grep Deep Dive

The text-search tool everyone reaches for

grep stands for "globally search a regular expression and print." It searches input for lines matching a pattern. The default basic regular expression (BRE) syntax is awkward; almost every modern user adds -E for extended (ERE) or -P for Perl-style (PCRE).

The flags you'll wear out

• -i — case-insensitive.
• -r / -R — recurse into directories.
• -n — show line numbers.
• -v — invert: show non-matching lines.
• -l — only filenames containing matches.
• -c — count of matches.
• -A 3 — 3 lines of context after each match. -B 3 before, -C 3 both.
• -E — extended regex (alternation, +, ?, {n,m} without backslashes).
• --include='*.py' / --exclude — file filters.
• --color=auto — highlight matches.

Common patterns

• Recursive case-insensitive: grep -rin 'TODO' .
• Files with matches only: grep -rl 'pattern' src/
• Skip binaries: grep -rn --binary-files=without-match 'pattern' .
• Count per file: grep -rc 'TODO' . shows file:count pairs.
• Extended regex: grep -E 'foo|bar' file

The exit code

0 = at least one match, 1 = no match, 2 = error. That makes grep -q 'pattern' file a perfect existence check in scripts: if grep -q ERROR build.log; then ...; fi.

When grep is the wrong tool

For multi-line patterns, structured data (JSON/YAML), or symbol-aware search, reach for the modern alternatives in the next track: ripgrep (rg), jq, ast-grep.