opusdev/vector-similarity-api

1/**2 * Estimate decoded byte length of a data:// URL *without* allocating large buffers.3 * - For base64: compute exact decoded size using length and padding;4 *               handle %XX at the character-count level (no string allocation).5 * - For non-base64: use UTF-8 byteLength of the encoded body as a safe upper bound.6 *7 * @param {string} url8 * @returns {number}9 */10export default function estimateDataURLDecodedBytes(url) {11  if (!url || typeof url !== 'string') return 0;12  if (!url.startsWith('data:')) return 0;13 14  const comma = url.indexOf(',');15  if (comma < 0) return 0;16 17  const meta = url.slice(5, comma);18  const body = url.slice(comma + 1);19  const isBase64 = /;base64/i.test(meta);20 21  if (isBase64) {22    let effectiveLen = body.length;23    const len = body.length; // cache length24 25    for (let i = 0; i < len; i++) {26      if (body.charCodeAt(i) === 37 /* '%' */ && i + 2 < len) {27        const a = body.charCodeAt(i + 1);28        const b = body.charCodeAt(i + 2);29        const isHex =30          ((a >= 48 && a <= 57) || (a >= 65 && a <= 70) || (a >= 97 && a <= 102)) &&31          ((b >= 48 && b <= 57) || (b >= 65 && b <= 70) || (b >= 97 && b <= 102));32 33        if (isHex) {34          effectiveLen -= 2;35          i += 2;36        }37      }38    }39 40    let pad = 0;41    let idx = len - 1;42 43    const tailIsPct3D = (j) =>44      j >= 2 &&45      body.charCodeAt(j - 2) === 37 && // '%'46      body.charCodeAt(j - 1) === 51 && // '3'47      (body.charCodeAt(j) === 68 || body.charCodeAt(j) === 100); // 'D' or 'd'48 49    if (idx >= 0) {50      if (body.charCodeAt(idx) === 61 /* '=' */) {51        pad++;52        idx--;53      } else if (tailIsPct3D(idx)) {54        pad++;55        idx -= 3;56      }57    }58 59    if (pad === 1 && idx >= 0) {60      if (body.charCodeAt(idx) === 61 /* '=' */) {61        pad++;62      } else if (tailIsPct3D(idx)) {63        pad++;64      }65    }66 67    const groups = Math.floor(effectiveLen / 4);68    const bytes = groups * 3 - (pad || 0);69    return bytes > 0 ? bytes : 0;70  }71 72  return Buffer.byteLength(body, 'utf8');73}74