tm_back/internal/document_scraper/service.go

package document_scraper

import (
	"context"
	"fmt"
	"time"
	"tm/internal/tender"
	"tm/pkg/logger"
	"tm/pkg/response"
	"tm/ted"
)

// Service defines business logic for the document scraper API
type Service interface {
	// ListPendingTenders retrieves tenders that need document scraping
	ListPendingTenders(ctx context.Context, limit, offset int) (*DocumentScraperListResponse, *response.Meta, error)
	// GetTenderByNoticeID retrieves a specific tender by its notice publication ID
	GetTenderByNoticeID(ctx context.Context, noticeID string) (*DocumentScraperTenderResponse, error)
}

type service struct {
	tenderRepo tender.TenderRepository
	tedParser  *ted.TEDParser
	logger     logger.Logger
}

// NewService creates a new document scraper service
func NewService(tenderRepo tender.TenderRepository, logger logger.Logger) Service {
	return &service{
		tenderRepo: tenderRepo,
		tedParser:  ted.NewTEDParser(),
		logger:     logger,
	}
}

// buildTenderResponse creates a DocumentScraperTenderResponse from a tender
func (s *service) buildTenderResponse(_ context.Context, t *tender.Tender) *DocumentScraperTenderResponse {
	documentURL := t.DocumentURI
	if documentURL == "" {
		documentURL = t.TenderURL
	}

	return &DocumentScraperTenderResponse{
		ID:                  t.ID.Hex(),
		NoticePublicationID: t.NoticePublicationID,
		DocumentURL:         documentURL,
		Title:               t.Title,
		Description:         t.Description,
	}
}

// ListPendingTenders retrieves tenders that have not yet been scraped for documents.
// Only returns tenders from Sweden with active deadlines (deadline not yet reached).
func (s *service) ListPendingTenders(ctx context.Context, limit, offset int) (*DocumentScraperListResponse, *response.Meta, error) {
	s.logger.Info("Listing pending tenders for document scraper", map[string]interface{}{
		"limit":  limit,
		"offset": offset,
	})

	now := time.Now().Unix()

	// Fetch a larger batch to account for filtering, then apply Sweden + active deadline filters
	// We need to over-fetch since the repository doesn't filter by country/deadline
	fetchLimit := limit * 5
	if fetchLimit > 500 {
		fetchLimit = 500
	}

	var filteredTenders []tender.Tender
	currentOffset := offset
	hasMore := true

	for hasMore {
		tenders, _, err := s.tenderRepo.GetUnScrapedTenders(ctx, fetchLimit, currentOffset)
		if err != nil {
			s.logger.Error("Failed to list pending tenders", map[string]interface{}{
				"error": err.Error(),
			})
			return nil, nil, fmt.Errorf("failed to list pending tenders: %w", err)
		}

		if len(tenders) == 0 {
			hasMore = false
			break
		}

		for i := range tenders {
			if tenders[i].CountryCode == "SWE" && tenders[i].TenderDeadline > now {
				filteredTenders = append(filteredTenders, tenders[i])
			}
			if len(filteredTenders) >= limit {
				break
			}
		}

		if len(filteredTenders) >= limit {
			filteredTenders = filteredTenders[:limit]
			hasMore = false
		} else {
			currentOffset += fetchLimit
		}
	}

	// Convert to response
	result := &DocumentScraperListResponse{
		Tenders: make([]DocumentScraperTenderResponse, len(filteredTenders)),
	}

	for i := range filteredTenders {
		result.Tenders[i] = *s.buildTenderResponse(ctx, &filteredTenders[i])
	}

	// Build pagination metadata
	meta := &response.Meta{
		Total:  len(filteredTenders),
		Limit:  limit,
		Offset: offset,
	}
	if meta.Limit > 0 {
		meta.Page = (offset / meta.Limit) + 1
		meta.Pages = 1
	}

	s.logger.Info("Pending tenders retrieved successfully", map[string]interface{}{
		"count": len(filteredTenders),
	})

	return result, meta, nil
}

// GetTenderByNoticeID retrieves a specific tender by its notice publication ID.
// Only returns tenders from Sweden with active deadlines (deadline not yet reached).
func (s *service) GetTenderByNoticeID(ctx context.Context, noticeID string) (*DocumentScraperTenderResponse, error) {
	s.logger.Info("Getting tender by notice ID for document scraper", map[string]interface{}{
		"notice_id": noticeID,
	})

	t, err := s.tenderRepo.GetByNoticePublicationID(ctx, noticeID)
	if err != nil {
		s.logger.Error("Failed to get tender by notice ID", map[string]interface{}{
			"notice_id": noticeID,
			"error":     err.Error(),
		})
		return nil, fmt.Errorf("failed to get tender: %w", err)
	}

	if t == nil {
		return nil, fmt.Errorf("tender not found")
	}

	// Only return tenders from Sweden with active deadlines
	now := time.Now().Unix()
	if t.CountryCode != "SWE" || t.TenderDeadline <= now {
		s.logger.Info("Tender filtered out: not Sweden or deadline passed", map[string]interface{}{
			"notice_id":       noticeID,
			"country_code":    t.CountryCode,
			"tender_deadline": t.TenderDeadline,
			"now":             now,
		})
		return nil, fmt.Errorf("tender not found")
	}

	result := s.buildTenderResponse(ctx, t)

	s.logger.Info("Tender retrieved successfully for document scraper", map[string]interface{}{
		"notice_id": noticeID,
		"tender_id": t.ID.Hex(),
	})

	return result, nil
}