@reportwright/pdf 0.1.0-beta.4 → 0.1.0-beta.5

package/dist/index.js

@@ -15070,7 +15070,25 @@ function pageText(src, pageDict, resources, raw = false) {
   let chars3 = 0;
   let moved = true;
   const fonts = /* @__PURE__ */ new Map();
-  const add2 = (x2, y, end, size, text3, d0, d1, d2, d3) => {
+  let run2 = [];
+  let runs = false;
+  const flush = () => {
+    if (run2.length) {
+      PT[PT.length - 1] += run2.join("");
+      run2 = [];
+    }
+  };
+  const add2 = (x2, y, end, size, text3, d0, d1, d2, d3, glyph = false) => {
+    const n = PT.length - 1;
+    if (glyph && runs && !moved && !(PG[n] & 1) && PD[n] === d3 && PC[n] === d2 && PS[n] === size && d0 >= PA[n] && d0 >= PAE[n] - 0.01 && d0 - PAE[n] <= Math.max(size, 1) * 0.25 && !RTL2.test(text3)) {
+      if ((chars3 += text3.length) > MAX_TEXT_CHARS) throw new PdfReadError("E_BUDGET", `A page makes more than ${MAX_TEXT_CHARS.toLocaleString("en")} characters of text`);
+      run2.push(text3);
+      if (run2.length >= 4096) flush();
+      PAE[n] = Math.max(PAE[n], d1);
+      return;
+    }
+    flush();
+    runs = glyph && !raw && !RTL2.test(text3);
     if (PT.length >= budget.maxTextItems) throw new PdfReadError("E_BUDGET", `A page has more than ${budget.maxTextItems.toLocaleString("en")} pieces of text (budget.maxTextItems)`);
     if ((chars3 += text3.length) > MAX_TEXT_CHARS) throw new PdfReadError("E_BUDGET", `A page makes more than ${MAX_TEXT_CHARS.toLocaleString("en")} characters of text`);
     if (raw) {
@@ -15136,7 +15154,7 @@ function pageText(src, pageDict, resources, raw = false) {
       }
       const n = PT.length - 1;
       if (text3.includes("⁠")) text3 = text3.replaceAll("⁠", "");
-      if (text3) add2(x2, y, end, size, text3, d0, d1, d2, d3);
+      if (text3) add2(x2, y, end, size, text3, d0, d1, d2, d3, true);
       else if (n >= 0 && PAE[n] < d1 && PD[n] === d3 && Math.abs(PC[n] - d2) < 0.01) {
         if (raw) PE[n] = end;
         PAE[n] = d1;
@@ -15302,6 +15320,7 @@ function pageText(src, pageDict, resources, raw = false) {
       default:
     }
   }
+  flush();
   if (raw) return PT.map((text3, i2) => ({ x: PX[i2], y: PY[i2], end: PE[i2], size: PS[i2], text: text3, gap: !!(PG[i2] & 1) }));
   const out = [];
   for (let i2 = 0; i2 < PT.length; ) {
@@ -16502,7 +16521,8 @@ async function run(self, sink, o) {
   if (level && o.encrypt) throw new TypeError("save: PDF/A forbids encryption: pass pdfa: false to drop the PDF/A claim and encrypt");
   const keepStruct = single && o.keepStructure !== false && !self.edits.flatten;
   const info = self.info();
-  const edited = !!self.edits.info || level?.conformance !== self.pdfa()?.conformance;
+  const tooBig = ["title", "author", "subject", "keywords"].some((k) => typeof info[k] === "string" && info[k].length > INFO_MAX);
+  const edited = !!self.edits.info || tooBig || level?.conformance !== self.pdfa()?.conformance;
   const uaLost = !keepStruct && /pdfuaid:part/.test(self.xmp() ?? "");
   const srcXmp = self.edits.xmp ?? (edited || uaLost || !level && self.pdfa() ? null : self.xmp());
   const created = o.creationDate ?? (info.creationDate instanceof Date ? info.creationDate : /* @__PURE__ */ new Date());
@@ -16938,17 +16958,19 @@ async function saveIncremental(self, o = {}) {
     keepStruct: () => true
   });
   const override = async (n, v) => put(n, await copier.text(self, v), null, v instanceof Map ? v : null);
-  let present;
-  const objects = [];
-  try {
-    present = findActive(self, (n, v) => {
+  let scan = SCANS.get(src);
+  if (!scan) try {
+    const objects2 = [];
+    const present2 = findActive(self, (n, v) => {
       const d = v instanceof PdfStream ? v.dict : v;
-      if (d instanceof Map && (d.has("ByteRange") || d.has("Perms") || d.has("Lock") || R(d.get("FT")) === "Sig" || d.has("V") || nestedSigs(d).length)) objects.push([n, v]);
+      if (d instanceof Map && (d.has("ByteRange") || d.has("Perms") || d.has("Lock") || R(d.get("FT")) === "Sig" || d.has("V") || nestedSigs(d).length)) objects2.push([n, v]);
     });
+    SCANS.set(src, scan = { present: present2, objects: objects2 });
   } catch (e) {
     throw new PdfReadError("E_ACTIVE", `${who}: the file could not be checked for active content (${/** @type {Error} */
     e.message}); use save() to rewrite the file without it (this invalidates existing signatures)`);
   }
+  const { present, objects } = scan;
   checkSigned(self, sig, o, objects, who);
   if (present.length && !o.keepActiveContent) throw new PdfReadError("E_ACTIVE", `${who}: the file has active content (${present.map((p) => `${p.what} ×${p.count}`).join(", ")}), which an incremental update cannot remove (the original bytes stay). Use save() to rewrite the file without it (this invalidates existing signatures), or pass keepActiveContent: true to keep it knowingly`);
   const fills = await fillReplacements(self, { pdfa: !!self.pdfa(), warn: (m) => warnings.push(m) });
@@ -17400,7 +17422,7 @@ function checkSigned(self, sig, o, objects, who) {
     }
   }
 }
-var SIGN_KEYS, MAX_WIDGET_SHARE, TRANSFORMS, isSigDict;
+var SIGN_KEYS, SCANS, MAX_WIDGET_SHARE, TRANSFORMS, isSigDict;
 var init_incremental = __esm({
   "src/reader/incremental.js"() {
     init_lexer();
@@ -17416,6 +17438,7 @@ var init_incremental = __esm({
     init_serialize();
     init_serialize();
     SIGN_KEYS = ["field", "signer", "reason", "location", "contactInfo", "name", "date", "timestamp", "subFilter", "reserve", "hash"];
+    SCANS = /* @__PURE__ */ new WeakMap();
     MAX_WIDGET_SHARE = 0.4;
     TRANSFORMS = /* @__PURE__ */ new Set(["DocMDP", "FieldMDP", "UR", "UR3"]);
     isSigDict = (d) => d instanceof Map && d.has("ByteRange") && (d.has("Contents") || d.has("Type") || d.has("Filter"));

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@reportwright/pdf",
-  "version": "0.1.0-beta.4",
+  "version": "0.1.0-beta.5",
   "description": "A streaming PDF writer for Node and browsers with no runtime dependencies: flat memory, PDF/A-1b to 3a, PDF/UA-1, Factur-X, embedded TrueType with Unicode text.",
   "type": "module",
   "license": "MIT",