creator-hub/internal/platform/douyin/share.go

package douyin

import (
	"bytes"
	"encoding/json"
)

const routerDataMarker = "window._ROUTER_DATA"

// ParseShareTargetHTML extracts only the author identity needed to add a
// competitor. It intentionally ignores the work's playback and engagement data.
func ParseShareTargetHTML(body []byte, expectedWorkKey string) (TargetProfile, bool) {
	if len(body) == 0 || len(body) > 2<<20 || !workKeyPattern.MatchString(expectedWorkKey) {
		return TargetProfile{}, false
	}
	root, ok := extractRouterData(body)
	if !ok {
		return TargetProfile{}, false
	}
	loaderData, ok := root["loaderData"].(map[string]any)
	if !ok {
		return TargetProfile{}, false
	}
	page, ok := loaderData["video_(id)/page"].(map[string]any)
	if !ok {
		return TargetProfile{}, false
	}
	videoInfo, ok := page["videoInfoRes"].(map[string]any)
	if !ok {
		return TargetProfile{}, false
	}
	items, ok := videoInfo["item_list"].([]any)
	if !ok || len(items) == 0 {
		return TargetProfile{}, false
	}
	item, ok := items[0].(map[string]any)
	if !ok {
		return TargetProfile{}, false
	}
	if workKey, ok := stringValue(item["aweme_id"]); ok && workKey != expectedWorkKey {
		return TargetProfile{}, false
	}
	author, ok := item["author"].(map[string]any)
	if !ok {
		return TargetProfile{}, false
	}
	return parseShareAuthor(author)
}

func extractRouterData(body []byte) (map[string]any, bool) {
	marker := []byte(routerDataMarker)
	for offset := 0; ; {
		index := bytes.Index(body[offset:], marker)
		if index < 0 {
			return nil, false
		}
		index += offset + len(marker)
		equals := bytes.IndexByte(body[index:], '=')
		if equals < 0 {
			return nil, false
		}
		start := index + equals + 1
		for start < len(body) && (body[start] == ' ' || body[start] == '\n' || body[start] == '\r' || body[start] == '\t') {
			start++
		}
		if start >= len(body) || body[start] != '{' {
			offset = index
			continue
		}
		end, ok := jsonObjectEnd(body, start)
		if !ok {
			return nil, false
		}
		var root map[string]any
		if json.Unmarshal(body[start:end], &root) == nil {
			return root, true
		}
		offset = end
	}
}

func jsonObjectEnd(body []byte, start int) (int, bool) {
	depth := 0
	inString := false
	escaped := false
	for index := start; index < len(body); index++ {
		char := body[index]
		if inString {
			if escaped {
				escaped = false
			} else if char == '\\' {
				escaped = true
			} else if char == '"' {
				inString = false
			}
			continue
		}
		switch char {
		case '"':
			inString = true
		case '{':
			depth++
		case '}':
			depth--
			if depth == 0 {
				return index + 1, true
			}
		}
	}
	return 0, false
}

func parseShareAuthor(author map[string]any) (TargetProfile, bool) {
	secUID, ok := stringValue(author["sec_uid"])
	if !ok || !keyPattern.MatchString(secUID) {
		return TargetProfile{}, false
	}
	uid := stringOrEmpty(author["uid"])
	uniqueID := stringOrEmpty(author["unique_id"])
	if uid != "" && !keyPattern.MatchString(uid) || uniqueID != "" && !keyPattern.MatchString(uniqueID) {
		return TargetProfile{}, false
	}
	return TargetProfile{
		UID:       uid,
		SecUID:    secUID,
		UniqueID:  uniqueID,
		Nickname:  stringOrEmpty(author["nickname"]),
		AvatarURL: avatarURL(author),
	}, true
}

func stringValue(value any) (string, bool) {
	result, ok := value.(string)
	return result, ok && result != ""
}

func stringOrEmpty(value any) string {
	result, _ := value.(string)
	return result
}

func avatarURL(author map[string]any) string {
	for _, key := range []string{"avatar_thumb", "avatar_larger", "avatar_medium", "avatar"} {
		value := author[key]
		if direct, ok := value.(string); ok && direct != "" {
			return direct
		}
		container, ok := value.(map[string]any)
		if !ok {
			continue
		}
		urls, ok := container["url_list"].([]any)
		if !ok || len(urls) == 0 {
			continue
		}
		if result, ok := urls[0].(string); ok {
			return result
		}
	}
	return ""
}