move from file storage to sqlite3

This commit is contained in:
2025-01-03 01:00:06 +01:00
parent fbca771479
commit 98655fd1fb
16 changed files with 150 additions and 389 deletions

View File

@@ -2,13 +2,16 @@ package main
import (
"crowsnest/internal/model"
"crowsnest/internal/data"
"crowsnest/internal/model/sqlite"
"database/sql"
"fmt"
"log"
"regexp"
"strings"
"time"
"github.com/gocolly/colly/v2"
_ "github.com/mattn/go-sqlite3"
)
func spiegelCollector(results *map[string]*model.Article) *colly.Collector {
@@ -21,7 +24,6 @@ func spiegelCollector(results *map[string]*model.Article) *colly.Collector {
// create entry if not behind paywall
paywall_false_pattern := regexp.MustCompile("\"paywall\":{\"attributes\":{\"is_active\":false")
c.OnResponse(func(r *colly.Response) {
if paywall_false_pattern.Match(r.Body) {
url := r.Request.URL.String()
(*results)[url] = &model.Article{
@@ -86,7 +88,6 @@ func spiegelCollector(results *map[string]*model.Article) *colly.Collector {
// cascade
c.OnHTML("a[href]", func(e *colly.HTMLElement) {
fmt.Println(e.Attr("href"))
e.Request.Visit(e.Attr("href"))
})
@@ -99,14 +100,15 @@ func main() {
c.Visit("https://www.spiegel.de/")
// data store
fds, _ := data.NewFileDatastore("./persistence/spiegel.json")
repo, _ := data.NewDefaultRepository[*model.Article](fds, "article")
db, err := sql.Open("sqlite3", "./persistence/app.db")
if err != nil { log.Fatal(err) }
db_articles := &sqlite.ArticleModel{ DB: db }
counter := 0
for _, val := range res {
counter++
repo.Create(val)
db_articles.Insert(val)
}
fmt.Println(counter)
}