feat(feeds): improve RSS size parsing (#1367)

* fix(feeds): Parse multiple sizes.

* refactor: Test_pullSizeFromDescription

* refactor: make test human readable

added helper function

* multi

* Agnewwwwww

* .

* humanize

* humanize

---------

Co-authored-by: soup <soup@r4tio.dev>

internal/feed/rss.go

@@ -22,6 +22,7 @@ import (
 var (
 	rxpSize      = regexp.MustCompile(`(?mi)(([0-9.]+)\s*(b|kb|kib|kilobyte|mb|mib|megabyte|gb|gib|gigabyte|tb|tib|terabyte))`)
 	rxpFreeleech = regexp.MustCompile(`(?mi)(\bfreeleech\b)`)
+	rxpHTML      = regexp.MustCompile(`(?mi)<.*?>`)
 )
 
 type RSSJob struct {
@@ -190,7 +191,7 @@ func (j *RSSJob) processItem(item *gofeed.Item) *domain.Release {
 		}
 
 		if element.ContentLength > 0 {
-			if uint64(element.ContentLength) != rls.Size {
+			if uint64(element.ContentLength) > rls.Size {
 				rls.Size = uint64(element.ContentLength)
 			}
 		}
@@ -210,10 +211,8 @@ func (j *RSSJob) processItem(item *gofeed.Item) *domain.Release {
 		rls.Description = item.Description
 
 		if rls.Size == 0 {
-			hrSize := readSizeFromDescription(item.Description)
+			readSizeFromDescription(item.Description, rls)
-			rls.ParseSizeBytesString(hrSize)
+			j.Log.Trace().Msgf("Set new size %d from description", rls.Size)
-
-			j.Log.Trace().Msgf("Set new size %d from description %s", rls.Size, hrSize)
 		}
 	}
 
@@ -326,13 +325,11 @@ func isFreeleech(str []string) bool {
 }
 
 // readSizeFromDescription get size from description
-func readSizeFromDescription(str string) string {
+func readSizeFromDescription(str string, r *domain.Release) {
-	matches := rxpSize.FindStringSubmatch(str)
+	clean := rxpHTML.ReplaceAllString(str, " ")
-	if matches == nil {
+	for _, sz := range rxpSize.FindAllString(clean, -1) {
-		return ""
+		r.ParseSizeBytesString(sz)
 	}
-
-	return matches[1]
 }
 
 // itemCustomElement

internal/feed/rss_test.go

@@ -10,6 +10,7 @@ import (
 	"github.com/autobrr/autobrr/internal/domain"
 	"github.com/autobrr/autobrr/internal/release"
 
+	"github.com/dustin/go-humanize"
 	"github.com/mmcdole/gofeed"
 	"github.com/rs/zerolog"
 	"github.com/stretchr/testify/assert"
@@ -221,25 +222,70 @@ func Test_isMaxAge(t *testing.T) {
 }
 
 func Test_readSizeFromDescription(t *testing.T) {
-	type args struct {
+	t.Parallel()
-		str string
+
-	}
 	tests := []struct {
 		name string
-		args args
+		str  string
 		want string
 	}{
-		{name: "size", args: args{"Size: 12GB"}, want: "12GB"},
+		{
-		{name: "size_1", args: args{"Size: 12 GB"}, want: "12 GB"},
+			name: "with size in GB",
-		{name: "size_2", args: args{"Size: 12 GiB"}, want: "12 GiB"},
+			str:  "Size: 12GB",
-		{name: "size_3", args: args{"Size: 537 MiB"}, want: "537 MiB"},
+			want: "12GB",
-		{name: "size_4", args: args{"<strong>Size</strong>: 20.48 GiB<br>"}, want: "20.48 GiB"},
+		},
-		{name: "size_5", args: args{"file.name-GROUP / 20.48 GiB / x265"}, want: "20.48 GiB"},
+		{
-		{name: "size_6", args: args{"<strong>Uploaded</strong>: 38 minutes ago<br>"}, want: ""},
+			name: "with size in GB with space",
+			str:  "Size: 12 GB",
+			want: "12GB",
+		},
+		{
+			name: "with size in GiB",
+			str:  "Size: 12 GiB",
+			want: "12GiB",
+		},
+		{
+			name: "with size in MiB",
+			str:  "Size: 537 MiB",
+			want: "537MiB",
+		},
+		{
+			name: "with HTML tags",
+			str:  "<strong>Size</strong>: 20.48 GiB<br>",
+			want: "20.48GiB",
+		},
+		{
+			name: "with additional text",
+			str:  "file.name-GROUP / 20.48 GiB / x265",
+			want: "20.48GiB",
+		},
+		{
+			name: "without size info",
+			str:  "<strong>Uploaded</strong>: 38 minutes ago<br>",
+			want: "0B",
+		},
+		{
+			name: "multiple sizes",
+			str:  "<strong>Uploaded</strong>: 38B minutes ago<br>Size: 32GB",
+			want: "32GB",
+		},
 	}
+
 	for _, tt := range tests {
+		tt := tt
 		t.Run(tt.name, func(t *testing.T) {
-			assert.Equalf(t, tt.want, readSizeFromDescription(tt.args.str), "readSizeFromDescription(%v)", tt.args.str)
+			t.Parallel()
+
+			wantBytes, err := humanize.ParseBytes(tt.want)
+			if err != nil {
+				t.Fatalf("Failed to parse size string %q: %v", tt.want, err)
+			}
+
+			r := &domain.Release{}
+			readSizeFromDescription(tt.str, r)
+			if r.Size != wantBytes {
+				t.Errorf("readSizeFromDescription(%q) got %v bytes, want %v bytes", tt.str, r.Size, wantBytes)
+			}
 		})
 	}
 }