unroll / tools /make_manifest.py
SharpMeow's picture
Multi-scroll support: build.py and fetch.py take any open-data scroll (per-scroll voxel size, X-ray scan, renumbering by radius, optional ink, a second 1 um ink layer), make_manifest.py, the scroll picker's prize eligible badge and eligible-only toggle (window.unrollScrolls), and a 'Why N ink layers?' box written from each manifest's ink layers
cf20894
Raw History Blame Contribute Delete
7.69 kB
"""Write scrolls/<id>/manifest.json and the registry entry for a scroll built by tools/build.py.
python tools/make_manifest.py PHerc0814
Text here is our own wording: no titles, authors or work names, nothing copied from the data browser.
The prize field mirrors the "prize eligible" badge on scrollprize.org (see https://scrollprize.org/prizes).
"""
import json
import os
import sys
ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
CITE = ("Giorgio Angelotti, Stephen Parsons, Sean Johnson, Elian Rafael Dal Prà, Johannes Rudolph, Paul Tafforeau, "
"Alessandro Mirone, Paul Henderson, Hendrik Schilling, Forrest McDonald, David Josey, Youssef Nader, C. Seth Parker, "
"W. Brent Seales. Vesuvius Challenge - CT Scans of Herculaneum Papyri. Vesuvius Challenge.")
EDUCELAB = ("EduceLab-Scrolls: Parsons, S., Parker, C. S., Chapman, C., Hayashida, M., & Seales, W. B. (2023). "
"EduceLab-Scrolls: Verifiable Recovery of Text from Herculaneum Papyri using X-ray CT. ArXiv [Cs.CV]. "
"https://doi.org/10.48550/arXiv.2304.02084. Data used here were obtained from the EduceLab-Scrolls dataset.")
CHECKED = "2026-10-16"
# per scroll: display name, holder, scan of the meshes, X-ray scan, ink layers, prize sets
S = {
"PHerc0814": dict(name="PHerc. 814", holder="Biblioteca Nazionale di Napoli", vox=9.362, ct=9.362, scan="ESRF",
ink=2.399, ink1=1.129),
"PHerc0009B": dict(name="PHerc. 9B", holder="Biblioteca Nazionale di Napoli", vox=8.64, ct=8.64, scan="ESRF", ink=2.401),
"PHerc0343P": dict(name="PHerc. 343P", holder="Biblioteca Nazionale di Napoli", vox=8.64, ct=8.64, scan="ESRF", ink=2.215),
"PHerc0841": dict(name="PHerc. 841", holder="Biblioteca Nazionale di Napoli", vox=9.366, ct=9.366, scan="ESRF", ink=2.403),
"PHerc0172": dict(name="PHerc. 172 (Scroll 5)", holder="Bodleian Library, Oxford", vox=7.91, ct=7.91, scan="Diamond Light Source",
ink=7.91),
"PHerc1667": dict(name="PHerc. 1667 (Scroll 4)", holder="Biblioteca Nazionale di Napoli", vox=7.91, ct=2.399, scan="ESRF",
ink=2.399, ink1=1.129, educelab=True),
"PHerc1447": dict(name="PHerc. 1447", holder="Biblioteca Nazionale di Napoli", vox=8.64, ct=8.64, scan="ESRF",
prize=["2027 Grand Prize"]),
"PHerc0800": dict(name="PHerc. 800", holder="Biblioteca Nazionale di Napoli", vox=8.64, ct=8.64, scan="ESRF",
prize=["2027 Grand Prize", "First Letters"]),
}
def um(x):
return f"{x:g} µm"
def main(sid):
c = S[sid]
d = f"data/{sid}"
M = json.load(open(f"{ROOT}/{d}/scroll.json"))
ids = [w["id"] for w in M["windings"]]
n = len(ids)
layers = [{
"id": "xray", "role": "ct", "label": "X‑ray", "atlas": f"{d}/atlas_ct.webp", "atlasFallback": f"{d}/atlas_ct.jpg",
"sizeHint": os.path.getsize(f"{ROOT}/{d}/atlas_ct.webp"),
"title": "Raw X-ray CT brightness on each segment. No ink detection.",
"hint": "Raw X-ray CT brightness, no ink detection.",
"source": f"Vesuvius Challenge surface volumes, {um(c['ct'])} scan", "resolutionUM": c["ct"],
}]
if c.get("ink"):
layers.append({
"id": "ink", "role": "ink", "label": f"Published ink ({um(c['ink'])})", "atlas": f"{d}/atlas_ink.webp",
"atlasFallback": f"{d}/atlas_ink.jpg", "sizeHint": os.path.getsize(f"{ROOT}/{d}/atlas_ink.webp"),
"hi": f"{d}/hi/{{id}}.jpg",
"title": f"Vesuvius Challenge's published ink detection on the {um(c['ink'])} scan.",
"hint": f"Published ink ({um(c['ink'])}): Vesuvius Challenge detection on the {um(c['ink'])} scan.",
"closeupLabel": f"Published ink ({um(c['ink'])})",
"source": "Vesuvius Challenge ink detection (ink-detection/downsampled), open data bucket",
"resolutionUM": c["ink"],
})
if c.get("ink1") and os.path.isdir(f"{ROOT}/{d}/ink1"):
cov = [i for i in ids if os.path.exists(f"{ROOT}/{d}/ink1/{i}.jpg")]
layers.append({
"id": "ink1", "role": "overlay", "label": f"Published ink ({um(c['ink1'])})",
"windings": cov if len(cov) < n else "all",
"texture": f"{d}/ink1/{{id}}.jpg", "hi": f"{d}/hi/{{id}}_1um.jpg", "sizeHint": 400000,
"title": f"Vesuvius Challenge's published ink detection on the finer {um(c['ink1'])} scan, on {len(cov)} of {n} segments.",
"hint": f"Published ink ({um(c['ink1'])}): Vesuvius Challenge detection on the {um(c['ink1'])} scan, {len(cov)} of {n} segments.",
"closeupLabel": f"Published ink ({um(c['ink1'])})", "readoutLabel": "INK 1 µm",
"bandTitle": f"Segments with published ink on the {um(c['ink1'])} scan",
"source": "Vesuvius Challenge ink detection (ink-detection/downsampled), open data bucket",
"resolutionUM": c["ink1"],
})
notes = ("Meshes, X-ray and ink images come only from the open data bucket vesuvius-challenge-open-data "
f"(s3://vesuvius-challenge-open-data/{sid}/), licensed CC BY-NC 4.0: attribution, non-commercial use. "
"This page shows images only: no transcriptions or readings of the text.")
if c.get("prize"):
notes += " This scroll is prize eligible on scrollprize.org; no ink model is run on it here."
credits = {"text": "Vesuvius Challenge", "url": f"https://scrollprize.org/data_browser/{sid}", "citation": CITE,
"holder": c["holder"]}
if c.get("educelab"):
credits["citation"] = CITE + " " + EDUCELAB
k = len([l for l in layers if l["role"] != "ct"])
kind = "X-ray only, no ink layer" if not k else ("one published ink layer" if k == 1 else f"{['', '', 'two', 'three', 'four'][k] if k < 5 else k} published ink layers")
man = {
"schema": "unroll-scroll/1", "id": sid, "name": c["name"],
"description": f"A carbonized Herculaneum papyrus scanned at {c['scan']}. {n} traced segments from the Vesuvius Challenge open data, {kind}.",
"voxelMM": c["vox"] / 1000, "base": "../../",
"geometry": {"format": "grid-int16", "index": f"{d}/scroll.json", "bin": f"{d}/scroll.bin",
"sizeHint": os.path.getsize(f"{ROOT}/{d}/scroll.bin"),
"source": f"Vesuvius Challenge {um(c['vox'])} tifxyz meshes, decimated by tools/build.py"},
"windings": ids, "outermost": "last",
"layers": layers,
"credits": credits,
"dataTerms": {"license": "CC BY‑NC 4.0", "url": "https://creativecommons.org/licenses/by-nc/4.0/", "notes": notes},
"notes": "Segment ids w001 and up are ours, ordered by median radius (innermost first); scroll.json keeps each source segment.",
}
os.makedirs(f"{ROOT}/scrolls/{sid}", exist_ok=True)
json.dump(man, open(f"{ROOT}/scrolls/{sid}/manifest.json", "w"), indent=1, ensure_ascii=False)
reg = json.load(open(f"{ROOT}/scrolls/index.json"))
entry = {"id": sid, "name": c["name"], "manifest": f"scrolls/{sid}/manifest.json",
"prize": {"eligible": bool(c.get("prize")), "sets": c.get("prize", []),
"source": "https://scrollprize.org/prizes", "checked": CHECKED}}
reg["scrolls"] = [e for e in reg["scrolls"] if e["id"] != sid] + [entry]
for e in reg["scrolls"]:
e.setdefault("prize", {"eligible": False, "sets": [], "source": "https://scrollprize.org/prizes", "checked": CHECKED})
json.dump(reg, open(f"{ROOT}/scrolls/index.json", "w"), indent=1, ensure_ascii=False)
print("wrote", sid, n, "segments,", [l["id"] for l in layers])
if __name__ == "__main__":
main(sys.argv[1])