[stevevanhooser]

MATLAB half of the ndr.reader.video symmetry tests. The Python half and the Python port are in VH-Lab/NDR-python#33. Both use branch claude/clever-bardeen-hf5jgx, so NDR-python's test-symmetry.yml checks out this branch.

Changes

• tools/tests/+ndr/+symmetry/+makeArtifacts/+reader/readVideo.m: runs the frame API on the two example videos and writes metadata.json and readFrames.json. These record:
  • numframes, framesize, dimensionorder, datatype, frametimes, t0_t1, epochclock, channels and image metadata;
  • exact pixels of the first and last RGB frame per channel, a SelectC read, and an out-of-order greyscale read.
• tools/tests/+ndr/+symmetry/+readArtifacts/+reader/readVideo.m: verifies those artifacts from both sources (matlabArtifacts, pythonArtifacts). It mirrors NDR-python's tests/symmetry/{make,read}_artifacts/reader/test_read_video.py.
• example_data/example_video.avi: 6-frame 12×16 uncompressed 24-bit RGB, 10 fps. R = 20*frame + col, G = 10*row + frame, B = 200 - 10*frame - col.
• example_data/example_video_grey.avi: the same layout as 8-bit with a 256-entry grey palette ('Grayscale AVI' storage). Pixel value is 30*frame + 4*row + col.
• example_data/Readme.md: documents both files. The README stays identical to NDR-python's.

Design notes

• Shared, byte-identical inputs: the two AVIs are generated by NDR-python's tools/make_example_data.py, like the other synthetic fixtures. That script writes the AVI bytes directly, laid out as VideoWriter writes them: BI_RGB frames stored bottom-up.
• Uncompressed: VideoReader can read these on every platform (MPEG-4 needs GStreamer on Linux), and since nothing is lossy the two ports must agree pixel for pixel.
• Time tolerance: frame times are compared to 1e-6 s rather than the 1e-9 the TIFF readers use, because each port measures them through its own decoder rather than reading a sidecar file.
• Already useful: the bottom-up greyscale fixture exposed a bug on the Python side. MATLAB-written greyscale AVIs came back upside down. It is fixed in NDR-python#33.

Verification

I wrote this without access to MATLAB. NDR-python's cross-language symmetry run on this branch (run 39476490054, attempt 2) is the first time these files ran, and they passed:

• Stage 1: MATLAB makeArtifacts/readVideo passed (7/7 makeArtifacts tests).
• Stage 2: Python read the MATLAB artifacts: metadata, image metadata and frames for both videos all passed.
• Stage 4: MATLAB readArtifacts/readVideo passed for both SourceType=matlabArtifacts and SourceType=pythonArtifacts (14/14 readArtifacts tests).

VideoReader reads the grey-palette AVI as Grayscale (C=1), matching Python pixel for pixel.

🤖 Generated with Claude Code

https://claude.ai/code/session_01QjkvVXsdCVw6NNe5SnGNyX

[github-actions]

Test Results

415 tests  +3   412 ✅ +3   37s ⏱️ -7s
 54 suites +2     3 💤 ±0 
  1 files   ±0     0 ❌ ±0 

Results for commit a07ed0b. ± Comparison against base commit 00cbbf8.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 54.94%. Comparing base (00cbbf8) to head (a07ed0b).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #144      +/-   ##
==========================================
+ Coverage   54.88%   54.94%   +0.06%     
==========================================
  Files         164      164              
  Lines        9598     9598              
==========================================
+ Hits         5268     5274       +6     
+ Misses       4330     4324       -6     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.