domain

type Clone struct {
	Hash         string          `json:"hash"`
	Type         string          `json:"type"`
	Similarity   float64         `json:"similarity"`
	LineCount    int             `json:"line_count"`
	TokenCount   int             `json:"token_count"`
	SavedLines   int             `json:"saved_lines,omitempty"` // lines removable by deduplicating: (instances-1) × line_count
	Scope        string          `json:"scope,omitempty"`       // same-file | cross-file | cross-dir
	Instances    []CloneInstance `json:"instances"`
	TestProdSpan bool            `json:"test_prod_span,omitempty"`
	Suppressed   bool            `json:"suppressed,omitempty"`
	Baseline     bool            `json:"baseline,omitempty"` // true when covered by the --baseline file (known debt)
	ChurnScore   int             `json:"churn_score,omitempty"`
	// Rule names the .dupehound.yml rule that judged this clone ("scan" for
	// the base settings) and RuleMinTokens the threshold it applied. Both are
	// set only when rules are in use, so output without rules is unchanged.
	Rule          string `json:"rule,omitempty"`
	RuleMinTokens int    `json:"rule_min_tokens,omitempty"`
}

type CloneInstance struct {
	File        string   `json:"file"`
	StartLine   int      `json:"start_line"`
	EndLine     int      `json:"end_line"`
	Function    string   `json:"function,omitempty"` // best-effort enclosing function/method name
	Lines       []string `json:"lines"`
	IsTest      bool     `json:"is_test,omitempty"`
	FileCommits int      `json:"file_commits_in_window,omitempty"`
}

type Config struct {
	Scan  ScanConfig `yaml:"scan"`
	Rules []Rule     `yaml:"rules"`
}

type DeadFunc struct {
	File     string `json:"file"`
	Line     int    `json:"line"`
	Name     string `json:"name"`
	Language string `json:"language"`
	// Confidence is "high" for a function that can only be called from code
	// that was scanned (unexported, private, module-local) and "low" for one
	// that is public: unused here, but possibly API for callers elsewhere.
	Confidence string `json:"confidence,omitempty"`
}

type FileStats struct {
	File           string  `json:"file"`
	TotalLines     int     `json:"total_lines"`
	DuplicateLines int     `json:"duplicate_lines"`
	DuplicationPct float64 `json:"duplication_pct"`
}

type Language struct {
	Name        string
	Extensions  []string
	LineComment string
	BlockStart  string
	BlockEnd    string
}

type Report struct {
	TotalFiles        int     `json:"total_files"`
	ScannedFiles      int     `json:"scanned_files"`
	SkippedFiles      int     `json:"skipped_files,omitempty"`       // binary files skipped
	SkippedLargeFiles int     `json:"skipped_large_files,omitempty"` // files over --max-file-size
	SkippedByRule     int     `json:"skipped_by_rule,omitempty"`     // files matched by a `skip: true` rule
	TotalClones       int     `json:"total_clones"`
	TotalLines        int     `json:"total_lines"`
	DuplicateLines    int     `json:"duplicate_lines"`
	DuplicationPct    float64 `json:"duplication_pct"`
	SavedLines        int     `json:"saved_lines,omitempty"` // total lines removable by deduplicating all reported clones
	SuppressedClones  int     `json:"suppressed_clones,omitempty"`
	// Partial is true when detection was capped (--max-pairs hit or a fuzzy
	// bucket was truncated by --max-bucket): type-1/2 results are complete but
	// some type-3 clones may be missing. PartialReason says which guard fired.
	Partial       bool   `json:"partial,omitempty"`
	PartialReason string `json:"partial_reason,omitempty"`
	// Baseline ratchet (--baseline): known clones are recorded debt, new ones fail the scan.
	BaselineFile   string      `json:"baseline_file,omitempty"`
	BaselineKnown  int         `json:"baseline_known,omitempty"`
	BaselineNew    int         `json:"baseline_new,omitempty"`
	FileStats      []FileStats `json:"file_stats,omitempty"`
	Clones         []Clone     `json:"clones"`
	DeadFunctions  []DeadFunc  `json:"dead_functions,omitempty"`
	NewClones      []Clone     `json:"new_clones,omitempty"`
	SinceDiffRef   string      `json:"since_diff_ref,omitempty"`
	SinceDiffFiles int         `json:"since_diff_files,omitempty"`
}

type Rule struct {
	Name       string   `yaml:"name"`
	Paths      []string `yaml:"paths"`
	MinTokens  int      `yaml:"min-tokens"`
	Similarity float64  `yaml:"similarity"`
	Skip       bool     `yaml:"skip"`
}

type ScanConfig struct {
	Path      string   `yaml:"path"`
	MinTokens int      `yaml:"min-tokens"`
	Exclude   []string `yaml:"exclude"`
	Include   []string `yaml:"include"`
	Language  string   `yaml:"language"`
	Format    string   `yaml:"format"`
	Output    string   `yaml:"output"`
	ExitZero  bool     `yaml:"exit-zero"`
	Top       int      `yaml:"top"`
	Baseline  string   `yaml:"baseline"` // path to a baseline file written with --write-baseline
	Scope     string   `yaml:"scope"`    // file | function: where type-1/2 clones may lie
	// MinCloneLines is a pointer so that an explicit 0 (no floor) in the config
	// can be told apart from the key being absent.
	MinCloneLines *int `yaml:"min-clone-lines"`
}