Crate pdfrum_anyrender 

pdfrum-anyrender

anyrender scenes written as vector PDF pages through pdfrum-edit. Paths stay paths, gradients stay gradients, text stays text in embedded, subsetted fonts, and images keep their source encoding where PDF allows (a JPEG as-is, an opaque PNG’s compressed data as-is).

A page is painted into anyrender’s recording Scene first, by whatever renders through anyrender. write() then embeds every face and image the scenes use and replays each scene onto a pdfrum canvas: anyrender’s push/pop layers become nested saved graphics states, the only shape a pdfrum canvas lets a content stream take.

use anyrender::Scene;
use pdfrum_anyrender::{GlyphArea, Page, Sources, write};
use peniko::kurbo::{Affine, Rect, Size};

let page = Page {
    size: Size::new(200.0, 100.0),
    scene: Scene::default(),
    placement: Affine::IDENTITY,
    clip: Rect::new(0.0, 0.0, 200.0, 100.0),
    area: GlyphArea::Anywhere,
};
let written = write(&[page], &Sources::default())?;
assert!(written.bytes.starts_with(b"%PDF-"));

anyrender’s glyph runs carry glyph ids, not the text they were shaped from. Hand the text over in Sources::texts, keyed by the run as the scene holds it (RunKey), to make words selectable; a run without an entry falls back to the font’s own cmap, reversed (GlyphTally counts both).

What is simplified

Box shadows are dropped; filters and backdrop filters are ignored; compositing operators other than source-over paint as source-over; a sweep gradient, and a gradient on a stroke or on text, paints as its stops’ average colour; gradients pad at their ends whatever extend mode they name, and interpolate in sRGB; an image brush draws once, clipped to its shape.

Test font

tests/fixtures/noto-serif-latin.ttf is a Latin subset of Noto Serif, licensed under the SIL Open Font License 1.1 (tests/fixtures/OFL-noto-serif.txt). It is a test fixture and is not part of the published crate’s code.

Structs

EncodedImage

An image’s bytes as they arrived, before anything decoded them.

GlyphSource

One glyph’s share of the source text: the cluster it belongs to (glyphs of one cluster share it) and that cluster’s text.

GlyphTally

What the pages did with the glyphs they were given.

ImageSources

Encoded sources keyed by the id of the decoded blob a brush paints (peniko::Blob::id): a renderer decodes once and paints that blob wherever the image appears, so the id names the image across every draw.

Page

One page: its sheet, what is painted on it, and where.

RunKey

A glyph run as the painter sees it: the font face, the size and every glyph’s id and position, bit for bit. Two runs with the same key draw the same thing.

RunText

The text of one run and, for each glyph in drawing order, the byte range of its cluster in that text. Glyphs of one cluster share a range; pdfrum writes it once, as /ActualText when the glyphs cannot carry it in the font’s ToUnicode map.

RunTexts

Run texts by key. When two runs share a key but not a text (only possible when two code points share a glyph and everything else about the runs matches), the first recorded wins: record in document order.

Sources

Everything a scene may look up while it paints; built once per document, shared by pages.

Written

A written PDF and what became of its glyphs.

Enums

GlyphArea

Where a page’s glyphs may land, in its scene’s coordinates.

ImageCodec

How an image’s source is encoded.

Functions

write

pages as one PDF, their text and images looked up in sources.