/**
 * Pure, DOM-free SMS encoding + segmentation helpers.
 *
 * SMS messages are plain text, but how many "segments" (billable parts) a
 * message costs depends on its characters:
 *
 * - **GSM-7**: the default 7-bit alphabet. 160 chars in a single segment, 153
 *   per segment once the message is split (the extra 7 bits per segment go to
 *   the multipart UDH header). A handful of characters (`^ { } \ [ ] ~ | €` and
 *   form-feed) live in an "extension table" and cost **two** characters each.
 * - **UCS-2**: used when any character is outside GSM-7 (e.g. emoji, many
 *   non-Latin scripts). 70 chars single, 67 per segment when split. Counted in
 *   UTF-16 code units, so emoji built from surrogate pairs cost two.
 *
 * Reference: 3GPP TS 23.038.
 */

/** Which physical encoding the carrier will use for the message. */
export type SmsEncoding = 'GSM-7' | 'UCS-2';

/** A full segmentation report for a piece of text. */
export interface SmsSegmentInfo {
  /** Encoding the message forces. */
  encoding: SmsEncoding;
  /** Weighted character count (GSM-7 extension chars count as 2). */
  length: number;
  /** Number of billable segments. */
  segments: number;
  /** Capacity of a single segment for the active encoding (160 or 70). */
  maxSingle: number;
  /** Capacity per segment once the message is multipart (153 or 67). */
  maxMulti: number;
  /** Characters left before the next segment boundary. */
  remaining: number;
}

/** GSM-7 default alphabet (each char costs one septet). Includes space. */
const GSM7_BASIC = new Set(
  '@£$¥èéùìòÇ\nØø\rÅåΔ_ΦΓΛΩΠΨΣΘΞÆæßÉ !"#¤%&\'()*+,-./0123456789:;<=>?¡ABCDEFGHIJKLMNOPQRSTUVWXYZÄÖÑܧ¿abcdefghijklmnopqrstuvwxyzäöñüà'.split(
    '',
  ),
);

/** GSM-7 extension table (each char costs two septets: ESC + char). */
const GSM7_EXTENDED = new Set(['^', '{', '}', '\\', '[', ']', '~', '|', '€', '\f']);

const SINGLE = { 'GSM-7': 160, 'UCS-2': 70 } as const;
const MULTI = { 'GSM-7': 153, 'UCS-2': 67 } as const;

/**
 * Determine whether `text` can be sent as GSM-7 or must fall back to UCS-2.
 */
export function detectEncoding(text: string): SmsEncoding {
  for (const ch of text) {
    if (!GSM7_BASIC.has(ch) && !GSM7_EXTENDED.has(ch)) {
      return 'UCS-2';
    }
  }
  return 'GSM-7';
}

/**
 * Weighted length of `text` for the given encoding. In GSM-7, extension-table
 * characters count as two. In UCS-2, the count is in UTF-16 code units (so a
 * surrogate-pair emoji counts as two).
 */
export function countChars(text: string, encoding: SmsEncoding): number {
  if (encoding === 'UCS-2') {
    return text.length;
  }
  let count = 0;
  for (const ch of text) {
    count += GSM7_EXTENDED.has(ch) ? 2 : 1;
  }
  return count;
}

/**
 * Produce a full segmentation report for `text`: encoding, weighted length,
 * segment count, per-segment capacity and remaining characters.
 */
export function segment(text: string): SmsSegmentInfo {
  const encoding = detectEncoding(text);
  const length = countChars(text, encoding);
  const maxSingle = SINGLE[encoding];
  const maxMulti = MULTI[encoding];

  let segments: number;
  if (length === 0) {
    segments = 0;
  } else if (length <= maxSingle) {
    segments = 1;
  } else {
    segments = Math.ceil(length / maxMulti);
  }

  const capacity = segments <= 1 ? maxSingle : segments * maxMulti;
  const remaining = segments === 0 ? maxSingle : capacity - length;

  return { encoding, length, segments, maxSingle, maxMulti, remaining };
}