Vito Sansevero
58157052c2
Port utils/inspection_response.py to the browser: a strict parse with Python's Decimal spelling, duplicate-key rejection, NaN/Infinity as null and integers past 2^53-1 as text; the prompt, workflow and carrier projection with the depth-64 and 20,000-node ceilings; _metadata; and the 8 MiB omission steps (workflow graph, then prompt graph, then all raw) measured as UTF-8 bytes. local-inspection.mjs builds the inspection_version 1 wire for one chosen File from the Task 1 and Task 2 readers (observation selected-local-file, interim unresolved roles, unavailable library), fits it to the response budget, and exposes localFileAdapter for the shared controller, inspector and viewer: handles are session-owned by a private token, so a tampered or hand-made descriptor is rejected and blob: URLs are never resolved through the API client. projectInspectionPayload admits selected-local-file only for a local-file identity. A Python oracle test compares projectMetadata with _metadata byte for byte over the parity PNG corpus, the committed video fixtures and tag maps, 75 committed adversarial record sets (projection-cases.json) and node-ceiling sets built at test time.
116 lines
7.5 KiB
Markdown
116 lines
7.5 KiB
Markdown
# Browser media fixtures
|
|
|
|
`tiny.webm` is an original, silent, one-second VP8 color test pattern generated
|
|
for this repository. It contains no user media and is dedicated to the public
|
|
domain under CC0-1.0.
|
|
|
|
Reproduce with FFmpeg:
|
|
|
|
```sh
|
|
ffmpeg -f lavfi -i testsrc=size=32x32:rate=4:duration=1 \
|
|
-c:v libvpx -pix_fmt yuv420p -an tiny.webm
|
|
```
|
|
|
|
Browser tests use the committed file; FFmpeg is not required to run them.
|
|
|
|
# Local Inspect header fixtures
|
|
|
|
`local-inspect/tiny.jpg` (48x32) and `local-inspect/tiny.webp` (40x24) are original
|
|
synthetic FFmpeg color test patterns, under 4 KB, dedicated to the public domain
|
|
under CC0-1.0. `tests/js/png-local-reader.test.js` reads their dimensions with the
|
|
browser header reader and `tests/test_local_inspect_parity.py` checks the same two
|
|
files against Pillow.
|
|
|
|
Reproduce with FFmpeg:
|
|
|
|
```sh
|
|
ffmpeg -f lavfi -i testsrc=size=48x32:rate=1 -frames:v 1 -q:v 10 tiny.jpg
|
|
ffmpeg -f lavfi -i testsrc=size=40x24:rate=1 -frames:v 1 -c:v libwebp -quality 40 tiny.webp
|
|
```
|
|
|
|
# Local Inspect video fixtures
|
|
|
|
`local-inspect/core-faststart.mp4`, `core-moov-end.mp4`, `core.mov`, `core.m4v`,
|
|
`core.mkv`, `core-tags.webm`, `legacy-comment.mp4` and `legacy-comment.webm` are
|
|
0.2-second 16x16 black clips (each under 4 KB) carrying an original synthetic
|
|
ComfyUI graph in their container tags, dedicated to the public domain under
|
|
CC0-1.0. The `core-*` files carry one tag per key as core SaveVideo/SaveWEBM write
|
|
them (MP4/MOV/M4V through `-movflags use_metadata_tags`; Matroska upper-cases the
|
|
names); `core-faststart.mp4` also carries `parameters` and
|
|
`comfyassets_prompt_pairs`. The `legacy-comment.*` files carry the old
|
|
VideoHelperSuite `comment` JSON envelope (an ilst `©cmt` atom in MP4, a `COMMENT`
|
|
tag in WebM). `video-tags-v1.json` records `ffprobe -show_entries format_tags -of
|
|
json` for each file (ffmpeg 9.0.2); `tests/test_utils_video_metadata.py`
|
|
`VideoFixtureFfprobeTests` re-checks it when ffprobe is installed, and
|
|
`tests/js/workspace-local-video.test.js` plus `tests/test_local_inspect_parity.py`
|
|
hold the browser video reader to the same maps.
|
|
|
|
Reproduce with FFmpeg, from `tests/fixtures/local-inspect`:
|
|
|
|
```sh
|
|
PROMPT='{"3":{"class_type":"KSampler","inputs":{"seed":7,"steps":20,"cfg":8.0,"sampler_name":"euler","scheduler":"normal","denoise":1.0,"model":["4",0],"positive":["6",0],"negative":["7",0],"latent_image":["5",0]}},"4":{"class_type":"CheckpointLoaderSimple","inputs":{"ckpt_name":"synthetic-v1.safetensors"}},"5":{"class_type":"EmptyLatentImage","inputs":{"width":512,"height":512,"batch_size":1}},"6":{"class_type":"CLIPTextEncode","inputs":{"text":"a red fox in snow, detailed fur","clip":["4",1]}},"7":{"class_type":"CLIPTextEncode","inputs":{"text":"blurry, lowres","clip":["4",1]}},"8":{"class_type":"VAEDecode","inputs":{"samples":["3",0],"vae":["4",2]}},"9":{"class_type":"SaveVideo","inputs":{"filename_prefix":"video/synthetic","video":["8",0]}}}'
|
|
WORKFLOW='{"last_node_id":9,"nodes":[{"id":3,"type":"KSampler","widgets_values":[7,"fixed",20,8.0,"euler","normal",1.0]},{"id":6,"type":"CLIPTextEncode","widgets_values":["a red fox in snow, detailed fur"]},{"id":7,"type":"CLIPTextEncode","widgets_values":["blurry, lowres"]}],"links":[],"version":0.4}'
|
|
PARAMETERS=$'a red fox in snow, detailed fur\nNegative prompt: blurry, lowres\nSteps: 20, Sampler: Euler, CFG scale: 8.0, Seed: 7, Size: 512x512'
|
|
PAIRS='{"version":1,"nodes":{"6":{"records":[{"positive_text":"a red fox in snow, detailed fur","negative_text":"blurry, lowres","negative_effective_text":"blurry, lowres","prepend_text":"","append_text":""}]}}}'
|
|
COMMENT=$(PROMPT="$PROMPT" WORKFLOW="$WORKFLOW" python3 -c 'import json,os; print(json.dumps({"prompt": os.environ["PROMPT"], "workflow": json.loads(os.environ["WORKFLOW"])}))')
|
|
SRC=(-v quiet -y -f lavfi -i "color=c=black:s=16x16:d=0.2")
|
|
ffmpeg "${SRC[@]}" -movflags use_metadata_tags+faststart -metadata prompt="$PROMPT" -metadata workflow="$WORKFLOW" -metadata parameters="$PARAMETERS" -metadata comfyassets_prompt_pairs="$PAIRS" core-faststart.mp4
|
|
ffmpeg "${SRC[@]}" -movflags use_metadata_tags -metadata prompt="$PROMPT" -metadata workflow="$WORKFLOW" core-moov-end.mp4
|
|
ffmpeg "${SRC[@]}" -movflags use_metadata_tags -metadata prompt="$PROMPT" -metadata workflow="$WORKFLOW" core.mov
|
|
ffmpeg "${SRC[@]}" -movflags use_metadata_tags -metadata prompt="$PROMPT" -metadata workflow="$WORKFLOW" core.m4v
|
|
ffmpeg "${SRC[@]}" -metadata prompt="$PROMPT" -metadata workflow="$WORKFLOW" core.mkv
|
|
ffmpeg "${SRC[@]}" -c:v libvpx -metadata prompt="$PROMPT" -metadata workflow="$WORKFLOW" core-tags.webm
|
|
ffmpeg "${SRC[@]}" -metadata comment="$COMMENT" legacy-comment.mp4
|
|
ffmpeg "${SRC[@]}" -c:v libvpx -metadata comment="$COMMENT" legacy-comment.webm
|
|
for f in core-faststart.mp4 core-moov-end.mp4 core.mov core.m4v core.mkv core-tags.webm legacy-comment.mp4 legacy-comment.webm; do
|
|
ffprobe -v quiet -show_entries format_tags -of json "$f" # -> video-tags-v1.json "fixtures"[f] = .format.tags
|
|
done
|
|
```
|
|
|
|
# Local Inspect projection cases
|
|
|
|
`local-inspect/projection-cases.json` holds synthetic record sets (prompt, workflow,
|
|
parameters and pair-record texts, optional `record_formats`, a video `carrier_text`
|
|
and `limited` records) that exercise every branch of the server's inspection
|
|
projection: Python's Decimal spellings and exact-construction limits, the 4300-digit
|
|
integer limit, duplicate keys, NaN/Infinity, the depth-64 and node ceilings, prompt
|
|
node key filtering and the carrier rule. `records: null` stands for an unreadable
|
|
file. `expect` pins both graph statuses; `needs` names an interpreter feature a case
|
|
depends on. `tests/js/workspace-local-projection.test.js` runs them through
|
|
`web/js/workspace/local-projection.mjs`, and `tests/test_local_inspect_parity.py`
|
|
asserts that its answer equals `utils.inspection_response._metadata` byte for byte.
|
|
All texts are original synthetic content, dedicated to the public domain under
|
|
CC0-1.0.
|
|
|
|
# Inspection parity fixtures
|
|
|
|
`inspection_parity/` holds small synthetic PNGs (`f01b.png` ... `f24.png`, each
|
|
under 32 KiB) and `vectors.json`, used by the #240 inspection parity tests. Every
|
|
graph, prompt text, model name and filename prefix in them is original synthetic
|
|
content written for this repository. Nothing comes from user output. They are
|
|
dedicated to the public domain under CC0-1.0.
|
|
|
|
`tests/inspection_parity.py` builds every fixture from code. It writes the
|
|
`prompt` and `workflow` records the way core ComfyUI `SaveImage` does (Pillow
|
|
`PngInfo.add_text`, `compress_level=4`, a 64x64 constant colour image), and
|
|
splices extra raw `tEXt` chunks with real CRC32 for the duplicate-chunk shapes
|
|
`f15` and `f16`. The larger shapes `f01`, `f04` and `f05` are built at test
|
|
time instead of committed. Reproduce the committed files with:
|
|
|
|
```sh
|
|
python3 -m tests.inspection_parity --write tests/fixtures/inspection_parity
|
|
```
|
|
|
|
`tests/test_inspection_parity_fixtures.py` compares the decoded text chunks with
|
|
the builders (not bytes, because zlib output differs between platforms).
|
|
|
|
`vectors.json` holds an `inspection` and a `default` vector for every fixture,
|
|
committed or built at test time (test-time fixtures are built into a temporary
|
|
directory). `inspection` is the workspace inspector path, with the facts the
|
|
workspace shows. `default` pins only the resolver's default profile, run on the
|
|
role records decoded from the PNG text chunks. Ingestion's own readers are covered
|
|
by their own suites. The expectations start as the observed baseline. A change
|
|
that alters an outcome edits the vector in the same commit.
|
|
`tests/js/legacy-metadata-characterization.test.js` pins what
|
|
the legacy viewers show on the same files.
|