Discussion
As I described yesterday, Papyrix's page-cache can defer to logic for laying out paragraphs of plaintext into a layout/rendering tree. So what objects might a Papyrix layout be constructed from?
There's a PageElement superclass with x/y coordinates, as well as PageLine & PageImage subclasses both wrapping a block object. While a Page wraps an array of such elements.
There's a method to test whether any of those are PageImages, & a method to get min/max coords on both axis.
1/?
There's a ParsedText class tracking an array of words & an array of their text-styles, the block's text style, indentation level & flag, hyphenation flag, greedy-breaking flag, & text-direction flag.
There's a helper checking if a word ends in certain latinate punctuation chars. A utility for locating all the soft-hyphen chars in a UTF-8 string alongside ones for removing them, checking if a word ends in one, or replacing the trailing one with a hard hyphen.
3/4?
There's helpers for splitting around a given hyphen converting it into an actual one, & removing hyphens from the prefix but not suffix.
There's a CJK charset test. A formula calculating cubic-ratio in the difference between line lengths, alongside one for calculating "demerits" from that without penalizing the last line.
There's a method for appending a word to the parsed text ignoring empty words & handling CJK characters specially as individual words.
Then there's the layout method!
4/5?
To apply layout to a ParsedText object after some validation Papyrix tidies up after interrupted work looking for soft-hyphens to collapse, takes a shortcut for excessively long paragraphs, pre-hyphenates (via a helper with multiple branches/loops) words that are so wide they'll never fit, calculates the width of each word (via a helper & renderer method) taking into account indentation, runs one of 2 methods for the core logic, checks whether the work's aborted, & ...
5/5.2 For today!
These objects have Render methods deferring to their inner objects, as well as (de)serialization methods (aided by a support library I've already studied) so it can be cached in the SD card.
That said PageImages might clear its background 1st.
Pages might traverse its tree to gather a histogram of required glyphs, & sort that transient array to prefetch them.
There's a RenderConfig struct specifying various text-layout parameters.
2/3?