@eifi1/ui-kit 0.54.0 → 0.56.0

package/src/lib/person-name.ts

@@ -1,155 +0,0 @@
-/**
- * A person's name as the reader's language writes it — the one rule of
- * docs/auth-harmonization.md §3.2 for every place a name is shown: the top bar, the
- * account menu, user lists, avatars, feedback, the greeting in a mail.
- *
- * WHY A FUNCTION AND NOT `${first} ${last}`. The three apps move from one free-text
- * `display_name` to `first_name` + `last_name` (§2.3), and the moment a name has two
- * parts, their ORDER is a language rule, not a data rule:
- *
- *  - "First Last" in de-CH, en, fr, it and es — "Ada Example";
- *  - **"Last First" in hu** — Hungarian puts the family name first: "Example Ada";
- *  - **"LastFirst" with no space in zh** — a Chinese name is written family name first
- *    and run together: 李 + 小龙 → "李小龙". Only a name written in CJK characters,
- *    though: BOTH parts must hold Han (or kana, or hangul) characters. A Latin name read
- *    in Chinese stays as it is written — "Ada Lovelace", not "LovelaceAda" (the
- *    coordinator's refinement of §3.2, 2026-10-06).
- *
- * WHOSE language (§3.2, §10.6): on screen the READER's, so this takes the locale of the
- * UI, never the locale stored with the person. A mail uses the recipient's and a frozen
- * stamp the author's — that is server-kit's `full_name(first, last, locale)`, which
- * follows exactly these rules so a name reads the same in the app and in its mails.
- *
- * NOTHING IS GUESSED. Both parts are trimmed, and an empty part simply drops out: a
- * migrated keksdose or Kurvenschmiede user keeps the whole old display name as the
- * first name and an empty last name until they complete it (§3.3), and "Ada Example"
- * with no last name must still read "Ada Example" — not "Ada Example " and not a split
- * at the space. Inner spaces are the person's own and stay as typed.
- *
- * Only the language subtag counts ("de-CH" → "de", "zh-Hant-TW" → "zh"), and a language
- * this does not know reads "First Last", the order of the kit's five other languages.
- */
-
-/** The two parts of a person's name. Either may be missing or blank: a migrated row
- *  has an empty `last` until it is completed (§3.3), and APIs send `null` for "none". */
-export interface PersonName {
-  /** The given name — `first_name`. */
-  first?: string | null;
-  /** The family name — `last_name`. */
-  last?: string | null;
-}
-
-type NameOrder = "given-first" | "family-first" | "family-first-joined";
-
-/** A part written in CJK characters: Han, and kana or hangul where they occur (a
- *  Japanese or Korean name in a Chinese UI). One such character is enough. */
-const CJK = /\p{Script=Han}|\p{Script=Hiragana}|\p{Script=Katakana}|\p{Script=Hangul}/u;
-
-function language(locale: string | undefined): string {
-  return (locale ?? "").split(/[-_]/)[0]!.toLowerCase();
-}
-
-/**
- * The order the reader's language writes THIS name in. See the module note: hu puts the
- * family name first for every name; zh runs a CJK name together family-first and leaves
- * any other name as it is written; the rest write the given name first. A part that is
- * empty passes the CJK test, so a one-part name orders by the part it has. Called with
- * at least one part filled in.
- */
-function nameOrder(locale: string | undefined, first: string, last: string): NameOrder {
-  const lang = language(locale);
-  if (lang === "hu") return "family-first";
-  if (lang === "zh" && (!first || CJK.test(first)) && (!last || CJK.test(last))) {
-    return "family-first-joined";
-  }
-  return "given-first";
-}
-
-function clean(part: string | null | undefined): string {
-  return (part ?? "").trim();
-}
-
-/**
- * The full name in the reader's order: "Ada Example" (en, de, fr, it, es), "Example Ada"
- * (hu), "李小龙" (zh, for a name in CJK characters — "Ada Lovelace" stays so). Trimmed;
- * with one part empty, the other alone; with both empty, `""` — so a caller can fall
- * back with `||` (`formatPersonName(p, l) || email`).
- *
- * `locale` is the READER's — the UI language, `useKitLocale()` in a component. Left out,
- * the given-name-first order.
- */
-export function formatPersonName(name: PersonName, locale?: string): string {
-  const first = clean(name.first);
-  const last = clean(name.last);
-  if (!last) return first;
-  if (!first) return last;
-  switch (nameOrder(locale, first, last)) {
-    case "family-first":
-      return `${last} ${first}`;
-    case "family-first-joined":
-      return `${last}${first}`;
-    default:
-      return `${first} ${last}`;
-  }
-}
-
-/** One segmenter for every avatar on the page, made on first use. `null`: the runtime
- *  has none, and code points have to do. */
-let graphemes: Intl.Segmenter | null | undefined;
-
-/** The first user-perceived character — "É" whether it was typed as one code point or
- *  as E + a combining accent, a whole emoji rather than half a surrogate pair. */
-function firstGrapheme(text: string): string {
-  graphemes ??=
-    typeof Intl !== "undefined" && "Segmenter" in Intl
-      ? new Intl.Segmenter(undefined, { granularity: "grapheme" })
-      : null;
-  if (graphemes) {
-    const segment = graphemes.segment(text)[Symbol.iterator]().next();
-    if (!segment.done) return segment.value.segment;
-  }
-  return [...text][0] ?? "";
-}
-
-/** Upper case under the reader's rules (Turkish "i" → "İ"), and plain upper case for a
- *  locale tag the runtime refuses rather than a thrown RangeError in a render. */
-function upper(text: string, locale: string | undefined): string {
-  try {
-    return text.toLocaleUpperCase(locale);
-  } catch {
-    return text.toUpperCase();
-  }
-}
-
-/**
- * The initials for an avatar, in the same order as {@link formatPersonName}:
- *
- *  - "AE" for Ada Example in en, de, fr, it and es;
- *  - "EA" in hu, family name first;
- *  - **one character in zh for a name in CJK characters, the family name's first** —
- *    "李" for 李小龙. Two Chinese characters are not initials, they are most of a name.
- *    A Latin name in zh is "AE", as it is written.
- *
- * With only one part (a migrated user's whole old display name in `first`, §3.3), the
- * initials are what {@link avatarInitials} has always made of a display name: the first
- * letters of its first two words ("Ada Example" → "AE"), or the first two letters of a
- * single word ("Ada" → "AD") — so nobody's avatar changes when their account migrates.
- * In zh, a part in CJK characters gives its first character.
- *
- * `""` when both parts are blank; {@link UserAvatar} then falls back to the email.
- */
-export function personInitials(name: PersonName, locale?: string): string {
-  const first = clean(name.first);
-  const last = clean(name.last);
-  if (!first && !last) return "";
-  const order = nameOrder(locale, first, last);
-  if (order === "family-first-joined") return upper(firstGrapheme(last || first), locale);
-  if (!first || !last) {
-    const words = (first || last).split(/\s+/);
-    const initials =
-      words.length >= 2 ? firstGrapheme(words[0]!) + firstGrapheme(words[1]!) : [...words[0]!].slice(0, 2).join("");
-    return upper(initials, locale);
-  }
-  const pair = order === "family-first" ? [last, first] : [first, last];
-  return upper(firstGrapheme(pair[0]!) + firstGrapheme(pair[1]!), locale);
-}

package/src/lib/phone.ts

@@ -1,208 +0,0 @@
-/**
- * Phone numbers — E.164 in, grouped out — for kastlan's contacts, without a
- * libphonenumber dependency.
- *
- * kastlan stored whatever was typed ("079 …", "+41 (0)79 …", "0041 79 …") and could
- * neither dial nor de-duplicate it. What it needs is one canonical form to store —
- * E.164, "+41210000001" — and a readable one to show. libphonenumber gets both right
- * for every country, at ~150 kB of metadata in every app's bundle; the kit's apps are
- * Swiss and their neighbours, so this is **light per-country rules** for the six that
- * matter and a generic shape check for the rest:
- *
- *  - **CH, LI, DE, AT, FR, IT**: the national trunk prefix is dropped (a leading 0 —
- *    but NOT for Italy, whose fixed-line numbers keep their 0 after +39, and not for
- *    Liechtenstein, which has none), and the remaining national number has to be a
- *    plausible length for the country.
- *  - **Anything else** typed with its own "+" (or "00"): the digits, 8 to 15 in all,
- *    are taken as an E.164 number as they stand.
- *
- * **The limitation:** a plausible length is not a valid number. There is no check of
- * area codes or number ranges, so a well-formed number that does not exist passes, and
- * a German or Austrian number (whose area codes run from one to five digits) is
- * grouped only where the split is known (mobile prefixes, the largest cities) and
- * otherwise shown in one block. A number these rules cannot read is not an error —
- * {@link parsePhone} returns `null` and the field keeps the text as typed.
- */
-
-/** The countries with their own rules. */
-export type PhoneCountryCode = "CH" | "LI" | "DE" | "AT" | "FR" | "IT";
-
-/** A country with rules, or `"other"`: a number typed with its own "+". */
-export type PhoneCountry = PhoneCountryCode | "other";
-
-interface CountryRule {
-  /** The country calling code, without the "+". */
-  dial: string;
-  /** A national number starts with a trunk 0 that is dropped after the country code. */
-  trunk: boolean;
-  /** The national significant number itself may start with 0 (Italy's fixed lines). */
-  zeroStart?: boolean;
-  /** The national significant number's length, inclusive. */
-  min: number;
-  max: number;
-  /** Groups for display, given the national significant number. */
-  group: (nsn: string) => string[];
-}
-
-/** `nsn` cut into the given sizes; whatever is left is one last group. */
-function cut(nsn: string, sizes: number[]): string[] {
-  const out: string[] = [];
-  let at = 0;
-  for (const size of sizes) {
-    if (at >= nsn.length) break;
-    out.push(nsn.slice(at, at + size));
-    at += size;
-  }
-  if (at < nsn.length) out.push(nsn.slice(at));
-  return out;
-}
-
-const RULES: Record<PhoneCountryCode, CountryRule> = {
-  // 9 digits, 2-3-2-2 ("21 000 00 01"); the 08xx/09xx service numbers 3-3-3.
-  CH: {
-    dial: "41",
-    trunk: true,
-    min: 9,
-    max: 9,
-    group: (n) => (/^[89]/.test(n) ? cut(n, [3, 3, 3]) : cut(n, [2, 3, 2, 2])),
-  },
-  // 7 digits, 3-2-2 ("100 00 01"), and 9 for some mobile ranges (3-3-3). No trunk prefix.
-  LI: {
-    dial: "423",
-    trunk: false,
-    min: 7,
-    max: 9,
-    group: (n) => (n.length === 7 ? cut(n, [3, 2, 2]) : n.length === 9 ? cut(n, [3, 3, 3]) : [n]),
-  },
-  // Area codes of two to five digits: only the mobile prefixes and the four
-  // two-digit cities are split; anything else is one block rather than a wrong split.
-  DE: {
-    dial: "49",
-    trunk: true,
-    min: 6,
-    max: 13,
-    group: (n) =>
-      /^15/.test(n) ? cut(n, [4]) : /^1[67]/.test(n) ? cut(n, [3]) : /^(30|40|69|89)/.test(n) ? cut(n, [2]) : [n],
-  },
-  // Vienna is "1"; mobile numbers start 6xx; other area codes vary, so one block.
-  AT: {
-    dial: "43",
-    trunk: true,
-    min: 6,
-    max: 13,
-    group: (n) => (/^6/.test(n) ? cut(n, [3]) : /^1/.test(n) ? cut(n, [1]) : [n]),
-  },
-  // 9 digits, 1-2-2-2-2 ("1 99 00 00 01").
-  FR: { dial: "33", trunk: true, min: 9, max: 9, group: (n) => cut(n, [1, 2, 2, 2, 2]) },
-  // The 0 of a fixed line is part of the number ("+39 06 …"); mobiles start with 3.
-  IT: {
-    dial: "39",
-    trunk: false,
-    zeroStart: true,
-    min: 6,
-    max: 11,
-    group: (n) => (/^3/.test(n) ? cut(n, [3, 3]) : /^0[26]/.test(n) ? cut(n, [2, 4]) : [n]),
-  },
-};
-
-/** The country calling codes with rules, "+" left off: `{ CH: "41", … }`. */
-export const PHONE_DIAL_CODES: Readonly<Record<PhoneCountryCode, string>> = Object.freeze({
-  CH: RULES.CH.dial,
-  LI: RULES.LI.dial,
-  DE: RULES.DE.dial,
-  AT: RULES.AT.dial,
-  FR: RULES.FR.dial,
-  IT: RULES.IT.dial,
-});
-
-/** The countries with rules, in the order a picker offers them by default. */
-export const PHONE_COUNTRIES: readonly PhoneCountryCode[] = Object.freeze(["CH", "LI", "DE", "AT", "FR", "IT"]);
-
-/** Longest dial code first, so "+423" is Liechtenstein before anything shorter. */
-const BY_DIAL = [...PHONE_COUNTRIES].sort((a, b) => RULES[b].dial.length - RULES[a].dial.length);
-
-/** What {@link parsePhone} reads out of a number. */
-export interface ParsedPhone {
-  /** The number in E.164: "+", the country code, the national number. */
-  e164: string;
-  /** The country whose rules read it, or `null` for a generic "+…" number. */
-  country: PhoneCountryCode | null;
-  /** The national significant number — the digits after the country code. For a
-   *  generic number, every digit after the "+" (the country code is not split off). */
-  national: string;
-}
-
-/** A national significant number that passes `rule`, or `null`. */
-function national(digits: string, rule: CountryRule): string | null {
-  const nsn = rule.trunk && digits.startsWith("0") ? digits.slice(1) : digits;
-  // A national number starts with 0 only where 0 is part of it (Italy's fixed lines).
-  if (nsn.startsWith("0") && !rule.zeroStart) return null;
-  if (nsn.length < rule.min || nsn.length > rule.max) return null;
-  return nsn;
-}
-
-/**
- * Read a typed number. With a leading "+" or "00" the number carries its own country
- * code — one of the six with rules, or any other (8–15 digits in all, the E.164
- * limit). Without one it is a national number of `country` (default `"CH"`), its
- * trunk 0 optional; with `country="other"` a number without "+" cannot be read.
- *
- * Spaces, hyphens, dots, slashes and parentheses are ignored ("(030) 1234567"), and
- * so is the "(0)" of "+41 (0)21 …". Anything else — a letter, an extension ("ext. 12"), a note —
- * and the result is `null`: the text is not a number these rules can read, and a field
- * keeps it as it is rather than guess.
- */
-export function parsePhone(input: string, country: PhoneCountry = "CH"): ParsedPhone | null {
-  const text = input.trim();
-  if (text === "" || !/^\+?[\d\s().\-/]+$/.test(text)) return null;
-  let digits = text.replace(/\D/g, "");
-  let international = text.startsWith("+");
-  if (!international && digits.startsWith("00")) {
-    international = true;
-    digits = digits.slice(2);
-  }
-  if (international) {
-    // "+41 (0)21 …": the bracketed trunk prefix written for callers inside the country.
-    digits = text
-      .replace(/\(\s*0\s*\)/g, "")
-      .replace(/\D/g, "")
-      .replace(/^00/, "");
-    if (digits.startsWith("0")) return null;
-    const known = BY_DIAL.find((c) => digits.startsWith(RULES[c].dial));
-    if (known) {
-      const rule = RULES[known];
-      const nsn = national(digits.slice(rule.dial.length), rule);
-      return nsn === null ? null : { e164: `+${rule.dial}${nsn}`, country: known, national: nsn };
-    }
-    if (digits.length < 8 || digits.length > 15) return null;
-    return { e164: `+${digits}`, country: null, national: digits };
-  }
-  if (country === "other") return null;
-  const rule = RULES[country];
-  const nsn = national(digits, rule);
-  return nsn === null ? null : { e164: `+${rule.dial}${nsn}`, country, national: nsn };
-}
-
-/** The national number grouped by its country's rules ("21 000 00 01"). */
-export function formatNationalPhone(nsn: string, country: PhoneCountryCode): string {
-  return RULES[country].group(nsn).join(" ");
-}
-
-/**
- * A number for display: an E.164 number (or anything {@link parsePhone} reads with a
- * "+") grouped by its country's rules — "+41 21 000 00 01" — and a generic one as
- * "+" and its digits. Anything it cannot read comes back exactly as it went in, so a
- * list of contacts can format every row without first sorting out the free text.
- */
-export function formatPhone(value: string): string {
-  const parsed = value.trim().startsWith("+") ? parsePhone(value, "other") : null;
-  if (!parsed) return value;
-  if (!parsed.country) return parsed.e164;
-  return `+${RULES[parsed.country].dial} ${formatNationalPhone(parsed.national, parsed.country)}`;
-}
-
-/** Whether `value` is in E.164 shape: "+" and 8–15 digits, nothing else. Shape only —
- *  what {@link PhoneInput} emits for every number it could read. */
-export function isE164(value: string): boolean {
-  return /^\+[1-9]\d{7,14}$/.test(value);
-}

package/src/lib/plain-text.ts

@@ -1,22 +0,0 @@
-import { isValidElement } from "react";
-import type { ReactNode } from "react";
-
-/**
- * The text of a label node: a string or number as is, an element by its children,
- * recursively, whitespace collapsed. A `<Tooltip label="…">Spend</Tooltip>` label yields
- * "Spend" — the text the reader sees, not the bubble's. `undefined` when there is no text
- * to find (an icon), so the caller falls back to a name of its own.
- *
- * Package-private — not re-exported from the barrel. StatTile names its sparkline by it;
- * a field names its `help` "?" by it ("About “<label>”", round 0.36).
- */
-export function plainText(node: ReactNode): string | undefined {
-  const walk = (n: ReactNode): string => {
-    if (typeof n === "string" || typeof n === "number") return String(n);
-    if (Array.isArray(n)) return n.map(walk).join("");
-    if (isValidElement<{ children?: ReactNode }>(n)) return walk(n.props.children);
-    return "";
-  };
-  const text = walk(node).replace(/\s+/g, " ").trim();
-  return text === "" ? undefined : text;
-}

package/src/lib/presence.ts

@@ -1,18 +0,0 @@
-import type { ReactNode } from "react";
-
-/**
- * Whether a node says anything — the one copy of the rule (DUP-11, refactor audit
- * 2026-10: it was written fourteen times as `hasContent` / `hasMessage` / `hasReason` /
- * `present` / `given`, and FeedbackComposer's own `!= null && !== false` drifted from it
- * and took `""` for a reason).
- *
- * `undefined`, `null`, `false` and `""` are what a caller's `cond && "…"` or
- * `touched && errors.x` evaluates to when there is nothing to say; none of them is
- * content, and pointing a control's `aria-describedby` at one would describe it with an
- * empty node. A numeric `0` is content.
- *
- * Package-private — not re-exported from the barrel.
- */
-export function hasContent(node: ReactNode): boolean {
-  return node !== undefined && node !== null && node !== false && node !== "";
-}