scanner

type FileEntry struct {
	// Path is the absolute or relative path to the file.
	Path string

	// Language is the detected programming language (e.g., "go", "typescript").
	Language string

	// Size is the file size in bytes.
	Size int64

	// Content holds the file bytes when PreloadContent is enabled.
	// nil when content was not preloaded.
	Content []byte
}

type FileScanner struct {
	// contains filtered or unexported fields
}

type ScanOptions struct {
	// RootPath is the directory or file to scan.
	RootPath string

	// SupportedExtensions maps file extensions to language names.
	SupportedExtensions map[string]string

	// IgnoreFiles is the list of ignore file paths (default: [".gitignore"]).
	IgnoreFiles []string

	// IncludePatterns is a list of glob patterns to include.
	// If non-empty, only files matching at least one pattern are included.
	// Supports doublestar (**) patterns.
	IncludePatterns []string

	// ExcludePatterns is a list of glob patterns to exclude.
	// Files matching any pattern are excluded.
	// Supports doublestar (**) patterns.
	ExcludePatterns []string

	// ChangedFiles is an optional whitelist of file paths (relative to RootPath).
	// When non-nil, only files in this list are included in scan results.
	// Used by --changed and --since flags to restrict scanning to git-changed files.
	ChangedFiles map[string]bool

	// IncludeHidden determines whether to include hidden files (dotfiles).
	IncludeHidden bool

	// MaxFileSize is the maximum file size in bytes to include.
	MaxFileSize int64

	// PreloadContent reads file content during scan so downstream consumers
	// (e.g., the extractor) can skip a redundant os.ReadFile call.
	PreloadContent bool

	// MaxTotalPreloadSize limits the total bytes preloaded into memory when
	// PreloadContent is true. Once this budget is exceeded, remaining files
	// are included in the scan results but with Content set to nil (the
	// extractor will fall back to on-demand os.ReadFile). A value of 0 means
	// no limit. Default: 0.
	MaxTotalPreloadSize int64
}

type ScanResult struct {
	// Files is the list of matched files.
	Files []FileEntry

	// TotalSize is the sum of all matched file sizes.
	TotalSize int64

	// SkippedCount is the number of files skipped (too large, unsupported, etc.).
	SkippedCount int

	// Warnings contains non-fatal issues encountered during scanning.
	Warnings []string
}

type Scanner interface {
	// Scan performs the scan and returns scan results.
	// The context allows cancellation of long-running scans.
	Scan(ctx context.Context) (*ScanResult, error)
}