Add TED Scraper Configuration and Cron Scheduler Implementation
- Introduced a new configuration file for the TED scraper, defining database connection settings, logging preferences, and scraping parameters. - Implemented a CronScheduler to manage scheduled tasks for TED operations, utilizing the robfig/cron library for scheduling. - Added new entities and structures for handling TED XML data, including eForms and tender-related information, enhancing the scraper's functionality. - Updated the main application to initialize the TED scraper with the new configuration and set up graceful shutdown handling. - Removed the deprecated scraping implementation to streamline the codebase and focus on the new architecture.
This commit is contained in:
+172
@@ -0,0 +1,172 @@
|
||||
package ted
|
||||
|
||||
import (
|
||||
"encoding/xml"
|
||||
"fmt"
|
||||
"strings"
|
||||
"tm/pkg/notification"
|
||||
"tm/pkg/xmlparser"
|
||||
)
|
||||
|
||||
type ParserType string
|
||||
|
||||
const (
|
||||
ParserTypeContractNotice ParserType = "ContractNotice"
|
||||
ParserTypeContractAwardNotice ParserType = "ContractAwardNotice"
|
||||
ParserTypePriorInformationNotice ParserType = "PriorInformationNotice"
|
||||
ParserTypeBusinessRegistrationInformationNotice ParserType = "BusinessRegistrationInformationNotice"
|
||||
)
|
||||
|
||||
type TEDParser struct {
|
||||
Notification notification.SDK
|
||||
NoticeParser xmlparser.Parser[ContractNotice]
|
||||
}
|
||||
|
||||
// NewTEDParser creates a new TED parser
|
||||
func NewTEDParser() *TEDParser {
|
||||
options := xmlparser.DefaultParserOptions()
|
||||
options.MaxSize = 50 * 1024 * 1024 // 50MB limit for TED XML files
|
||||
options.IgnoreUnknownElements = true // TED XML has many optional fields
|
||||
|
||||
parser := xmlparser.NewParser[ContractNotice](options)
|
||||
return &TEDParser{
|
||||
NoticeParser: parser,
|
||||
}
|
||||
}
|
||||
|
||||
// ParseAnyNotice parses XML content and detects the notice type automatically
|
||||
func (tp *TEDParser) Parse(content []byte) (interface{}, string, ParserType, error) {
|
||||
// First, detect the root element type
|
||||
rootElement, err := detectRootElement(content)
|
||||
if err != nil {
|
||||
return nil, "", ParserType(""), fmt.Errorf("failed to detect root element: %w", err)
|
||||
}
|
||||
|
||||
// Parse based on the detected type
|
||||
switch ParserType(rootElement) {
|
||||
case ParserTypeContractNotice:
|
||||
notice, err := tp.parseContractNotice(content)
|
||||
return notice, notice.GetID(), ParserTypeContractNotice, err
|
||||
case ParserTypeContractAwardNotice:
|
||||
notice, err := tp.parseContractAwardNotice(content)
|
||||
return notice, notice.GetID(), ParserTypeContractAwardNotice, err
|
||||
case ParserTypePriorInformationNotice:
|
||||
notice, err := tp.parsePriorInformationNotice(content)
|
||||
return notice, notice.GetID(), ParserTypePriorInformationNotice, err
|
||||
case ParserTypeBusinessRegistrationInformationNotice:
|
||||
notice, err := tp.parseBusinessRegistrationInformationNotice(content)
|
||||
return notice, notice.GetID(), ParserTypeBusinessRegistrationInformationNotice, err
|
||||
default:
|
||||
return nil, "", ParserType(""), fmt.Errorf("unsupported notice type: %s", rootElement)
|
||||
}
|
||||
}
|
||||
|
||||
// detectRootElement detects the root XML element name
|
||||
func detectRootElement(content []byte) (string, error) {
|
||||
decoder := xml.NewDecoder(strings.NewReader(string(content)))
|
||||
for {
|
||||
token, err := decoder.Token()
|
||||
if err != nil {
|
||||
return "", err
|
||||
}
|
||||
if startElement, ok := token.(xml.StartElement); ok {
|
||||
return startElement.Name.Local, nil
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// parseContractNotice parses a ContractNotice
|
||||
func (tp *TEDParser) parseContractNotice(content []byte) (*ContractNotice, error) {
|
||||
options := xmlparser.DefaultParserOptions()
|
||||
options.MaxSize = 50 * 1024 * 1024
|
||||
options.IgnoreUnknownElements = true
|
||||
|
||||
parser := xmlparser.NewParser[ContractNotice](options)
|
||||
notice, err := parser.ParseBytes(content)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
return notice, nil
|
||||
}
|
||||
|
||||
// parseContractAwardNotice parses a ContractAwardNotice
|
||||
func (tp *TEDParser) parseContractAwardNotice(content []byte) (*ContractAwardNotice, error) {
|
||||
options := xmlparser.DefaultParserOptions()
|
||||
options.MaxSize = 50 * 1024 * 1024
|
||||
options.IgnoreUnknownElements = true
|
||||
|
||||
parser := xmlparser.NewParser[ContractAwardNotice](options)
|
||||
notice, err := parser.ParseBytes(content)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
return notice, nil
|
||||
}
|
||||
|
||||
// parsePriorInformationNotice parses a PriorInformationNotice
|
||||
func (tp *TEDParser) parsePriorInformationNotice(content []byte) (*PriorInformationNotice, error) {
|
||||
options := xmlparser.DefaultParserOptions()
|
||||
options.MaxSize = 50 * 1024 * 1024
|
||||
options.IgnoreUnknownElements = true
|
||||
|
||||
parser := xmlparser.NewParser[PriorInformationNotice](options)
|
||||
notice, err := parser.ParseBytes(content)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
return notice, nil
|
||||
}
|
||||
|
||||
// parseBusinessRegistrationInformationNotice parses a BusinessRegistrationInformationNotice
|
||||
func (tp *TEDParser) parseBusinessRegistrationInformationNotice(content []byte) (*BusinessRegistrationInformationNotice, error) {
|
||||
options := xmlparser.DefaultParserOptions()
|
||||
options.MaxSize = 50 * 1024 * 1024
|
||||
options.IgnoreUnknownElements = true
|
||||
|
||||
parser := xmlparser.NewParser[BusinessRegistrationInformationNotice](options)
|
||||
notice, err := parser.ParseBytes(content)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
return notice, nil
|
||||
}
|
||||
|
||||
// GetID returns the contract notice ID
|
||||
func (cn ContractNotice) GetID() string {
|
||||
return cn.ID
|
||||
}
|
||||
|
||||
// GetID returns the contract award notice ID
|
||||
func (can ContractAwardNotice) GetID() string {
|
||||
return can.ID
|
||||
}
|
||||
|
||||
// GetID returns the prior information notice ID
|
||||
func (pin PriorInformationNotice) GetID() string {
|
||||
return pin.ID
|
||||
}
|
||||
|
||||
// GetID returns the business registration information notice ID
|
||||
func (brin BusinessRegistrationInformationNotice) GetID() string {
|
||||
return brin.ID
|
||||
}
|
||||
|
||||
// Validate validates the parsed model
|
||||
func (cn ContractNotice) Validate() error {
|
||||
return nil
|
||||
}
|
||||
|
||||
// Validate validates the parsed model
|
||||
func (can ContractAwardNotice) Validate() error {
|
||||
return nil
|
||||
}
|
||||
|
||||
// Validate validates the parsed model
|
||||
func (pin PriorInformationNotice) Validate() error {
|
||||
return nil
|
||||
}
|
||||
|
||||
// Validate validates the parsed model
|
||||
func (brin BusinessRegistrationInformationNotice) Validate() error {
|
||||
return nil
|
||||
}
|
||||
Reference in New Issue
Block a user