Release date: 14th July 2020
Next FormVu release: 25th August 2020
A note about OpenJDK 8: If you are currently running FormVu using OpenJDK 8 we strongly recommend updating to version 11 or newer, or using an alternative JDK provider.
Due to bugs in the OpenJDK 8 JVM implementation, running FormVu with OpenJDK 8 can cause lower-quality rendering, missing image content, higher memory usage and potentially JVM crashes.
- [HTML-3387] - Clip images in SVG to allow JPG to be used to reduce file size
This is a large change to the generation of images in FormVu (FDF conversion only).
Instead of clipping the images in advance and writing out as PNG files (in order to support alpha transparency), FormVu now writes out unclipped JPEG images and clips them in SVG.
This has allowed us to achieve a significant reduction in the file size of generated images in SVG text modes. This reduces image sizes on average by over 80% for PDF files in our test suite.
This change can result in more prominent fractional white lines in a small subset of files. Read more about it here.
Please let us know if you rely on the old solution so that we know if we should continue to support it with a configuration option.
[HTML-3391] - Subset embedded fonts
Fonts now only include glyphs which are used in the file, which can considerably reduce file sizes.
[HTML-3378] - Reduce number of fonts required and improve text output
Large reduction in the number of fonts generated by PDF files that reuse the same base font file for many fonts. This change also improves the extraction values where multiple glyphs share a common extraction value in the same font.
This is achieved by splitting the text block and creating a new font instead of remapping the extraction value to achieve the correct appearance.
Combined with HTML-3391, the file size of generated fonts has been reduced on average by over 75% for PDF files in our test suite.
- [HTML-3389] - Links not appearing in converted HTML content
Improved regex detection of hyperlinks in text
[HTML-3392] - SVG Shapetext is not appearing in Safari 13
Adds a workaround for a regression that appeared in Safari 13 where shapetext may not be displayed on first page load
[HTML-3159] - Characters using unsuitable unicode range
Fixes characters that incorrectly appeared as question marks due to being in the surrogate range
[CORE-3614] - Issue with blending on images (should be transparent but appear opaque)
Fixes blend not being passed to mask code
[CORE-3636] - /MediaBox value not correctly read in customer file
Fixes exception which prevented conversion
[CORE-3646] - FontMatrix from CFF table not properly applied in OTF font
Fixes font being drawn too large
[CORE-3617] - Avoid memory issues by limiting use of smoothing on very large images
Prevents OutOfMemoryError in files with very large images
[HTML-3313] - Font is not output due to lack of table data
Fixed by drawing text as shapetext when there is insufficient data in the font