Displaying Arabic city names in AGSTextSymbol

[ChrisJenkins1]

Hi,

I am having difficulty getting Arabic city names to display in AGSTextSymbol using ArcGIS 10.2, the names appears with each character as a square box, is there any reason why this shouldn't work? I have checked that my city names are correct and they can be displayed in a UITableView in Arabic without difficulty, but not on the map.

[ChrisJenkins1]

I have attached a minimal test project which clearly demonstrates this problem. I also stumbled upon a solution, if the fontFamily of the AGSTextSymbol is set to "Arial" the text renders, if fontFamily is left unset it does not. So it would appear that the font AGSTextSymbol uses by default (presumably the system font) is not usable for Arabic. Could I please have some confirmation that this is the case and advice as to which font family I should be using to ensure the text renders in all languages?

[ChrisJenkins1]

I have also noticed that Japanese characters do not get rendered either. Choosing Arial as the font family does not work in this case, I was forced to pick "Seiti SC" instead in order to get it working. This basically means it is not possible to have mixed Japanese/Roman labels whilst retaining any control over what font is used for the roman characters (roman characters displayed using Seiti SC don't look like the system font). It also means I must know what characters are in the labels before assigning the appropriate font family as otherwise non-roman characters won't render and roman characters will look odd. Previous versions of the SDK (v2.x) did not have these limitations.

[DiveshGoyal]

Thanks for reporting this. We'll look into it.

[ChrisJenkins1]

Thanks Divesh, have you had any luck reproducing the effect?

[ChrisJenkins1]

Hi, it's been over a month since I reported this issue, is anyone at ESRI still looking into it? is there a solution?

[DiveshGoyal]

Chris,

We confirm that this is indeed a problem, and we suspect this is happening because of lack of font substitution / fallback in the SDK. Most fonts do not have a complete set of unicode code points and so UI frameworks, web browsers etc substitute an alternative font that has the missing glyphs. I cannot offer any workarounds at the moment but we will continue to investigate and hopefully fix it soon.

[ChrisJenkins1]

FYI: this also happens for Chinese, so I suspect every non-Roman character set will suffer the same way. Interesting that using Apple's UI controls, or indeed any Core Graphics text rendering, does not encounter this problem, it seems possible to request the system font and render English, Arabic, Chinese and Japanese without any extra effort.

Anyway, thanks for looking into it, I do now have a situation where I need to display mixed language labels without being able to know which labels require which character set, so a fix is slightly more urgent than before, if that makes any difference!

[DiveshGoyal]

Thanks for the confirmation regarding Chinese. Yes, this issue probably affects more languages.

I encourage you to contact Esri Support and report this issue to them along with its impact on your project and timelines. They'll help ensure this gets resolved in a timely manner.

[ChrisJenkins1]

Divesh, has this issue been addressed in the recent 10.2.1 update?