How to compare character ignoring case in primitive types

The Character class of Java API has various functions you can use.

You can convert your char to lowercase at both sides:

Character.toLowerCase(name1.charAt(i)) == Character.toLowerCase(name2.charAt(j))

There are also a methods you can use to verify if the letter is uppercase or lowercase:

Character.isUpperCase('P')
Character.isLowerCase('P') 

You can't actually do the job quite right with toLowerCase, either on a string or in a character. The problem is that there are variant glyphs in either upper or lower case, and depending on whether you uppercase or lowercase your glyphs may or may not be preserved. It's not even clear what you mean when you say that two variants of a lower-case glyph are compared ignoring case: are they or are they not the same? (Note that there are also mixed-case glyphs: \u01c5, \u01c8, \u01cb, \u01f2 or Dž, Lj, Nj, Dz, but any method suggested here will work on those as long as they should count as the same as their fully upper or full lower case variants.)

There is an additional problem with using Char: there are some 80 code points not representable with a single Char that are upper/lower case variants (40 of each), at least as detected by Java's code point upper/lower casing. You therefore need to get the code points and change the case on these.

But code points don't help with the variant glyphs.

Anyway, here's a complete list of the glyphs that are problematic due to variants, showing how they fare against 6 variant methods:

1. Character toLowerCase
2. Character toUpperCase
3. String toLowerCase
4. String toUpperCase
5. String equalsIgnoreCase
6. Character toLowerCase(toUpperCase) (or vice versa)

For these methods, S means that the variants are treated the same as each other, D means the variants are treated as different from each other.

Behavior     Unicode                             Glyphs
===========  ==================================  =========
1 2 3 4 5 6  Upper  Lower  Var Up Var Lo Vr Lo2  U L u l l2
- - - - - -  ------ ------ ------ ------ ------  - - - - -
D D D D S S  \u0049 \u0069 \u0130 \u0131         I i İ ı   
S D S D S S  \u004b \u006b \u212a                K k K     
D S D S S S  \u0053 \u0073        \u017f         S s   ſ   
D S D S S S  \u039c \u03bc        \u00b5         Μ μ   µ   
S D S D S S  \u00c5 \u00e5 \u212b                Å å Å     
D S D S S S  \u0399 \u03b9        \u0345 \u1fbe  Ι ι   ͅ ι 
D S D S S S  \u0392 \u03b2        \u03d0         Β β   ϐ   
D S D S S S  \u0395 \u03b5        \u03f5         Ε ε   ϵ   
D D D D S S  \u0398 \u03b8 \u03f4 \u03d1         Θ θ ϴ ϑ   
D S D S S S  \u039a \u03ba        \u03f0         Κ κ   ϰ   
D S D S S S  \u03a0 \u03c0        \u03d6         Π π   ϖ   
D S D S S S  \u03a1 \u03c1        \u03f1         Ρ ρ   ϱ   
D S D S S S  \u03a3 \u03c3        \u03c2         Σ σ   ς   
D S D S S S  \u03a6 \u03c6        \u03d5         Φ φ   ϕ   
S D S D S S  \u03a9 \u03c9 \u2126                Ω ω Ω     
D S D S S S  \u1e60 \u1e61        \u1e9b         Ṡ ṡ   ẛ   

Complicating this still further is that there is no way to get the Turkish I's right (i.e. the dotted versions are different than the undotted versions) unless you know you're in Turkish; none of these methods give correct behavior and cannot unless you know the locale (i.e. non-Turkish: i and I are the same ignoring case; Turkish, not).

Overall, using toUpperCase gives you the closest approximation, since you have only five uppercase variants (or four, not counting Turkish).

You can also try to specifically intercept those five troublesome cases and call toUpperCase(toLowerCase(c)) on them alone. If you choose your guards carefully (just toUpperCase if c < 0x130 || c > 0x212B, then work through the other alternatives) you can get only a ~20% speed penalty for characters in the low range (as compared to ~4x if you convert single characters to strings and equalsIgnoreCase them) and only about a 2x penalty if you have a lot in the danger zone. You still have the locale problem with dotted I, but otherwise you're in decent shape. Of course if you can use equalsIgnoreCase on a larger string, you're better off doing that.

Here is sample Scala code that does the job:

def elevateCase(c: Char): Char = {
  if (c < 0x130 || c > 0x212B) Character.toUpperCase(c)
  else if (c == 0x130 || c == 0x3F4 || c == 0x2126 || c >= 0x212A)
    Character.toUpperCase(Character.toLowerCase(c))
  else Character.toUpperCase(c)
}