Nima Nakhostin
1f6d3fd1cb
- Added an entrypoint script to the scraper Dockerfile to handle one-time scraping with date range support. - Introduced a new ContentXML field in the Notice entity to store XML content, improving data structure for notices. - Updated the scraper logic to map XML content to the tender entity, enhancing data processing capabilities.
277 lines
14 KiB
Go
277 lines
14 KiB
Go
package notice
|
|
|
|
import (
|
|
"fmt"
|
|
"strconv"
|
|
"strings"
|
|
"time"
|
|
"tm/pkg/mongo"
|
|
)
|
|
|
|
// Tender represents a tender/contract notice entity
|
|
type Notice struct {
|
|
mongo.Model `bson:",inline"`
|
|
ContractNoticeID string `bson:"contract_notice_id" json:"contract_notice_id"`
|
|
NoticePublicationID string `bson:"notice_publication_id" json:"notice_publication_id"`
|
|
ContractFolderID string `bson:"contract_folder_id" json:"contract_folder_id"`
|
|
NoticeTypeCode string `bson:"notice_type_code" json:"notice_type_code"`
|
|
NoticeSubTypeCode string `bson:"notice_sub_type_code" json:"notice_sub_type_code"`
|
|
NoticeLanguageCode string `bson:"notice_language_code" json:"notice_language_code"`
|
|
IssueDate int64 `bson:"issue_date" json:"issue_date"`
|
|
IssueTime int64 `bson:"issue_time" json:"issue_time"`
|
|
TenderDeadline int64 `bson:"tender_deadline" json:"tender_deadline"`
|
|
PublicationDate int64 `bson:"publication_date" json:"publication_date"`
|
|
SubmissionDeadline int64 `bson:"submission_deadline" json:"submission_deadline"`
|
|
ApplicationDeadline int64 `bson:"application_deadline" json:"application_deadline"`
|
|
GazetteID string `bson:"gazette_id" json:"gazette_id"`
|
|
Title string `bson:"title" json:"title"`
|
|
Description string `bson:"description" json:"description"`
|
|
ProcurementTypeCode string `bson:"procurement_type_code" json:"procurement_type_code"`
|
|
ProcedureCode string `bson:"procedure_code" json:"procedure_code"`
|
|
MainClassification string `bson:"main_classification" json:"main_classification"`
|
|
AdditionalClassifications []string `bson:"additional_classifications" json:"additional_classifications"`
|
|
EstimatedValue float64 `bson:"estimated_value" json:"estimated_value"`
|
|
Currency string `bson:"currency" json:"currency"`
|
|
Duration string `bson:"duration" json:"duration"`
|
|
DurationUnit string `bson:"duration_unit" json:"duration_unit"`
|
|
PlaceOfPerformance string `bson:"place_of_performance" json:"place_of_performance"`
|
|
CountryCode string `bson:"country_code" json:"country_code"`
|
|
RegionCode string `bson:"region_code" json:"region_code"`
|
|
CityName string `bson:"city_name" json:"city_name"`
|
|
PostalCode string `bson:"postal_code" json:"postal_code"`
|
|
DocumentURI string `bson:"document_uri" json:"document_uri"`
|
|
TenderURL string `bson:"tender_url" json:"tender_url"`
|
|
SubmissionURL string `bson:"submission_url" json:"submission_url"`
|
|
BuyerOrganization *Organization `bson:"buyer_organization" json:"buyer_organization"`
|
|
ReviewOrganization *Organization `bson:"review_organization,omitempty" json:"review_organization,omitempty"`
|
|
Organizations []Organization `bson:"organizations" json:"organizations"`
|
|
SelectionCriteria []SelectionCriterion `bson:"selection_criteria" json:"selection_criteria"`
|
|
OfficialLanguages []string `bson:"official_languages" json:"official_languages"`
|
|
Status TenderStatus `bson:"status" json:"status"`
|
|
Source TenderSource `bson:"source" json:"source"`
|
|
SourceFileURL string `bson:"source_file_url" json:"source_file_url"`
|
|
SourceFileName string `bson:"source_file_name" json:"source_file_name"`
|
|
ProcessingMetadata ProcessingMetadata `bson:"processing_metadata" json:"processing_metadata"`
|
|
ContentXML string `bson:"content_xml" json:"content_xml"`
|
|
|
|
// Status-specific fields
|
|
CancellationReason string `bson:"cancellation_reason,omitempty" json:"cancellation_reason,omitempty"`
|
|
CancellationDate int64 `bson:"cancellation_date,omitempty" json:"cancellation_date,omitempty"`
|
|
AwardDate int64 `bson:"award_date,omitempty" json:"award_date,omitempty"`
|
|
AwardedValue float64 `bson:"awarded_value,omitempty" json:"awarded_value,omitempty"`
|
|
WinningTenderer *Organization `bson:"winning_tenderer,omitempty" json:"winning_tenderer,omitempty"`
|
|
ContractNumber string `bson:"contract_number,omitempty" json:"contract_number,omitempty"`
|
|
SuspensionReason string `bson:"suspension_reason,omitempty" json:"suspension_reason,omitempty"`
|
|
SuspensionDate int64 `bson:"suspension_date,omitempty" json:"suspension_date,omitempty"`
|
|
Modifications []TenderModification `bson:"modifications,omitempty" json:"modifications,omitempty"`
|
|
AwardedEntities []Awarded `bson:"awarded,omitempty" json:"awarded,omitempty"`
|
|
}
|
|
|
|
// Organization represents organization information
|
|
type Organization struct {
|
|
ID string `bson:"id" json:"id"`
|
|
Name string `bson:"name" json:"name"`
|
|
CompanyID string `bson:"company_id,omitempty" json:"company_id,omitempty"`
|
|
WebsiteURI string `bson:"website_uri,omitempty" json:"website_uri,omitempty"`
|
|
ContactName string `bson:"contact_name,omitempty" json:"contact_name,omitempty"`
|
|
ContactTelephone string `bson:"contact_telephone,omitempty" json:"contact_telephone,omitempty"`
|
|
ContactEmail string `bson:"contact_email,omitempty" json:"contact_email,omitempty"`
|
|
ContactFax string `bson:"contact_fax,omitempty" json:"contact_fax,omitempty"`
|
|
Address Address `bson:"address" json:"address"`
|
|
Role string `bson:"role,omitempty" json:"role,omitempty"`
|
|
}
|
|
|
|
// Address represents address information
|
|
type Address struct {
|
|
StreetName string `bson:"street_name,omitempty" json:"street_name,omitempty"`
|
|
CityName string `bson:"city_name,omitempty" json:"city_name,omitempty"`
|
|
PostalZone string `bson:"postal_zone,omitempty" json:"postal_zone,omitempty"`
|
|
CountrySubentityCode string `bson:"country_subentity_code,omitempty" json:"country_subentity_code,omitempty"`
|
|
Department string `bson:"department,omitempty" json:"department,omitempty"`
|
|
Region string `bson:"region,omitempty" json:"region,omitempty"`
|
|
CountryCode string `bson:"country_code" json:"country_code"`
|
|
}
|
|
|
|
// SelectionCriterion represents selection criteria
|
|
type SelectionCriterion struct {
|
|
TypeCode string `bson:"type_code" json:"type_code"`
|
|
Description string `bson:"description,omitempty" json:"description,omitempty"`
|
|
LanguageID string `bson:"language_id,omitempty" json:"language_id,omitempty"`
|
|
}
|
|
|
|
// ProcessingMetadata contains metadata about how the tender was processed
|
|
type ProcessingMetadata struct {
|
|
ScrapedAt int64 `bson:"scraped_at" json:"scraped_at"`
|
|
ProcessedAt int64 `bson:"processed_at" json:"processed_at"`
|
|
ProcessingVersion string `bson:"processing_version" json:"processing_version"`
|
|
ParsingErrors []string `bson:"parsing_errors,omitempty" json:"parsing_errors,omitempty"`
|
|
ValidationErrors []string `bson:"validation_errors,omitempty" json:"validation_errors,omitempty"`
|
|
EnrichmentData map[string]string `bson:"enrichment_data,omitempty" json:"enrichment_data,omitempty"`
|
|
TranslatedData map[string]string `bson:"translated_data,omitempty" json:"translated_data,omitempty"`
|
|
TranslatedAt int64 `bson:"translated_at,omitempty" json:"translated_at,omitempty"`
|
|
Processed bool `bson:"processed" json:"processed"`
|
|
}
|
|
|
|
// TenderModification represents a modification made to a tender
|
|
type TenderModification struct {
|
|
ModificationDate int64 `bson:"modification_date" json:"modification_date"` // Unix milliseconds
|
|
ModificationReason string `bson:"modification_reason" json:"modification_reason"`
|
|
Description string `bson:"description,omitempty" json:"description,omitempty"`
|
|
LanguageID string `bson:"language_id,omitempty" json:"language_id,omitempty"`
|
|
}
|
|
|
|
// AwardedEntity represents an entity that has been awarded a contract
|
|
type Awarded struct {
|
|
Name string `bson:"name" json:"name"`
|
|
Address string `bson:"address,omitempty" json:"address,omitempty"`
|
|
Country string `bson:"country,omitempty" json:"country,omitempty"`
|
|
Amount float64 `bson:"amount" json:"amount"`
|
|
Currency string `bson:"currency,omitempty" json:"currency,omitempty"`
|
|
Share float64 `bson:"share,omitempty" json:"share,omitempty"` // Percentage share for multiple winners
|
|
AwardDate int64 `bson:"award_date" json:"award_date"` // Unix milliseconds
|
|
ContractID string `bson:"contract_id,omitempty" json:"contract_id,omitempty"` // Contract reference ID
|
|
TenderID string `bson:"tender_id,omitempty" json:"tender_id,omitempty"` // Tender reference ID
|
|
LotID string `bson:"lot_id,omitempty" json:"lot_id,omitempty"` // Lot reference ID
|
|
CompanyID string `bson:"company_id,omitempty" json:"company_id,omitempty"` // Company registration ID
|
|
OrganizationID string `bson:"organization_id,omitempty" json:"organization_id,omitempty"` // Organization reference ID
|
|
}
|
|
|
|
// TenderStatus represents the status of a tender
|
|
type TenderStatus string
|
|
|
|
const (
|
|
TenderStatusActive TenderStatus = "active"
|
|
TenderStatusExpired TenderStatus = "expired"
|
|
TenderStatusCancelled TenderStatus = "cancelled"
|
|
TenderStatusAwarded TenderStatus = "awarded"
|
|
TenderStatusDraft TenderStatus = "draft"
|
|
TenderStatusClosed TenderStatus = "closed"
|
|
TenderStatusModified TenderStatus = "modified"
|
|
TenderStatusSuspended TenderStatus = "suspended"
|
|
TenderStatusPublished TenderStatus = "published"
|
|
)
|
|
|
|
// TenderSource represents the source of tender data
|
|
type TenderSource string
|
|
|
|
const (
|
|
TenderSourceTEDScraper TenderSource = "ted_scraper"
|
|
TenderSourceManual TenderSource = "manual"
|
|
TenderSourceAPI TenderSource = "api"
|
|
TenderSourceBulkImport TenderSource = "bulk_import"
|
|
)
|
|
|
|
// TenderStatistics represents tender statistics
|
|
type TenderStatistics struct {
|
|
TotalTenders int64 `json:"total_tenders"`
|
|
ActiveTenders int64 `json:"active_tenders"`
|
|
ExpiredTenders int64 `json:"expired_tenders"`
|
|
TendersByCountry map[string]int64 `json:"tenders_by_country"`
|
|
TendersByType map[string]int64 `json:"tenders_by_type"`
|
|
TendersByClassification map[string]int64 `json:"tenders_by_classification"`
|
|
AverageEstimatedValue float64 `json:"average_estimated_value"`
|
|
TotalEstimatedValue float64 `json:"total_estimated_value"`
|
|
LastUpdated int64 `json:"last_updated"` // Unix milliseconds
|
|
}
|
|
|
|
// GetID returns the tender ID
|
|
func (t *Notice) GetID() string {
|
|
return t.ID.Hex()
|
|
}
|
|
|
|
// IsActive returns true if the tender is active
|
|
func (t *Notice) IsActive() bool {
|
|
return t.Status == TenderStatusActive
|
|
}
|
|
|
|
// IsExpired returns true if the tender deadline has passed
|
|
func (t *Notice) IsExpired() bool {
|
|
if t.Status == TenderStatusExpired {
|
|
return true
|
|
}
|
|
|
|
if t.TenderDeadline == 0 {
|
|
return false
|
|
}
|
|
|
|
// Check if deadline (in milliseconds) is in the past
|
|
now := time.Now().Unix()
|
|
return t.TenderDeadline < now
|
|
}
|
|
|
|
// GetTenderURL returns the TED tender URL
|
|
func (t *Notice) GetTenderURL() string {
|
|
if t.TenderURL != "" {
|
|
return t.TenderURL
|
|
}
|
|
|
|
if t.NoticePublicationID != "" {
|
|
return generateTenderURL(t.NoticePublicationID)
|
|
}
|
|
|
|
return ""
|
|
}
|
|
|
|
// generateTenderURL generates the TED tender URL from NoticePublicationID
|
|
func generateTenderURL(noticePublicationID string) string {
|
|
// Remove leading zeros and format for URL
|
|
// Example: 00522942-2025 -> https://ted.europa.eu/en/notice/-/detail/522942-2025
|
|
parts := strings.Split(noticePublicationID, "-")
|
|
if len(parts) != 2 {
|
|
return ""
|
|
}
|
|
|
|
// Convert to int and back to string to remove leading zeros
|
|
if num, err := strconv.Atoi(parts[0]); err == nil {
|
|
cleanID := fmt.Sprintf("%d-%s", num, parts[1])
|
|
return fmt.Sprintf("https://ted.europa.eu/en/notice/-/detail/%s", cleanID)
|
|
}
|
|
|
|
return ""
|
|
}
|
|
|
|
// GetMainOrganization returns the buyer organization
|
|
func (t *Notice) GetMainOrganization() *Organization {
|
|
return t.BuyerOrganization
|
|
}
|
|
|
|
// HasEstimatedValue returns true if the tender has an estimated value
|
|
func (t *Notice) HasEstimatedValue() bool {
|
|
return t.EstimatedValue > 0
|
|
}
|
|
|
|
// GetFormattedEstimatedValue returns formatted estimated value with currency
|
|
func (t *Notice) GetFormattedEstimatedValue() string {
|
|
if !t.HasEstimatedValue() {
|
|
return ""
|
|
}
|
|
|
|
if t.Currency != "" {
|
|
return fmt.Sprintf("%.2f %s", t.EstimatedValue, t.Currency)
|
|
}
|
|
|
|
return fmt.Sprintf("%.2f", t.EstimatedValue)
|
|
}
|
|
|
|
// UpdateStatus updates the tender status and updated timestamp
|
|
func (t *Notice) UpdateStatus(status TenderStatus) {
|
|
t.Status = status
|
|
t.UpdatedAt = time.Now().Unix()
|
|
}
|
|
|
|
// GetDaysUntilDeadline returns the number of days until the tender deadline
|
|
func (t *Notice) GetDaysUntilDeadline() int {
|
|
if t.TenderDeadline == 0 {
|
|
return 0
|
|
}
|
|
|
|
now := time.Now().Unix()
|
|
if t.TenderDeadline <= now {
|
|
return 0 // Already expired
|
|
}
|
|
|
|
diffMillis := t.TenderDeadline - now
|
|
diffDays := diffMillis / (24 * 60 * 60 * 1000)
|
|
return int(diffDays)
|
|
}
|