Regular expression for persian(arabic) letters without any numbers

Question

In Java i'm looking for a regular expression that accepts any Persian( or Arabic ) letters except any Persian ( or Arabic) numbers. In order to have only letters i found a very good regular expression:

[\u0600-\u065F\u066A-\u06EF\u06FA-\u06FF]

although it is true and works for me, But we know that we can use the \\p{L}+ as a regular expression which accepts all letters from all languages in the world, and in my case ( Arabic - Persian ) i can modified it and use [\\p{InArabic}]+$.

But by using [\\p{InArabic}]+$ not only all Arabic(Persian) letters are going to be accepted but also Arabic numbers are acceptable too, like ۱ ۲.

So my question is how can i modify [\\p{InArabic}]+$ to just accept letters not numbers, or in other word how can i restrict [\\p{InArabic}]+$ to not accept any numbers?

Please Notice that the Persian(Arabic) numbers are like these: ۱ ۲ ۳ ۴ ۵ ۶ ۷ ۸ ۹ ۰

score 22 · Accepted Answer · answered May 08 '15 at 05:46

22

You can use the following regex:

"[\\p{InArabic}&&\\PN]"

\p{InArabic} matches any character in Unicode Block Arabic (from U+0600 to U+06FF)

\PN matches any character not belonging to any of the Number category (note the capital P).

Intersecting the 2 sets give the desired result: both digit ranges (U+0660 to U+0669) and (U+06F0 to U+06F9) are excluded.

Testing code

for (int i = 0x600; i <= 0x6ff; i++) {
    String c = "" + (char) i;
    System.out.println(Integer.toString(i, 16) + " " + c.matches("[\\p{InArabic}&&\\PN]"));
}

answered May 08 '15 at 05:46

nhahtdh

55,989
15
126
162

are you sure it should be like this? `[\\p{InArabic}&&\\PN] ` because my IDE shows me a red line behind of \\PN :( It says : character category expected. – Elyas Hadizadeh May 08 '15 at 05:51
@ElyasHadizadeh: I have tested the regex on my machine (Java 7 and Java 8) before posting. Note that I'm specifying the regex inside string literal. Remove one ``\`` if you need the raw form. – nhahtdh May 08 '15 at 05:52
yes you are true and it works as a charm, but it's very strange that my IDE draws a red line behind of \\PN. ( I am using IntelliJ IDEA 13.0.4, with Java 7 and Java 8, in both cases it shows red line behind of \\PN, but when i compile and run the application it works correctly. – Elyas Hadizadeh May 08 '15 at 06:07
Thank you for you answer, What is your IDE? do you know why do i get such a bug? – Elyas Hadizadeh May 08 '15 at 06:07
@ElyasHadizadeh: I use Eclipse, but Eclipse doesn't do any validity detection on regex. The most it does is check syntax of string literal. – nhahtdh May 08 '15 at 06:13
aha ! that's the reason, because the IntelliJ do validity detection on regex, and it seems that `"[\\p{InArabic}&&\\PN]"` is very strange for it. but you know it just shows a red line under it. in the compiling and running it doesn't have any problem. any way thank you buddy – Elyas Hadizadeh May 08 '15 at 06:18

score 7 · Answer 2 · edited May 23 '17 at 12:02

7

You can use character class subtraction, which is a rather obscure feature:

[\p{InArabic}&&[^۰-۹]]

Working example: http://ideone.com/jChGem

edited May 23 '17 at 12:02

Community

1
1

answered May 08 '15 at 05:02

Kobi

135,331
41
252
292

Hey buddy, thank you, unfortunately your regular expression is just acceptable in .Net technologies and i am looking for a Java Programming Language. – Elyas Hadizadeh May 08 '15 at 05:05
unfortunately your edited answer does not work, you know? i think Java does not support [^۰-۹], i don't know :-?? – Elyas Hadizadeh May 08 '15 at 05:11
1

@ElyasHadizadeh - Weird... Seems to be working for me: http://ideone.com/jChGem . Can you please post your code in the question? – Kobi May 08 '15 at 05:28
You need to also exclude the range `[٠-٩]`, current regex only excludes the code points from U+06F0 to U+06F9 – nhahtdh May 08 '15 at 05:36
@nhahtdh - Isn't that exactly the range I have? What am I missing? – Kobi May 08 '15 at 05:47
The range I suggested `[٠-٩]` is the range U+0660 to U+0669. It is missing in your answer. – nhahtdh May 08 '15 at 05:49
@nhahtdh - Oh, got it. lookalikes. – Kobi May 08 '15 at 05:51

Regular expression for persian(arabic) letters without any numbers

2 Answers2

Testing code