duplicate-detection-rules

Duplicate Detection Rules

Core Principle

Duplication is the enemy of change. When you change one, you must change all copies. Miss one, and you have a bug.

---

Detection Categories

1. Clone Code

Type	Definition	Similarity	Severity
Type-1	Exact clone (whitespace ignored)	100%	CRITICAL
Type-2	Same structure, different names/literals	>90%	HIGH
Type-3	Similar with some modifications	>70%	MEDIUM

Details: Read("references/clone-patterns.md")

2. Role Duplication

Detection Signals:

Signal	Description	Probability
Similar names	*Manager, *Handler, *Service	High
Same interface	Implements same interface	Very High
Similar methods	3+ same method names	High

Synonym Groups to Watch:

• Retrieve: get, fetch, retrieve, find, load
• Create: create, make, build, generate, new
• Validate: validate, check, verify, ensure

3. Variable/Constant Duplication

Type	Detection Criteria
Magic numbers	Same value used 2+ times
Duplicate strings	Same string (5+ chars) used 3+ times
Config values	Same config in multiple places

---

Detection Thresholds

Type	Threshold	Report
Type-1 Clone	5 lines	Always
Type-2 Clone	90% similar	Always
Type-3 Clone	70% similar	10+ lines
Role duplication	3 same methods	Always
Magic numbers	2+ usages	Always

---

Exceptions (Allowed Duplication)

Situation	Reason
Test code	Maintain test independence
Generated code	Auto-generation tool responsibility
External interface	Contract compliance required
Performance code	Intentional inlining

Not Considered Duplication:

• Code 2 lines or less
• Idiomatic patterns (e.g., if (err) return;)
• Language boilerplate

---

Fix Strategies

Detailed patterns: Read("references/fix-patterns.md")

Duplication Type	Strategy
Clone code	Extract common function
Role duplication	Consolidate classes
Constants	Centralize config