database

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

type BunDocument struct {
	bun.BaseModel `bun:"table:documents,alias:d"`

	ID             int       `bun:"id,pk,autoincrement"`
	Name           string    `bun:"name,notnull"`
	Path           string    `bun:"path,notnull,unique"`
	IngressTime    time.Time `bun:"ingress_time,notnull,default:current_timestamp"`
	Folder         string    `bun:"folder,notnull"`
	Hash           string    `bun:"hash,notnull"`
	ULID           string    `bun:"ulid,notnull,unique"` // Stored as string in DB
	DocumentType   string    `bun:"document_type,notnull"`
	FullText       string    `bun:"full_text,nullzero"`
	URL            string    `bun:"url,nullzero"`
	FullTextSearch string    `bun:"full_text_search,type:tsvector,nullzero"` // PostgreSQL-specific
	CreatedAt      time.Time `bun:"created_at,notnull,default:current_timestamp"`
	UpdatedAt      time.Time `bun:"updated_at,notnull,default:current_timestamp"`
}

type BunJob struct {
	bun.BaseModel `bun:"table:jobs,alias:j"`

	ID          string     `bun:"id,pk"` // ULID as string
	Type        string     `bun:"type,notnull"`
	Status      string     `bun:"status,default:'pending'"`
	Progress    int        `bun:"progress,default:0"`
	CurrentStep string     `bun:"current_step,default:''"`
	TotalSteps  int        `bun:"total_steps,default:0"`
	Message     string     `bun:"message,default:''"`
	Error       string     `bun:"error,nullzero"`
	Result      string     `bun:"result,nullzero"`
	CreatedAt   time.Time  `bun:"created_at,notnull,default:current_timestamp"`
	UpdatedAt   time.Time  `bun:"updated_at,notnull,default:current_timestamp"`
	StartedAt   *time.Time `bun:"started_at,nullzero"`
	CompletedAt *time.Time `bun:"completed_at,nullzero"`
}

type BunServerConfig struct {
	bun.BaseModel `bun:"table:server_config,alias:sc"`

	ID                   int       `bun:"id,pk"`
	ListenAddrIP         string    `bun:"listen_addr_ip,default:''"`
	ListenAddrPort       string    `bun:"listen_addr_port,notnull,default:'8000'"`
	IngressPath          string    `bun:"ingress_path,notnull,default:''"`
	IngressDelete        bool      `bun:"ingress_delete,notnull,default:false"`
	IngressMoveFolder    string    `bun:"ingress_move_folder,notnull,default:''"`
	IngressPreserve      bool      `bun:"ingress_preserve,notnull,default:true"`
	DocumentPath         string    `bun:"document_path,notnull,default:''"`
	NewDocumentFolder    string    `bun:"new_document_folder,default:''"`
	NewDocumentFolderRel string    `bun:"new_document_folder_rel,default:''"`
	WebUIPass            bool      `bun:"web_ui_pass,notnull,default:false"`
	ClientUsername       string    `bun:"client_username,default:''"`
	ClientPassword       string    `bun:"client_password,default:''"`
	PushBulletToken      string    `bun:"pushbullet_token,default:''"`
	TesseractPath        string    `bun:"tesseract_path,default:''"`
	UseReverseProxy      bool      `bun:"use_reverse_proxy,notnull,default:false"`
	BaseURL              string    `bun:"base_url,default:''"`
	IngressInterval      int       `bun:"ingress_interval,notnull,default:10"`
	NewDocumentNumber    int       `bun:"new_document_number,notnull,default:5"`
	ServerAPIURL         string    `bun:"server_api_url,default:''"`
	CreatedAt            time.Time `bun:"created_at,notnull,default:current_timestamp"`
	UpdatedAt            time.Time `bun:"updated_at,notnull,default:current_timestamp"`
}

type BunWordCloudMetadata struct {
	bun.BaseModel `bun:"table:word_cloud_metadata,alias:wcm"`

	ID                  int        `bun:"id,pk"`
	LastFullCalculation *time.Time `bun:"last_full_calculation,nullzero"`
	TotalDocsProcessed  int        `bun:"total_documents_processed,default:0"`
	TotalWordsIndexed   int        `bun:"total_words_indexed,default:0"`
	Version             int        `bun:"version,default:1"`
	CreatedAt           time.Time  `bun:"created_at,notnull,default:current_timestamp"`
	UpdatedAt           time.Time  `bun:"updated_at,notnull,default:current_timestamp"`
}

type BunWordFrequency struct {
	bun.BaseModel `bun:"table:word_frequencies,alias:wf"`

	Word        string    `bun:"word,pk"`
	Frequency   int       `bun:"frequency,default:1"`
	LastUpdated time.Time `bun:"last_updated,default:current_timestamp"`
}

type Document struct {
	StormID      int // ID field (kept as StormID for backward compatibility)
	Name         string
	Path         string // full path to the file
	IngressTime  time.Time
	Folder       string
	Hash         string
	ULID         ulid.ULID // Have a smaller (than hash) id that can be used in URL's, hopefully speed things up
	DocumentType string    // type of document (pdf, txt, etc)
	FullText     string
	URL          string
}

type EphemeralPostgresDB struct {
	*PostgresDB
	// contains filtered or unexported fields
}

type Job struct {
	ID          ulid.ULID  `json:"id"`
	Type        JobType    `json:"type"`
	Status      JobStatus  `json:"status"`
	Progress    int        `json:"progress"`         // 0-100
	CurrentStep string     `json:"currentStep"`      // Human-readable current step
	TotalSteps  int        `json:"totalSteps"`       // Total number of steps
	Message     string     `json:"message"`          // Status message
	Error       string     `json:"error,omitempty"`  // Error message if failed
	Result      string     `json:"result,omitempty"` // JSON result data
	CreatedAt   time.Time  `json:"createdAt"`
	UpdatedAt   time.Time  `json:"updatedAt"`
	StartedAt   *time.Time `json:"startedAt,omitempty"`
	CompletedAt *time.Time `json:"completedAt,omitempty"`
}

type JobStatus string

type JobSummary struct {
	FilesProcessed int    `json:"filesProcessed"`
	FilesTotal     int    `json:"filesTotal"`
	BytesProcessed int64  `json:"bytesProcessed"`
	Errors         int    `json:"errors"`
	Details        string `json:"details,omitempty"`
}

type JobType string

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

type Repository interface {
	Close() error
	SaveDocument(doc *Document) error
	GetDocumentByID(id int) (*Document, error)
	GetDocumentByULID(ulid string) (*Document, error)
	GetDocumentByPath(path string) (*Document, error)
	GetDocumentByHash(hash string) (*Document, error)
	GetNewestDocuments(limit int) ([]Document, error)
	GetNewestDocumentsWithPagination(page int, pageSize int) ([]Document, int, error)
	GetAllDocuments() ([]Document, error)
	GetDocumentsByFolder(folder string) ([]Document, error)
	DeleteDocument(ulid string) error
	UpdateDocumentURL(ulid string, url string) error
	UpdateDocumentFolder(ulid string, folder string) error
	SaveConfig(config *config.ServerConfig) error
	GetConfig() (*config.ServerConfig, error)
	SearchDocuments(searchTerm string) ([]Document, error)
	ReindexSearchDocuments() (int, error)
	// Word cloud methods
	GetTopWords(limit int) ([]WordFrequency, error)
	GetWordCloudMetadata() (*WordCloudMetadata, error)
	RecalculateAllWordFrequencies() error
	UpdateWordFrequencies(docID string) error
	// Job tracking methods
	CreateJob(jobType JobType, message string) (*Job, error)
	UpdateJobProgress(jobID ulid.ULID, progress int, currentStep string) error
	UpdateJobStatus(jobID ulid.ULID, status JobStatus, message string) error
	UpdateJobError(jobID ulid.ULID, errorMsg string) error
	CompleteJob(jobID ulid.ULID, result string) error
	GetJob(jobID ulid.ULID) (*Job, error)
	GetRecentJobs(limit, offset int) ([]Job, error)
	GetActiveJobs() ([]Job, error)
	DeleteOldJobs(olderThan time.Duration) (int, error)
}

type WordCloudMetadata struct {
	LastCalculation    time.Time `json:"lastCalculation"`
	TotalDocsProcessed int       `json:"totalDocsProcessed"`
	TotalWordsIndexed  int       `json:"totalWordsIndexed"`
	Version            int       `json:"version"`
}

type WordFrequency struct {
	Word      string    `json:"word"`
	Frequency int       `json:"frequency"`
	Updated   time.Time `json:"updated"`
}

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