Skip to content

LetterQ

Status: Stable

documented, exercised by the test suite and/or worked examples, with no known limitations recorded.

Description

LetterQ["string"]

Gives True if every character of "string" is a letter (any script), and False otherwise. LetterQ[""] is True.

Examples (14)

Every input below was run against the current Mathilda build and its output recorded.

Basic examples (4)

In[1]:= LetterQ["αβγ"]
Out[1]= True

In[2]:= DigitQ["\:0662"]
Out[2]= False

In[3]:= UpperCaseQ["ABC"]
Out[3]= True

In[4]:= LowerCaseQ["abc1"]
Out[4]= False

Applications (10)

In[5]:= LetterQ["abc"]
Out[5]= True

One non-letter makes it False

In[6]:= LetterQ["abc1"]
Out[6]= False

Letters of any script count

In[7]:= LetterQ["αβγ日本語"]
Out[7]= True

Unicode category Lo: the feminine ordinal is a letter

In[8]:= LetterQ["ª"]
Out[8]= True

A Roman-numeral letterlike symbol is Nl, not a letter

In[9]:= LetterQ["Ⅻ"]
Out[9]= False

A combining accent is a separate, non-letter character

In[10]:= LetterQ["e\:0301"]
Out[10]= False

Vacuously True

In[11]:= LetterQ[""]
Out[11]= True

In[12]:= Select[{"cat", "r2d2", "naïve", "hello world"}, LetterQ]
Out[12]= {"cat", "naïve"}

Inside a string pattern it becomes a character class

In[13]:= StringCases["a1B2ç3", _?LetterQ]
Out[13]= {"a", "B", "ç"}

Not Listable: a list is not a string

In[14]:= LetterQ[{"a", "b"}]
Out[14]= LetterQ[{"a", "b"}]

Implementation notes

Algorithm. builtin_letterq calls charq_builtin(res, "LetterQ", CQ_LETTER), which requires one String argument (otherwise LetterQ::string via strings_arg_message, unevaluated), then decodes the UTF-8 code points one at a time with u8_decode and returns False at the first one for which char_ok(CQ_LETTER, cp) fails — so the scan short-circuits — and True if none does (vacuously for ""). char_ok answers ASCII code points inline (A–Z, a–z) and otherwise binary-searches k_letter_ranges with in_ranges. Inside a string pattern, predicate_class in string_pattern.c translates _?LetterQ to the PCRE class [[:alpha:]] instead of emitting a per-character callout.

Data structures. k_letter_ranges: sorted, disjoint [lo, hi] code-point ranges covering Unicode general categories L* (Lu, Ll, Lt, Lm, Lo), generated from Unicode 16.0.0 by tools/gen_unicode_tables.py into src/strings/unicode_tables.inc.

Complexity / limits. O(L) for an ASCII string; otherwise O(L log R) over the R letter ranges. Combining marks (Mn/Mc) and letter-like numbers (Nl) are not letters. Characters are code points, not grapheme clusters. Not Listable; no packed-array or Compile[] surface (the argument is a String).

Attributes: Protected.

References

See also: DigitQ, UpperCaseQ, LowerCaseQ

Notes & additional examples

Notes

  • True when every character (code point) belongs to a Unicode letter category L* (Lu, Ll, Lt, Lm, Lo), using Unicode 16.0 ranges; ASCII is tested inline.
  • Characters are code points, not grapheme clusters, so a decomposed e + U+0301 is not all letters.
  • A non-string argument emits LetterQ::string and stays unevaluated; it is not Listable (use Map or AllTrue).
  • In a string pattern _?LetterQ is translated directly to a PCRE character class rather than a per-character callout.
  • See also: LetterCharacter, DigitQ, UpperCaseQ, LowerCaseQ.