database

type AnalyzedArticle struct {
	URL     string      `gorm:"column:url" json:"url"`
	Title   *string     `gorm:"column:title" json:"title,omitempty"`
	Rating  *float64    `gorm:"column:rating" json:"rating,omitempty"`
	Authors StringArray `gorm:"column:authors;type:text[]" json:"authors,omitempty"`
	PubDate time.Time   `gorm:"column:pub_date" json:"pub_date"`
}

type AnalyzedItem struct {
	Hash               string           `json:"hash"`
	URL                string           `json:"url"`
	ThinkResult        *ThinkResult     `gorm:"type:jsonb"`
	Sentiments         *SentimentScores `gorm:"type:jsonb" json:"sentiments,omitempty"`
	SentimentsDeframed *SentimentScores `gorm:"type:jsonb" json:"sentiments_deframed,omitempty"`
	MediaContent       *MediaContent    `gorm:"type:jsonb" json:"media,omitempty"`
	ThinkRating        float64          `json:"rating"`
	Authors            StringArray      `gorm:"type:text[]" json:"authors,omitempty"`
	PubDate            time.Time        `json:"pubDate"`
}

type Base struct {
	ID        uuid.UUID      `gorm:"primary_key;type:uuid;default:uuid_generate_v4()"`
	CreatedAt time.Time      `gorm:"not null;default:now()"`
	UpdatedAt time.Time      `gorm:"not null;default:now()"`
	DeletedAt gorm.DeletedAt `gorm:"index"`
}

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

type DomainComparison struct {
	Classification string  `gorm:"column:classification" json:"classification"`
	RankGroup      int     `gorm:"column:rank_group" json:"rank_group"`
	TrendTopic     string  `gorm:"column:trend_topic" json:"trend_topic"`
	ScoreA         float64 `gorm:"column:score_a" json:"score_a"`
	ScoreB         float64 `gorm:"column:score_b" json:"score_b"`
}

type Feed struct {
	Base
	URL               string        `gorm:"index"`                 // we can't enforce uniqueness here (because of the soft deletes)
	RootDomain        *string       `gorm:"index"`                 // example.com
	PortalUrl         *string       `gorm:"type:text"`             // web.foo.bar/whatever
	Language          *string       `gorm:"type:char(2)"`          // ISO 639-1 language code
	EnforceFeedDomain bool          `gorm:"not null;default:true"` // item url must be from our URL
	Enabled           bool          `gorm:"not null;default:false;index"`
	Polling           bool          `gorm:"not null;default:false"`
	Mining            bool          `gorm:"not null;default:false"`
	ResolveItemUrl    bool          `gorm:"not null;default:false"`
	LastSyncedAt      *time.Time    `gorm:"index"`
	Categories        StringArray   `gorm:"type:text[];not null;default:'{}'"`
	FeedSchedule      *FeedSchedule `gorm:"foreignKey:ID;references:ID"`
}

type FeedSchedule struct {
	ID                 uuid.UUID  `gorm:"primaryKey;type:uuid"` // this is a FK to Feed.ID
	CreatedAt          time.Time  `gorm:"not null;default:now()"`
	UpdatedAt          time.Time  `gorm:"not null;default:now()"`
	NextThinkerAt      *time.Time `gorm:"index"`
	ThinkerLockedUntil *time.Time
	NextMiningAt       *time.Time `gorm:"index"`
	MiningLockedUntil  *time.Time
}

type Item struct {
	ID              uuid.UUID     `gorm:"primaryKey;type:uuid"`
	CreatedAt       time.Time     `gorm:"not null;default:now()"`
	UpdatedAt       time.Time     `gorm:"not null;default:now()"`
	Hash            string        `gorm:"type:char(64);uniqueIndex:idx_hash_feed_url;uniqueIndex:idx_hash_feed;not null"`
	FeedID          uuid.UUID     `gorm:"type:uuid;index;uniqueIndex:idx_feed_url;uniqueIndex:idx_hash_feed_url;uniqueIndex:idx_hash_feed;not null"`
	Feed            Feed          `gorm:"foreignKey:FeedID;references:ID;constraint:OnUpdate:CASCADE,OnDelete:CASCADE;"`
	URL             string        `gorm:"index;uniqueIndex:idx_feed_url;uniqueIndex:idx_hash_feed_url;not null"`
	Language        *string       `gorm:"type:char(2)"` // ISO 639-1 language code
	Content         string        `gorm:"type:text;not null"`
	PubDate         time.Time     `gorm:"not null;index;default:now()"`
	MediaContent    *MediaContent `gorm:"type:jsonb"`
	ThinkResult     *ThinkResult  `gorm:"type:jsonb"`
	ThinkError      *string       `gorm:"type:text;null"`
	ThinkErrorCount int           `gorm:"not null;default:0"`
	ThinkRating     float64       `gorm:"not null;default:0.0"`
	Categories      StringArray   `gorm:"type:text[];not null;default:'{}'"`
	Authors         StringArray   `gorm:"type:text[];not null;default:'{}'"`
}

type Lifecycle struct {
	TimeSlice time.Time `gorm:"column:time_slice" json:"time_slice"`
	Frequency int64     `gorm:"column:frequency" json:"frequency"`
	Velocity  int64     `gorm:"column:velocity" json:"velocity"`
}

type MediaContent struct {
	// Technical attributes required for <img src> or <video src>
	URL    string `xml:"url,attr" json:"url,omitempty"`
	Type   string `xml:"type,attr" json:"type,omitempty"`     // e.g., "image/jpeg"
	Medium string `xml:"medium,attr" json:"medium,omitempty"` // e.g., "image" or "video"

	// Dimensions are essential for preventing HTML Layout Shift (CLS)
	Height int `xml:"height,attr" json:"height,omitempty"`
	Width  int `xml:"width,attr" json:"width,omitempty"`

	// Descriptive text used for HTML 'alt' attributes and captions
	Title       string `xml:"title" json:"title,omitempty"`
	Description string `xml:"description" json:"description,omitempty"`

	// Essential for Video: Used for the <video poster="..."> attribute
	// We use a pointer so it is nil if no thumbnail exists
	Thumbnail *MediaThumbnail `xml:"thumbnail" json:"thumbnail,omitempty"`

	// Optional Copyright for the content
	Credit string `xml:"credit" json:"credit,omitempty"`
}

type MediaThumbnail struct {
	URL    string `xml:"url,attr" json:"url,omitempty"`
	Height int    `xml:"height,attr" json:"height,omitempty"`
	Width  int    `xml:"width,attr" json:"width,omitempty"`
}

type Repository interface {
	FindFeedByUrl(u *url.URL) (*Feed, error)
	FindFeedByUrlAndAvailability(u *url.URL, onlyEnabled bool) (*Feed, error)
	FindFeedById(feedID uuid.UUID) (*Feed, error)
	UpsertFeed(feed *Feed) error
	// FindItemsByUrl retrieves all items associated with a specific URL.
	// As per the specification (Syndication), a single URL can legitimately appear in multiple feeds.
	// Therefore, the system allows multiple Item records for the same URL, distinguished by their FeedID.
	// This means the same content (probably different Hashes) can exist multiple times if it is syndicated across different feeds.
	// Feed.EnforceFeedDomain = will enforce only items with the same base domain as the Feed URL
	FindItemsByUrl(u *url.URL) ([]Item, error)
	FindItemsByRootDomain(rootDomain string, limit int) ([]Item, error)
	GetAllFeeds(deleted bool) ([]Feed, error)
	DeleteFeedById(id uuid.UUID) error
	PurgeFeedById(id uuid.UUID) error
	EnqueueSync(id uuid.UUID, pollingInterval time.Duration) error
	EnqueueMine(id uuid.UUID, miningInterval time.Duration) error
	RemoveMine(id uuid.UUID) error
	RemoveSync(id uuid.UUID) error
	BeginFeedUpdate(lockDuration time.Duration) (*Feed, error)
	EndFeedUpdate(id uuid.UUID, jobErr error, pollingInterval time.Duration) error
	GetPendingItems(feedID uuid.UUID, hashes []string, maxRetries int) (map[string]int, error)
	GetItemsByHashes(feedID uuid.UUID, hashes []string) ([]Item, error)
	BeginThinkerBatch(limit int, lockDuration time.Duration) ([]Item, error)
	BeginThinkerFixerBatch(limit int, since time.Time, minErrorCount int, maxErrorCount int, lockDuration time.Duration) ([]Item, error)
	UpsertItem(item *Item) error
	UpsertItemWithTrendInvalidation(item *Item) error
	FindFeedScheduleById(feedID uuid.UUID) (*FeedSchedule, error)
	CreateFeedSchedule(feedID uuid.UUID) error
	GetTopTrendByDomain(domain string, language string, date *time.Time, days int) ([]TrendMetric, error)
	GetContextByDomain(term string, domain string, language string, date *time.Time, days int) ([]TrendContext, error)
	GetLifecycleByDomain(term string, domain string, language string, date *time.Time, days int) ([]Lifecycle, error)
	GetDomainComparison(domainA string, domainB string, language string, date *time.Time, days int, utilityThreshold float64, outlierRatioThreshold float64, limit int) ([]DomainComparison, error)
	GetArticlesByTrend(term string, domain string, date *time.Time, days int, offset int, limit int) ([]AnalyzedArticle, error)
	GetSentimentsByTrend(term string, domain string, date *time.Time, days int) (*SentimentItem, error)
	FindAnalyzedItemsByRootDomain(rootDomain string, limit int) ([]AnalyzedItem, error)
	FindFirstAnalyzedItemByUrl(u *url.URL) (*AnalyzedItem, error)
}

type Sentiment struct {
	Valence   float64 `json:"v,omitempty"`   // VAD: 1.0..9.0, 5.0 neutral
	Arousal   float64 `json:"a,omitempty"`   // VAD: 1.0..9.0, 5.0 neutral
	Dominance float64 `json:"d,omitempty"`   // VAD: 1.0..9.0, 5.0 neutral
	Joy       float64 `json:"j,omitempty"`   // BE5: 1.0..5.0, 1.0 absent/neutral
	Anger     float64 `json:"a_n,omitempty"` // BE5: 1.0..5.0, 1.0 absent/neutral
	Sadness   float64 `json:"s,omitempty"`   // BE5: 1.0..5.0, 1.0 absent/neutral
	Fear      float64 `json:"f,omitempty"`   // BE5: 1.0..5.0, 1.0 absent/neutral
	Disgust   float64 `json:"d_g,omitempty"` // BE5: 1.0..5.0, 1.0 absent/neutral
}

type SentimentItem struct {
	Sentiments         *SentimentScores `json:"sentiments,omitempty"`
	SentimentsDeframed *SentimentScores `json:"sentiments_deframed,omitempty"`
}

type SentimentScores struct {
	Valence   float64 `gorm:"column:valence" json:"valence,omitempty"`
	Arousal   float64 `gorm:"column:arousal" json:"arousal,omitempty"`
	Dominance float64 `gorm:"column:dominance" json:"dominance,omitempty"`
	Joy       float64 `gorm:"column:joy" json:"joy,omitempty"`
	Anger     float64 `gorm:"column:anger" json:"anger,omitempty"`
	Sadness   float64 `gorm:"column:sadness" json:"sadness,omitempty"`
	Fear      float64 `gorm:"column:fear" json:"fear,omitempty"`
	Disgust   float64 `gorm:"column:disgust" json:"disgust,omitempty"`
}

type StringArray []string

type ThinkResult struct {
	TitleOriginal               string  `json:"title_original,omitempty"`
	DescriptionOriginal         string  `json:"description_original,omitempty"`
	TitleCorrected              string  `json:"title_corrected,omitempty"`
	TitleCorrectionReason       string  `json:"title_correction_reason,omitempty"`
	DescriptionCorrected        string  `json:"description_corrected,omitempty"`
	DescriptionCorrectionReason string  `json:"description_correction_reason,omitempty"`
	Framing                     float64 `json:"framing,omitempty"`
	FramingReason               string  `json:"framing_reason,omitempty"`
	Clickbait                   float64 `json:"clickbait,omitempty"`
	ClickbaitReason             string  `json:"clickbait_reason,omitempty"`
	Persuasive                  float64 `json:"persuasive,omitempty"`
	PersuasiveReason            string  `json:"persuasive_reason,omitempty"`
	HyperStimulus               float64 `json:"hyper_stimulus,omitempty"`
	HyperStimulusReason         string  `json:"hyper_stimulus_reason,omitempty"`
	Speculative                 float64 `json:"speculative,omitempty"`
	SpeculativeReason           string  `json:"speculative_reason,omitempty"`
	Overall                     float64 `json:"overall,omitempty"`
	OverallReason               string  `json:"overall_reason,omitempty"`
	Category                    string  `json:"category,omitempty"`
}

type Trend struct {
	ItemID             uuid.UUID   `gorm:"primaryKey;type:uuid"` // FK to ItemID
	FeedID             uuid.UUID   `gorm:"type:uuid;not null"`   // FK to FeedID
	Language           string      `gorm:"not null"`
	PubDate            time.Time   `gorm:"not null"`
	CategoryStems      StringArray `gorm:"type:text[]"`
	NounStems          StringArray `gorm:"type:text[]"`
	VerbStems          StringArray `gorm:"type:text[]"`
	AdjectiveStems     StringArray `gorm:"type:text[]"`
	RootDomain         string      `gorm:"not null"`
	Sentiments         *Sentiment  `gorm:"type:jsonb;not null;default:'{}'"`
	SentimentsDeframed *Sentiment  `gorm:"type:jsonb;not null;default:'{}'"`
}

type TrendContext struct {
	Context   string `gorm:"column:context_word" json:"context"`
	Frequency int64  `gorm:"column:frequency" json:"frequency"`
}

type TrendMetric struct {
	TrendTopic   string    `gorm:"column:trend_topic" json:"trend_topic"`
	Frequency    int64     `gorm:"column:frequency" json:"frequency"`
	Utility      int64     `gorm:"column:utility" json:"utility"`
	OutlierRatio float64   `gorm:"column:outlier_ratio" json:"outlier_ratio"`
	TimeSlice    time.Time `gorm:"column:time_slice" json:"time_slice"`
}