theseus/bundled-addons/pdf-editor/lib/save.js

// Turning the editing session back into a PDF, with pdf-lib.
//
// The source bytes are never mutated. Every save re-loads the original and
// replays the session onto that fresh copy, so a botched save cannot poison
// the next one and the file on disk stays whatever it was.
//
// Step order is load-bearing:
//
//   1. form values     — before anything strips a widget
//   2. flatten         — replaces a redacted page's content stream, which
//                        would wipe marks drawn in step 3
//   3. marks           — drawn against ORIGINAL page indices, which are still
//                        valid because nothing has moved yet
//   4. page tree       — reorder and delete last, so step 3's indices held
//   5. rotation        — independent of order, done in the same pass
//
// Marks are drawn with pdf-lib's operator helpers rather than drawSvgPath.
// drawSvgPath takes an SVG-shaped path with y running downwards from an
// anchor, so every mark would need a second coordinate convention and a
// matching inverse in the overlay. Raw operators take the user-space points we
// already store, unchanged.

import {
  PDFDocument, PDFName, StandardFonts, rgb, degrees,
  pushGraphicsState, popGraphicsState, moveTo, lineTo, stroke, fill,
  setStrokingColor, setFillingColor, setLineWidth, setLineCap, setLineJoin,
  LineCapStyle, LineJoinStyle,
} from "../vendor/pdf-lib/pdf-lib.esm.min.js";

import { arrowBarbs, signatureStrokes, textLayout, RULE_OFFSET } from "./shape.js";

/** #rrggbb -> pdf-lib rgb(). */
function colorOf(hex) {
  const m = /^#?([0-9a-f]{6})$/i.exec(String(hex || "#000000"));
  const n = m ? parseInt(m[1], 16) : 0;
  return rgb(((n >> 16) & 255) / 255, ((n >> 8) & 255) / 255, (n & 255) / 255);
}

// Helvetica is a WinAnsi font: it simply cannot encode most of Unicode.
// Embedding a font that could would mean shipping a several-MB typeface for a
// feature most stamps never need, so instead we substitute and say so.
const WIN_ANSI_OK = /^[\x20-\x7E\xA0-\xFF‘’“”–—•€…‰‹›ŒœŠšŸŽžƒˆ˜™†‡‰]*$/;
export function winAnsiSafe(text) {
  const s = String(text ?? "");
  if (WIN_ANSI_OK.test(s)) return { text: s, dropped: 0 };
  let dropped = 0;
  const out = Array.from(s).map((ch) => {
    if (WIN_ANSI_OK.test(ch)) return ch;
    dropped++;
    return "?";
  }).join("");
  return { text: out, dropped };
}

/** True when any text stamp carries characters Helvetica cannot write. */
export function unencodableStamps(model) {
  return model.annots.filter((a) => a.kind === "text" && winAnsiSafe(a.text).dropped > 0).length;
}

// ---- mark drawing -----------------------------------------------------
function strokePolyline(page, pts, color, width) {
  if (pts.length < 2) return;
  const ops = [
    pushGraphicsState(),
    setStrokingColor(colorOf(color)),
    setLineWidth(width),
    setLineCap(LineCapStyle.Round),
    setLineJoin(LineJoinStyle.Round),
    moveTo(pts[0][0], pts[0][1]),
  ];
  for (let i = 1; i < pts.length; i++) ops.push(lineTo(pts[i][0], pts[i][1]));
  ops.push(stroke(), popGraphicsState());
  page.pushOperators(...ops);
}

function fillRect(page, x, y, w, h, color, opacity) {
  if (opacity != null && opacity < 1) {
    // drawRectangle is the only path that can set a fill opacity, because it
    // registers the ExtGState for us.
    page.drawRectangle({ x, y, width: w, height: h, color: colorOf(color), opacity, borderWidth: 0 });
    return;
  }
  page.pushOperators(
    pushGraphicsState(), setFillingColor(colorOf(color)),
    moveTo(x, y), lineTo(x + w, y), lineTo(x + w, y + h), lineTo(x, y + h),
    fill(), popGraphicsState());
}

function drawMark(page, a, font) {
  switch (a.kind) {
    case "highlight":
      for (const r of a.rects || []) {
        if (r.w <= 0 || r.h <= 0) continue;
        fillRect(page, r.x, r.y, r.w, r.h, a.color, a.opacity ?? 0.38);
      }
      return;
    case "pen":
      strokePolyline(page, a.pts || [], a.color, a.width || 2);
      return;
    case "underline": case "strikeout": {
      // Same offset the overlay drew it at, from the same constant.
      const frac = RULE_OFFSET[a.kind];
      const w = Math.max(0.5, a.width || 1.4);
      for (const r of a.rects || []) {
        const y = r.y + r.h * frac;
        strokePolyline(page, [[r.x, y], [r.x + r.w, y]], a.color, w);
      }
      return;
    }
    case "rect":
      page.drawRectangle({ x: a.x, y: a.y, width: a.w, height: a.h,
        borderColor: colorOf(a.color), borderWidth: a.width || 2,
        color: a.fill ? colorOf(a.color) : undefined,
        opacity: a.fill ? (a.opacity ?? 0.25) : 0 });
      return;
    case "ellipse":
      page.drawEllipse({ x: a.x + a.w / 2, y: a.y + a.h / 2,
        xScale: Math.max(0.5, a.w / 2), yScale: Math.max(0.5, a.h / 2),
        borderColor: colorOf(a.color), borderWidth: a.width || 2,
        color: a.fill ? colorOf(a.color) : undefined,
        opacity: a.fill ? (a.opacity ?? 0.25) : 0 });
      return;
    case "line":
      strokePolyline(page, [[a.x1, a.y1], [a.x2, a.y2]], a.color, a.width || 2);
      return;
    case "arrow": {
      const w = a.width || 2;
      strokePolyline(page, [[a.x1, a.y1], [a.x2, a.y2]], a.color, w);
      const b = arrowBarbs(a.x1, a.y1, a.x2, a.y2, w);
      strokePolyline(page, [[b[0].x, b[0].y], [a.x2, a.y2], [b[1].x, b[1].y]], a.color, w);
      return;
    }
    case "text": {
      const size = a.size || 12;
      const f = typeof font === "function" ? font(a) : font;
      for (const line of textLayout(a)) {
        const { text } = winAnsiSafe(line.text);
        if (!text) continue;
        page.drawText(text, { x: line.x, y: line.y, size, font: f, color: colorOf(a.color) });
      }
      return;
    }
    case "signature":
      for (const s of signatureStrokes(a)) strokePolyline(page, s, a.color, a.width || 1.6);
      return;
    case "redact":
      // Opaque, no border, no transparency — a redaction that can be seen
      // through is not a redaction.
      fillRect(page, a.x, a.y, a.w, a.h, "#000000", 1);
      return;
    default:
      console.warn("save: unknown mark kind", a.kind);
  }
}

// ---- form fields ------------------------------------------------------
function applyFormValues(doc, values, font, report) {
  if (!values.length) return;
  let form;
  try { form = doc.getForm(); }
  catch (e) { report.warn.push("This file's form could not be opened, so field values were not saved."); return; }
  const byName = new Map();
  try { for (const f of form.getFields()) byName.set(f.getName(), f); }
  catch (e) { report.warn.push("This file's field list could not be read."); return; }

  let wrote = 0, missed = 0;
  for (const { name, value } of values) {
    const f = byName.get(name);
    if (!f) { missed++; continue; }
    try {
      if (typeof f.setText === "function") {
        const { text, dropped } = winAnsiSafe(value == null ? "" : String(value));
        if (dropped) report.substituted += dropped;
        f.setText(text);
      } else if (typeof f.check === "function") {
        const on = value === true || value === "On" || (typeof value === "string" && value !== "Off" && value !== "");
        if (on) f.check(); else f.uncheck();
      } else if (typeof f.select === "function") {
        if (value == null || value === "") { if (typeof f.clear === "function") f.clear(); }
        else f.select(Array.isArray(value) ? value[0] : String(value));
      } else { missed++; continue; }
      // Per-field so one field Helvetica cannot draw does not abort the save.
      try { if (typeof f.updateAppearances === "function") f.updateAppearances(font); } catch {}
      wrote++;
    } catch (e) {
      missed++;
      console.warn(`form field "${name}":`, e?.message || e);
    }
  }
  report.fieldsWritten = wrote;
  if (missed) report.warn.push(`${missed} field${missed === 1 ? "" : "s"} could not be written back.`);
}

// ---- flattening -------------------------------------------------------
// A redaction box that only paints over text is theatre: the text is still in
// the content stream and still comes out of a copy-paste. The only fix that
// pdf-lib alone can reach is to replace the page's whole content stream with a
// picture of it, which is what this does. Annotations (form widgets, links) are
// deliberately KEPT — they are not part of the content stream, so they were
// never the leak, and dropping them would quietly break a form.
async function flattenPage(doc, page, pngBytes, report) {
  const img = await doc.embedPng(pngBytes);
  const box = page.getCropBox();
  // Empty array, not a fresh empty stream: pdf-lib appends its own stream to
  // this array on the first draw call, so the original marks are simply gone.
  page.node.set(PDFName.of("Contents"), doc.context.obj([]));
  page.drawImage(img, { x: box.x, y: box.y, width: box.width, height: box.height });
  report.flattened++;
}

// ---- page tree --------------------------------------------------------
function rebuildPages(doc, originalPages, srcOrder) {
  const wanted = srcOrder.map((i) => originalPages[i]).filter(Boolean);
  if (!wanted.length) throw new Error("a PDF needs at least one page");
  for (let i = doc.getPageCount() - 1; i >= 0; i--) doc.removePage(i);
  // addPage re-registers a page whose ref was dropped by removePage, so the
  // same page objects (and the widget annotations that point at them) come
  // back intact in the new order.
  for (const p of wanted) doc.addPage(p);
}

/**
 * @param {object}   o
 * @param {Uint8Array} o.originalBytes  untouched source
 * @param {DocModel} o.model
 * @param {{name:string,value:any}[]} o.formValues harvested from pdf.js
 * @param {boolean}  o.flatten          rebuild redacted pages as images
 * @param {(uid:string)=>Promise<Uint8Array>} o.rasterize renders one page to PNG
 * @returns {Promise<{bytes:Uint8Array, report:object}>}
 */
export async function buildPdf({ originalBytes, model, formValues = [], flatten = true, rasterize }) {
  const report = { flattened: 0, marks: 0, fieldsWritten: 0, substituted: 0, warn: [] };
  const doc = await PDFDocument.load(originalBytes, {
    ignoreEncryption: true,
    // Leave the file's own Producer/ModDate alone. A privacy-minded browser
    // should not stamp "this was edited, with this library, at this time" into
    // every document that passes through it.
    updateMetadata: false,
  });
  // Four variants, because bold and italic are separate fonts in a PDF rather
  // than a style applied to one. Embedded once up front; unused ones cost a
  // few hundred bytes of font dictionary and nothing else.
  const fonts = {
    regular: await doc.embedFont(StandardFonts.Helvetica),
    bold: await doc.embedFont(StandardFonts.HelveticaBold),
    italic: await doc.embedFont(StandardFonts.HelveticaOblique),
    boldItalic: await doc.embedFont(StandardFonts.HelveticaBoldOblique),
  };
  const font = fonts.regular;
  const fontFor = (a) => a.bold && a.italic ? fonts.boldItalic
    : a.bold ? fonts.bold
    : a.italic ? fonts.italic
    : fonts.regular;
  const originalPages = doc.getPages();

  // 1. form values
  applyFormValues(doc, formValues, font, report);

  // 2. flatten redacted pages
  const redacted = model.redactedPages();
  if (flatten && redacted.size && typeof rasterize === "function") {
    for (const uid of redacted) {
      const src = model.page(uid).src;
      const page = originalPages[src];
      if (!page) continue;
      try { await flattenPage(doc, page, await rasterize(uid), report); }
      catch (e) {
        report.warn.push(`Page ${src + 1} could not be rebuilt as an image; its redaction only covers the text.`);
        console.warn("flatten failed:", e);
      }
    }
  } else if (redacted.size && !flatten) {
    report.warn.push("Redactions cover the text but do not remove it — it is still extractable from the saved file.");
  }

  // 3. marks, against original indices
  for (const uid of model.visible()) {
    const page = originalPages[model.page(uid).src];
    if (!page) continue;
    // Redaction boxes go on top of everything else on their page.
    const marks = model.annotsFor(uid);
    for (const a of marks) if (a.kind !== "redact") { drawMark(page, a, fontFor); report.marks++; }
    for (const a of marks) if (a.kind === "redact") { drawMark(page, a, fontFor); report.marks++; }
  }

  // 4. page tree
  const srcOrder = model.visible().map((uid) => model.page(uid).src);
  if (model.pagesChanged()) {
    rebuildPages(doc, originalPages, srcOrder);
    report.pagesRebuilt = true;
  }

  // 5. rotation — applied to the page objects, so order does not matter
  for (const uid of model.visible()) {
    const extra = model.extraRotation(uid);
    if (!extra) continue;
    const page = originalPages[model.page(uid).src];
    if (!page) continue;
    try { page.setRotation(degrees((page.getRotation().angle + extra) % 360)); }
    catch (e) { console.warn("setRotation:", e?.message); }
  }

  const bytes = await doc.save({ useObjectStreams: true });
  return { bytes, report };
}