go install github.com/tnyeanderson/yamlgrep@latestPipe a multidoc YAML to this command and give it grep-ish arguments, and only
the matching documents from the YAML will be printed. It's like grep -C, but
with the perfect amount of context.
Take the following demo.yaml:
---
id: 1
first_name: Kin
last_name: Stockwell
email: [email protected]
---
id: 2
first_name: Serene
last_name: Edgeller
email: [email protected]
---
id: 3
first_name: Hollis
last_name: Syrie
email: [email protected]Let's find all the YAML documents that contain ell:
$ cat /tmp/demo.yaml | yamlgrep ell
---
id: 1
first_name: Kin
last_name: Stockwell
email: [email protected]
---
id: 2
first_name: Serene
last_name: Edgeller
email: [email protected]
Under the hood, yamlgrep calls whatever grep is in the user's $PATH
(usually the system's). This means you can use all the same arguments you
normally would with grep. However, arguments like -C are ignored, because
this program only cares if the grep matched or not, and it ignores grep's
output (both to stdout and stderr).
This started as a simple bash script, which I thought was really elegant. I wrote it quickly after being frustrated with YAML searching tools at work, and it all worked fine with small YAML files at home. But once I tried using it on a very large YAML response, I learned just how INCREDIBLY SLOW bash scripts can be, even if they seem so nice. I was waiting minutes for my elegant script to complete.
So I rewrote it in go and it became 1.5 billion times faster. That is not
an exaggeration: the original script is retained in the testdata directory
and the benchmarks are available for you to run yourself:
$ go test -bench=.
goos: linux
goarch: amd64
pkg: github.com/tnyeanderson/yamlgrep
cpu: AMD Ryzen 5 2600 Six-Core Processor
BenchmarkYAMLGrep-12 1000000000 0.7599 ns/op
BenchmarkYAMLGrep_Bash-12 1 1172747556 ns/op
PASS
ok github.com/tnyeanderson/yamlgrep 37.227s