What makes a good test string for testing web forms for unicode compatibility?

Question

What test text do you try and type into your web forms to check that they handle all the edge cases properly (especially Unicode and XSS style problems).

I am particularly interested in good Unicode strings that may do something odd if they are mis-encoded when they are displayed again.

Text that contains potentially problematic characters, like quotes, <, > etc would also be interesting.

Look at http://stackoverflow.com/questions/1319022/really-good-bad-utf-8-example-test-data — The Disintegrator, Aug 27 '09 at 19:17
That has some useful "bad" character sequences. I was also after valid sequences that turn into something obvious when they are mishandled. — Rik Heywood, Aug 27 '09 at 19:24
oh, that's easy á Á ñ ç those ones look horribly bad when mishanded — The Disintegrator, Aug 28 '09 at 05:13

score 22 · Accepted Answer · answered Aug 27 '09 at 19:58

22

Your idea of HTML-sensitive characters is a good start. I also like using characters that are kind of readable, but are still Unicode. When I was doing this kind of testing for tabblo.com, I used this string:

Testing «ταБЬℓσ»: 1<2 & 4+1>3, now 20% off!

This has HTML-sensitive characters, ASCII, upper-half ISO characters, and multi-byte Unicode characters.

answered Aug 27 '09 at 19:58

Ned Batchelder

364,293
75
561
662

Not bad UTF-8 data though, which would be useful. – Alix Axel Aug 18 '10 at 21:10

score 11 · Answer 2 · answered Aug 27 '09 at 20:05

11

Turkey testing!

http://www.moserware.com/2008/02/does-your-code-pass-turkey-test.html

This is actually pretty advanced internationalization testing, not for the faint of heart, including date formatting, percent calculations, upper/lowercase translations, etc.

answered Aug 27 '09 at 20:05

willoller

7,106
1
35
63

Not really good test. Would fail in Poland. For example these are valid dates in Poland: 31.12.2018, 31/12/2018, 2018-12-31, 31.12. And these are valid numbers: 1234,56 and 1 234,56 and 1.234,56 – Tom Nov 13 '18 at 20:10
The point is that if i18n is fully implemented for Turkey, it will work many places worldwide, including Poland, since dates, money, and decimals are all handled using i18nionalized resources/libs/variables/strings. You would just set the region to Poland. – willoller Nov 13 '18 at 20:27

score 9 · Answer 3 · edited Mar 20 '17 at 10:18

9

These smilies from SuperUser.com are pretty cool for testing your unicode support as well...

https://superuser.com/questions/52671/how-do-i-create-unicode-smilies-like

٩(-̮̮̃-̃)۶ ٩(●̮̮̃•̃)۶ ٩(͡๏̯͡๏)۶ ٩(-̮̮̃•̃).

edited Mar 20 '17 at 10:18

Community

1
1

answered Oct 29 '09 at 14:51

Rik Heywood

13,816
9
61
81

score 2 · Answer 4 · answered Aug 27 '09 at 19:43

2

Well, this is a bit of a brute force approach, but if you wanted to start from some well formed Unicode and add some errors, a great resources for the real stuff is here: http://www.unicode.org/charts.

answered Aug 27 '09 at 19:43

John Lockwood

3,787
29
27

What makes a good test string for testing web forms for unicode compatibility?

4 Answers4

٩(-̮̮̃-̃)۶ ٩(●̮̮̃•̃)۶ ٩(͡๏̯͡๏)۶ ٩(-̮̮̃•̃).