Version 3.10.1 includes a couple of fixes:

• Fix in lazy reader for a bug affecting some files
• Relax package check removing module requirement

Version 3.10.0 includes a number of fixes, including substantial performance improvements.

Highlights

• Image handling refactored and optimized for speed and memory use.
• Support for images with BPC other than 8, more generic handling.
• Partial form handling, flattening a subset of fields
• ApplyArea text added to extractor to enable extracting text from a defined region.
• Fix applied in drawing signature appearance recengle, closing #369

Note that UniPDF requires an electronic license code to operate. A free trial can be obtained at https://unidoc.io/

Version 3.9.0 includes multiple new features and improvements:

• Improvements in text extraction quality supporting complex layouts such as multi columns.
• Multiple improvements in form filling and appearance generation as well as added test coverage.
• Improvements in font subsetting
• More options for PDF optimization including automatic font cleanup and content stream reduction.

Pull requests included:

• #389 Add FDF support for form fill rendering test (#389) (@adrg)
• #387 Add form fill render tests (#387) (@adrg)
• #386 Refactor creator render test functionality (#386) (@adrg)
• #366 Text extraction code for columns. (#366) (@peterwilliams97)
• #385 Form fill field rotation (#385) (@adrg)
• #382 Add cached Stream method for CMap objects (#382) (@adrg)
• #381 Add fill and stroke colors for extracted text marks (#381) (@adrg)
• #377 Skip referenced pages which are not present in the catalog (#377) (@adrg)
• #362 Font subsetting and font optimization improvements (#362) (@gunnsth)

Version 3.8.0 highlights fixes and improvements in form filling. Custom fonts can now be used when filling forms. In addition contains fixes for field appearance generation. Improvements in multi-byte CMap handling for text extraction.

Pull requests included:

• #370 Combo field appearance (#370) (@adrg)
• #368 Add configurable font fallback options for form fields (#368) (@adrg)
• #367 Fix outline null object check (#367) (@adrg)
• #353 Handle multibyte entries in CMaps. (#353) (@peterwilliams97)
• #364 Account for rotation when generating flattened text field appearances (@adrg)

Version 3.7.1 includes a number of fixes.

Pull requests included:

• #349 Account for parent CTM when calculating positions of extracted forms (#349) (@adrg)
• #359 Use page indirect object for internal outline destinations (#359) (@adrg)
• #357 Account for inverted annotation rects when calculation appearance bounds (#357) (@adrg)
• #356 Add reader method for checking if the AcroForm needs repair (#356) (@adrg)
• #351 Add reader AcroForm repair functionality (#351) (@adrg)
• #348 Fix table right-aligned content (#348) (@adrg)
• #346 Subsetting fixes (#346) (@gunnsth)
• #344 Fix CMap generation and serialization for composite fonts (#344) (@adrg)
• #342 Fix incorrect render of invoice totals (#342) (@adrg)

Version 3.7.0 introduced font subsetting of composite fonts which enables significant reduction of file sizes for symbolic languages. List of updates:

• Example of font subsetting: In our test case the output PDF with Chinese text was reduced from 6.2MB to 45kB. See our example testcase
• Invoice component improved with more customization and can now support multi-page spanning text.
• Bugfixes and improvements in form filling support.

Pull requests included:

• #335 Support for subsetting fonts (#335) (@gunnsth)
• #334 Invoice component improvements (#334) (@adrg)
• #331 Add support for multi-block styled paragraphs (#331) (@adrg)
• #328 Form fill fixes (#328) (@adrg)

Version 3.6.2 adds digital timestamping support, as well as support for creating and reading PDF page labels.
There are also a number of bugfixes.

Pull requests included:

• #301 Add timestamp signature handler (#301) (@a5i)
• #325 Add low level PageLabels support (#325) (@adrg)
• #319 Add NewCompositePdfFontFromTTF to load composite TTF from memory (@gunnsth)
• #324 Cache raw CMap data (#324) (@adrg)
• #320 Fix error handling in Writer (@gunnsth)
• #315 Fix for crash in CCITT decoder. Resolves #314 (#315) (@gunnsth)
• #313 Double quote content stream operator fixes (#313) (@adrg)
• #309 Prevent Type 0 function evaluation crash (#309) (@adrg)
• #308 Prevent recursion when building invalid outline tree (#308) (@adrg)
• #305 Prevent panics (#305) (@adrg)

Version 3.6.0 introduces a JBIG2 encoder that can be used for efficient compression of bitonal images and masks. In addition, there are notable bugfixes.

Pull requests included:

• #294 Paragraph components fixes (#294) (@adrg)
• #288 JBIG2 Encoder support for inserting binary images into PDF (#288) (@kucjac)
• #291 Prevent crashing when processing content stream (#291) (@adrg)
• #264 JBIG2 Generic Encoder (#264) (@kucjac)
• #286 Use NRGBA when loading model.Image instances from Go images (@adrg)
• #281 Write out multi-encoder filters correctly (@peterwilliams97)

Version 3.5.0 adds initial support for rendering, allowing converting PDF to image formats. In addition, there are a few bug fixes with font/encodings and indexed colorspace.

Pull requests included:

• #273 Parse ttf encoding subtable 31 after subtable 10 (#273) (@adrg)
• #272 Add basic glyph metrics support for Type 0 CID fonts (#272) (@adrg)
• #266 Add basic image rendering support (#266) (@adrg)
• #259 Fixed PdfColorspaceSpecialIndexed.ImageToRGB() (#259) (@peterwilliams97)
• #257 Add PdfFont text encoding methods (#257) (@adrg)