remove model files for responses

This commit is contained in:
2025-01-07 12:32:38 +01:00
parent ce10e1e62b
commit 321ccfe44d
6 changed files with 20 additions and 155 deletions

View File

@@ -3,6 +3,5 @@ package collectors
import "crowsnest/internal/model/database"
type Collector struct {
Responses *database.ResponseModel
Articles *database.ArticleModel
Articles *database.ArticleModel
}

View File

@@ -17,7 +17,7 @@ func (c *Collector) CollectSpiegel() {
collycollector := colly.NewCollector(
colly.AllowedDomains("www.spiegel.de", "spiegel.de"),
colly.CacheDir("./persistence/spiegel_cache"),
colly.MaxDepth(5),
colly.MaxDepth(3),
)
// store articles
@@ -42,7 +42,7 @@ func (c *Collector) CollectSpiegel() {
})
// go through archive
startDate := time.Date(2025, time.January, 1, 0, 0, 0, 0, time.UTC)
startDate := time.Date(2020, time.January, 1, 0, 0, 0, 0, time.UTC)
currentDate := time.Now()
for date := startDate; date.Before(currentDate) || date.Equal(currentDate); date = date.AddDate(0, 0, 1) {

View File

@@ -19,7 +19,7 @@ func (c *Collector) CollectZeit() {
collycollector := colly.NewCollector(
colly.AllowedDomains("www.zeit.de", "zeit.de"),
colly.CacheDir("./persistence/zeit_cache"),
colly.MaxDepth(5),
colly.MaxDepth(3),
)
// store articles
@@ -44,7 +44,7 @@ func (c *Collector) CollectZeit() {
})
// go through archive
startDate := time.Date(2025, time.January, 1, 0, 0, 0, 0, time.UTC)
startDate := time.Date(2020, time.January, 1, 0, 0, 0, 0, time.UTC)
//startDate := time.Date(1946, time.January, 1, 0, 0, 0, 0, time.UTC)
currentDate := time.Now()

View File

@@ -6,6 +6,7 @@ import (
"database/sql"
"log"
"os"
"sync"
_ "github.com/lib/pq"
)
@@ -23,10 +24,20 @@ func main() {
// collect websites
coll := collectors.Collector{
Responses: &database.ResponseModel{DB: db},
Articles: &database.ArticleModel{DB: db},
Articles: &database.ArticleModel{DB: db},
}
coll.SpiegelCollect()
//coll.Zeit()
var wg sync.WaitGroup
wg.Add(2)
go func() {
defer wg.Done()
coll.CollectSpiegel()
}()
go func() {
defer wg.Done()
go coll.CollectZeit()
}()
wg.Wait()
}