Better error message for unseekable formats (#1647)

Author: NicolasHug

src/torchcodec/_core/SingleStreamDecoder.cpp

@@ -12,6 +12,7 @@
 #include <iostream>
 #include <limits>
 #include <numeric>
+#include <sstream>
 #include <string_view>
 #include "Demuxer.h"
 #include "Metadata.h"
@@ -24,6 +25,29 @@ extern "C" {
 
 namespace facebook::torchcodec {
 
+namespace {
+
+// FFmpeg reports "this seek cannot be performed" as a bare -1, i.e. EPERM,
+// which renders as the very misleading "Operation not permitted". It covers
+// both a target that the demuxer can't reach and a demuxer with no seeking
+// support whatsoever.
+std::string get_seek_error_message(
+    const AVFormatContext* format_context,
+    int64_t desired_pts,
+    int status) {
+  std::stringstream ss;
+  ss << "Could not seek file to pts=" << desired_pts << ": "
+     << get_ffmpeg_error_string_from_error_code(status) << ".";
+  if (status == AVERROR(EPERM)) {
+    ss << " This is either because that timestamp is out of range, or because"
+       << " the '" << format_context->iformat->name << "' format does not"
+       << " support seeking.";
+  }
+  return ss.str();
+}
+
+} // namespace
+
 // --------------------------------------------------------------------------
 // CONSTRUCTORS, INITIALIZATION, DESTRUCTORS
 // --------------------------------------------------------------------------
@@ -350,9 +374,7 @@ void SingleStreamDecoder::scan_file_and_update_metadata_and_index() {
   // Reset the seek-cursor back to the beginning.
   int status = avformat_seek_file(format_context_.get(), 0, INT64_MIN, 0, 0, 0);
   STD_TORCH_CHECK(
-      status >= 0,
-      "Could not seek file to pts=0: ",
-      get_ffmpeg_error_string_from_error_code(status));
+      status >= 0, get_seek_error_message(format_context_.get(), 0, status));
 
   // Sort all frames by their pts.
   sort_all_frames();
@@ -1511,10 +1533,7 @@ bool SingleStreamDecoder::maybe_seek_to_before_desired_pts() {
       0);
   STD_TORCH_CHECK(
       status >= 0,
-      "Could not seek file to pts=",
-      std::to_string(desired_pts),
-      ": ",
-      get_ffmpeg_error_string_from_error_code(status));
+      get_seek_error_message(format_context_.get(), desired_pts, status));
 
   decode_stats_.num_flushes++;
   device_interface_->flush();

test/resources/sine_mono_mp3.swf

@@ -143,6 +143,7 @@
     TESTSRC2_ODD_WIDTH_VP9,
     TESTSRC2_ODD_WIDTH_VP9_10BIT,
     TRANSPARENT_GIF,
+    UNSEEKABLE_SWF,
     WAV_ODD_DATA_TRAILING_CHUNK,
 )
 
@@ -2861,6 +2862,14 @@ def test_fresh_decoder_seek(self, tmp_path):
         AudioEncoder(torch.rand(1, 1000), sample_rate=16000).to_file(path)
         AudioDecoder(path).get_all_samples()
 
+    def test_unseekable_format(self):
+        decoder = AudioDecoder(UNSEEKABLE_SWF.path)
+        samples = decoder.get_all_samples()
+        assert samples.data.shape == (1, 89856)
+
+        with pytest.raises(RuntimeError, match="'swf' format does not support seeking"):
+            decoder.get_samples_played_in_range(start_seconds=1)
+
     @pytest.mark.parametrize("asset", (NASA_AUDIO, NASA_AUDIO_MP3))
     @pytest.mark.parametrize("stop_seconds", (None, "duration", 99999999))
     def test_get_all_samples_with_range(self, asset, stop_seconds):

test/test_decoders.py

@@ -1710,3 +1710,20 @@ def sample_format(self) -> str:
         )
     },
 )
+
+# Generated with:
+# ffmpeg -y -f lavfi -i "sine=frequency=440:duration=2" -c:a mp3 -b:a 32k -ar 44100 -ac 1 test/resources/sine_mono_mp3.swf
+UNSEEKABLE_SWF = TestAudio(
+    filename="sine_mono_mp3.swf",
+    default_stream_index=0,
+    frames={0: {}},
+    stream_infos={
+        0: TestAudioStreamInfo(
+            sample_rate=44_100,
+            num_channels=1,
+            duration_seconds=2.2739909297052154,
+            num_frames=78,
+            sample_format="fltp",
+        )
+    },
+)