Refactor Scraper Configuration and Command Line Handling
- Removed command line flag parsing for date range scraping in the scraper's main function. - Introduced configuration fields for one-time scraping and date range (FromDate, ToDate) in the ScraperConfig struct. - Updated the main function to utilize the new configuration fields, improving the clarity and maintainability of the scraping logic. - This change enhances the scraper's functionality by centralizing configuration management and reducing command line complexity.
This commit is contained in:
@@ -24,4 +24,7 @@ type ScraperConfig struct {
|
|||||||
DownloadDir string `env:"TED_DOWNLOAD_DIR" envDefault:"./downloads"`
|
DownloadDir string `env:"TED_DOWNLOAD_DIR" envDefault:"./downloads"`
|
||||||
CleanupAfter time.Duration `env:"TED_CLEANUP_AFTER" envDefault:"24h"`
|
CleanupAfter time.Duration `env:"TED_CLEANUP_AFTER" envDefault:"24h"`
|
||||||
ScrapingInterval string `env:"TED_SCRAPING_INTERVAL" envDefault:"* * 10 * * *"`
|
ScrapingInterval string `env:"TED_SCRAPING_INTERVAL" envDefault:"* * 10 * * *"`
|
||||||
|
OneTime bool `env:"TED_ONE_TIME" envDefault:"false"`
|
||||||
|
FromDate string `env:"TED_FROM_DATE" envDefault:"02/01/2025"`
|
||||||
|
ToDate string `env:"TED_TO_DATE" envDefault:"13/09/2025"`
|
||||||
}
|
}
|
||||||
|
|||||||
+7
-16
@@ -2,7 +2,6 @@ package main
|
|||||||
|
|
||||||
import (
|
import (
|
||||||
"context"
|
"context"
|
||||||
"flag"
|
|
||||||
"fmt"
|
"fmt"
|
||||||
"os"
|
"os"
|
||||||
"os/signal"
|
"os/signal"
|
||||||
@@ -12,14 +11,6 @@ import (
|
|||||||
)
|
)
|
||||||
|
|
||||||
func main() {
|
func main() {
|
||||||
// Define command line flags
|
|
||||||
var (
|
|
||||||
fromDate = flag.String("from", "", "Start date for scraping (format: 02/01/2006)")
|
|
||||||
toDate = flag.String("to", "", "End date for scraping (format: 02/01/2006)")
|
|
||||||
oneTime = flag.Bool("onetime", false, "Run once with date range instead of scheduled mode")
|
|
||||||
)
|
|
||||||
flag.Parse()
|
|
||||||
|
|
||||||
// Load configuration
|
// Load configuration
|
||||||
config, err := bootstrap.InitConfig()
|
config, err := bootstrap.InitConfig()
|
||||||
if err != nil {
|
if err != nil {
|
||||||
@@ -48,28 +39,28 @@ func main() {
|
|||||||
notificationService := bootstrap.InitNotificationService(config.Notification, appLogger)
|
notificationService := bootstrap.InitNotificationService(config.Notification, appLogger)
|
||||||
|
|
||||||
// Check if one-time date range scraping is requested
|
// Check if one-time date range scraping is requested
|
||||||
if *oneTime {
|
if config.TED.OneTime {
|
||||||
if *fromDate == "" || *toDate == "" {
|
if config.TED.FromDate == "" || config.TED.ToDate == "" {
|
||||||
appLogger.Error("Both -from and -to dates are required for one-time scraping", map[string]interface{}{})
|
appLogger.Error("Both TED_FROM_DATE and TED_TO_DATE are required for one-time scraping", map[string]interface{}{})
|
||||||
fmt.Println("Usage: ./scraper -onetime -from 12/01/2025 -to 13/09/2025")
|
fmt.Println("Usage: ./scraper -onetime -from 12/01/2025 -to 13/09/2025")
|
||||||
os.Exit(1)
|
os.Exit(1)
|
||||||
}
|
}
|
||||||
|
|
||||||
// Parse dates
|
// Parse dates
|
||||||
from, err := time.Parse("02/01/2006", *fromDate)
|
from, err := time.Parse("02/01/2006", config.TED.FromDate)
|
||||||
if err != nil {
|
if err != nil {
|
||||||
appLogger.Error("Invalid from date format", map[string]interface{}{
|
appLogger.Error("Invalid from date format", map[string]interface{}{
|
||||||
"from_date": *fromDate,
|
"from_date": config.TED.FromDate,
|
||||||
"error": err.Error(),
|
"error": err.Error(),
|
||||||
})
|
})
|
||||||
fmt.Println("Date format should be: DD/MM/YYYY (e.g., 12/01/2025)")
|
fmt.Println("Date format should be: DD/MM/YYYY (e.g., 12/01/2025)")
|
||||||
os.Exit(1)
|
os.Exit(1)
|
||||||
}
|
}
|
||||||
|
|
||||||
to, err := time.Parse("02/01/2006", *toDate)
|
to, err := time.Parse("02/01/2006", config.TED.ToDate)
|
||||||
if err != nil {
|
if err != nil {
|
||||||
appLogger.Error("Invalid to date format", map[string]interface{}{
|
appLogger.Error("Invalid to date format", map[string]interface{}{
|
||||||
"to_date": *toDate,
|
"to_date": config.TED.ToDate,
|
||||||
"error": err.Error(),
|
"error": err.Error(),
|
||||||
})
|
})
|
||||||
fmt.Println("Date format should be: DD/MM/YYYY (e.g., 13/09/2025)")
|
fmt.Println("Date format should be: DD/MM/YYYY (e.g., 13/09/2025)")
|
||||||
|
|||||||
Reference in New Issue
Block a user