more work done

scripts/_tmp_probe_sections.js

@@ -0,0 +1,111 @@
+const fs = require("fs");
+
+const ALLOWED_U5 = new Set([0x20, 0x22, 0x30]);
+
+function readU32LE(buffer, offset) {
+  return buffer.readUInt32LE(offset);
+}
+
+function readU16LE(buffer, offset) {
+  return buffer.readUInt16LE(offset);
+}
+
+function isStructuredCandidate(record) {
+  if (record[0] >= 0x200) {
+    return false;
+  }
+  if (record[1] === 0 && record[2] === 0) {
+    return false;
+  }
+  if (record[1] >= 0x4000 || record[2] >= 0x4000) {
+    return false;
+  }
+  if (record[3] > 0x20 || record[4] > 0x04) {
+    return false;
+  }
+  return ALLOWED_U5.has(record[5]);
+}
+
+function probeFile(filePath) {
+  const data = fs.readFileSync(filePath);
+  const headerSize = readU32LE(data, 0);
+  const audioSize = readU32LE(data, 4);
+  const sectionSizes = [];
+  for (let offset = 8; offset < 0x38; offset += 4) {
+    sectionSizes.push(readU32LE(data, offset));
+  }
+
+  let cursor = headerSize + audioSize;
+  const sections = [];
+  for (let index = 0; index < sectionSizes.length; index += 1) {
+    const size = sectionSizes[index];
+    const start = cursor;
+    const end = start + size;
+    const bytes = data.subarray(start, end);
+    cursor = end;
+
+    let rowCount = null;
+    let rootHits = 0;
+    let bulkHits = 0;
+
+    if (bytes.length >= 4) {
+      rowCount = readU32LE(bytes, 0);
+      for (let rowIndex = 0; rowIndex < Math.min(rowCount, 5000); rowIndex += 1) {
+        const base = 4 + rowIndex * 24;
+        if (base + 24 > bytes.length) {
+          break;
+        }
+        const words = [];
+        for (let wordIndex = 0; wordIndex < 12; wordIndex += 1) {
+          words.push(readU16LE(bytes, base + wordIndex * 2));
+        }
+        const left = [words[4], words[5], words[0], words[1], words[2], words[3]];
+        const right = [words[10], words[11], words[6], words[7], words[8], words[9]];
+        if (isStructuredCandidate(left)) {
+          rootHits += 1;
+        }
+        if (isStructuredCandidate(right)) {
+          rootHits += 1;
+        }
+      }
+    }
+
+    const usableSize = bytes.length - (bytes.length % 24);
+    for (let offset = 0; offset < usableSize; offset += 24) {
+      for (const sideOffset of [0, 12]) {
+        const base = offset + sideOffset;
+        const record = [
+          readU16LE(bytes, base),
+          readU16LE(bytes, base + 2),
+          readU16LE(bytes, base + 4),
+          readU16LE(bytes, base + 6),
+          readU16LE(bytes, base + 8),
+          readU16LE(bytes, base + 10)
+        ];
+        if (isStructuredCandidate(record)) {
+          bulkHits += 1;
+        }
+      }
+    }
+
+    sections.push({
+      index,
+      start,
+      size,
+      rowCount,
+      rootHits,
+      bulkHits
+    });
+  }
+
+  return {
+    filePath,
+    headerSize,
+    audioSize,
+    sections
+  };
+}
+
+for (const filePath of process.argv.slice(2)) {
+  console.log(JSON.stringify(probeFile(filePath), null, 2));
+}