Merge pull request #21683 from calixteman/fix/quad-regex-writer

Don't write numbers in exponential notation when saving a pdf

src/core/writer.js

@@ -148,6 +148,31 @@ async function writeArray(array, buffer, transform) {
   buffer.push("]");
 }
 
+// The exponential notation isn't valid in a PDF, hence a number is always
+// written with all its digits.
+function numberToPDFString(value) {
+  // `toFixed` uses the exponential notation from 1e21 on, so such a number is
+  // written thanks to BigInt: it's necessarily an integer, and `isInteger` also
+  // rules out NaN and ±Infinity for which BigInt would throw.
+  if (Number.isInteger(value) && Math.abs(value) >= 1e21) {
+    return BigInt(value).toString();
+  }
+
+  // Below that limit `toFixed(10)` never uses the exponential notation (unlike
+  // `toString` which uses it under 1e-6) and it rounds the value: it always
+  // adds 10 decimals, hence scan them backwards to remove the trailing zeros,
+  // and then the dot itself when none of the decimals is left.
+  const str = value.toFixed(10);
+  let end = str.length;
+  while (str[end - 1] === "0") {
+    end--;
+  }
+  if (str[end - 1] === ".") {
+    end--;
+  }
+  return str.slice(0, end);
+}
+
 async function writeValue(value, buffer, transform) {
   if (value instanceof Name) {
     buffer.push(`/${escapePDFName(value.name)}`);
@@ -165,9 +190,7 @@ async function writeValue(value, buffer, transform) {
     // matrices (e.g. [0.000008 0 0 0.000008 0 0]).
     // The numbers must be "rounded" only when pdf.js is producing them and the
     // current transformation matrix is well known.
-    // toFixed(10) avoids scientific notation and rounds; the replace removes
-    // trailing zeros (and a trailing dot for integers).
-    buffer.push(value.toFixed(10).replace(/\.?0+$/, ""));
+    buffer.push(numberToPDFString(value));
   } else if (typeof value === "boolean") {
     buffer.push(value.toString());
   } else if (value instanceof Dict) {

test/unit/writer_spec.js

@@ -295,10 +295,20 @@ describe("Writer", function () {
     });
 
     it("should not use scientific notation for very large numbers", async function () {
-      // JavaScript produces scientific notation above ~1e21 but such values
-      // are unlikely in PDFs; values below that threshold must be plain.
+      // JavaScript's toString() and toFixed() produce scientific notation from
+      // 1e21 on, which is invalid PDF: such a number must be written with all
+      // its digits, which are the exact ones of the underlying double.
       expect(await serialize(1e10)).toEqual("10000000000");
       expect(await serialize(1.5e6)).toEqual("1500000");
+      expect(await serialize(1e20)).toEqual("100000000000000000000");
+      expect(await serialize(1e21)).toEqual("1000000000000000000000");
+      // Removing the trailing zeros of the exponent used to change the value:
+      // "1e+30" was written "1e+3" and "1e+100" was written "1e+1".
+      expect(await serialize(1e30)).toEqual("1000000000000000019884624838656");
+      expect(await serialize(-1e30)).toEqual(
+        "-1000000000000000019884624838656"
+      );
+      expect((await serialize(1e100)).length).toEqual(101);
     });
 
     it("should round to at most 10 decimal places", async function () {