feat: parallel download of deb packages

.github/workflows/debian-cache.yml

@@ -68,9 +68,7 @@ jobs:
 
         def get_kernels() -> bool:
             status, output = getstatusoutput(
-                "./out-of-tree --log-level=warn "
-                "distro debian get-deb "
-                "--ignore-cached --max=16"
+                "./out-of-tree distro debian fetch --max=16"
             )
             logging.info(output)
             return status == 0

distro.go

@@ -1,11 +1,14 @@
 package main
 
 import (
+	"context"
 	"os"
 	"path/filepath"
 	"regexp"
+	"time"
 
 	"github.com/cavaliergopher/grab/v3"
+	"github.com/remeh/sizedwaitgroup"
 	"github.com/rs/zerolog/log"
 
 	"code.dumpstack.io/tools/out-of-tree/cache"
@@ -20,8 +23,8 @@ type DistroCmd struct {
 }
 
 type DebianCmd struct {
-	Cache  DebianCacheCmd  `cmd:"" help:"populate cache"`
-	GetDeb DebianGetDebCmd `cmd:"" help:"download deb packages"`
+	Cache DebianCacheCmd `cmd:"" help:"populate cache"`
+	Fetch DebianFetchCmd `cmd:"" help:"download deb packages"`
 }
 
 type DebianCacheCmd struct {
@@ -47,16 +50,81 @@ func (cmd *DebianCacheCmd) Run() (err error) {
 	return
 }
 
-type DebianGetDebCmd struct {
+type DebianFetchCmd struct {
 	Path   string `help:"path to download directory" type:"existingdir" default:"./"`
 	Regexp string `help:"match deb pkg names by regexp" default:".*"`
 
-	IgnoreCached bool `help:"ignore packages found on remote mirror"`
+	IgnoreMirror bool `help:"ignore check if packages on the mirror"`
 
 	Max int `help:"do not download more than X" default:"100500"`
+
+	Threads int `help:"parallel download threads" default:"8"`
+
+	Timeout time.Duration `help:"timeout for each download" default:"1m"`
+
+	swg        sizedwaitgroup.SizedWaitGroup
+	hasResults bool
 }
 
-func (cmd DebianGetDebCmd) Run() (err error) {
+func (cmd *DebianFetchCmd) fetch(pkg snapshot.Package) {
+	flog := log.With().
+		Str("pkg", pkg.Deb.Name).
+		Logger()
+
+	defer cmd.swg.Done()
+
+	if !cmd.IgnoreMirror {
+		flog.Debug().Msg("check mirror")
+		found, _ := cache.PackageURL(config.Debian, pkg.Deb.URL)
+		if found {
+			flog.Info().Msg("found on the mirror")
+			return
+		}
+	}
+
+	target := filepath.Join(cmd.Path, filepath.Base(pkg.Deb.URL))
+
+	if fs.PathExists(target) {
+		flog.Debug().Msg("already exists")
+		return
+	}
+
+	tmp, err := os.MkdirTemp(cmd.Path, "tmp-")
+	if err != nil {
+		flog.Fatal().Err(err).Msg("mkdir")
+		return
+	}
+	defer os.RemoveAll(tmp)
+
+	flog.Info().Msg("fetch")
+	flog.Debug().Msg(pkg.Deb.URL)
+
+	ctx, cancel := context.WithTimeout(context.Background(), cmd.Timeout)
+	defer cancel()
+
+	req, err := grab.NewRequest(tmp, pkg.Deb.URL)
+	if err != nil {
+		flog.Warn().Err(err).Msg("cannot create request")
+		return
+	}
+	req = req.WithContext(ctx)
+
+	resp := grab.DefaultClient.Do(req)
+	if err := resp.Err(); err != nil {
+		flog.Warn().Err(err).Msg("request cancelled")
+		return
+	}
+
+	err = os.Rename(resp.Filename, target)
+	if err != nil {
+		flog.Fatal().Err(err).Msg("mv")
+	}
+
+	cmd.hasResults = true
+	cmd.Max--
+}
+
+func (cmd *DebianFetchCmd) Run() (err error) {
 	re, err := regexp.Compile(cmd.Regexp)
 	if err != nil {
 		log.Fatal().Err(err).Msg("regexp")
@@ -79,54 +147,18 @@ func (cmd DebianGetDebCmd) Run() (err error) {
 		}
 	}
 
-	tmp, err := os.MkdirTemp(cmd.Path, "tmp-")
-	if err != nil {
-		return
-	}
-	defer os.RemoveAll(tmp)
-
-	hasresults := false
-
+	cmd.swg = sizedwaitgroup.New(cmd.Threads)
 	for _, pkg := range packages {
 		if cmd.Max <= 0 {
 			break
 		}
 
-		if cmd.IgnoreCached {
-			log.Debug().Msgf("check cache for %s", pkg.Deb.Name)
-			found, _ := cache.PackageURL(config.Debian, pkg.Deb.URL)
-			if found {
-				log.Debug().Msgf("%s already cached", pkg.Deb.Name)
-				continue
-			}
-		}
-
-		target := filepath.Join(cmd.Path, filepath.Base(pkg.Deb.URL))
-
-		if fs.PathExists(target) {
-			log.Info().Msgf("%s already exists", pkg.Deb.URL)
-			continue
-		}
-
-		log.Info().Msgf("downloading %s", pkg.Deb.URL)
-
-		resp, err := grab.Get(tmp, pkg.Deb.URL)
-		if err != nil {
-			err = nil
-			log.Warn().Err(err).Msg("download")
-			continue
-		}
-
-		err = os.Rename(resp.Filename, target)
-		if err != nil {
-			log.Fatal().Err(err).Msg("mv")
-		}
-
-		hasresults = true
-		cmd.Max--
+		cmd.swg.Add()
+		go cmd.fetch(pkg)
 	}
+	cmd.swg.Wait()
 
-	if !hasresults {
+	if !cmd.hasResults {
 		log.Fatal().Msg("no packages found to download")
 	}
 	return