Boost Date Time Parsing string

Question

I have looked at many examples seem to address this simple case. The string I want to parse is:

"2012-06-01 16:45:34 EDT"

I have tried to create a local_time_input_facet with the folloiwng: "%Y-%m-%d %H:%M:%S %Z"

The zone pointer of the local_date_time object is always not set. Reading the documentation is confusing:

%Z *! Full time zone name (output only). This flag is ignored when using the time_facet with a ptime.

"EDT" // Eastern Daylight Time

Has anyone done this before?

UPDATE: I have updated the code to illustrate the problem a little better:

using namespace std;
using namespace boost::local_time;

int main()
{
    stringstream ss;

    // Set up the input datetime format.
    local_time_input_facet *input_facet 
    = new local_time_input_facet("%Y-%m-%d %H:%M:%S %ZP");
    ss.imbue(std::locale(ss.getloc(), input_facet));

    local_date_time ldt(not_a_date_time),ldt1(not_a_date_time);

    // Read a time into ldt
    ss.str("2012-06-01 17:45:34 EDT");
    ss >> ldt;

    ss.str("2012-06-01 17:45:34 CDT");
        ss >> ldt1;
        std::cerr << (ldt - ldt1).total_seconds() << std::endl;

    // Write the time to stdout.
    cout << "Full Time:\t"   << ldt.to_string() << endl;
    cout << "Local time:\t"  << ldt.local_time() << endl;
    cout << "Time zone:\t"   << ldt.zone_as_posix_string() << endl;
    cout << "Zone abbrev:\t" << ldt.zone_abbrev() << endl;
    cout << "Zone offset:\t" << ldt.zone_abbrev(true) << endl;

    cout << "Full Time:\t"   << ldt1.to_string() << endl;
    cout << "Local time:\t"  << ldt1.local_time() << endl;
    cout << "Time zone:\t"   << ldt1.zone_as_posix_string() << endl;
    cout << "Zone abbrev:\t" << ldt1.zone_abbrev() << endl;
    cout << "Zone offset:\t" << ldt1.zone_abbrev(true) << endl;

    return 0;
}

OUTPUT:

0
Full Time:  2012-Jun-01 17:45:34 EDT
Local time: 2012-Jun-01 17:45:34
Time zone:  EDT+00
Zone abbrev:    EDT
Zone offset:    +0000
Full Time:  2012-Jun-01 17:45:34 CDT
Local time: 2012-Jun-01 17:45:34
Time zone:  CDT+00
Zone abbrev:    CDT
Zone offset:    +0000

Answer 1

The Bug

According to boost's documentation here: http://www.boost.org/doc/libs/1_57_0/doc/html/date_time/date_time_io.html#date_time.format_flags

%Z is:

Full time zone name (output only).

It also says to %ZP:

Posix time zone string (available to both input and output).

So you need to change %Z to %ZP. Now your timestamps will parse. However, you'll notice that the zone offset will not be set.

Posix Time Zone Strings

For a Posix time zone string, you'll need to specify at least the zone abbreviation and the offset from UTC, e.g. EST-5.

A full Posix time zone string is formatted as:

"std offset dst [offset],start[/time],end[/time]"

with no spaces, according to http://www.boost.org/doc/libs/1_57_0/doc/html/date_time/local_time.html#date_time.local_time.posix_time_zone

Here are some examples of full Posix time zone strings for EST and PST:

EST-5EDT,M3.2.0,M11.1.0
PST-8PDT,M4.1.0,M10.1.0

This contains the information on when Daylight Savings Time is in effect.

However, you may be able to get away with EDT-4 in your case, depending on what you're doing with it.

It should be noted that as simple as Posix time zones are, they are limited in that they don't account for historical changes in time zone rules. I think it's best to just avoid working with time zones in the first place.

What if I can't control the input format?

As the OP noted in the comments, the offsets are not listed in his input timestamps. One solution is to read the zone abbreviation from the end of the input timestamp (e.g. "EDT"), and check it against a map of known zone abbreviations:

std::map<std::string, std::string> zone_map;
zone_map["EST"] = "EST-5EDT,M4.1.0,M10.5.0";  // Eastern Standard Time
zone_map["EDT"] = zone_map["EST"];            // Eastern Daylight Time
zone_map["PST"] = "PST-8PDT,M4.1.0,M10.1.0";  // Pacific Standard Time
zone_map["PDT"] = zone_map["PST"];            // Pacific Daylight Time
// ...

(Note the DST zones above should be the same as the standard time zones.)

You might also get away with with just storing simple UTC offsets:

zone_map["EST"] = "EST-5";  // Eastern Standard Time
zone_map["EDT"] = "EDT-4";  // Eastern Daylight Time
// ...

Working Example

Here's an example that uses a built-in timezone database:

#include <map>
#include <string>
#include <sstream>
#include <iostream>
#include <boost/date_time/local_time/local_time.hpp>

using namespace boost::local_time;

int main()
{
    // A little database of time zones.
    std::map<std::string, std::string> zone_map;
    zone_map["EST"] = "EST-5EDT,M4.1.0,M10.5.0";  // Eastern Standard Time
    zone_map["EDT"] = zone_map["EST"];            // Eastern Daylight Time
    zone_map["PST"] = "PST-8PDT,M4.1.0,M10.1.0";  // Pacific Standard Time
    zone_map["PDT"] = zone_map["PST"];            // Pacific Daylight Time
    // ...

    // This is our input timestamp.
    std::string timestamp = "2012-06-01 16:45:34 EDT";

    // Replace time zone abbrev with full Posix time zone.
    const size_t abbrev_pos = timestamp.find_last_of(' ') + 1;
    const std::string abbrev = timestamp.substr(abbrev_pos);
    timestamp.replace(abbrev_pos, std::string::npos, zone_map[abbrev]);

    std::cout << "Time stamp with full timezone: " << timestamp << std::endl;

    // Set up the input datetime format.
    local_time_input_facet *input_facet = new local_time_input_facet("%Y-%m-%d %H:%M:%S %ZP");
    std::stringstream ss;
    ss.imbue(std::locale(ss.getloc(), input_facet));

    // This is our output date time.
    local_date_time ldt(not_a_date_time);

    // Read the timestamp into ldt.
    ss.str(timestamp);
    ss >> ldt;

    // Write the time to stdout.
    std::cout << "Full Time:\t"   << ldt.to_string() << std::endl
              << "Local time:\t"  << ldt.local_time() << std::endl
              << "Time zone:\t"   << ldt.zone_as_posix_string() << std::endl
              << "Zone abbrev:\t" << ldt.zone_abbrev() << std::endl
              << "Zone offset:\t" << ldt.zone_abbrev(true) << std::endl;

    return 0;
}

This outputs:

Time stamp with full timezone: 2012-06-01 16:45:34 EST-5EDT,M4.1.0,M10.5.0
Full Time:      2012-Jun-01 16:45:34 EDT
Local time:     2012-Jun-01 16:45:34
Time zone:      EST-05EDT+01,M4.1.0/02:00,M10.5.0/02:00
Zone abbrev:    EDT
Zone offset:    -0400

While the solution here may not be ideal, it's the only way I can see of doing it.