Fix inconsistent fingerprints

.github/workflows/main.yml

@@ -28,6 +28,7 @@ jobs:
 
       - name: Install system dependencies
         run: |
+          sudo apt-get update
           sudo apt install --yes --no-install-recommends \
           gstreamer1.0-plugins-good \
           gstreamer1.0-plugins-bad \

acoustid.py

@@ -221,13 +221,21 @@ def fingerprint(samplerate, channels, pcmiter, maxlength=MAX_AUDIO_LENGTH):
         fper = chromaprint.Fingerprinter()
         fper.start(samplerate, channels)
 
-        position = 0  # Samples of audio fed to the fingerprinter.
-        for block in pcmiter:
-            fper.feed(block)
-            position += len(block) // 2  # 2 bytes/sample.
-            if position >= endposition:
+        position = 0
+        while position < endposition:
+            try:
+                block = next(pcmiter)
+            except StopIteration:
+                # No more data
                 break
 
+            # Calculate remaining samples needed
+            remaining = endposition - position
+            # Feed only up to remaining samples
+            bytes_to_feed = min(len(block), remaining * 2)
+            fper.feed(block[:bytes_to_feed])
+            position += bytes_to_feed // 2
+
         return fper.finish()
     except chromaprint.FingerprintError:
         raise FingerprintGenerationError("fingerprint calculation failed")

test/test_acoustid.py

@@ -257,6 +257,47 @@ def raise_error(*_):
         with pytest.raises(acoustid.FingerprintGenerationError):
             acoustid.fingerprint(44100, 2, [b"\x00\x00"])
 
+    @pytest.mark.parametrize(
+        "block_sizes",
+        [
+            (2, 4),
+        ],
+    )
+    def test_fingerprint_inconsistent_block_sizes(
+        self, monkeypatch, block_sizes: list[int]
+    ):
+        """Different block sizes may produce inconsistent results."""
+        from unittest.mock import Mock
+
+        def chunk_by_size(lst, size):
+            for i in range(0, len(lst), size):
+                yield lst[i : i + size]
+
+        data = bytes(range(100))  # b'\x00\x01\x02...\x63'
+
+        monkeypatch.setattr("chromaprint.Fingerprinter.start", Mock())
+        monkeypatch.setattr("chromaprint.Fingerprinter.finish", Mock())
+
+        values = []
+        for b in block_sizes:
+            mock = Mock()
+            monkeypatch.setattr("chromaprint.Fingerprinter.feed", mock)
+
+            acoustid.fingerprint(
+                1,
+                1,
+                chunk_by_size(data, b),
+                5,
+                # Twice the maxlength is consumed/feed -> 5*2 = 10
+            )
+
+            flattened_bytes = b"".join(c.args[0] for c in mock.call_args_list)
+            values.append(flattened_bytes)
+
+        # Regardless of the block chunking
+        # we always feed the same bytes!
+        assert len(set(values)) == 1
+
 
 class TestSubmissions:
     @pytest.fixture(autouse=True)