Osclass Support Forums

General osclass questions => Report bug => Topic started by: dsf on March 20, 2023, 07:29:17 PM

Title: GEO for Bangladesh
Post by: dsf on March 20, 2023, 07:29:17 PM
I added Bangladesh geo to a multi-country Osclass.

The Cities show many duplicates, same with accents, same without.

Thanks
Title: Re: GEO for Bangladesh
Post by: dsf on March 20, 2023, 07:49:15 PM
Some cities are even 3 times:

Title: Re: GEO for Bangladesh
Post by: MB Themes on March 20, 2023, 11:39:17 PM
That is problem.
Need to check if there is reliable function in osclass that can compare strings ignoring accent
Title: Re: GEO for Bangladesh
Post by: dsf on March 21, 2023, 08:24:38 AM
Not accent issue only.

For example check DUBAI geo.

Ḩiyāwah

Murbaḩ

Check how it is handled at

https://www.dubizzle.com/
Title: Re: GEO for Bangladesh
Post by: Wiz on March 21, 2023, 03:40:48 PM
I don't think this is an osclass issue as the quality of geo DBs leave little to desire and are difficult to work with for countries other than US/EU ... I ended up using gmaps paid API to populate correct names of cities, regions, locations, etc. in multi-languages including correct coordinates.

Blame the data source/provider.
Title: Re: GEO for Bangladesh
Post by: dsf on March 21, 2023, 03:52:43 PM
Quote from: Wiz on March 21, 2023, 03:40:48 PM
I don't think this is an osclass issue as the quality of geo DBs leave little to desire and are difficult to work with for countries other than US/EU ... I ended up using gmaps paid API to populate correct names of cities, regions, locations, etc. in multi-languages including correct coordinates.

Blame the data source/provider.

I don't blame anyone. Just reporting something that can easily be fixed.

Otherwise you end up having an unsuable dropdown or autofill option for cities and also a huge database with dups.
Title: Re: GEO for Bangladesh
Post by: Wiz on March 21, 2023, 04:25:53 PM
I wasn’t saying you were, just stating a fact. A quick search on Google shows many hits for developers dumping geonames sources as they’re inaccurate, use accents, riddled with duplicates and spelling errors, wrong coordinates and the list goes on and on.

Best to build from scratch using mapbox, gmaps or go commercial with a ready-made DB.

Open-source while great is usually not on par with commercial solutions.

Title: Re: GEO for Bangladesh
Post by: MB Themes on March 22, 2023, 08:43:10 PM
Quote from: dsf on March 20, 2023, 07:49:15 PM
Some cities are even 3 times:

Just updated script and regenerated Bangladesh, around 2k cities removed.

Original (some examples):
Code: [Select]
(4213277, 4337210, 'BD', 'Adābāria', 1, 'adabaria', 23.99315, 88.80688),
(4213278, 4337210, 'BD', 'Ādabāria', 1, 'adabaria', 23.8028, 89.16724),
(4213279, 4337210, 'BD', 'Ādābāria', 1, 'adabaria', 22.3906, 90.45925),
(4213280, 4337229, 'BD', 'Adabaria', 1, 'adabaria', 22.21691, 90.19591),
(4213263, 4337210, 'BD', 'Adampur', 1, 'adampur', 22.38265, 90.40251),
(4213264, 14287936, 'BD', 'Ādampur', 1, 'adampur', 24.9967, 90.15321),
(4213266, 4337179, 'BD', 'Ādampur', 1, 'adampur', 23.62119, 89.83618),
(4213268, 4337200, 'BD', 'Ādampur', 1, 'adampur', 23.43333, 91.05),
(4213271, 4337179, 'BD', 'Adampur', 1, 'adampur', 23.29666, 89.87274),

After refresh:
Code: [Select]
(4213277, 4337210, 'BD', 'Adābāria', 1, 'adabaria', 23.99315, 88.80688),
(4213280, 4337229, 'BD', 'Adabaria', 1, 'adabaria', 22.21691, 90.19591),
(4213266, 4337179, 'BD', 'Ādampur', 1, 'adampur', 23.62119, 89.83618),
(4213268, 4337200, 'BD', 'Ādampur', 1, 'adampur', 23.43333, 91.05),

Notice that cities looks like identical/duplicated, but they are actually in different regions thats visible from coordinates as well.
Title: Re: GEO for Bangladesh
Post by: dsf on March 22, 2023, 09:19:08 PM
It is not good to includ accents of any kind.

See here:

https://www.locanto.com.bd/

We have -> Barisāl

But should be -> Barisal

Title: Re: GEO for Bangladesh
Post by: MB Themes on March 22, 2023, 09:43:10 PM
We cannot decide which form is correct and will use first available.
Title: Re: GEO for Bangladesh
Post by: dsf on March 22, 2023, 10:27:18 PM
Quote from: MB Themes on March 22, 2023, 09:43:10 PM
We cannot decide which form is correct and will use first available.

ok, but why REinvent the wheel? Why not follow what others do?

Can you write with your keyboard the area "Barisāl"? Because i can't. I don't have the "ā". And most people do not.

And that is not only for Bangladesh, UAE also for example and who knows what else.
Title: Re: GEO for Bangladesh
Post by: dsf on March 22, 2023, 10:31:15 PM
Ok, if anyone wants to remove the "ā" or whatever, he can download the GEO SQL, load it in some editor and do a search and replace.

Simple.
Title: Re: GEO for Bangladesh
Post by: MB Themes on March 23, 2023, 06:01:01 PM
In slovakia we have áéíóúä those you may not have on keyboard, but everyone here has it.
Is it correct to write "Banovce" instead of "Bánovce"?
It is not, its not gramatically correct...
Title: Re: GEO for Bangladesh
Post by: dsf on March 23, 2023, 06:12:52 PM
I see. I don't know what could be the right thing to do.

Because if I want to come to Bánovce or buy something locally, how do i write it? I can't.

How do Slovakian classified sites handle this?
Title: Re: GEO for Bangladesh
Post by: MB Themes on March 23, 2023, 08:54:36 PM
People on Slovakia use SK keyboard that has these characters.