Goondex/internal/scraper/merger/performer_merger.go

package merger

import (
	"fmt"
	"strings"

	"git.leaktechnologies.dev/stu/Goondex/internal/model"
	"git.leaktechnologies.dev/stu/Goondex/internal/scraper/adultemp"
)

// MergePerformerData intelligently combines data from multiple sources
// Priority: TPDB data is primary, Adult Empire fills in gaps or provides additional context
func MergePerformerData(tpdbPerformer *model.Performer, adultempData *adultemp.PerformerData) *model.Performer {
	merged := tpdbPerformer

	// Fill in missing fields from Adult Empire
	if merged.Birthday == "" && adultempData.Birthdate != "" {
		merged.Birthday = adultempData.Birthdate
	}

	if merged.Ethnicity == "" && adultempData.Ethnicity != "" {
		merged.Ethnicity = adultempData.Ethnicity
	}

	if merged.Country == "" && adultempData.Country != "" {
		merged.Country = adultempData.Country
	}

	if merged.HairColor == "" && adultempData.HairColor != "" {
		merged.HairColor = adultempData.HairColor
	}

	if merged.EyeColor == "" && adultempData.EyeColor != "" {
		merged.EyeColor = adultempData.EyeColor
	}

	if merged.Measurements == "" && adultempData.Measurements != "" {
		merged.Measurements = adultempData.Measurements
	}

	// Height: prefer TPDB if available, otherwise use Adult Empire
	if merged.Height == 0 && adultempData.Height != "" {
		// Parse height from Adult Empire format (e.g., "168 cm")
		// This is already converted by the Adult Empire scraper
		// We just need to extract the numeric value
		var height int
		if _, err := fmt.Sscanf(adultempData.Height, "%d cm", &height); err == nil {
			merged.Height = height
		}
	}

	// Bio: Combine if both exist, otherwise use whichever is available
	if merged.Bio == "" && adultempData.Biography != "" {
		merged.Bio = adultempData.Biography
	} else if merged.Bio != "" && adultempData.Biography != "" {
		// If both exist and are different, append Adult Empire bio
		if !strings.Contains(merged.Bio, adultempData.Biography) {
			merged.Bio = merged.Bio + "\n\n[Adult Empire]: " + adultempData.Biography
		}
	}

	// Aliases: Merge unique aliases
	if len(adultempData.Aliases) > 0 {
		aliasesStr := strings.Join(adultempData.Aliases, ", ")
		if merged.Aliases == "" {
			merged.Aliases = aliasesStr
		} else {
			// Add new aliases that aren't already present
			existingAliases := strings.Split(merged.Aliases, ",")
			existingMap := make(map[string]bool)
			for _, alias := range existingAliases {
				existingMap[strings.TrimSpace(alias)] = true
			}

			for _, newAlias := range adultempData.Aliases {
				trimmed := strings.TrimSpace(newAlias)
				if !existingMap[trimmed] {
					merged.Aliases += ", " + trimmed
				}
			}
		}
	}

	// Image URL: prefer TPDB, but keep Adult Empire as fallback reference
	// We don't override TPDB images as they're generally higher quality
	if merged.ImageURL == "" && adultempData.Image != "" {
		merged.ImageURL = adultempData.Image
	}

	return merged
}

// ShouldMerge determines if two performers are likely the same person
// Returns true if names match closely enough
func ShouldMerge(performer1Name, performer2Name string) bool {
	name1 := strings.ToLower(strings.TrimSpace(performer1Name))
	name2 := strings.ToLower(strings.TrimSpace(performer2Name))

	// Exact match
	if name1 == name2 {
		return true
	}

	// Check if one name is contained in the other
	// (e.g., "Riley Reid" and "Riley Red" should not match,
	//  but "Riley Reid" and "Reid, Riley" should)
	words1 := strings.Fields(name1)
	words2 := strings.Fields(name2)

	// If all words from one name are in the other, consider it a match
	matchCount := 0
	for _, word1 := range words1 {
		for _, word2 := range words2 {
			if word1 == word2 {
				matchCount++
				break
			}
		}
	}

	// At least 70% of words must match
	threshold := 0.7
	maxWords := len(words1)
	if len(words2) > maxWords {
		maxWords = len(words2)
	}

	return float64(matchCount)/float64(maxWords) >= threshold
}