Difference: Tasks (1 vs. 9)

• orphans/widows
• html5 parser
• small caps (and real SC font)

• Tables
• Floats
• Links
• SVG images
• Hyphenation
• Penalty for breaking at soft hyphen
• CSS details:
  • min-width, max-width, width, height, etc
  • background shorthand property
  • inline-block
  • text-decoration, inline borders and background

• mutool convert
• Simplify text extraction.
• Improve HTML output: semantic and visual variant outputs.
• CBZ output.
• SVG output.
• XPS output.

• Raster images (also how to get image data from SVG in an EPUB; share the fz_archive?).
• Text.
• Everything else.

• A non-AA rendering mode that never has AA blending artifacts, i.e. "pixel perfect".
• Separate text and path AA settings (non-AA paths, AA text)
• No dropouts in path rendering (touch every pixel rule)
• No overlaps in image and shading rendering (no grid fitting)

Thin pdf_page struct in pdf interpreter; base everything off lazy pdf_obj accesses.

mutool clean sanitize mode should clean annotations too!

Generic reusable text shaping functions, for use with SVG, annotation creation, and GUIs. Usable with JS and JNI too, to create fz_text objects to draw.

mupdf-gl:

• Shaping in text widgets (search field, outline, info box)
• Form filling
• Annotation editing
• RTL/manga mode (affects smart scrolling)
• Facing side-by-side page views
• Printing on win32, linux

Transfer functions.

PDF operator processing should be exposed as an API. To C, JNI and JS.

Link and action rework:

• fz_link and fz_link_dest should be opaque structs with subclasses for each document type.
• actions vs. links -- all pdf links are actions, but actions are pdf specific.

• Expose forms and values to JS/JNI.
• Add 'action' as a type.

• Use printf to create annotation streams (don't bother with display lists).
• Use text shaping when updating text.

High priority tasks:

        * new JNI interface, and new android sample viewer
        * pdf page and annotation editing API, with pdf output from fz_device calls
        * form filling API

Incomplete features...

        * make I/O consistently use our interfaces (fz_stream/fz_output)
        * javascript bindings can be simplified with mujs; no need for abstraction layer
        * link destinations need an X,Y page position
        * pdf operator filtering should be exposed to library for pdfwrite editing

OUTPUT

        * SVG output
        * XPS output
        * PDF output from graphics
                - needs a decent interface
                - create PDF from scratch
                - create content stream within existing PDF
                - create font objects and use proper text encoding
        * text extraction heuristics need lots of work
                HTML text output for reflow needs improvement.
                The current mess of tables is unsuitable for text to speech.
                We should use real html with CSS styling.
                Separate HTML output variant for faithful page reproduction can be done
                with css absolute positioning.

EPUB

        * SVG parser -- needed for EPUB
        * EPUB layout needs a lot of work
                - floating images
                - tables
                - css details

FORMS
        * form filling parity for all viewers
        * form input/display/updating need proper library API
        * form actions are missing huge chunks of functionality (only Goto and JS implemented)
        * more missing PDF javascript interfaces
        * form fields use wrong encodings
        * form text layout and appearance synthesis with correct encodings
        * digital signatures -- update crypto library (use libressl, not openssl)
        * digital signature API usable as library
        * form submission API to tell client what to do
        * saving and loading form data (FDF)

EDITING
        * pdf low level editing API needs polishing and robustness work
        * make document editing/creation robust -- it's fragile as hell
                - create blank PDF document from scratch
                - create, delete, rearrange PDF pages using library API
                - copy pages between PDF documents
                - create new pages and fill with buffers
                - create new pages and fill with pdf output module
        * all file writing needs fixing (don't clobber on errors)

ANNOTATIONS
        * annotation creation API usable from library, not just android app
                - create high level
                - use pdf output / printf-style content stream editing api
        * annotation appearance synthesis for all annotation types

ANDROID
        * real library interface from JNI that isn't tied to android app
        * pluggable android viewer component/activity using JNI interface