osmWorkflow/scripts/glb-digest.js

#!/usr/bin/env node
"use strict";

// Structural digest of a GLB, for the osmassets refactor parity check.
//
// Byte-comparing the GLB is too strict: Blender packs images in hash-map order
// and the buffer padding shifts with it, so two runs of identical code can
// differ.  This reads the glTF JSON chunk instead and reports the parts that
// carry meaning downstream in Cesium — node/mesh/material identity and PBR
// values — plus buffer lengths as a coarse size check.
//
//   node scripts/glb-digest.js <file.glb> [--out digest.json]

const fs = require("fs");
const path = require("path");

function readGlbJson(file) {
  const buffer = fs.readFileSync(file);
  if (buffer.length < 12 || buffer.readUInt32LE(0) !== 0x46546c67) {
    throw new Error(`Not a GLB (bad magic): ${file}`);
  }
  const total = buffer.readUInt32LE(8);
  let offset = 12;
  while (offset + 8 <= Math.min(total, buffer.length)) {
    const chunkLength = buffer.readUInt32LE(offset);
    const chunkType = buffer.readUInt32LE(offset + 4);
    const start = offset + 8;
    if (chunkType === 0x4e4f534a) {
      return JSON.parse(buffer.slice(start, start + chunkLength).toString("utf8"));
    }
    offset = start + chunkLength;
  }
  throw new Error(`No JSON chunk found in ${file}`);
}

function round(value) {
  if (typeof value === "number") return Number(value.toFixed(6));
  if (Array.isArray(value)) return value.map(round);
  return value;
}

function materialDigest(material) {
  const pbr = material.pbrMetallicRoughness || {};
  return {
    name: material.name || null,
    baseColorFactor: round(pbr.baseColorFactor || null),
    metallicFactor: round(pbr.metallicFactor ?? null),
    roughnessFactor: round(pbr.roughnessFactor ?? null),
    hasBaseColorTexture: Boolean(pbr.baseColorTexture),
    hasNormalTexture: Boolean(material.normalTexture),
    emissiveFactor: round(material.emissiveFactor || null),
    emissiveStrength: round(
      material.extensions?.KHR_materials_emissive_strength?.emissiveStrength ?? null,
    ),
    alphaMode: material.alphaMode || null,
    doubleSided: material.doubleSided ?? null,
  };
}

function digest(file) {
  const gltf = readGlbJson(file);
  return digestGltf(gltf, { file: path.basename(file), fileBytes: fs.statSync(file).size });
}

function digestGltf(gltf, fileInfo = {}) {
  const meshes = (gltf.meshes || []).map((mesh) => ({
    name: mesh.name || null,
    primitives: (mesh.primitives || []).map((primitive) => ({
      material: primitive.material ?? null,
      attributes: Object.keys(primitive.attributes || {}).sort(),
      // Vertex/index counts live on the accessors; they are the real geometry
      // fingerprint and stay stable regardless of buffer layout.
      count: gltf.accessors?.[primitive.attributes?.POSITION]?.count ?? null,
      indices: gltf.accessors?.[primitive.indices]?.count ?? null,
      triangles: primitiveTriangleCount(primitive, gltf.accessors || []),
    })),
  }));
  const sourceSummary = summarizeNodeSources(gltf.nodes || [], meshes);
  const images = (gltf.images || [])
    .map((image) => ({
      name: image.name || null,
      mimeType: image.mimeType || null,
      bytes: gltf.bufferViews?.[image.bufferView]?.byteLength ?? 0,
    }))
    .sort((a, b) => b.bytes - a.bytes || String(a.name).localeCompare(String(b.name)));
  const triangles = meshes.reduce(
    (total, mesh) => total + mesh.primitives.reduce((sum, primitive) => sum + primitive.triangles, 0),
    0,
  );
  const renderTriangles = sourceSummary.reduce((total, source) => total + source.triangles, 0);
  const embeddedImageBytes = images.reduce((total, image) => total + image.bytes, 0);
  return {
    file: fileInfo.file || null,
    fileBytes: fileInfo.fileBytes ?? 0,
    counts: {
      nodes: (gltf.nodes || []).length,
      meshes: meshes.length,
      materials: (gltf.materials || []).length,
      images: (gltf.images || []).length,
      accessors: (gltf.accessors || []).length,
      triangles,
      renderTriangles,
    },
    embeddedImageBytes,
    sourceSummary,
    extensionsUsed: (gltf.extensionsUsed || []).slice().sort(),
    buffers: (gltf.buffers || []).map((buffer) => buffer.byteLength),
    nodes: (gltf.nodes || [])
      .map((node) => ({
        name: node.name || null,
        mesh: node.mesh ?? null,
        translation: round(node.translation || null),
        rotation: round(node.rotation || null),
        scale: round(node.scale || null),
        extras: node.extras ?? null,
      }))
      .sort((a, b) => String(a.name).localeCompare(String(b.name))),
    meshes: meshes.sort((a, b) => String(a.name).localeCompare(String(b.name))),
    materials: (gltf.materials || [])
      .map(materialDigest)
      .sort((a, b) => String(a.name).localeCompare(String(b.name))),
    images,
  };
}

function primitiveTriangleCount(primitive, accessors) {
  if (primitive.mode !== undefined && primitive.mode !== 4) return 0;
  const count = primitive.indices === undefined
    ? accessors[primitive.attributes?.POSITION]?.count
    : accessors[primitive.indices]?.count;
  return Number.isFinite(count) ? Math.floor(count / 3) : 0;
}

function summarizeNodeSources(nodes, meshes) {
  const summaries = new Map();
  for (const node of nodes) {
    if (node.mesh === undefined || !meshes[node.mesh]) continue;
    const source = nodeSource(node.name || meshes[node.mesh].name || "");
    const summary = summaries.get(source) || { source, nodes: 0, meshInstances: 0, triangles: 0 };
    const triangles = meshes[node.mesh].primitives.reduce((sum, primitive) => sum + primitive.triangles, 0);
    summary.nodes += 1;
    summary.meshInstances += 1;
    summary.triangles += triangles;
    summaries.set(source, summary);
  }
  return [...summaries.values()].sort((a, b) => b.triangles - a.triangles || a.source.localeCompare(b.source));
}

function nodeSource(name) {
  const value = String(name).toLowerCase();
  if (value.includes("building")) return "buildings";
  if (/(shapespark|tree|bush|shrub|grass|branch|trunk|leaf)/.test(value)) return "vegetation";
  if (/(road|sidewalk|lane|crosswalk|stop|intersection)/.test(value)) return "roads";
  if (/(lake|water)/.test(value)) return "water";
  if (value.includes("fountain")) return "fountain";
  if (/(vehicle|car)/.test(value)) return "vehicles";
  return "other";
}

function main() {
  const argv = process.argv.slice(2);
  const file = argv.find((arg) => !arg.startsWith("--"));
  if (!file) {
    console.error("usage: node scripts/glb-digest.js <file.glb> [--out digest.json]");
    process.exit(1);
  }
  const outIndex = argv.indexOf("--out");
  const result = digest(path.resolve(file));
  const text = `${JSON.stringify(result, null, 2)}\n`;
  if (outIndex >= 0 && argv[outIndex + 1]) {
    const out = path.resolve(argv[outIndex + 1]);
    fs.mkdirSync(path.dirname(out), { recursive: true });
    fs.writeFileSync(out, text);
    console.log(`GLB digest: ${out}`);
  } else {
    process.stdout.write(text);
  }
}

if (require.main === module) {
  main();
}

module.exports = {
  digest,
  digestGltf,
  readGlbJson,
};