cli

type AppConfig struct {
	AQLQuery             string
	ArangoUser           string // For authorship in gRPC updates
	NumProcessingWorkers int
	NumGrpcWorkers       int
	Logger               *logrus.Entry
	Metrics              *ProcessingMetrics // Add this field
}

type ArangoProperty struct {
	Name  string `json:"name"`
	Value string `json:"value"`
}

type ArangoResultDoc struct {
	ID    string           `json:"id"` // This is dbx.accession, likely the feature_id
	Props []ArangoProperty `json:"props"`
}

type BatchGeneProductJob struct {
	GeneProducts []ProcessedGeneProduct
}

type BatchGeneProductResult struct {
	GeneID         string
	Success        bool
	Message        string
	Error          error
	ProcessedCount int
	SkippedCount   int
}

type CSVFileConstraints struct {
	MaxFileSizeBytes int64
	MaxRecords       int
}

type CSVRecordWriter interface {
	WriteRecord(record GeneDataRecord) error
	ShouldSkip(record GeneDataRecord) bool
	Close() error
}

type CSVWriterCreationParams struct {
	GeneProductOutputFile     string `validate:"required,min=1" json:"gene_product_output_file"`
	GeneDescriptionOutputFile string `validate:"required,min=1" json:"gene_description_output_file"`
}

type CellTextExtractor func(*goquery.Selection) O.Option[string]

type ConcurrentValidationParams struct {
	ValidationParams GeneValidationParams
	Records          [][]string
	Context          context.Context
}

type FeatureAnnotationAppConfig struct {
	Ctx              context.Context
	NumPubmedWorkers int
	NumGrpcWorkers   int
	Logger           *logrus.Entry
	Metrics          *FeatureAnnotationMetrics
}

type FeatureAnnotationMetrics struct {
	TotalProcessed int64
	SuccessCount   int64
	ErrorCount     int64
	StartTime      time.Time

	TotalFetchedFromArango      int64
	AllArangoDocsFetched        bool
	JobsSubmittedToPubmedPool   int64
	JobsCompletedFromPubmedPool int64
	JobsSubmittedToGrpcPool     int64
	JobsCompletedFromGrpcPool   int64
	// contains filtered or unexported fields
}

type FileContext struct {
	Setup  SetupConfig
	File   *os.File
	Reader *csv.Reader
	Error  error // To propagate errors
}

type Gene struct {
	FeatureID int    `json:"feature_id"`
	GeneID    string `json:"gene_id"`
	Name      string `json:"name"`
	CreatedBy string `json:"created_by"`
}

type GeneDataRecord struct {
	GeneID      string `validate:"required,min=1" json:"gene_id"`
	GeneProduct string `                          json:"gene_product"`
	Description string `                          json:"description"`
}

type GeneDescription struct {
	GeneID      string `validate:"required,min=1"`
	Description string `validate:"required,min=1"`
}

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

type GeneDescriptionCSVWriterParams struct {
	Filename string `validate:"required,min=1" json:"filename"`
}

type GeneDescriptionRequest struct {
	GeneID      string `validate:"required,min=1"`
	Description string `validate:"required,min=1"`
	User        string `validate:"required,email"`
}

type GeneGeneralInformationQuery struct {
	GeneGeneralInformation *struct {
		ID          string `graphql:"id"`
		Description string `graphql:"description"`
	} `graphql:"geneGeneralInformation(gene: $gene)"`
}

type GeneInfo struct {
	Name      string `json:"name"`
	GeneID    string `json:"gene_id"`
	FeatureID int64  `json:"feature_id"`
	CreatedBy string `json:"created_by"`
}

type GeneProcessingAction int

type GeneProcessingContext struct {
	Config ProcessingConfig
	GeneID string
}

type GeneProcessingResult struct {
	GeneID string
	Action GeneProcessingAction
}

type GeneProduct struct {
	GeneID  string
	Product string
}

type GeneProductAppConfig struct {
	Ctx              context.Context
	LegacyDatabase   string
	NumLegacyWorkers int
	NumGrpcWorkers   int
	Logger           *logrus.Entry
	Metrics          *GeneProductMetrics
}

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

type GeneProductCSVWriterParams struct {
	Filename string `validate:"required,min=1" json:"filename"`
}

type GeneProductMetrics struct {
	TotalProcessed int64
	SuccessCount   int64
	ErrorCount     int64
	SkippedCount   int64
	StartTime      time.Time

	TotalFetchedFromArango      int64
	AllArangoDocsFetched        bool
	JobsSubmittedToLegacyPool   int64
	JobsCompletedFromLegacyPool int64
	JobsSubmittedToGrpcPool     int64
	JobsCompletedFromGrpcPool   int64
	// contains filtered or unexported fields
}

type GeneProductResult struct {
	GeneProduct string     `json:"gene_product"`
	CreatedBy   string     `json:"created_by"`
	CreatedOn   LegacyTime `json:"created_on"`
}

type GeneValidationParams struct {
	InputFile    string `validate:"required,file_exists,max_file_size"`
	OutputReport string `validate:"required,endswith=.json"`
	GraphQLURL   string `validate:"required,url,startswith=https"`
	Timeout      int    `validate:"gte=10,lte=300"`
	Workers      int    `validate:"gte=1,lte=20"`
}

type GeneWithPubmed struct {
	Gene
	Pubmeds    []string
	Skip       bool
	SkipReason string
}

type GraphQLClientConfig struct {
	URL        string
	HTTPClient *http.Client
	Headers    map[string]string
}

type GraphQLClientOption func(*GraphQLClientConfig)

type GrpcAnnotationResult struct {
	GeneID  string
	Success bool
	Skipped bool
	Message string
	Error   error
}

type GrpcSynonymResult struct {
	GeneID  string
	Success bool
	Message string
	Error   error
}

type GrpcUpdateResult struct {
	GeneID  string
	Success bool
	Message string
	Error   error
}

type LegacyTime struct {
	time.Time
}

type LoadGeneDescriptionParams struct {
	InputFile string `validate:"required,file"`
	Workers   int    `validate:"gte=1,lte=50"`
	BatchSize int    `validate:"gte=1,lte=1000"`
	User      string `validate:"required,email"`
}

type LoadGeneProductParams struct {
	InputFiles []string `validate:"required,dive,file"`
	Workers    int      `validate:"gte=1,lte=50"`
	BatchSize  int      `validate:"gte=1,lte=1000"`
	User       string   `validate:"required,email"`
}

type LoadHypotheticalGeneProductsParams struct {
	User   string
	Client pb.FeatureAnnotationServiceClient
}

type ParseUnknowmeDataParams struct {
	InputFiles            []string `validate:"required,min=1,dive,file" json:"input_files"`
	GeneProductOutput     string   `validate:"required,min=1"           json:"gene_product_output"`
	GeneDescriptionOutput string   `validate:"required,min=1"           json:"gene_description_output"`
}

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

type ParsingOption func(*ParsingConfig)

type PipelineResult struct {
	File        *os.File
	Setup       SetupConfig
	UpdateCount int
	Error       error
}

type ProcessSingleRecordParams struct {
	Record             []string
	FeaturePropIDIndex int
	ValueIndex         int
	RowNumForLogging   int // Actual row number in the CSV file for logging
	Logger             *logrus.Entry
}

type ProcessedGeneData struct {
	GeneID            string
	StrippedPropsText []StrippedProperty
}

type ProcessedGeneProduct struct {
	GeneID      string
	GeneName    string
	GeneProduct string
	CreatedBy   string
	CreatedOn   time.Time
}

type ProcessingConfig struct {
	GeneIDs []string
	User    string
	Client  pb.FeatureAnnotationServiceClient
}

type ProcessingContext struct {
	FileContext
	FeaturePropIDIndex int
	ValueIndex         int
}

type ProcessingMetrics struct {
	TotalProcessed int64
	SuccessCount   int64
	ErrorCount     int64
	StartTime      time.Time

	// TotalFetchedFromArango stores the total number of items fetched by queryArango.
	// This field is set by queryArango once the total count is known.
	TotalFetchedFromArango int64
	// AllArangoDocsFetched is a flag set to true by queryArango after all documents
	// have been fetched and sent to the processing pipeline.
	AllArangoDocsFetched bool
	// Intermediate tracking counters for detailed pipeline monitoring
	JobsSubmittedToHTMLPool   int64
	JobsCompletedFromHTMLPool int64
	JobsSubmittedToGrpcPool   int64
	JobsCompletedFromGrpcPool int64
	// contains filtered or unexported fields
}

type ProcessingResult struct {
	TotalRecordsProcessed     int
	GeneProductRecordsWritten int
	DescriptionRecordsWritten int
}

type ProcessingStats struct {
	Created int
	Updated int
	Skipped int
	Total   int
}

type SetupConfig struct {
	Logger         *logrus.Entry
	DBH            *arangomanager.Database
	CSVFilePath    string
	CollectionName string
	BatchSize      int
	Delimiter      string
	Workers        int
}

type SingleGeneValidationParams struct {
	Client  *graphql.Client
	Record  []string
	Context context.Context
}

type StrippedProperty struct {
	OriginalName string
	StrippedText string
}

type SubmitBatchAndLogParams struct {
	Setup            *SetupConfig
	Docs             []map[string]string
	Logger           *logrus.Entry
	BatchDescription string
}

type SynonymAppConfig struct {
	Ctx            context.Context
	NumGrpcWorkers int
	Logger         *logrus.Entry
	Metrics        *SynonymMetrics
}

type SynonymData struct {
	Name     string   `json:"name"`
	GeneID   string   `json:"gene_id"`
	Synonyms []string `json:"synonyms"`
}

type SynonymMetrics struct {
	TotalProcessed int64
	SuccessCount   int64
	NotFoundCount  int64
	ErrorCount     int64
	StartTime      time.Time

	TotalFetchedFromArango    int64
	AllArangoDocsFetched      bool
	JobsSubmittedToGrpcPool   int64
	JobsCompletedFromGrpcPool int64
	// contains filtered or unexported fields
}

type ValidationError struct {
	Errors validator.ValidationErrors
}

type ValidationReport struct {
	TotalGenes    int                `json:"total_genes"`
	MatchCount    int                `json:"match_count"`
	MismatchCount int                `json:"mismatch_count"`
	ErrorCount    int                `json:"error_count"`
	Results       []ValidationResult `json:"results"`
	GeneratedAt   time.Time          `json:"generated_at"`
}

type ValidationResult struct {
	GeneID         string `json:"gene_id"`
	CSVDescription string `json:"csv_description"`
	GQLDescription string `json:"gql_description"`
	Match          bool   `json:"match"`
	Error          string `json:"error,omitempty"`
}

type WithAction struct {
	WithAnnotation
	Action GeneProcessingAction
}

type WithAnnotation struct {
	GeneProcessingContext
	Annotation O.Option[*pb.FeatureAnnotation]
}