Voiced/src/pdf/engine/measure.ts

import { create, type Font } from "fontkit";
import { FONT_MANIFEST } from "../fonts/manifest";
import { familySources, FONT_STACK, stack, type FamilySource } from "../fonts/stack";
import { resolveFace, roleToken, type RoleName, type TypeToken } from "../fonts/roles";
import { userFamilySources } from "../fonts/userFontStore";

export interface MeasureInput {
  text: string;
  /** A registered family name (after resolveFace). */
  face: string;
  weight: number;
  italic?: boolean;
  size: number;
  /** pt added after every glyph, including the last one (textkit does the same). */
  letterSpacing?: number;
}

export type WrapInput = Omit<MeasureInput, "text">;

export interface Measurer {
  /** Advance width of a single line in pt, with kerning and font fallback as textkit applies them. */
  measureWidth(input: MeasureInput): number;
  /**
   * Greedy word-wrap that predicts textkit's line breaks. Estimate for planning only: textkit's
   * Knuth-Plass breaker lets spaces shrink on an overfull line, so it can fit a word this routine
   * pushes to the next line (and the other way round for very narrow boxes).
   */
  wrapLines(text: string, width: number, input: WrapInput): string[];
  /** Code points that no family in the stack covers (they would render as .notdef). */
  missing(text: string, face?: string): number[];
}

const userFonts = new Map<string, Font>();

/**
 * The fontkit font of a user-imported family at a weight and style (the file react-pdf resolves for it), or
 * undefined when `familyName` is not a user family. Parsed once per file and shared by every measurer.
 */
export function userFontkitFont(familyName: string, weight: number, italic: boolean): Font | null | undefined {
  const sources = userFamilySources(familyName);
  if (sources.length === 0) return undefined;
  const src = resolveSource(
    sources.map((s) => ({ file: s.id, fontWeight: s.fontWeight, fontStyle: s.fontStyle })),
    weight,
    italic ? "italic" : "normal",
  );
  if (!src) return null;
  let font = userFonts.get(src.file);
  if (!font) {
    font = create((sources.find((s) => s.id === src.file) as { bytes: Uint8Array }).bytes) as Font;
    userFonts.set(src.file, font);
  }
  return font;
}

/** react-pdf's FontFamily.resolve, so the measurer reads the same file the renderer embeds. */
function resolveSource(sources: FamilySource[], weight: number, style: "normal" | "italic"): FamilySource | undefined {
  const styled = sources.filter((s) => s.fontStyle === style);
  const exact = styled.find((s) => s.fontWeight === weight);
  if (exact) return exact;
  const byWeight = (a: FamilySource, b: FamilySource) => a.fontWeight - b.fontWeight;
  const lt = styled.filter((s) => s.fontWeight < weight).sort(byWeight);
  const gt = styled.filter((s) => s.fontWeight > weight).sort(byWeight);
  if (weight >= 400 && weight <= 500) {
    const left = styled.filter((s) => s.fontWeight <= weight);
    const right = styled.filter((s) => s.fontWeight > 500);
    const fit = styled.filter((s) => s.fontWeight >= weight && s.fontWeight <= 500);
    return fit[0] ?? left[left.length - 1] ?? right[0];
  }
  if (weight < 400) return lt[lt.length - 1] ?? gt[0];
  return gt[0] ?? lt[lt.length - 1];
}

const SOFT_HYPHEN = 0xad;

/**
 * `loadBytes` receives a path relative to the fonts folder ("ibm-plex-sans/IBMPlexSans-Regular.ttf")
 * and must be synchronous; in a worker, preload the files first and serve them from memory.
 */
export function createMeasurer(loadBytes: (file: string) => Uint8Array): Measurer {
  const fonts = new Map<string, Font>();

  function load(familyName: string, weight: number, italic: boolean): Font | null {
    const own = userFontkitFont(familyName, weight, italic);
    if (own !== undefined) return own;
    const fam = FONT_MANIFEST.families.find((f) => f.family === familyName);
    if (!fam) return null;
    const src = resolveSource(familySources(fam), weight, italic ? "italic" : "normal");
    if (!src) return null;
    const file = `${fam.folder}/${src.file}`;
    let font = fonts.get(file);
    if (!font) {
      font = create(loadBytes(file));
      fonts.set(file, font);
    }
    return font;
  }

  function stackFonts(face: string, weight: number, italic: boolean): Font[] {
    return stack(face)
      .map((f) => load(f, weight, italic))
      .filter((f): f is Font => f !== null);
  }

  // Mirrors textkit's fontSubstitution: the first stack font with the glyph wins, and the last stack
  // font takes code points nobody covers.
  function runsOf(text: string, fontStack: Font[]): Array<{ font: Font; text: string }> {
    const runs: Array<{ font: Font; text: string }> = [];
    let last: Font | null = null;
    for (const ch of text) {
      const cp = ch.codePointAt(0) as number;
      let font: Font | undefined;
      if (cp === SOFT_HYPHEN && last) font = last;
      else font = [...fontStack, ...(last ? [last] : [])].find((f) => f.hasGlyphForCodePoint(cp));
      font ??= fontStack[fontStack.length - 1];
      const tail = runs[runs.length - 1];
      if (tail && tail.font === font) tail.text += ch;
      else runs.push({ font, text: ch });
      last = font;
    }
    return runs;
  }

  function measureWidth({ text, face, weight, italic = false, size, letterSpacing = 0 }: MeasureInput): number {
    if (!text) return 0;
    const fontStack = stackFonts(face, weight, italic);
    let width = 0;
    for (const run of runsOf(text, fontStack)) {
      // 'ltr' as textkit passes it, so fontkit does not reverse anything.
      const laid = run.font.layout(run.text, undefined, undefined, undefined, "ltr");
      const scale = size / run.font.unitsPerEm;
      for (const pos of laid.positions) width += pos.xAdvance * scale + letterSpacing;
    }
    return width;
  }

  function wrapLines(text: string, width: number, input: WrapInput): string[] {
    const lines: string[] = [];
    // Break opportunities: after a space run, and after a hyphen that is followed by more text.
    const tokens = text.match(/[^ -]*-(?=[^ ])|[^ ]+ *|^ +/g) ?? [];
    let line = "";
    for (const token of tokens) {
      const candidate = line + token;
      if (!line || measureWidth({ ...input, text: candidate.trimEnd() }) <= width + 1e-6) {
        line = candidate;
      } else {
        lines.push(line.trimEnd());
        line = token;
      }
    }
    if (line) lines.push(line.trimEnd());
    return lines;
  }

  function missing(text: string, face = FONT_STACK[0]): number[] {
    const fontStack = stackFonts(face, 400, false);
    const out = new Set<number>();
    for (const ch of text) {
      const cp = ch.codePointAt(0) as number;
      if (cp === SOFT_HYPHEN || cp === 0x200c || cp === 0x200d) continue;
      if (!fontStack.some((f) => f.hasGlyphForCodePoint(cp))) out.add(cp);
    }
    return [...out];
  }

  return { measureWidth, wrapLines, missing };
}

let defaultMeasurer: Measurer | null = null;

/** The render core installs one measurer (after preloading font bytes); primitives read it. */
export function setMeasurer(m: Measurer | null): void {
  defaultMeasurer = m;
}

export function getMeasurer(): Measurer {
  if (!defaultMeasurer) throw new Error("No text measurer installed; call setMeasurer(createMeasurer(...)) first");
  return defaultMeasurer;
}

/** The text as it will be drawn: caps applied. */
export const drawnText = (token: TypeToken, text: string) => (token.caps ? text.toUpperCase() : text);

/** Width of `text` set in a role/token at an optional size override, tracking and caps included. */
export function measureToken(
  m: Measurer,
  role: RoleName | TypeToken,
  text: string,
  opts: { size?: number; weight?: number; italic?: boolean; family?: string } = {},
): number {
  const t = roleToken(role);
  const size = opts.size ?? t.size;
  return m.measureWidth({
    text: drawnText(t, text),
    face: opts.family ?? resolveFace(t.face),
    weight: opts.weight ?? t.weight,
    italic: opts.italic ?? t.italic ?? false,
    size,
    letterSpacing: (t.tracking ?? 0) * size,
  });
}