eleven fields that describe the time zone. Returns true if zone_spec successfully added to database