Compression

gzip, deflate and zlib are built into cljgo — no library to install, and they work the same in the REPL and inside a compiled 6 MB binary.

Start small: shrink a string, get it back

(require '[cljg.compress :as z])

(def packed (z/gzip "hello, hello, hello, hello"))

(z/gunzip packed {:as :string})

Output:

hello, hello, hello, hello

gzip returns compressed bytes; gunzip reverses it. {:as :string} says “give me text back, not bytes”.

See it actually working

Repetitive data compresses dramatically. Watch the byte counts:

(def report (apply str (repeat 200 "quarterly numbers ")))

(count report)
(count (z/gzip report))

Output:

3600
67

3600 characters in, 67 bytes out — because gzip notices the repetition. This is why compressed HTTP responses and log archives are small.

The other codecs

Same shape, different algorithm — deflate/inflate and zlib-compress/zlib-decompress:

(= report (z/inflate (z/deflate report) {:as :string}))

Output:

true

Which do I pick?

• gzip — files and HTTP; the one you usually want
• deflate — the raw algorithm, smallest framing
• zlib — when another system says “zlib format”

All three round-trip anything you put in. For big inputs there are streaming versions (gunzip-stream, …) that decompress as you read — they appear again in the chapter on streams.