ripgrep（ripgrep 搜索）

Ripgrep (rg) - Fast Text Search Tool

Overview

Ripgrep is a line-oriented search tool that recursively searches directories for regex patterns. It's 10-100x faster than grep and respects .gitignore by default. Use it instead of grep, find, or manually reading large files.

Core principle: When you need to find text in files, use ripgrep. Don't read entire files into context when you can search them.

When to Use

Use ripgrep when:

• Searching for text patterns across a codebase or directory
• Finding all occurrences of a function, variable, or string
• Searching through books, documentation, or large text files
• Files are too large to read fully into context
• Looking for specific content in many files at once
• Finding files that contain (or don't contain) certain patterns
• Extracting matching lines for analysis

Don't use when:

• You need the full file content (use Read tool)
• Simple glob pattern matching for filenames only (use Glob tool)
• You need structured data extraction (consider jq, awk)

Quick Reference

File Filtering

By File Type

Ripgrep has built-in file type definitions. Use -t to include, -T to exclude:

# Search only Python files
rg -t py "def main"

# Search only JavaScript and TypeScript
rg -t js -t ts "import"

# Exclude test files
rg -T test "function"

# List all known types
rg --type-list

Common types: py, js, ts, rust, go, java, c, cpp, rb, php, html, css, json, yaml, md, txt, sh

By Glob Pattern

# Only .tsx files
rg -g "*.tsx" "useState"

# Exclude node_modules (in addition to gitignore)
rg -g "!node_modules/**" "pattern"

# Only files in src directory
rg -g "src/**" "pattern"

# Multiple globs
rg -g "*.js" -g "*.ts" "pattern"

# Case insensitive globs
rg --iglob "*.JSON" "pattern"

By File Size

# Skip files larger than 1MB
rg --max-filesize 1M "pattern"

Directory Control

# Limit depth
rg --max-depth 2 "pattern"

# Search hidden files (dotfiles)
rg --hidden "pattern"

# Follow symlinks
rg -L "pattern"

# Ignore all ignore files (.gitignore, etc.)
rg --no-ignore "pattern"

# Progressive unrestricted (-u can stack up to 3 times)
rg -u "pattern"      # --no-ignore
rg -uu "pattern"     # --no-ignore --hidden
rg -uuu "pattern"    # --no-ignore --hidden --binary

Context Options

# Lines after match
rg -A 5 "pattern"

# Lines before match
rg -B 5 "pattern"

# Lines before and after
rg -C 5 "pattern"

# Print entire file on match (passthrough mode)
rg --passthru "pattern"

Output Formats

# Just filenames with matches
rg -l "pattern"

# Files without matches
rg --files-without-match "pattern"

# Count matches per file
rg -c "pattern"

# Count total matches (not lines)
rg --count-matches "pattern"

# Only the matched text (not full line)
rg -o "pattern"

# JSON output (for parsing)
rg --json "pattern"

# Vim-compatible output (file:line:col:match)
rg --vimgrep "pattern"

# With statistics
rg --stats "pattern"

Regex Patterns

Ripgrep uses Rust regex syntax by default:

# Alternation
rg "foo|bar"

# Character classes
rg "[0-9]+"
rg "[a-zA-Z_][a-zA-Z0-9_]*"

# Word boundaries
rg "\bword\b"

# Quantifiers
rg "colou?r"           # 0 or 1
rg "go+gle"            # 1 or more
rg "ha*"               # 0 or more
rg "x{2,4}"            # 2 to 4 times

# Groups
rg "(foo|bar)baz"

# Lookahead/lookbehind (requires -P for PCRE2)
rg -P "(?<=prefix)content"
rg -P "content(?=suffix)"

Multiline Matching

# Enable multiline mode
rg -U "start.*\nend"

# Dot matches newline too
rg -U --multiline-dotall "start.*end"

# Match across lines
rg -U "function\s+\w+\([^)]*\)\s*\{"

Replacement (Preview Only)

Ripgrep can show what replacements would look like (doesn't modify files):

# Simple replacement
rg "old" -r "new"

# Using capture groups
rg "(\w+)@(\w+)" -r "$2::$1"

# Remove matches (empty replacement)
rg "pattern" -r ""

Searching Special Files

Compressed Files

# Search in gzip, bzip2, xz, lz4, lzma, zstd files
rg -z "pattern" file.gz
rg -z "pattern" archive.tar.gz

Binary Files

# Include binary files
rg --binary "pattern"

# Treat binary as text (may produce garbage)
rg -a "pattern"

Large Files

For files too large to read into context:

# Search and show only matching lines
rg "specific pattern" large_file.txt

# Limit matches to first N per file
rg -m 10 "pattern" huge_file.log

# Show byte offset for large file navigation
rg -b "pattern" large_file.txt

# Use with head/tail for pagination
rg "pattern" large_file.txt | head -100

Performance Tips

1. Be specific with paths - Don't search from root when you know the subdir
2. Use file types - -t py is faster than -g "*.py"
3. Use fixed strings - -F when you don't need regex
4. Limit depth - --max-depth when you know structure
5. Let gitignore work - Don't use --no-ignore unless needed
6. Use word boundaries - -w is optimized

Common Patterns

Find function definitions

# Python
rg "def \w+\(" -t py

# JavaScript/TypeScript
rg "(function|const|let|var)\s+\w+\s*=" -t js -t ts
rg "^\s*(async\s+)?function" -t js

# Go
rg "^func\s+\w+" -t go

Find imports/requires

# Python
rg "^(import|from)\s+" -t py

# JavaScript
rg "^(import|require\()" -t js

# Go
rg "^import\s+" -t go

Find TODO/FIXME comments

rg "(TODO|FIXME|HACK|XXX):"

Find error handling

# Python
rg "except\s+\w+:" -t py

# JavaScript
rg "\.catch\(|catch\s*\(" -t js

Find class definitions

# Python
rg "^class\s+\w+" -t py

# JavaScript/TypeScript
rg "^(export\s+)?(default\s+)?class\s+\w+" -t js -t ts

Search in books/documents

# Find chapter headings
rg "^(Chapter|CHAPTER)\s+\d+" book.txt

# Find quoted text
rg '"[^"]{20,}"' document.txt

# Find paragraphs containing word
rg -C 2 "keyword" book.txt

Combining with Other Tools

# Find files, then search
rg --files | xargs rg "pattern"

# Search and count by file
rg -c "pattern" | sort -t: -k2 -rn

# Search and open in editor
rg -l "pattern" | xargs code

# Extract unique matches
rg -o "\b[A-Z]{2,}\b" | sort -u

# Search multiple patterns from file
rg -f patterns.txt

Exit Codes

Useful for scripting:

if rg -q "pattern" file.txt; then
    echo "Found"
fi

Common Mistakes

When to Prefer Other Tools