Text CLI (grep / sed / awk) 速查表
搭配實際管線的 Unix 文字處理指令:GNU grep/ripgrep 搜尋、sed 串流編輯、awk 欄位與彙總,以及 cut、sort、uniq 與 tr。
運作原理
Unix text tools act like stream filters: each tool reads lines from files or stdin, does one small job (search, slice columns, replace, count, sort) and writes lines to stdout so you can snap them together with `|`.
生存指南
最先需要掌握的少數核心指令。
41 個指令
Search lines (grep)
grep prints lines matching a pattern. Add -n for line numbers, -i for case-insensitive and -E for extended regex.
官方文件:grep usage- 必備
Search recursively and print file paths and line numbers.
grep -rn "<pattern>" <dir>建構指令
grep -rn "<pattern>" <dir>輸入的值會依原樣插入。若包含空白或特殊字元,請自行加上引號。
使用時機
Find every occurrence of a function, error code or config key across a directory tree.
指令解析
- -r
- recurse into subdirectories
- -n
- prefix matches with 1-based line numbers
- <pattern>
- search regex or text
- <dir>
- directory to search (use `.` for current)
範例
printf "200 GET /api\n500 POST /pay\n200 GET /api\n404 GET /old\n" | grep -n "GET"執行結果
1:200 GET /api 3:200 GET /api 4:404 GET /old替代指令
- rg "<pattern>" (ripgrep, skips .gitignore automatically)
另請參閱
Transform text streams (sed)
sed edits lines on the fly. Test without `-i` first so you preview changes on stdout before modifying files.
官方文件:sed commandsProcess columns & aggregate (awk)
awk splits every line into fields ($1, $2, ...) automatically, running `{ action }` whenever `condition` is true.
官方文件:GNU Awk User GuideSort, count, cut & translate
Fast single-purpose coreutils for frequency counts, column slicing and character cleanup.
官方文件:Operating on sorted filesFind files & batch commands (find, xargs)
Locate files by name, size or age and pass them safely to downstream commands using NUL delimiters.
官方文件:GNU Findutils Manual在 KitLocal 工具中試用
術語表
- stdin / stdout
- Standard input and output streams wired together by pipes (`|`).
- ERE (-E)
- Extended Regular Expressions: `+`, `?`, `|`, `()` work without extra backslashes.
- Field ($1, $NF)
- In awk, columns split by whitespace (or `-F`): `$0` is the whole line, `$NF` is the last column.
- NUL delimiter (-0)
- Separating paths with `\0` instead of newlines so spaces and quotes never break pipelines.