Ignore bits above the precision when encoding

.github/workflows/pytest-builds.yml

@@ -25,6 +25,7 @@ jobs:
       with:
         python-version: ${{ matrix.python-version }}
         architecture: ${{ matrix.arch }}
+        allow-prereleases: true
 
     - name: Install package and dependencies
       run: |
@@ -59,6 +60,7 @@ jobs:
       uses: actions/setup-python@v5
       with:
         python-version: ${{ matrix.python-version }}
+        allow-prereleases: true
 
     - name: Install package and dependencies
       run: |
@@ -82,7 +84,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ['3.9', '3.10', '3.11', '3.12', '3.13']
+        python-version: ['3.9', '3.10', '3.11', '3.12', '3.13', '3.14']
 
     steps:
     - uses: actions/checkout@v4
@@ -93,6 +95,7 @@ jobs:
       uses: actions/setup-python@v5
       with:
         python-version: ${{ matrix.python-version }}
+        allow-prereleases: true
 
     - name: Install package and dependencies
       run: |
@@ -106,13 +109,14 @@ jobs:
         pytest --cov openjpeg openjpeg/tests
 
     - name: Install pydicom dev and rerun pytest (3.10+)
-      if: ${{ contains('3.10 3.11 3.12 3.13', matrix.python-version) }}
+      if: ${{ contains('3.10 3.11 3.12 3.13 3.14', matrix.python-version) }}
       run: |
         pip install pylibjpeg
         pip install git+https://github.com/pydicom/pydicom
         pytest --cov openjpeg openjpeg/tests
 
     - name: Switch to current pydicom release and rerun pytest
+      if: ${{ contains('3.10 3.11 3.12 3.13', matrix.python-version) }}
       run: |
         pip uninstall -y pydicom
         pip install pydicom pylibjpeg

.github/workflows/release-wheels.yml

@@ -57,6 +57,9 @@ jobs:
           - os: windows-latest
             python: 313
             platform_id: win32
+          - os: windows-latest
+            python: 314
+            platform_id: win32
 
           # Windows 64 bit
           - os: windows-latest
@@ -74,6 +77,9 @@ jobs:
           - os: windows-latest
             python: 313
             platform_id: win_amd64
+          - os: windows-latest
+            python: 314
+            platform_id: win_amd64
 
           # Linux 64 bit manylinux2014
           - os: ubuntu-latest
@@ -96,6 +102,10 @@ jobs:
             python: 313
             platform_id: manylinux_x86_64
             manylinux_image: manylinux2014
+          - os: ubuntu-latest
+            python: 314
+            platform_id: manylinux_x86_64
+            manylinux_image: manylinux2014
 
           # Linux aarch64
           - os: ubuntu-latest
@@ -113,6 +123,9 @@ jobs:
           - os: ubuntu-latest
             python: 313
             platform_id: manylinux_aarch64
+          - os: ubuntu-latest
+            python: 314
+            platform_id: manylinux_aarch64
 
           # MacOS 13 x86_64
           - os: macos-13
@@ -130,6 +143,9 @@ jobs:
           - os: macos-13
             python: 313
             platform_id: macosx_x86_64
+          - os: macos-13
+            python: 314
+            platform_id: macosx_x86_64
 
     steps:
     - uses: actions/checkout@v4
@@ -194,6 +210,9 @@ jobs:
         - os: macos-14
           python: 313
           platform_id: macosx_arm64
+        - os: macos-14
+          python: 314
+          platform_id: macosx_arm64
 
     steps:
     - uses: actions/checkout@v4

docs/changes/v2.5.0.rst

@@ -0,0 +1,10 @@
+.. _v2.5.0:
+
+2.5.0
+=====
+
+Changes
+.......
+
+* Bits above the precision are now ignored when encoding (:issue:`104`)
+* Supported Python versions are 3.9 to 3.14.

lib/interface/encode.c

@@ -781,20 +781,41 @@ extern int EncodeBuffer(
     OPJ_UINT64 nr_pixels = rows * columns;
     char *data = PyBytes_AsString(src);
     if (bytes_per_pixel == 1) {
+        unsigned char value;
+        unsigned char unsigned_mask = 0xFF >> (8 - bits_stored);
+        unsigned char signed_mask = 0xFF << bits_stored;
+        unsigned char bit_flag = 1 << (bits_stored - 1);
+        unsigned short do_masking = bits_stored < 8;
         for (OPJ_UINT64 ii = 0; ii < nr_pixels; ii++)
         {
             for (p = 0; p < samples_per_pixel; p++)
             {
                 // comps[...].data[...] is OPJ_INT32 -> int32_t
-                image->comps[p].data[ii] = is_signed ? (signed char) *data : (unsigned char) *data;
+                value = (unsigned char) *data;
                 data++;
+
+                if (do_masking) {
+                    // Unsigned: zero out bits above `precision`
+                    // Signed: zero out bits above `precision` if value >= 0, otherwise
+                    //  set them to one
+                    if (is_signed && (bit_flag & value)) {
+                        value = value | signed_mask;
+                    } else {
+                        value = value & unsigned_mask;
+                    }
+                }
+
+                image->comps[p].data[ii] = is_signed ? (signed char) value : value;
             }
         }
     } else if (bytes_per_pixel == 2) {
         unsigned short value;
         unsigned char temp1;
         unsigned char temp2;
-
+        unsigned short unsigned_mask = 0xFFFF >> (16 - bits_stored);
+        unsigned short signed_mask = 0xFFFF << bits_stored;
+        unsigned short bit_flag = 1 << (bits_stored - 1);
+        unsigned short do_masking = bits_stored < 16;
         for (OPJ_UINT64 ii = 0; ii < nr_pixels; ii++)
         {
             for (p = 0; p < samples_per_pixel; p++)
@@ -803,7 +824,16 @@ extern int EncodeBuffer(
                 data++;
                 temp2 = (unsigned char) *data;
                 data++;
+
                 value = (unsigned short) ((temp2 << 8) + temp1);
+                if (do_masking) {
+                    if (is_signed && (bit_flag & value)) {
+                        value = value | signed_mask;
+                    } else {
+                        value = value & unsigned_mask;
+                    }
+                }
+
                 image->comps[p].data[ii] = is_signed ? (signed short) value : value;
             }
         }
@@ -813,7 +843,10 @@ extern int EncodeBuffer(
         unsigned char temp2;
         unsigned char temp3;
         unsigned char temp4;
-
+        unsigned long unsigned_mask = 0xFFFFFFFF >> (32 - bits_stored);
+        unsigned long signed_mask = 0xFFFFFFFF << bits_stored;
+        unsigned long bit_flag = 1 << (bits_stored - 1);
+        unsigned short do_masking = bits_stored < 32;
         for (OPJ_UINT64 ii = 0; ii < nr_pixels; ii++)
         {
             for (p = 0; p < samples_per_pixel; p++)
@@ -827,13 +860,16 @@ extern int EncodeBuffer(
                 temp4 = (unsigned char) * data;
                 data++;
 
-                value = (unsigned long)((temp4 << 24) + (temp3 << 16) + (temp2 << 8) + temp1);
-
-                if (is_signed) {
-                    image->comps[p].data[ii] = (long) value;
-                } else {
-                    image->comps[p].data[ii] = value;
+                value = (unsigned long) ((temp4 << 24) + (temp3 << 16) + (temp2 << 8) + temp1);
+                if (do_masking) {
+                    if (is_signed && (bit_flag & value)) {
+                        value = value | signed_mask;
+                    } else {
+                        value = value & unsigned_mask;
+                    }
                 }
+
+                image->comps[p].data[ii] = is_signed ? (long) value : value;
             }
         }
     }

openjpeg/tests/test_decode.py

@@ -3,8 +3,8 @@
 from io import BytesIO
 
 try:
-    from pydicom.encaps import generate_pixel_data_frame
-    from pydicom.pixel_data_handlers.util import (
+    from pydicom.encaps import generate_frames
+    from pydicom.pixels.utils import (
         reshape_pixel_array,
         pixel_dtype,
     )
@@ -72,10 +72,10 @@ def test_version():
     assert 5 == version[1]
 
 
-def generate_frames(ds):
+def get_frame_generator(ds):
     """Return a frame generator for DICOM datasets."""
     nr_frames = ds.get("NumberOfFrames", 1)
-    return generate_pixel_data_frame(ds.PixelData, nr_frames)
+    return generate_frames(ds.PixelData, number_of_frames=nr_frames)
 
 
 def test_get_format_raises():
@@ -91,7 +91,7 @@ def test_bad_decode():
     """Test trying to decode bad data."""
     index = get_indexed_datasets("1.2.840.10008.1.2.4.90")
     ds = index["966.dcm"]["ds"]
-    frame = next(generate_frames(ds))
+    frame = next(get_frame_generator(ds))
     msg = r"Error decoding the J2K data: failed to decode image"
     with pytest.raises(RuntimeError, match=msg):
         decode(frame)
@@ -108,7 +108,7 @@ def test_decode_bytes(self):
         """Test decoding using bytes."""
         index = get_indexed_datasets("1.2.840.10008.1.2.4.90")
         ds = index["MR_small_jp2klossless.dcm"]["ds"]
-        frame = next(generate_frames(ds))
+        frame = next(get_frame_generator(ds))
         assert isinstance(frame, bytes)
         arr = decode(frame)
         assert arr.flags.writeable
@@ -126,7 +126,7 @@ def test_decode_filelike(self):
         """Test decoding using file-like."""
         index = get_indexed_datasets("1.2.840.10008.1.2.4.90")
         ds = index["MR_small_jp2klossless.dcm"]["ds"]
-        frame = BytesIO(next(generate_frames(ds)))
+        frame = BytesIO(next(get_frame_generator(ds)))
         assert isinstance(frame, BytesIO)
         arr = decode(frame)
         assert arr.flags.writeable
@@ -144,7 +144,7 @@ def test_decode_bad_type_raises(self):
         """Test decoding using invalid type raises."""
         index = get_indexed_datasets("1.2.840.10008.1.2.4.90")
         ds = index["MR_small_jp2klossless.dcm"]["ds"]
-        frame = tuple(next(generate_frames(ds)))
+        frame = tuple(next(get_frame_generator(ds)))
         assert not hasattr(frame, "tell") and not isinstance(frame, bytes)
 
         msg = (
@@ -159,7 +159,7 @@ def test_decode_bad_format_raises(self):
         """Test decoding using invalid jpeg format raises."""
         index = get_indexed_datasets("1.2.840.10008.1.2.4.90")
         ds = index["MR_small_jp2klossless.dcm"]["ds"]
-        frame = next(generate_frames(ds))
+        frame = next(get_frame_generator(ds))
 
         msg = r"Unsupported 'j2k_format' value: 3"
         with pytest.raises(ValueError, match=msg):
@@ -170,7 +170,7 @@ def test_decode_reshape_true(self):
         """Test decoding using invalid jpeg format raises."""
         index = get_indexed_datasets("1.2.840.10008.1.2.4.90")
         ds = index["US1_J2KR.dcm"]["ds"]
-        frame = next(generate_frames(ds))
+        frame = next(get_frame_generator(ds))
 
         arr = decode(frame)
         assert arr.flags.writeable
@@ -205,7 +205,7 @@ def test_decode_reshape_false(self):
         """Test decoding using invalid jpeg format raises."""
         index = get_indexed_datasets("1.2.840.10008.1.2.4.90")
         ds = index["US1_J2KR.dcm"]["ds"]
-        frame = next(generate_frames(ds))
+        frame = next(get_frame_generator(ds))
 
         arr = decode(frame, reshape=False)
         assert arr.flags.writeable
@@ -216,7 +216,7 @@ def test_signed_error(self):
         """Regression test for #30."""
         index = get_indexed_datasets("1.2.840.10008.1.2.4.90")
         ds = index["693_J2KR.dcm"]["ds"]
-        frame = next(generate_frames(ds))
+        frame = next(get_frame_generator(ds))
 
         arr = decode(frame)
         assert -2000 == arr[0, 0]
@@ -372,7 +372,7 @@ def test_jpeg2000r(self, fname, info):
         # info: (rows, columns, spp, bps)
         index = get_indexed_datasets("1.2.840.10008.1.2.4.90")
         ds = index[fname]["ds"]
-        frame = next(generate_frames(ds))
+        frame = next(get_frame_generator(ds))
         arr = decode(BytesIO(frame), reshape=False)
         assert arr.flags.writeable
 
@@ -406,7 +406,7 @@ def test_jpeg2000i(self, fname, info):
         index = get_indexed_datasets("1.2.840.10008.1.2.4.91")
         ds = index[fname]["ds"]
 
-        frame = next(generate_frames(ds))
+        frame = next(get_frame_generator(ds))
         arr = decode(BytesIO(frame), reshape=False)
         assert arr.flags.writeable