proximity

package
v1.2.1 Latest Latest
Warning

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

 Go to latest Published: Jul 1, 2025 License: MIT Imports: 4 Imported by: 0

Documentation

Overview

Package proximity implements proximity-based PI detection enhancement. It analyzes surrounding text for PI-related keywords and patterns to increase confidence in PI detection when relevant context is found nearby.

Index

Examples

Constants

This section is empty.

Variables

This section is empty.

Functions

This section is empty.

Types

type ContextAnalyzer 

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

ContextAnalyzer provides utilities for analyzing the context around potential PI matches

func NewContextAnalyzer 

func NewContextAnalyzer() *ContextAnalyzer

NewContextAnalyzer creates a new context analyzer

func (*ContextAnalyzer) AnalyzeSemanticContext 

func (ca *ContextAnalyzer) AnalyzeSemanticContext(content string, startIndex, endIndex int) SemanticAnalysis

AnalyzeSemanticContext performs semantic analysis of the context

func (*ContextAnalyzer) AnalyzeStructure 

func (ca *ContextAnalyzer) AnalyzeStructure(content string, startIndex, endIndex int) StructureAnalysis

AnalyzeStructure analyzes the structure type of content around a match

func (*ContextAnalyzer) CalculateContextWindow 

func (ca *ContextAnalyzer) CalculateContextWindow(content string, startIndex, endIndex, baseWindow int) (beforeWindow, afterWindow int)

CalculateContextWindow calculates the optimal context window size based on content

func (*ContextAnalyzer) CountKeywords 

func (ca *ContextAnalyzer) CountKeywords(text string, keywords []string) int

CountKeywords counts occurrences of keywords in the text as whole words

func (*ContextAnalyzer) ExtractSurroundingText 

func (ca *ContextAnalyzer) ExtractSurroundingText(content string, startIndex, endIndex, windowSize int) (before, after string)

ExtractSurroundingText extracts text before and after a match within a specified window

func (*ContextAnalyzer) FindNearestKeyword 

func (ca *ContextAnalyzer) FindNearestKeyword(content string, keywords []string, startIndex, endIndex int) (keyword string, distance int, found bool)

FindNearestKeyword finds the nearest keyword to a match position

func (*ContextAnalyzer) GetWordProximity 

func (ca *ContextAnalyzer) GetWordProximity(content, targetWord string, startIndex, endIndex int) (distance int, found bool)

GetWordProximity calculates the word distance between a target word and a match position

type PIContextInfo 

type PIContextInfo struct {
	Type     PIContextType `json:"type"`
	Keywords []string      `json:"keywords"`
	Distance int           `json:"distance"`
}

PIContextInfo contains information about detected PI context

type PIContextType 

type PIContextType string

PIContextType represents different types of PI context 

const (
	PIContextLabel         PIContextType = "label"         // "SSN:", "Tax File Number:"
	PIContextForm          PIContextType = "form"          // HTML forms, input fields
	PIContextDatabase      PIContextType = "database"      // SQL queries, database operations
	PIContextLog           PIContextType = "log"           // Log entries
	PIContextConfig        PIContextType = "config"        // Configuration files
	PIContextVariable      PIContextType = "variable"      // Variable assignments
	PIContextDocumentation PIContextType = "documentation" // Comments, docs
	PIContextProduction    PIContextType = "production"    // Regular production code
	PIContextTest          PIContextType = "test"          // Test/mock/sample data
)

type PatternMatcher 

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

PatternMatcher provides methods to identify various patterns that indicate PI context vs test data

func NewPatternMatcher 

func NewPatternMatcher() *PatternMatcher

NewPatternMatcher creates a new pattern matcher with compiled patterns

func (*PatternMatcher) AnalyzeContextType 

func (pm *PatternMatcher) AnalyzeContextType(text string) PIContextType

AnalyzeContextType determines the most likely context type

func (*PatternMatcher) ContainsTestDataKeywords 

func (pm *PatternMatcher) ContainsTestDataKeywords(text string) bool

ContainsTestDataKeywords checks if the text contains keywords that indicate test/mock data

func (*PatternMatcher) ExtractRelevantKeywords 

func (pm *PatternMatcher) ExtractRelevantKeywords(text string) []string

ExtractRelevantKeywords extracts keywords that are relevant for context analysis

func (*PatternMatcher) FindPIContextLabels 

func (pm *PatternMatcher) FindPIContextLabels(text string) []string

FindPIContextLabels finds PI context labels in the text

func (*PatternMatcher) GetPIContextConfidence 

func (pm *PatternMatcher) GetPIContextConfidence(text string) float64

GetPIContextConfidence returns a confidence score for PI context detection

func (*PatternMatcher) GetTestDataConfidence 

func (pm *PatternMatcher) GetTestDataConfidence(text string) float64

GetTestDataConfidence returns a confidence score for test data detection

func (*PatternMatcher) IsConfigurationContext 

func (pm *PatternMatcher) IsConfigurationContext(text string) bool

IsConfigurationContext checks if the text appears to be configuration related

func (*PatternMatcher) IsDatabaseContext 

func (pm *PatternMatcher) IsDatabaseContext(text string) bool

IsDatabaseContext checks if the text appears to be database query related

func (*PatternMatcher) IsDocumentationContext 

func (pm *PatternMatcher) IsDocumentationContext(text string) bool

IsDocumentationContext checks if the text appears to be documentation/comments

func (*PatternMatcher) IsFormFieldContext 

func (pm *PatternMatcher) IsFormFieldContext(text string) bool

IsFormFieldContext checks if the text appears to be form field related

func (*PatternMatcher) IsLogContext 

func (pm *PatternMatcher) IsLogContext(text string) bool

IsLogContext checks if the text appears to be log entry related

func (*PatternMatcher) IsVariableContext 

func (pm *PatternMatcher) IsVariableContext(text string) bool

IsVariableContext checks if the text appears to be variable assignment related

type ProximityDetector 

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

ProximityDetector analyzes the context around potential PI to improve detection accuracy

Example

ExampleProximityDetector demonstrates how to use the proximity detector 

detector := NewProximityDetector()

// Example 1: Test data
content1 := "// Test TFN: 123 456 789 for unit testing"
result1 := detector.AnalyzeContext(content1, "123 456 789", 12, 23)
fmt.Printf("Test data - Score: %.2f, Is Test: %t, Reason: %s\n",
	result1.Score, result1.IsTestData, result1.Reason)

// Example 2: Real PI with label
content2 := "Customer TFN: 123 456 789"
result2 := detector.AnalyzeContext(content2, "123 456 789", 14, 25)
fmt.Printf("Real PI - Score: %.2f, Is Test: %t, Reason: %s\n",
	result2.Score, result2.IsTestData, result2.Reason)

// Example 3: Form field
content3 := `<input type="text" name="tfn" value="123 456 789">`
result3 := detector.AnalyzeContext(content3, "123 456 789", 37, 48)
fmt.Printf("Form field - Score: %.2f, Context: %s, Reason: %s\n",
	result3.Score, result3.Context, result3.Reason)

Output:
Test data - Score: 0.10, Is Test: true, Reason: test data indicator
Real PI - Score: 0.90, Is Test: false, Reason: PI context label detected
Form field - Score: 0.80, Context: form, Reason: form field context

func NewProximityDetector 

func NewProximityDetector() *ProximityDetector

NewProximityDetector creates a new proximity detector

func (*ProximityDetector) AnalyzeContext 

func (pd *ProximityDetector) AnalyzeContext(content, match string, startIndex, endIndex int) ProximityResult

AnalyzeContext performs comprehensive context analysis around a potential PI match

func (*ProximityDetector) AnalyzeFile 

func (pd *ProximityDetector) AnalyzeFile(filename string, content string, findings []detection.Finding) []detection.Finding

AnalyzeFile performs proximity analysis on an entire file's findings

func (*ProximityDetector) CalculateProximityScore 

func (pd *ProximityDetector) CalculateProximityScore(distance int, contextType PIContextType) float64

CalculateProximityScore calculates a proximity score based on distance and context type

func (*ProximityDetector) EnhanceFinding 

func (pd *ProximityDetector) EnhanceFinding(finding *detection.Finding, content string)

EnhanceFinding enhances a detection finding with proximity analysis

func (*ProximityDetector) IdentifyPIContext 

func (pd *ProximityDetector) IdentifyPIContext(content, match string, startIndex, endIndex int) PIContextInfo

IdentifyPIContext identifies the type of context where PI appears

func (*ProximityDetector) IsTestData 

func (pd *ProximityDetector) IsTestData(filename, content, match string, startIndex, endIndex int) bool

IsTestData determines if the match appears to be test/mock/sample data

type ProximityResult 

type ProximityResult struct {
	Score      float64           `json:"score"`
	Reason     string            `json:"reason"`
	Context    PIContextType     `json:"context"`
	Keywords   []string          `json:"keywords"`
	Structure  StructureAnalysis `json:"structure"`
	Semantic   SemanticAnalysis  `json:"semantic"`
	IsTestData bool              `json:"is_test_data"`
}

ProximityResult represents the result of proximity context analysis

type SemanticAnalysis 

type SemanticAnalysis struct {
	Confidence float64  `json:"confidence"`
	Indicators []string `json:"indicators"`
	PITypes    []string `json:"pi_types"`
}

SemanticAnalysis contains semantic context information

type StructureAnalysis 

type StructureAnalysis struct {
	Type         StructureType `json:"type"`
	NestingLevel int           `json:"nesting_level"`
	ElementType  string        `json:"element_type,omitempty"`
}

StructureAnalysis contains information about content structure

type StructureType 

type StructureType string

StructureType represents different content structure types 

const (
	StructureJSON      StructureType = "json"
	StructureXML       StructureType = "xml"
	StructureHTML      StructureType = "html"
	StructureSQL       StructureType = "sql"
	StructureYAML      StructureType = "yaml"
	StructureCode      StructureType = "code"
	StructurePlainText StructureType = "plain_text"
	StructureURL       StructureType = "url"
)

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.