feat: better file sorting

pkg/gotenberg/sort.go

@@ -0,0 +1,69 @@
+package gotenberg
+
+import (
+	"regexp"
+	"sort"
+	"strconv"
+)
+
+// AlphanumericSort implements sort.Interface and helps to sort strings
+// alphanumerically.
+//
+// See https://github.com/gotenberg/gotenberg/issues/805.
+type AlphanumericSort []string
+
+func (s AlphanumericSort) Len() int {
+	return len(s)
+}
+
+func (s AlphanumericSort) Swap(i, j int) {
+	s[i], s[j] = s[j], s[i]
+}
+
+func (s AlphanumericSort) Less(i, j int) bool {
+	numI, restI := extractPrefix(s[i])
+	numJ, restJ := extractPrefix(s[j])
+
+	// Compares numerical prefixes if they exist.
+	if numI != -1 && numJ != -1 {
+		if numI != numJ {
+			return numI < numJ
+		}
+		// If numbers are equal, falls back to string comparison of the rest.
+		return restI < restJ
+	}
+
+	// If one has a numerical prefix and the other doesn't, the one with the
+	// number comes first.
+	if numI != -1 {
+		return true
+	}
+	if numJ != -1 {
+		return false
+	}
+
+	// If neither has a numerical prefix, compare as strings
+	return s[i] < s[j]
+}
+
+// extractPrefix attempts to extract a numerical prefix and the rest of the filename
+func extractPrefix(filename string) (int, string) {
+	matches := numPrefixRegexp.FindStringSubmatch(filename)
+	if len(matches) > 2 {
+		prefix, err := strconv.Atoi(matches[1])
+		if err == nil {
+			return prefix, matches[2]
+		}
+	}
+
+	// Returns -1 if no numerical prefix is found, indicating to just compare
+	// as strings.
+	return -1, filename
+}
+
+var numPrefixRegexp = regexp.MustCompile(`^(\d+)(.*)$`)
+
+// Interface guard.
+var (
+	_ sort.Interface = (*AlphanumericSort)(nil)
+)

pkg/gotenberg/sort_test.go

@@ -0,0 +1,34 @@
+package gotenberg
+
+import (
+	"reflect"
+	"sort"
+	"testing"
+)
+
+func TestAlphanumericSort(t *testing.T) {
+	for _, tc := range []struct {
+		scenario     string
+		values       []string
+		expectedSort []string
+	}{
+		{
+			scenario:     "numeric and letters",
+			values:       []string{"10qux.pdf", "2_baz.txt", "2_aza.txt", "1bar.pdf", "Afoo.txt", "Bbar.docx", "25zeta.txt", "3.pdf", "4_foo.pdf"},
+			expectedSort: []string{"1bar.pdf", "2_aza.txt", "2_baz.txt", "3.pdf", "4_foo.pdf", "10qux.pdf", "25zeta.txt", "Afoo.txt", "Bbar.docx"},
+		},
+		{
+			scenario:     "hrtime (PHP library)",
+			values:       []string{"245654773395259", "245654773395039", "245654773395149", "245654773394919", "245654773394369"},
+			expectedSort: []string{"245654773394369", "245654773394919", "245654773395039", "245654773395149", "245654773395259"},
+		},
+	} {
+		t.Run(tc.scenario, func(t *testing.T) {
+			sort.Sort(AlphanumericSort(tc.values))
+
+			if !reflect.DeepEqual(tc.values, tc.expectedSort) {
+				t.Fatalf("expected %+v but got: %+v", tc.expectedSort, tc.values)
+			}
+		})
+	}
+}

pkg/modules/api/formdata.go

@@ -11,6 +11,8 @@ import (
 	"time"
 
 	"go.uber.org/multierr"
+
+	"github.com/gotenberg/gotenberg/v8/pkg/gotenberg"
 )
 
 // FormData is a helper for validating and hydrating values from a
@@ -304,7 +306,7 @@ func (form *FormData) paths(extensions []string, target *[]string) *FormData {
 	}
 
 	// See https://github.com/gotenberg/gotenberg/issues/139.
-	sort.Strings(*target)
+	sort.Sort(gotenberg.AlphanumericSort(*target))
 
 	return form
 }