Queueing, prepping to fetch

A lot more to go, but this is the core.

Need to think about how to test the queue handlers.
This commit is contained in:
Matt Jadud
2025-11-30 21:29:30 -05:00
parent 06cdc68be7
commit f53639af2f
14 changed files with 258 additions and 105 deletions

39
internal/engine/entre.go Normal file
View File

@@ -0,0 +1,39 @@
package engine
import (
"context"
"fmt"
"log"
"time"
"git.jadud.com/jadudm/grosbeak/internal/liteq"
base "git.jadud.com/jadudm/grosbeak/internal/types"
)
func Entre(queue *liteq.JobQueue, ej *base.EntreJob) error {
n := time.Now()
var ignore_tag string
switch ej.UpdateFrequency {
case base.UPDATE_DAILY:
ignore_tag = fmt.Sprintf("%s|y%d-yd%d", ej.URL, n.Year(), n.YearDay())
case base.UPDATE_WEEKLY:
ignore_tag = fmt.Sprintf("%s|y%d-w%d", ej.URL, n.Year(), n.YearDay()/7)
case base.UPDATE_MONTHLY:
ignore_tag = fmt.Sprintf("%s|y%d-m%d", ej.URL, n.Year(), n.Month())
default:
ignore_tag = fmt.Sprintf("%s|y%d-yd%d", ej.URL, n.Year(), n.YearDay())
}
err := queue.QueueJob(context.Background(), liteq.QueueJobParams{
Queue: "fetch",
// This only works for things in the `queued` state
DedupingKey: liteq.IgnoreDuplicate(ignore_tag),
Job: ej.AsJson(),
})
if err != nil {
log.Printf("entre err %s: %s\n", ej.URL, err.Error())
}
return err
}

34
internal/engine/fetch.go Normal file
View File

@@ -0,0 +1,34 @@
package engine
import (
"context"
"log"
"git.jadud.com/jadudm/grosbeak/internal/domain64"
"git.jadud.com/jadudm/grosbeak/internal/liteq"
"git.jadud.com/jadudm/grosbeak/internal/types"
"github.com/jpillora/go-tld"
"github.com/tidwall/gjson"
)
func Fetch(d64m *domain64.Domain64Map) types.QueueWorker {
_f := func(ctx context.Context, job *liteq.Job) error {
url := gjson.Get(job.Job, "url").String()
turl, err := tld.Parse(url)
if err != nil {
// If we can't parse it, shut the job down as completed.
return nil
}
d64, err := d64m.URLToDomain64(turl)
if err != nil {
log.Printf("urltodomain64 err %s: %s\n", url, err.Error())
}
log.Printf("fetching %s 0x%016x\n", url, d64.ToInt64())
return nil
}
return _f
}