add RefreshAll with workers

This commit is contained in:
nytpu
2021-03-18 13:43:22 -06:00
parent 364e8fbc21
commit 326edc97c5
+73
View File
@@ -9,6 +9,7 @@ package fetch
import ( import (
"fmt" "fmt"
"net/url" "net/url"
"sync"
"golang.nytpu.com/comitium/core" "golang.nytpu.com/comitium/core"
) )
@@ -42,3 +43,75 @@ func Page(data *core.FullData, remote *url.URL, title string) error {
return fmt.Errorf("Unsupported protocol '%s'", remote.Scheme) return fmt.Errorf("Unsupported protocol '%s'", remote.Scheme)
} }
} }
type refreshJob struct {
typ string // "feed" or "page"
url url.URL
}
// RefreshAll will check all feeds and pages in a core.FullData for updates
func RefreshAll(data *core.FullData, numWorkers int) error {
var wg sync.WaitGroup
data.RLock()
numJobs := len(data.Feeds) + len(data.Pages)
if numJobs == 0 {
data.RUnlock()
return nil
}
if numWorkers < 1 {
numWorkers = 1
}
jobs := make(chan refreshJob, numJobs)
returns := make(chan error, numJobs)
// start workers but jobs is blocking for right now
for w := 0; w < numWorkers; w++ {
wg.Add(1)
go refreshWorker(&wg, jobs, returns, data)
}
// get all keys in maps
feedKeys := make([]url.URL, 0, len(data.Feeds))
for k := range data.Feeds {
// we know that the keys are already validated uris, so no err
u, _ := url.ParseRequestURI(k)
feedKeys = append(feedKeys, *u)
}
pageKeys := make([]url.URL, 0, len(data.Pages))
for k := range data.Pages {
u, _ := url.ParseRequestURI(k)
pageKeys = append(pageKeys, *u)
}
data.RUnlock()
for _, v := range feedKeys {
jobs <- refreshJob{"feed", v}
}
for _, v := range pageKeys {
jobs <- refreshJob{"page", v}
}
close(jobs)
wg.Wait()
for i := 0; i < numJobs; i++ {
if err := <-returns; err != nil {
return err
}
}
return nil
}
func refreshWorker(wg *sync.WaitGroup, jobs <-chan refreshJob, returns chan<- error, data *core.FullData) {
defer wg.Done()
for j := range jobs {
if j.typ == "feed" {
returns <- Feed(data, &j.url, "")
} else if j.typ == "page" {
returns <- Page(data, &j.url, "")
}
}
}