Structural Code Refactoring

This skill covers three complementary tools for structural code transformation. Use the right one for the job — they're not competitors, they're specialists.

Pick the Right Tool

comby — Structural Search and Replace

comby understands code shape: balanced delimiters, string boundaries, comment regions. Unlike sed, you write patterns that look like the code you're targeting.

Match:   request.get(':[url]')
Rewrite: request({ url: ':[url]', method: 'get' })

Wins over grep/sed when:

• The pattern spans multiple lines or contains nested brackets
• The same text appears in comments/strings and you want to skip it
• You need to capture and rearrange subexpressions
• You're working across multiple languages in one pass

Installation

which comby          # check first

brew install comby   # macOS
bash <(curl -sL get-comby.netlify.app)   # Ubuntu/Debian
cargo install comby  # any platform

# Docker — no install, prefix every command with:
# docker run --rm -it -v "$PWD":/src -w /src comby/comby

Linux PCRE note: if comby complains about libpcre, run: sudo ln -s /usr/lib/libpcre.so /usr/lib/libpcre.so.3

Hole Syntax

Holes are the wildcards in comby templates. Choose carefully to avoid over-matching.

:[x] respects nested delimiters: foo(:[args]) captures bar(a, b), c correctly even though there are inner parentheses.

Reusing a hole name enforces equality: (:[x], :[x]) only matches identical values.

String converters (rewrite side only): :[x.to_camelCase], :[x.to_snake_case], :[x.uppercase], :[x.lowercase], :[x.Capitalize], :[x.length]

Full reference: references/hole-syntax.md

Basic Usage

# Match-only (no changes)
comby 'console.log(:[msg])' '' .js -match-only
comby 'console.log(:[msg])' '' .js -match-only -json-lines   # structured output with line numbers

# Preview changes
comby 'var :[[n]] = :[v]' 'const :[[n]] = :[v]' .js -diff

# Apply
comby 'var :[[n]] = :[v]' 'const :[[n]] = :[v]' .js -i          # in-place
comby 'var :[[n]] = :[v]' 'const :[[n]] = :[v]' .js -review     # interactive y/n per match

# Scope to a directory
comby 'oldFunc(:[args])' 'newFunc(:[args])' .js -d src/

# Force a language parser
comby 'fmt.Println(:[args])' 'log.Printf("[DEBUG] %v", :[args])' -matcher .go -i

See references/language-matchers.md for all ~40 supported languages.

Standard Safe-Refactoring Loop

1. Search — preview matches, zero risk: comby 'pattern' '' .ext -d src/ -match-only
2. Inspect — verify hole captures with JSON: add -json-lines | jq '.matches[].environment'
3. Diff — see exact changes: comby 'pattern' 'rewrite' .ext -d src/ -diff
4. Apply — interactively or bulk: -review or -i
5. Verify — re-run match-only and confirm zero results

Multi-Pattern Migrations (templates directory)

For several related patterns (e.g., an API overhaul), use a templates directory:

migrations/
├── get-to-fetch/
│   ├── match      ← plain text match template
│   └── rewrite    ← plain text rewrite template
└── post-to-fetch/
    ├── match
    └── rewrite

comby -templates migrations/ .js -i

Name directories with 1-, 2- prefixes when order matters (more-specific patterns first).

JSON Pipeline for Scripting

# Extract all hardcoded strings from logger.error() calls with file + line
comby 'logger.error(":[msg]", :[rest])' '' .ts -d src/ -match-only -json-lines \
  | jq -r '.uri + ":" + (.matches[].range.start.line | tostring) + "  " + .matches[].environment.msg'

# Cover both quote styles
{
  comby 'logger.error(":[msg]":[?rest])' '' .ts -d src/ -match-only -json-lines
  comby "logger.error(':[msg]':[?rest])" '' .ts -d src/ -match-only -json-lines
} | jq -r '.uri + ":" + (.matches[].range.start.line | tostring)' | sort -u

JSON structure: { "uri": "...", "matches": [{ "range": { "start": { "line": N } }, "environment": { "varname": "..." } }] }

Pre-filtering Large Codebases

rg -l 'createStore' --include='*.js' \
  | xargs comby 'createStore(:[args])' 'configureStore(:[args])' -matcher .js -i

Advanced: Rules, Regex Holes, Parallelism

# where clauses filter after structural match
comby ':[obj].:[method](:[args])' ':[obj].:[method]_v2(:[args])' .js \
  -rule 'where :[method] != "config"'

# regex constraint in hole
comby 'setTimeout(:[fn], :[ms~[0-9]+])' 'setTimeout(:[fn], 0)' .js -i

# optional hole (matches with or without the argument)
comby 'db.query(:[sql], :[?cb])' 'db.queryAsync(:[sql])' .js -i

# parallel jobs for speed
comby 'pattern' 'rewrite' .py -i -jobs 8

comby Quick Reference

Common Pitfalls

• :[x] vs :[[x]]: use :[[x]] for identifiers; :[x] captures spaces and expressions too.
• Don't start templates with a hole: :[x] = val is slow. Lead with a literal token.
• Python / indentation-sensitive languages: comby doesn't model indentation. Always use -review or -diff first, and consider 2to3/pyupgrade for Python-specific migrations (see below).
• Empty rewrite = deletion: comby 'pattern' '' .ext -i deletes all matches.

2to3 — Python 2 → 3 Migration

2to3 ships with Python and is the right tool for Python 2 → 3 migrations. It is AST-based, meaning it understands the language precisely — it handles every edge case that comby would struggle with: print >> sys.stderr, trailing commas, bare print, unicode, exec, dict.iteritems(), and more.

Use 2to3 instead of comby when migrating Python 2 code. Use comby for cross-language or library-API refactoring in Python 3 codebases.

2to3 Installation

Included with Python. Verify:

2to3 --version    # should print "2to3 X.Y"

Common fixers

Each -f flag applies a specific transformation:

Run 2to3 --list-fixes for the full list.

2to3 Usage

# Preview diffs without writing (safe)
2to3 -f print lib/

# Apply print fixer to all .py files in lib/
2to3 -f print -w lib/

# Apply ALL Python 2 → 3 fixers at once
2to3 -w lib/

# Apply multiple specific fixers
2to3 -f print -f unicode -f dict -w lib/

# No backup files (safe if you're in git)
2to3 -f print -w -n lib/

Safe Migration Workflow

# 1. Commit first — gives you a clean rollback point
git add lib/ && git commit -m "chore: snapshot before Python 3 migration"

# 2. Preview what will change
2to3 -f print lib/ 2>&1 | less

# 3. Apply
2to3 -f print -w -n lib/    # -n skips .bak files since you're in git

# 4. Verify no statements remain
grep -rn '\bprint ' lib/ --include='*.py' | grep -v 'print('

# 5. Syntax check every file
find lib/ -name '*.py' | xargs python3 -m py_compile && echo "All OK"

# 6. Run tests
python -m pytest lib/

Edge cases 2to3 handles correctly (that comby/sed would not)

pyupgrade — Modernize Python 3 Code

pyupgrade upgrades Python 3 code to use modern idioms. It is not about Python 2 → 3 — it is about moving from old Python 3 style to current best practices. Use it on codebases that are already Python 3 but were written a long time ago.

pyupgrade Installation

pip install pyupgrade
# or: pipx install pyupgrade

pyupgrade Usage

# Upgrade to minimum Python 3.8+ idioms
find lib/ -name '*.py' | xargs pyupgrade --py38-plus

# Upgrade to Python 3.10+ (adds match statement patterns, etc.)
find lib/ -name '*.py' | xargs pyupgrade --py310-plus

# Preview mode (dry run)
pyupgrade --py38-plus --check myfile.py

What pyupgrade does

Combine with 2to3 for a full Python 2 → modern Python 3 migration

# Step 1: Migrate Python 2 → 3
2to3 -w -n lib/

# Step 2: Modernize Python 3 idioms
find lib/ -name '*.py' | xargs pyupgrade --py38-plus

# Step 3: Format
black lib/

# Step 4: Verify
python -m pytest lib/

Reference Files

• references/hole-syntax.md — Full comby hole type reference with examples for every variant
• references/language-matchers.md — All ~40 supported comby language matchers