Regular Expressions: Basic Characters and Character Classes

This lesson begins with how a pattern matches one character. Unless noted otherwise, a ‘regex’ code block contains the pattern itself, without JavaScript ’/…/’ literal delimiters. Runnable exercises use Python ‘re’ as the baseline; Unicode and ASCII differences are called out when character classes are involved.

1. Ordinary characters

Letters, digits, and most unreserved punctuation are ordinary regex characters. They match themselves and normally distinguish case. Adjacent ordinary characters match that text in order.

cat

Test text: ‘a cat and a catalog’

The pattern matches ‘cat’ twice: the independent word and the start of ‘catalog’. To match only a whole word, combine it with word boundaries.

2. Metacharacters

Metacharacters have special meaning. Common ones include ’.’, ’^’, ’$’, ’*’, ’+’, ’?’, ’(’, ’)’, ’[’, ’]’, ’{’, ’}’, ’|’, and backslash. They express rules such as any character, boundary, repetition, grouping, and alternation rather than matching literal text.

ca.t

Test text: ‘cat cart cast’

The period matches one arbitrary character by default, normally excluding newlines. Therefore ‘ca.t’ matches ‘cart’ and ‘cast’, but not the three-character string ‘cat’. To match a literal period, use ’.‘.

3. Escaping

A backslash escapes a metacharacter or introduces a special class or boundary such as ‘\d’ or ‘\b’. When a pattern is inside a programming-language string, it can also pass through a second string-escaping layer.

file\.txt

Test text: ‘file.txt fileXtxt’

’.’ matches the literal period, so only ‘file.txt’ matches. In Python write ‘re.compile(r”file.txt”)’; in PowerShell, prefer a single-quoted pattern such as ”file.txt”; in JavaScript, write ‘/file.txt/’. If you use an ordinary program string, inspect the host-language escaping layer too.

4. Character set: ‘[abc]’

A character set, also called a character class, says “match any one of these characters.” ‘[abc]’ matches one character at a time.

gr[ae]y

Test text: ‘gray grey gruy’

‘gr[ae]y’ matches ‘gray’ and ‘grey’, because that position permits ‘a’ or ‘e’; it does not match ‘gruy’. Most metacharacters lose their special role inside brackets, but ’]’, ’-’, ’^’, and backslash still need care.

5. Character range: ‘[a-z]’

Inside a set, ’-’ can express a range. ‘[a-z]’ represents any lower-case ASCII letter from ‘a’ through ‘z’; ranges can combine, as in ‘[A-Za-z0-9]’.

[A-Z][a-z]+

Test text: ‘Alice BOB alice Zhang’

It matches ‘Alice’ and ‘Zhang’: the first character must be upper-case and at least one following character must be lower-case. ‘BOB’ lacks following lower-case letters, and ‘alice’ does not begin upper-case. Ranges depend on engine and encoding; do not assume ‘[a-z]’ covers every letter in internationalized text.

6. Negated character set: ’[^0-9]’

When ’^’ is the first character in a set, it means “none of these characters.” ’[^0-9]’ matches one non-ASCII-digit character. Its meaning here differs from the start-of-line anchor because position determines the role.

[^0-9]+

Test text: ‘A12-B_3’

It matches two nondigit runs: ‘A’ and ‘-B_’. The trailing ’+’ merges adjacent nondigits into one match; ‘1’, ‘2’, and ‘3’ are not matched.

7. Digit class: ‘\d’

‘\d’ is a digit class. In ECMAScript it means ‘[0-9]’; Python Unicode ‘str’ patterns and .NET defaults also accept other Unicode decimal digits. For machine protocols, timestamps, and identifiers that allow ASCII digits only, write ‘[0-9]’ explicitly or deliberately use ‘re.ASCII’ in Python.

\d{4}

Test text: ‘No. 2026, batch 73’

It matches ‘2026’, because ‘{4}’ requires four consecutive digits. ‘73’ is only two digits.

8. Nondigit class: ‘\D’

‘\D’ matches one nondigit character, the complement of ‘\d’. It can extract or exclude digits, but it does not automatically prove that remaining text is alphabetic.

\D+

Test text: ‘v2.0-beta’

The successive results are ‘v’, ’.’, and ‘-beta’. Digits ‘2’ and ‘0’ are skipped; ’+’ joins adjacent nondigits into one match.

9. Word class: ‘\w’

‘\w’ matches a “word character,” where “word” is an engine term rather than natural-language tokenization. Python Unicode ‘str’ patterns include Unicode alphanumerics and underscore by default; ECMAScript’s basic set is close to ‘[A-Za-z0-9_]’; .NET follows Unicode categories. For machine identifiers, write the permitted set rather than relying on default ‘\w’.

\w+

Test text: ‘user_name-01’ followed by two non-ASCII names

Under basic ECMAScript semantics it matches ‘user_name’ and ‘01’. Default Python also matches Unicode word characters. Hyphen and whitespace are not word characters. This is why a pattern cannot be discussed separately from its engine and flags.

10. Nonword class: ‘\W’

‘\W’ matches one nonword character, the complement of ‘\w’. It is useful for locating punctuation, separators, and whitespace, but its exact range changes with the definition of ‘\w’.

\W+

Test text: ‘name@example.com!’

The results are ’@’, ’.’, and ’!’. Letters belong to ‘\w’, so this pattern does not match them.

11. Whitespace class: ‘\s’

‘\s’ matches one whitespace character such as a space, tab, or newline; whether it includes more Unicode whitespace depends on the engine. It is useful for irregular spaces and line breaks.

\s+

Test text: ‘name agecity’

It returns two whitespace pieces: three consecutive spaces and one tab. ’+’ merges contiguous whitespace rather than returning one character at a time.

12. Nonwhitespace class: ‘\S’

‘\S’ matches one nonwhitespace character, the complement of ‘\s’. It extracts whitespace-separated nonblank chunks, not natural-language words.

\S+

Test text: ’ onetwo three ’

The results are ‘one’, ‘two’, and ‘three’. Leading, trailing, and intervening spaces or tabs are not matched.

13. Period: ’.’

The period normally matches one arbitrary character, but most engines do not let it match a newline by default. In dotAll mode it normally can match newlines as well. Use ’.’ for a literal period.

a.c

Test text: ‘abc a-c ac’

It matches ‘abc’ and ‘a-c’: the period matches their middle character. ‘ac’ has no middle character and does not match. Do not use ’.*’ as a substitute for every structured rule; it easily produces overmatching and poor performance.

Practice and self-check

1. Write a character class that allows only ASCII hexadecimal characters, then test ‘0fA9’, ‘0x10’, and the empty string.
2. In Python, match full-width digits with ‘r”\d+”’ and ‘r”[0-9]+”’ and explain the difference.
3. Write a pattern that matches the literal text ‘a+b.c’; identify the characters that need escaping.
4. Is ’[^0-9]’ equivalent to “a letter”? Why not?
5. Why can ’.’ in default mode not reliably mean “any byte” or “any line of JSON”?

---

Previous: Task modeling and engine differences | Next: Boundaries and quantifiers.