← Shell / Bash Course12 / 13

Pattern Matching & Regex

Match text in Bash: filename globbing, the =~ regex operator inside [[ ]], and basic vs extended regular expressions with grep — knowing which is which.

Ad 728×90

Globbing vs regex

Why: these two look similar but are different languages. A glob matches FILENAMES: * is any text, ? is one character, [abc] is a character set. A regex matches inside TEXT and is what grep and =~ use. Confusing them is a common source of "why didn't that match" frustration.

Globs — expanded by the shell into matching filenames

ls *.txt            # every file ending in .txt

ls report-?.csv     # report-1.csv, report-a.csv (one char)

ls [abc]*.log       # files starting with a, b, or c

Regex inside [[ =~ ]]

Why: the =~ operator tests a string against an extended regular expression, right inside Bash — no grep needed. The captured groups land in the BASH_REMATCH array. Note: do not quote the pattern, or it is treated as literal text.

#!/usr/bin/env bash

read -r -p "Enter a date (YYYY-MM-DD): " input

if [[ "$input" =~ ^([0-9]{4})-([0-9]{2})-([0-9]{2})$ ]]; then
  echo "year is ${BASH_REMATCH[1]}"   # first capture group
else
  echo "not a valid date"
fi

Common regex building blocks

Note: the pieces you reach for most. ^ start, $ end, . any character, * zero-or-more, + one-or-more, [0-9] a digit, [a-z] a letter, \. a literal dot. Anchoring with ^ and $ is what turns "contains" into "exactly matches".

#!/usr/bin/env bash

# A few patterns worth recognizing:
#   ^error        line starts with "error"
#   failed$       line ends with "failed"
#   [0-9]+        one or more digits
#   ^\s*$         a blank line (only whitespace)
#   \.log$        ends with a literal ".log"

email="ada@example.com"
[[ "$email" =~ ^[^@]+@[^@]+\.[a-z]+$ ]] && echo "looks like an email"

Basic vs extended regex in grep

Why: plain grep uses Basic Regular Expressions, where +, ?, |, and () must be backslash-escaped to act as operators — surprising everyone. grep -E (extended) makes them work without escaping. Reach for -E by default. Note: grep and the text pipeline live in the Linux course lesson on pipes & text processing.

Extended regex (-E): operators work without escaping

grep -E "warn|error" app.log         # lines with either word

grep -E "^[0-9]{3}-[0-9]{4}$" phones  # match a phone format

-o prints only the match; -i ignores case; -c counts lines

grep -Eoi "[a-z]+@[a-z.]+" contacts.txt