Deterministic Output

PDFixa guarantees that the same sequence of API calls, with the same data, produces byte-identical PDF output on every run.

What makes a PDF non-deterministic

Most PDF libraries introduce at least one of these:

Source	Example
Timestamps	/CreationDate (D:20240315...) changes every run
Random IDs	Document ID is a random UUID embedded in the trailer
HashMap iteration	Resource dictionary keys in unpredictable order
Floating-point	Locale-dependent decimal formatting in content streams
Font subsetting	Glyphs subsetted in hash-map order

PDFixa eliminates all of these by design.

What PDFixa does

• No timestamps — Creation and modification dates are not written, or can be set to a fixed value.
• Stable IDs — Document ID is derived from content, not randomly generated.
• Ordered resources — Dictionaries, fonts, and image resources are written in insertion order.
• Fixed number format — PDF unit values are serialised with a fixed locale and precision.
• Deterministic font subsetting — Glyph selection and ordering are deterministic.

Verification

You can verify determinism in a test:

@Test
void pdfIsDeterministic() throws Exception {
    byte[] first  = generateReport(sampleData);
    byte[] second = generateReport(sampleData);

    assertArrayEquals(first, second);
}

Or compare SHA-256 hashes across deploys:

String hash = sha256Hex(generateReport(sampleData));
assertEquals("e3b0c44298fc...", hash);

Your responsibilities

PDFixa controls its own output, but you control the inputs. For byte-identical results:

Do	Avoid
Derive titles/authors from input	Instant.now() in metadata
Use consistent font files	Different font versions
Pass images as stable byte arrays	Re-encoding images with different quality
Sort collections before iterating	Iterating HashMap or Set

See Metadata for field-level guidance.