Olson Timezone DB

[Rick Kelly]

Is there a PB implementation for the Olson Timezone DB before I write one?

[Michael Mattias]

Heck I could not find anything but a .NET API.

I guess you get to do it yourself. But on the plus side....it should not be too hard...

File formats
The zoneinfo database is published as a set of text files which list the rules and zone transitions in a human-readable format. For use, these text files are compiled into a set of platform-independent binary files—one per time zone. The reference source code includes such a compiler called zic (zone information compiler), as well as code to read those files and use them in standard APIs such as localtime() and mktime().

Source: http://en.wikipedia.org/wiki/Zoneinfo#File_formats

... er, at least the part where you start with the human-readable format should not be too hard.

I'll guess the current compiled binary files use some kind of compression and the APIs decompress on the fly as needed.

There's a number of folks here who have 'big interest' in compression technology... you should be able to find a couple of relevant threads here to get you started.

MCM

[Rick Kelly]

Olson Timezone DB

I ran google for an hour and couldn't uncover anything related to various BASIC language flavors that I could co-opt. This is my first effort in using PB and have a date/time DLL with all kinds of goodies in it - only missing timezone support. I plan on uploading the source here when I'm done and move on to my first "real" project after this familiarization effort.

The Olson files look pretty straightforward.

1. Skip lines that start with '#'
2. Skip blank lines
3. Change all the tab chars to spaces
4. Compress two consecutive spaces until no more occurences are found
5. Split the line into an array
6. Remove possible trailing comments by looking for a '#' character
7. First array entry will tell you what to do

or something along similiar lines.

So now, the only question is the format for the output side. I could emit a series of DATA statements for a DLL one could compile that loaded some arrays, SQLlite, custom external file, etc and then exposed some functions for TZ retrievals of various sorts.

More thought obviously needed....

Rick

[Michael Mattias]

This is my first effort in using PB and have a date/time DLL with all kinds of goodies in it - only missing timezone support

Exactly what kind of "timezone support" do you need? There may be "another way" to do it, a method sans the Olsen DB.

[Rick Kelly]

Olson Timezone DB

I'm kind of a nut with regard to dates and times. The PB DLL I have now exposes about 45 functions and supports Gregorian, Chinese (with Korean, Vietnamese, Japanese sub variants), ISO, Julian, Hebrew, Islamic (observational), Persian, Hindu Solar and Lunar, etc calendars. Moonrise/set, Sunrise/set, Bearing, Distance, Lunar Illumination, Rules based calculation of Gregorian dates, Business date calcuations, Weekday, various types of date/time math, retrieval of TZInfo from Windows with option of getting the actual transition date/times, and a fistful of misc stuff that I just found interesting. I just want the ability to move a date/time from one timezone to another and the Olson db seems the best source for all the rules that have occurred. Windows API's are good for local to UTC and UTC to local and that's about all I want from the OS.

It's all good learning and fun to boot....

Rick

[John Gleason]

One way the code might look for those rules, tho I never saw any actual data, so I just made an example line:

#COMPILE EXE
#DIM ALL

FUNCTION PBMAIN () AS LONG

'1. Skip lines that start with '#'
'2. Skip blank lines
'3. Change all the tab chars to spaces
'4. Compress two consecutive spaces until no more occurences are found
'5. Split the line into an array
'6. Remove possible trailing comments by looking for a '#' character
'7. First array entry will tell you what to do

LOCAL x AS LONG, lineOdata AS STRING
'   OPEN file FOR INPUT AS #1
'   LINE INPUT lineOdata
    lineOdata = "123456  asdfgh 123456         123456 asdfgh" & $TAB & $TAB & $TAB & "123456  asdfgh 123456  asdfgh"
    ? "An example data line: " & $CRLF & $CRLF & lineOdata
    
    IF ASC(lineOdata, 1) <> 35 THEN                'if 1st char not = "#"
       IF lineOdata > "      " THEN                'and line isn't blank
          REPLACE ANY $TAB WITH $SPC IN lineOdata  'tabs change to spaces
          DO
             IF INSTR(lineOdata, "  ") THEN        '2 spaces change to one space until none left
                REPLACE "  " WITH $SPC IN lineOdata'     "              "                "
             ELSE
                EXIT DO
             END IF
          LOOP
          x = PARSECOUNT(lineOdata, $SPC)          'split to array
          DIM timeZonArr(1 TO x) AS STRING         '      "
          PARSE lineOdata, timeZonArr(), $SPC      '      "
          ? timeZonArr(1)                          '1st element
          ? timeZonArr(5)                          '5th element
       END IF
    END IF
    
END FUNCTION

[Michael Mattias]

I just want the ability to move a date/time from one timezone to another and the Olson db seems the best source for all the rules that have occurred. Windows API's are good for local to UTC and UTC to local and that's about all I want from the OS.

UTC is universal. If you know the source and destination timezones the WInAPI can get you the local times.

OTOH, if your real challenge is "what is the timezone adjustment between New York and Bangkok - today" that's where AFAIK you'll need some kind of non-intrinsic support such as the Olsen DB.

[Rick Kelly]

UTC is universal. If you know the source and destination timezones the WInAPI can get you the local times.

OTOH, if your real challenge is "what is the timezone adjustment between New York and Bangkok - today" that's where AFAIK you'll need some kind of non-intrinsic support such as the Olsen DB.

It seems one also has to know the rules for DST (daylight savings) in both locales in addition to the timezones active for that datetime. Many locales have a history of altering both for a myriad of reasons. Additionally, some countries based their transition dates for DST on non-Gregorian calendars. Since I'm in the neighborhood, so to speak, I thought it challenging to tackle all the above and the Olson files are simply a part of the possible solution as I dive into this area a bit deeper. I've looked into the WIN32 API and there are certainly some things that will be used and I'm also careful of the API dependencies I might use.

Even in it's current form, perhaps if I upload the work so far, it might assist you and others to get a idea of what I'm up to.

Rick

[Marc Giao]

Hi Rick,

I think you and our other PowerBASIC time lord and date guru, Egbert Zijlema, have a lot to talk about:

http://zijlema.basicguru.eu/

I am surprised he hasn't said hello yet

[Rick Kelly]

Hi Rick,

I think you and our other PowerBASIC time lord and date guru, Egbert Zijlema, have a lot to talk about:

http://zijlema.basicguru.eu/

I am surprised he hasn't said hello yet

That would be interesting! I'm the former date guru for Corel Paradox (http:\\www.thedbcommunity.com)

[Rick Kelly]

One way the code might look for those rules, tho I never saw any actual data, so I just made an example line:

#COMPILE EXE
#DIM ALL
 
FUNCTION PBMAIN () AS LONG
 
'1. Skip lines that start with '#'
'2. Skip blank lines
'3. Change all the tab chars to spaces
'4. Compress two consecutive spaces until no more occurences are found
'5. Split the line into an array
'6. Remove possible trailing comments by looking for a '#' character
'7. First array entry will tell you what to do
 
LOCAL x AS LONG, lineOdata AS STRING
'   OPEN file FOR INPUT AS #1
'   LINE INPUT lineOdata
    lineOdata = "123456  asdfgh 123456         123456 asdfgh" & $TAB & $TAB & $TAB & "123456  asdfgh 123456  asdfgh"
    ? "An example data line: " & $CRLF & $CRLF & lineOdata
 
    IF ASC(lineOdata, 1) <> 35 THEN                'if 1st char not = "#"
       IF lineOdata > "      " THEN                'and line isn't blank
          REPLACE ANY $TAB WITH $SPC IN lineOdata  'tabs change to spaces
          DO
             IF INSTR(lineOdata, "  ") THEN        '2 spaces change to one space until none left
                REPLACE "  " WITH $SPC IN lineOdata'     "              "                "
             ELSE
                EXIT DO
             END IF
          LOOP
          x = PARSECOUNT(lineOdata, $SPC)          'split to array
          DIM timeZonArr(1 TO x) AS STRING         '      "
          PARSE lineOdata, timeZonArr(), $SPC      '      "
          ? timeZonArr(1)                          '1st element
          ? timeZonArr(5)                          '5th element
       END IF
    END IF
 
END FUNCTION

As a quick proof on concept, I put together this quick code snippet:

#COMPILE EXE
#DIM ALL
#DEBUG DISPLAY ON
FUNCTION PBMAIN () AS LONG
LOCAL dwIndex           AS LONG
LOCAL dwCount           AS LONG
LOCAL dwTrailer         AS LONG
LOCAL sOlsonTimeZone    AS STRING
LOCAL arOlson()         AS STRING
LOCAL iFile             AS INTEGER
LOCAL oFile             AS INTEGER
LOCAL quFileSize        AS QUAD
    iFile = FREEFILE
    oFile = FREEFILE
    OPEN "northamerica" FOR BINARY AS #iFile
    OPEN "olson.txt" FOR OUTPUT AS #oFile
    quFileSize = LOF(#iFile)
    sOlsonTimeZone = SPACE$(quFileSize)
 
    GET #iFile, 1, sOlsonTimeZone
    CLOSE #iFile
 
    dwCount = PARSECOUNT(sOlsonTimeZone, $LF)
    DIM arOlson(1 TO dwCount) AS STRING
    PARSE sOlsonTimeZone, arOlson(), $LF
 
    FOR dwIndex = 1 TO dwCount
 
        sOlsonTimeZone = arOlson(dwIndex)
 
' Skip comment lines (start with #)
        IF ASC(sOlsonTimeZone,1) <> 35 THEN
 
' Skip blank lines
 
            IF LEN(sOlsonTimeZone) > 0 THEN
 
' Replace any tabs or carriage returns (not normally found) with spaces
                REPLACE ANY $TAB WITH $SPC IN sOlsonTimeZone
                REPLACE ANY $CR WITH $SPC IN sOlsonTimeZone
 
' Replace consecutive spaces with single space
                DO
                    IF INSTR(sOlsonTimeZone,"  ") THEN
                        REPLACE "  " WITH $SPC IN sOlsonTimeZone
                    ELSE
                        EXIT DO
                    END IF
                LOOP
 
' Remove leading spaces
                sOlsonTimeZone = LTRIM$(sOlsonTimeZone)
 
' Check if there are any trailing comments on the line and remove them if found
                dwTrailer = INSTR(sOlsonTimeZone,"#")
                IF dwTrailer > 0 THEN
                    sOlsonTimeZone = LEFT$(sOlsonTimeZone,dwTrailer - 1)
                END IF
' Remove trailing spaces
                sOlsonTimeZone = RTRIM$(sOlsonTimeZone)
 
' At this point, can do a parse on $SPC and the first array entry will
' direct any further action
 
                WRITE #oFile, sOlsonTimeZone
            END IF
        END IF
 
    NEXT
 
    CLOSE #oFile
END FUNCTION

I examined the output with Windows Notepad and initial inspection of the results seemed correct. The files I downloaded use linefeeds and not CRLF tags to separate records.

I'm thinking of emitting a series of DATA statements to be used as an include file and the DLL can load up some internal arrays to service the supporting timezone functions.

Rick