PDF Optimization: Pdf.Optimize

The operations

Most of the work (deduplicating, removing unused objects and resources, merging content streams, subsetting fonts, recompressing streams) is done in memory during the optimize call, so its savings are exact. Two options take effect when the file is written — objectStreams and dropOldRevisions — so their savings are estimates until the PDF is saved. JPEG, JPEG 2000, JBIG2 and CCITT image data are never re-encoded; their bytes pass through unchanged.

Quick start

An empty jsonIn applies the lossless preset: every option at its default.

Chilkat.Pdf pdf = new Chilkat.Pdf();
pdf.LoadFile("in.pdf");

Chilkat.JsonObject jsonIn = new Chilkat.JsonObject();
Chilkat.JsonObject jsonOut = new Chilkat.JsonObject();

if (pdf.Optimize("optimize", jsonIn, jsonOut)) {
    pdf.WriteFile("out.pdf");
    Console.WriteLine("Saved about " + jsonOut.IntOf("bytesSaved") + " bytes");
}

Audit first, optimize only if it is worth it

pdf.Optimize("audit", jsonIn, jsonOut);
int fileSize = jsonOut.IntOf("fileSize");
int potential = jsonOut.IntOf("estimatedSavings.total");
if (potential > fileSize / 10) {               // more than 10%
    pdf.Optimize("optimize", jsonIn, jsonOut);
    pdf.WriteFile("out.pdf");
}

The options (jsonIn)

General

Streams

Duplicates

Unused content

Page content

Fonts

File structure

Example: override individual switches

{
  "objectStreams": "disable",
  "dropOldRevisions": false
}

{
  "bytesSaved": 1181810,
  "bytesSavedIsEstimate": false,
  "actions": { "subsetFonts": { "count": 7, "bytesSaved": 942000 },
               "dedupeStreams": { "count": 39, "bytesSaved": 211800 }, ... },
  "skipped": []
}

{
  "pages": "1-24",
  "compressionLevel": 6,
  "subsetFonts": false
}

{
  "bytesSaved": 318540,
  "bytesSavedIsEstimate": true,
  "actions": { "dedupeStreams": { "count": 12, "bytesSaved": 201300 },
               "objectStreams": { "bytesSaved": 54300, "estimate": true }, ... },
  "skipped": []
}

The audit report (op = audit)

{ }

{
  "fileSize": 4831022,
  "pdfVersion": "1.4",
  "pdfa": "none",
  "usage": {
    "images":        { "bytes": 3120455, "count": 41 },
    "fonts":         { "bytes": 1204310, "count": 12 },
    "pageContent":   { "bytes": 288112,  "count": 60 },
    "formXObjects":  { "bytes": 61200,   "count": 9 },
    "annotations":   { "bytes": 8400,    "count": 22 },
    "structureTree": { "bytes": 51020 },
    "metadata":      { "bytes": 6400 },
    "embeddedFiles": { "bytes": 0,       "count": 0 },
    "oldRevisions":  { "bytes": 18210 },
    "overhead":      { "bytes": 72915 }
  },
  "estimatedSavings": {
    "total": 1290410,
    "subsetFonts": 942000,
    "dedupeStreams": 211800,
    "objectStreams": 54300,
    "recompressFlate": 41210,
    "dropOldRevisions": 18210,
    "removeUnreferenced": 12890,
    "coalesceContent": 10000
  },
  "skipped": []
}

The optimize report (op = optimize)

{ }

{
  "bytesSaved": 1236110,
  "bytesSavedIsEstimate": true,
  "actions": {
    "subsetFonts":        { "count": 7,  "bytesSaved": 942000 },
    "dedupeStreams":      { "count": 39, "bytesSaved": 211800 },
    "objectStreams":      { "bytesSaved": 54300, "estimate": true },
    "recompressFlate":    { "count": 88, "bytesSaved": 41210 },
    "dropOldRevisions":   { "bytesSaved": 18210, "estimate": true },
    "removeUnreferenced": { "count": 17, "bytesSaved": 12890 },
    "coalesceContent":    { "count": 12, "bytesSaved": 10000 }
  },
  "skipped": [
    { "option": "tidyContent", "pages": "12",
      "reason": "content could not be parsed; left unchanged" }
  ]
}

Signed, encrypted, PDF/A and tagged PDFs

Optimizing always rewrites the whole file, so it follows these rules; audit changes nothing and is always allowed.

• Signed PDFs. With the PreserveSignatures property true (the default), optimize fails on a signed PDF. With it false, the PDF is optimized and its signatures are invalidated.
• Encrypted PDFs. The PDF is saved with its original security unchanged; object streams are compatible with encryption. Because Chilkat does not ignore owner restrictions, optimize on an encrypted PDF requires owner rights or the modify permission (see Protection).
• PDF/A. Chilkat reads the PDF/A claim from the XMP metadata and stays within that part's rules: for PDF/A-1, objectStreams is skipped and reported; for every part, the output intent and XMP metadata are kept and fonts stay embedded (subsetting is allowed). replaceLegacyFilters also removes LZW, which PDF/A forbids.
• Forms. Fonts that form fields need to display newly typed text are not subset.
• Tagged PDF. The structure tree is never touched; tidyContent keeps every marked-content tag, and dedupeStreams never shares a page content stream that carries marked-content IDs between two pages.

Option reference

{
  "preset": "lossless",
  "pages": "all",
  "compressUncompressed": true,     "recompressFlate": true,
  "replaceLegacyFilters": true,     "imagePredictors": true,
  "compressionLevel": 9,
  "dedupeStreams": true,            "dedupeObjects": true,
  "mergeFontSubsets": false,
  "removeUnreferenced": true,       "removeUnusedResources": true,
  "removeUnusedLayers": true,
  "coalesceContent": true,          "tidyContent": true,
  "subsetFonts": true,              "compactType1Fonts": false,
  "objectStreams": "generate",      // generate | preserve | disable
  "dropOldRevisions": true
}

Reference documentation: Pdf.Optimize, Pdf.WriteFile. Related: ExtractText, TextReplace, Watermark, Protection, Pdf class overview.