analyzer

package
v0.0.0-...-ec117d0 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Jul 20, 2026 License: Apache-2.0 Imports: 19 Imported by: 0

Documentation

Index

Constants

View Source 
const DefaultAutoThreshold = 0.80

DefaultAutoThreshold is the default minimum confidence for auto-detection.

View Source 
const DefaultSkipThreshold = 0.50

DefaultSkipThreshold is the default minimum confidence to include at all.

Variables

This section is empty.

Functions

func IsValidContentType 

func IsValidContentType(ct catalog.ContentType) bool

IsValidContentType checks whether ct is a recognized content type.

func PreserveUserMetadata 

func PreserveUserMetadata(existing *registry.ManifestItem, detected *DetectedItem)

PreserveUserMetadata merges user-edited display metadata from an existing manifest item into a newly-analyzed DetectedItem. Only non-empty user values are preserved.

func ResolveReferences 

func ResolveReferences(path string, repoRoot string) []string

ResolveReferences finds files referenced by a content item. path is the item's primary file path (relative to repoRoot). repoRoot must be filepath.EvalSymlinks-resolved. Returns relative paths of referenced files that exist within repoRoot.

func SanitizeItem 

func SanitizeItem(item *DetectedItem)

SanitizeItem strips C0/C1 control characters from all string fields of a DetectedItem and truncates fields that exceed display length limits. Must be called after Classify, before dedup or audit writes.

func SaveScanAsConfig 

func SaveScanAsConfig(repoRoot string, cfg *ScanAsConfig) error

SaveScanAsConfig writes cfg to .syllago.yaml in repoRoot.

func ShouldTriggerInteractiveFallback 

func ShouldTriggerInteractiveFallback(result *AnalysisResult) bool

ShouldTriggerInteractiveFallback returns true if the result is sparse enough to warrant user-directed discovery. Only applicable in interactive mode.

func ToManifestItem 

func ToManifestItem(item *DetectedItem) registry.ManifestItem

ToManifestItem converts a DetectedItem to a registry.ManifestItem.

func WriteGeneratedManifest 

func WriteGeneratedManifest(name string, items []*DetectedItem) error

WriteGeneratedManifest writes a registry.yaml to the syllago cache directory for the named registry. This file is separate from the repo's own registry.yaml.

Types

type AnalysisConfig 

type AnalysisConfig struct {
	AutoThreshold float64                        // default DefaultAutoThreshold
	SkipThreshold float64                        // default DefaultSkipThreshold
	ExcludeDirs   []string                       // additional per-registry exclusions
	SymlinkPolicy string                         // "ask", "follow", "skip"
	Strict        bool                           // disables content-signal fallback
	ScanAsPaths   map[string]catalog.ContentType // user-directed: path prefix → type
	DebugSkips    bool                           // collect per-file skip reasons
	NoConfig      bool                           // suppress .syllago.yaml loading
}

AnalysisConfig controls analyzer behavior.

func DefaultConfig 

func DefaultConfig() AnalysisConfig

DefaultConfig returns the default analysis configuration.

type AnalysisResult 

type AnalysisResult struct {
	Auto        []*DetectedItem // confidence > AutoThreshold (excluding hooks/MCP)
	Confirm     []*DetectedItem // confidence in [SkipThreshold, AutoThreshold], plus ALL hooks/MCP
	Warnings    []string
	SkipReasons []SkipEntry // populated when DebugSkips=true
}

AnalysisResult holds the output of a full repository analysis.

func (*AnalysisResult) AllItems 

func (r *AnalysisResult) AllItems() []*DetectedItem

AllItems returns Auto and Confirm combined, Auto first.

func (*AnalysisResult) CountByType 

func (r *AnalysisResult) CountByType() map[catalog.ContentType]int

CountByType returns a map of ContentType to item count across Auto+Confirm.

func (*AnalysisResult) IsEmpty 

func (r *AnalysisResult) IsEmpty() bool

IsEmpty returns true if both Auto and Confirm are empty.

type Analyzer 

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

Analyzer orchestrates content discovery for a repository.

func New 

func New(config AnalysisConfig) *Analyzer

New creates an Analyzer with all built-in detectors registered.

func (*Analyzer) Analyze 

func (a *Analyzer) Analyze(repoDir string) (*AnalysisResult, error)

Analyze examines repoDir and returns classified content items. repoDir is resolved via filepath.EvalSymlinks before analysis.

type CandidateMatch 

type CandidateMatch struct {
	Path     string
	Pattern  DetectionPattern
	Detector ContentDetector
}

CandidateMatch records one detector's claim on a file path.

func MatchPatterns 

func MatchPatterns(paths []string, detectors []ContentDetector) []CandidateMatch

MatchPatterns evaluates all detectors' patterns against the path index. Returns one CandidateMatch per (detector, path) pair where the glob matched. paths are relative to repoRoot (as returned by Walk), normalized to forward slashes via filepath.ToSlash before matching (B4: ensures patterns work on all platforms since filepath.Match uses the OS separator on Windows).

type ClaudeCodeDetector 

type ClaudeCodeDetector struct{}

ClaudeCodeDetector detects Claude Code content.

func (*ClaudeCodeDetector) Classify 

func (d *ClaudeCodeDetector) Classify(path string, repoRoot string) ([]*DetectedItem, error)

func (*ClaudeCodeDetector) Patterns 

func (d *ClaudeCodeDetector) Patterns() []DetectionPattern

func (*ClaudeCodeDetector) ProviderSlug 

func (d *ClaudeCodeDetector) ProviderSlug() string

type ClaudeCodePluginDetector 

type ClaudeCodePluginDetector struct{}

ClaudeCodePluginDetector detects Claude Code Plugin content.

func (*ClaudeCodePluginDetector) Classify 

func (d *ClaudeCodePluginDetector) Classify(path string, repoRoot string) ([]*DetectedItem, error)

func (*ClaudeCodePluginDetector) Patterns 

func (d *ClaudeCodePluginDetector) Patterns() []DetectionPattern

func (*ClaudeCodePluginDetector) ProviderSlug 

func (d *ClaudeCodePluginDetector) ProviderSlug() string

type ClineDetector 

type ClineDetector struct{}

ClineDetector detects Cline content.

func (*ClineDetector) Classify 

func (d *ClineDetector) Classify(path string, repoRoot string) ([]*DetectedItem, error)

func (*ClineDetector) Patterns 

func (d *ClineDetector) Patterns() []DetectionPattern

func (*ClineDetector) ProviderSlug 

func (d *ClineDetector) ProviderSlug() string

type CodexDetector 

type CodexDetector struct{}

CodexDetector detects OpenAI Codex content.

func (*CodexDetector) Classify 

func (d *CodexDetector) Classify(path string, repoRoot string) ([]*DetectedItem, error)

func (*CodexDetector) Patterns 

func (d *CodexDetector) Patterns() []DetectionPattern

func (*CodexDetector) ProviderSlug 

func (d *CodexDetector) ProviderSlug() string

type ConfidenceCategory 

type ConfidenceCategory int

ConfidenceCategory partitions items for UI presentation. 

const (
	CategoryAuto    ConfidenceCategory = iota // > autoThreshold
	CategoryConfirm                           // >= skipThreshold and <= autoThreshold
	CategorySkip                              // < skipThreshold (never included in manifest)
)

type ConfidenceTier 

type ConfidenceTier string

ConfidenceTier classifies a content-signal item's confidence level. 

const (
	TierLow    ConfidenceTier = "Low confidence"
	TierMedium ConfidenceTier = "Medium confidence"
	TierHigh   ConfidenceTier = "High confidence"
	TierUser   ConfidenceTier = "User-asserted, no content signals"
)

func TierForItem 

func TierForItem(item *DetectedItem) ConfidenceTier

TierForItem returns the confidence tier label for a DetectedItem. User-directed items with zero content signals get a special label.

func TierForMeta 

func TierForMeta(confidence float64, method string) ConfidenceTier

TierForMeta applies tier logic from persisted metadata fields. Uses DetectionMethod instead of floating-point equality to identify user-directed items — avoids fragile float comparison at the 0.60 boundary.

type ContentDetector 

type ContentDetector interface {
	// ProviderSlug returns the detector's provider identifier.
	ProviderSlug() string
	// Patterns returns the glob patterns this detector recognizes.
	Patterns() []DetectionPattern
	// Classify inspects a candidate path and returns detected items.
	// Returns nil if the path matched a pattern but content inspection
	// shows it is not actually content.
	// repoRoot is filepath.EvalSymlinks-resolved before being passed in.
	Classify(path string, repoRoot string) ([]*DetectedItem, error)
}

ContentDetector is the interface every provider detector implements.

type ContentSignalDetector 

type ContentSignalDetector struct{}

ContentSignalDetector classifies files that pattern-based detectors did not match. It uses weighted signal scoring against static fingerprints only.

func (*ContentSignalDetector) Classify 

func (d *ContentSignalDetector) Classify(path, repoRoot string) ([]*DetectedItem, error)

Classify satisfies ContentDetector but is not used by this detector.

func (*ContentSignalDetector) ClassifyUnmatched 

func (d *ContentSignalDetector) ClassifyUnmatched(
	unmatchedPaths []string,
	repoRoot string,
	scanAsPaths map[string]catalog.ContentType,
	debugSkips bool,
) ([]*DetectedItem, []SkipEntry, error)

ClassifyUnmatched inspects files not matched by any pattern-based detector. unmatchedPaths are relative to repoRoot. scanAsPaths maps relative path prefixes to their user-specified content type (bypasses directory keyword filter). When debugSkips is true, skipped files are recorded in the returned skip entries.

func (*ContentSignalDetector) Patterns 

func (d *ContentSignalDetector) Patterns() []DetectionPattern

Patterns returns empty — the content-signal detector does not participate in MatchPatterns. It operates on unmatched files via ClassifyUnmatched.

func (*ContentSignalDetector) ProviderSlug 

func (d *ContentSignalDetector) ProviderSlug() string

type CopilotDetector 

type CopilotDetector struct{}

CopilotDetector detects VS Code Copilot content.

func (*CopilotDetector) Classify 

func (d *CopilotDetector) Classify(path string, repoRoot string) ([]*DetectedItem, error)

func (*CopilotDetector) Patterns 

func (d *CopilotDetector) Patterns() []DetectionPattern

func (*CopilotDetector) ProviderSlug 

func (d *CopilotDetector) ProviderSlug() string

type CursorDetector 

type CursorDetector struct{}

CursorDetector detects Cursor content.

func (*CursorDetector) Classify 

func (d *CursorDetector) Classify(path string, repoRoot string) ([]*DetectedItem, error)

func (*CursorDetector) Patterns 

func (d *CursorDetector) Patterns() []DetectionPattern

func (*CursorDetector) ProviderSlug 

func (d *CursorDetector) ProviderSlug() string

type DependencyRef 

type DependencyRef struct {
	Registry string
	Type     catalog.ContentType
	Name     string
}

DependencyRef references another content item this item depends on.

type DetectedItem 

type DetectedItem struct {
	Name          string
	Type          catalog.ContentType
	InternalLabel string // "hook-script", "hook-wiring", etc. Empty = use Type.
	Provider      string
	Path          string // primary file or directory (relative to repoRoot)
	ContentHash   string // SHA-256 of primary file content
	Confidence    float64
	Scripts       []string // referenced script files (relative to repoRoot)
	References    []string // other files needed (relative to repoRoot)
	Dependencies  []DependencyRef
	HookEvent     string
	HookIndex     int
	ConfigSource  string // where wiring was found (e.g., ".claude/settings.json")
	DisplayName   string
	Description   string
	// Providers holds alias paths for deduplicated items — paths of lower-confidence
	// duplicates that were suppressed in favor of this item (same name+type+hash).
	Providers []string
	// Signals records the content-signal scoring breakdown for audit logging.
	// Only populated for items from the ContentSignalDetector.
	Signals []SignalEntry
}

DetectedItem is the output of a detector's Classify call.

func DeduplicateItems 

func DeduplicateItems(items []*DetectedItem) (deduped []*DetectedItem, conflicts [][2]*DetectedItem)

DeduplicateItems processes classified items: 1. Suppresses hook-script items whose Path appears in another item's Scripts list. 2. Deduplicates same (type, name) items:

  • Same content hash: keep highest confidence, record other paths in Providers alias.
  • Different content hash: keep both (conflicts returned separately).

Returns deduplicated items and conflict pairs.

type DetectionPattern 

type DetectionPattern struct {
	Glob        string
	ContentType catalog.ContentType
	// InternalLabel overrides ContentType for detector-internal classification.
	// Use for "hook-script", "hook-wiring", "plugin-manifest", "output-style".
	// Empty means use ContentType directly.
	InternalLabel string
	Confidence    float64
}

DetectionPattern declares a glob pattern a detector recognizes.

type GeminiDetector 

type GeminiDetector struct{}

GeminiDetector detects Gemini CLI content.

func (*GeminiDetector) Classify 

func (d *GeminiDetector) Classify(path string, repoRoot string) ([]*DetectedItem, error)

func (*GeminiDetector) Patterns 

func (d *GeminiDetector) Patterns() []DetectionPattern

func (*GeminiDetector) ProviderSlug 

func (d *GeminiDetector) ProviderSlug() string

type ReanalysisResult 

type ReanalysisResult struct {
	Unchanged []*registry.ManifestItem // path exists, hash unchanged
	Changed   []string                 // relative paths where hash changed
	Missing   []string                 // relative paths that no longer exist
	Warnings  []string
}

ReanalysisResult holds the outcome of a sync hash comparison.

func DiffManifest 

func DiffManifest(manifest *registry.Manifest, repoRoot string) ReanalysisResult

DiffManifest compares the current filesystem state against an existing manifest. For each item in the manifest, it reads the file at item.Path (relative to repoRoot) and compares its SHA-256 against item.ContentHash. Items with empty ContentHash are treated as Unchanged (authored manifests without hashes).

type RooCodeDetector 

type RooCodeDetector struct{}

RooCodeDetector detects Roo Code content.

func (*RooCodeDetector) Classify 

func (d *RooCodeDetector) Classify(path string, repoRoot string) ([]*DetectedItem, error)

func (*RooCodeDetector) Patterns 

func (d *RooCodeDetector) Patterns() []DetectionPattern

func (*RooCodeDetector) ProviderSlug 

func (d *RooCodeDetector) ProviderSlug() string

type ScanAsConfig 

type ScanAsConfig struct {
	ScanAs []ScanAsEntry `yaml:"scan-as"`
}

ScanAsConfig is the parsed content of .syllago.yaml.

func LoadScanAsConfig 

func LoadScanAsConfig(repoRoot string) (*ScanAsConfig, error)

LoadScanAsConfig reads .syllago.yaml from repoRoot. Returns an empty config if the file does not exist.

func (*ScanAsConfig) ToPathMap 

func (c *ScanAsConfig) ToPathMap() map[string]catalog.ContentType

ToPathMap converts ScanAsConfig entries to the map format used by AnalysisConfig.

type ScanAsEntry 

type ScanAsEntry struct {
	Type catalog.ContentType `yaml:"type"`
	Path string              `yaml:"path"`
}

ScanAsEntry is one path-to-type mapping in .syllago.yaml.

type SignalEntry 

type SignalEntry struct {
	Signal string
	Weight float64
}

SignalEntry records one matched signal and its weight contribution. Exported for audit logging; populated by ContentSignalDetector.

type SkipEntry 

type SkipEntry struct {
	Path   string `json:"path"`
	Reason string `json:"reason"` // pre_filter_excluded, below_threshold, locked_conflict, walk_skipped
}

SkipEntry records why a file was not classified by the content-signal detector.

type SyllagoDetector 

type SyllagoDetector struct{}

SyllagoDetector detects content organized in syllago's canonical layout.

func (*SyllagoDetector) Classify 

func (d *SyllagoDetector) Classify(path string, repoRoot string) ([]*DetectedItem, error)

func (*SyllagoDetector) Patterns 

func (d *SyllagoDetector) Patterns() []DetectionPattern

func (*SyllagoDetector) ProviderSlug 

func (d *SyllagoDetector) ProviderSlug() string

type TopLevelDetector 

type TopLevelDetector struct{}

TopLevelDetector detects provider-agnostic content in top-level directories.

func (*TopLevelDetector) Classify 

func (d *TopLevelDetector) Classify(path string, repoRoot string) ([]*DetectedItem, error)

func (*TopLevelDetector) Patterns 

func (d *TopLevelDetector) Patterns() []DetectionPattern

func (*TopLevelDetector) ProviderSlug 

func (d *TopLevelDetector) ProviderSlug() string

type ValidationIssue 

type ValidationIssue struct {
	ItemName     string
	DeclaredType string
	DetectedType catalog.ContentType
	Path         string
	Severity     string // "warning", "error"
	Message      string
}

ValidationIssue describes a disagreement between registry.yaml and file content.

func ValidateManifest 

func ValidateManifest(m *registry.Manifest, repoRoot string) []ValidationIssue

ValidateManifest cross-checks an authored registry.yaml against actual file content.

type WalkResult 

type WalkResult struct {
	Paths    []string // all file paths relative to root
	Warnings []string
}

WalkResult holds all file paths collected during a walk.

func Walk 

func Walk(root string, extraExcludeDirs []string) WalkResult

Walk collects all non-excluded file paths under root. extraExcludeDirs are appended to defaultExcludeDirs. root must already be filepath.EvalSymlinks-resolved.

type WindsurfDetector 

type WindsurfDetector struct{}

WindsurfDetector detects Windsurf content.

func (*WindsurfDetector) Classify 

func (d *WindsurfDetector) Classify(path string, repoRoot string) ([]*DetectedItem, error)

func (*WindsurfDetector) Patterns 

func (d *WindsurfDetector) Patterns() []DetectionPattern

func (*WindsurfDetector) ProviderSlug 

func (d *WindsurfDetector) ProviderSlug() string

Source Files

View all Source files

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.