Overview
zgrep runs grep on gzip-compressed files, decompressing them on the fly, so you can search a rotated log such as app.log.2.gz without unpacking it on disk. It takes the options of grep. zegrep and zfgrep are the same with grep -E and grep -F, and this page answers for those names too.
Background
Today the name covers the GNU gzip script zgrep, and the zgrep on macOS, which the macOS manual lists in the same page as grep, egrep and fgrep. The BusyBox 1.37.0 build used for this page has no zgrep applet. They agree on passing grep options through and on reading compressed files by name.
Examples
Search a compressed file for a word:
printf 'a\nb\n' | gzip > f.gz && zgrep b f.gz
Count the matching lines with -c:
printf 'a\nb\n' | gzip > f.gz && zgrep -c a f.gz
Ignore case with -i:
printf 'a\nb\n' | gzip > f.gz && zgrep -i B f.gz
Use an extended regular expression through the zegrep name:
printf 'a\nb\n' | gzip > f.gz && zegrep 'a|b' f.gz | paste -sd, -
Search for a fixed string through the zfgrep name:
printf 'a\nb\n' | gzip > f.gz && zfgrep a f.gz
Common mistakes
zgrep reads the whole file every time and is slower than grep on the same text, so unpack a file you will search many times. With several files it prefixes each match with the file name, as grep does. It handles gzip files, and compressed files of other kinds (.bz2, .xz) need their own tools, such as bzgrep and xzgrep. A file that is not compressed is searched as plain text on GNU.
Portability
This page has no BusyBox view: BusyBox has no zgrep applet. The GNU versions are shell scripts, which the implementation probe never runs, so they are not identified automatically; the Linux guest ran them anyway. The Common examples were executed on GNU gzip scripts in the Linux guest and macOS system tools (see the development guide for the exact check).
References
Original explanations, with upstream documentation for further reading. FreeBSD references are comparison material, not a claim of Apple version equivalence.