How to determine a string is English or Persian?

Question

I have edittext in a form, I want that when the user inputs text into the edittext for my program to detect which language was inserted into the edittext.

Is there a way to determine a string is English or Persian?

I found this code for Arabic

public static boolean isProbablyArabic(String s) {
for (int i = 0; i < Character.codePointCount(s, 0, s.length()); i++) {
    int c = s.codePointAt(i);
    if (c >= 0x0600 && c <=0x06E0)
        return true;
}
return false;
}

but how can I change this code for Persian?

The character code range is probably robust enough for this simple either/or question. For more advanced approaches, look at some language-identification libraries. — tripleee, Apr 13 '14 at 08:02
@tripleee but I use persian or english language,arabic has some difference from persian — Saeed Hashemi, Apr 13 '14 at 08:12
If you expect Farsi input which does not use the Arabic script, you will need to update your question with more details. — tripleee, Apr 13 '14 at 08:18

score 6 · Accepted Answer · edited Jun 20 '20 at 09:12

6

All possible Unicode ranges for Persian (also for Urdu) alphabet:

0x0600 to 0x06FF
0xFB50 to 0xFDFF
0xFE70 to 0xFEFF

So if you want don't miss any char check all ranges. Hope helps you.

edited Jun 20 '20 at 09:12

Community

1
1

answered Jan 12 '16 at 10:03

afruzan

1,454
19
20

score 2 · Answer 2 · answered May 06 '16 at 20:23

You can know a string is english or persian by using Regex.

public static final Pattern VALID_NAME_PATTERN_REGEX = Pattern.compile("[a-zA-Z_0-9]+$");

public static boolean isEnglishWord(String string) {
    return VALID_NAME_PATTERN_REGEX.matcher(string).find();
}

this only works with words and numbers. if there is a character like '=' or '+' , the function would return false . you can fix that by editing the regex to match what you need .

score 1 · Answer 3 · answered Apr 13 '14 at 08:06

1

Why don't you evaluate it when keyboard is popup.. Means You can do it by getting the language of phone... here is the method useLocale.getDefault().getDisplayLanguage(); minSDK is 11 is required.

answered Apr 13 '14 at 08:06

Bhaskar

911
6
18

but maybe user copy and paste text ! – Saeed Hashemi Apr 13 '14 at 08:08
and program minSDK is 8 – Saeed Hashemi Apr 13 '14 at 08:09
if your app is online... the try to use google translation API – Bhaskar Apr 13 '14 at 08:10
You can Change those unicode points 0x0600... As per the standard unicode set url below.... http://www.utf8-chartable.de/unicode-utf8-table.pl http://www.tamasoft.co.jp/en/general-info/unicode.html – Bhaskar Apr 13 '14 at 08:19

score 1 · Answer 4 · answered Jan 12 '16 at 10:44

There's no exact way to determine what language your user is typing in unless you get really complicated, hence why the method example you've given is called isProbablyArabic rather than isArabic. If your users are writing exclusively in English or Farsi and nothing else, one option is to use a regex that looks to see if the user's text contains Western Roman characters ("^[a-zA-Z]*$"), if this returns false you can assume they've typed in Persian, though it could be anything that uses a different character set.

score 0 · Answer 5 · answered Apr 13 '14 at 08:20

Using characters' range is not a perfect way to detect some languages that have overlapped range e.g Arabic, Persian and Urdu. But, if you insist on this way, my suggestion is looking for especial characters that are language-specific. For example, گ or پ are in Persian but are not in Arabic. On the other hand, ئ or ة maybe more common in Arabic text than Persian. By counting these specific characters you can distinguish between Arabic, Persian and Urdu.

Although I've got good results from the mentioned method, using n-grams to detect a language is more popular and dependable. There are many libraries that do language detection task by this method.

How to determine a string is English or Persian?

5 Answers5

Linked