In the AE (United Arab Emirates) data, the sub-key for Sharjah carries an ARABIC SHADDA (U+0651) on a letter that cannot take one.
Fetched from the live endpoint just now:
https://chromium-i18n.appspot.com/ssl-aggregate-address/data/AE
-> https://www.gstatic.com/chrome/autofill/libaddressinput/chromium-i18n/ssl-aggregate-address/data/AE
"sub_keys": "أبو ظبي~إمارة الشارقةّ~الفجيرة~ام القيوين~إمارة دبيّ~إمارة رأس الخيمة~عجمان",
"sub_names": "أبو ظبي~الشارقة~الفجيرة~ام القيوين~دبي~رأس الخيمة~عجمان",
The problem
The Sharjah key ends ... U+0642 U+0629 U+0651, i.e. a shadda placed after the teh marbuta ة:
key إمارة الشارقةّ U+0625 U+0645 U+0627 U+0631 U+0629 U+0020
U+0627 U+0644 U+0634 U+0627 U+0631 U+0642 U+0629 U+0651
name الشارقة U+0627 U+0644 U+0634 U+0627 U+0631 U+0642 U+0629
A shadda marks gemination of a consonant. Teh marbuta is a word-final letter representing the feminine ending, and it is not geminated in this name — it cannot carry a shadda. In الشارقة the gemination is on the sheen ش, by sun-letter assimilation after the definite article ال. If the name is written with the shadda at all, the correct position is:
الشّارقة U+0627 U+0644 U+0634 U+0651 U+0627 U+0631 U+0642 U+0629
^ shadda on the sheen
This is also what the standard romanization "Ash Shāriqah" reflects — the doubled Sh.
Why I think it is a slip rather than a convention
The same file spells Dubai's shadda correctly: إمارة دبيّ, where the final yaa genuinely is geminated (دُبَيّ). So the data can and does place a shadda correctly one field away. Sharjah is the odd one out, not the result of a consistent policy.
User-visible effect
name is the clean الشارقة while key carries the mark, and in region_data_builder.cc the sub-region is added as AddSubRegion(key, name). So a dropdown displays الشارقة while the value that is stored and printed on the label is إمارة الشارقةّ. The user cannot see the anomaly in order to correct it.
It also means the natural spelling does not validate — an address submitting الشارقة or إمارة الشارقة does not match the key.
Two smaller things in the same record
- The
إمارة ("Emirate of") prefix is applied to only 3 of the 7 keys — Sharjah, Dubai and Ras al Khaimah have it; Abu Dhabi, Fujairah, Umm Al Quwain and Ajman do not.
ام القيوين is missing the hamza; the usual spelling is أم القيوين.
Note on provenance
I am fairly sure this form is inherited from the ISO 3166-2 table rather than authored here, so this is a report that the upstream source has it wrong, not a claim that it originated in this project. I mention it because #244 says reporting data issues here is the right route for server-side updates.
I have not verified whether the other RTL-script countries have the same class of issue; I only checked that across the aggregate data the AE entries were the ones where a key and its own name differ by a combining mark.
In the AE (United Arab Emirates) data, the sub-key for Sharjah carries an ARABIC SHADDA (U+0651) on a letter that cannot take one.
Fetched from the live endpoint just now:
The problem
The Sharjah key ends
... U+0642 U+0629 U+0651, i.e. a shadda placed after the teh marbuta ة:A shadda marks gemination of a consonant. Teh marbuta is a word-final letter representing the feminine ending, and it is not geminated in this name — it cannot carry a shadda. In
الشارقةthe gemination is on the sheen ش, by sun-letter assimilation after the definite articleال. If the name is written with the shadda at all, the correct position is:This is also what the standard romanization "Ash Shāriqah" reflects — the doubled Sh.
Why I think it is a slip rather than a convention
The same file spells Dubai's shadda correctly:
إمارة دبيّ, where the final yaa genuinely is geminated (دُبَيّ). So the data can and does place a shadda correctly one field away. Sharjah is the odd one out, not the result of a consistent policy.User-visible effect
nameis the cleanالشارقةwhilekeycarries the mark, and inregion_data_builder.ccthe sub-region is added asAddSubRegion(key, name). So a dropdown displaysالشارقةwhile the value that is stored and printed on the label isإمارة الشارقةّ. The user cannot see the anomaly in order to correct it.It also means the natural spelling does not validate — an address submitting
الشارقةorإمارة الشارقةdoes not match the key.Two smaller things in the same record
إمارة("Emirate of") prefix is applied to only 3 of the 7 keys — Sharjah, Dubai and Ras al Khaimah have it; Abu Dhabi, Fujairah, Umm Al Quwain and Ajman do not.ام القيوينis missing the hamza; the usual spelling isأم القيوين.Note on provenance
I am fairly sure this form is inherited from the ISO 3166-2 table rather than authored here, so this is a report that the upstream source has it wrong, not a claim that it originated in this project. I mention it because #244 says reporting data issues here is the right route for server-side updates.
I have not verified whether the other RTL-script countries have the same class of issue; I only checked that across the aggregate data the AE entries were the ones where a key and its own name differ by a combining mark.