Keep the PDF text layer aligned with line breaks, use one clock reading for raster and text layer, and test the page text (#1153)

This commit is contained in:
MarekSpodBiedry committed 2026-10-23 16:08:53 +02:00
1 parent cc6773c95f
commit 813378cc20
5 files changed
+89 -22

No files matched your search

+5 -4
View File
@@ -299,7 +299,7 @@ pub fn master_text_pieces(pres: &Presentation, m: &deckcraft_model::Master, pape
.body
.paragraphs
.get(run.para)
.map(|p| p.runs.iter().map(|r| r.text.as_str()).collect::<String>())
.map(|p| p.runs.iter().map(|r| r.content()).collect::<String>())
.unwrap_or_default()
.chars()
.skip(run.chars.0)
@@ -440,7 +440,8 @@ impl Exporter<'_> {
let k = b.width() / sw;
// Raster at the requested dpi relative to the box size on paper.
let scale = (b.width() / 72.0 * self.opts.dpi.clamp(36.0, 600.0)) / sw;
let ropts = RenderOpts { scale, threads: 4, ..Default::default() };
// One clock reading for the raster and the text layer, so a date field cannot differ between them.
let ropts = RenderOpts { scale, threads: 4, now: deckcraft_text::datetime::DateTime::now_local(), ..Default::default() };
// Two rasters per slide: the backdrop (background, master and layout graphics) is the same
// on every slide that shares its layout, so the writer stores it once, and the slide's own
// shapes draw over it (#845).
@@ -476,7 +477,7 @@ impl Exporter<'_> {
}
if self.opts.text_layer {
let view = Affine::translate((b.x0, b.y0)) * Affine::scale(k);
for pt in deckcraft_render::place_slide(pres, i).texts {
for pt in deckcraft_render::place_slide_with(pres, i, &ropts).texts {
self.text_layer(s, &pt, view, false)?;
}
let _ = sh;
@@ -494,7 +495,7 @@ impl Exporter<'_> {
.body
.paragraphs
.get(run.para)
.map(|p| p.runs.iter().map(|r| r.text.as_str()).collect::<String>())
.map(|p| p.runs.iter().map(|r| r.content()).collect::<String>())
.unwrap_or_default()
.chars()
.skip(run.chars.0)
+56 -5
View File
@@ -200,8 +200,6 @@ fn unembeddable_font_returns_error_instead_of_a_pdf_missing_text() {
#[cfg_attr(not(target_arch = "wasm32"), test)]
fn supported_unicode_is_shaped_and_recovered_from_the_page_text_layer() {
use deckcraft_model::{Shape, ShapeId, Xfrm};
use hayro_syntax::content::ops::TypedInstruction;
use hayro_syntax::object::{Dict, Object, Stream};
let text = "Café Ελληνικά Привет";
let face = FontDb::global().face("Ubuntu", "Regular");
@@ -223,8 +221,17 @@ fn supported_unicode_is_shaped_and_recovered_from_the_page_text_layer() {
assert!(runs.iter().all(|run| run.face.family == "Ubuntu" && !run.glyphs.is_empty() && run.glyphs.iter().all(|glyph| glyph.0 != 0)));
let bytes = export(&p, &opts(PageLayout::Slides)).unwrap();
let pdf = hayro_syntax::Pdf::new(bytes.clone()).unwrap();
let page = &pdf.pages()[0];
assert_eq!(page_text(&bytes, 0), text);
}
/// The text of page `index`, read the way a PDF reader does: the content stream's text-showing
/// operators decoded through the page font's ToUnicode map.
fn page_text(bytes: &[u8], index: usize) -> String {
use hayro_syntax::content::ops::TypedInstruction;
use hayro_syntax::object::{Dict, Object, Stream};
let pdf = hayro_syntax::Pdf::new(bytes.to_vec()).unwrap();
let page = &pdf.pages()[index];
let resources = page.raw().get::<Dict<'_>>(b"Resources").unwrap();
let fonts = resources.get::<Dict<'_>>(b"Font").unwrap();
assert_eq!(fonts.len(), 1);
@@ -275,7 +282,51 @@ fn supported_unicode_is_shaped_and_recovered_from_the_page_text_layer() {
_ => {}
}
}
assert_eq!(extracted, text);
extracted
}
fn ubuntu_runs(runs: Vec<deckcraft_model::text::Run>) -> TextBody {
let paragraph = deckcraft_model::text::Paragraph { runs, ..Default::default() };
let mut body = TextBody { paragraphs: vec![paragraph], ..Default::default() };
for run in &mut body.paragraphs[0].runs {
run.props.font = Some("Ubuntu".into());
run.props.size = Some(20.0);
}
body
}
fn deck_with_body(n: usize, body: TextBody) -> Presentation {
use deckcraft_model::{Shape, ShapeId, Xfrm};
let mut p = deck(n);
for slide in &mut p.slides {
Arc::make_mut(slide).shapes =
vec![Shape { id: ShapeId(9001), xfrm: Some(Xfrm::new(20.0, 20.0, 920.0, 100.0)), text: Some(body.clone()), ..Default::default() }];
}
p
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[cfg_attr(not(target_arch = "wasm32"), test)]
fn slide_number_field_is_recovered_from_the_page_text_layer() {
use deckcraft_model::text::{Run, RunKind};
let mut field = Run::new("<#>");
field.kind = RunKind::Field { field: "slidenum".into() };
let p = deck_with_body(3, ubuntu_runs(vec![Run::new("Page "), field]));
let bytes = export(&p, &opts(PageLayout::Slides)).unwrap();
for (i, want) in ["Page 1", "Page 2", "Page 3"].into_iter().enumerate() {
assert_eq!(page_text(&bytes, i), want);
}
}
#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
#[cfg_attr(not(target_arch = "wasm32"), test)]
fn line_break_keeps_the_text_layer_aligned_with_the_layout() {
use deckcraft_model::text::{Run, RunKind};
let mut brk = Run::new("");
brk.kind = RunKind::Break;
let p = deck_with_body(1, ubuntu_runs(vec![Run::new("one"), brk, Run::new("two")]));
let bytes = export(&p, &opts(PageLayout::Slides)).unwrap();
assert_eq!(page_text(&bytes, 0), "onetwo");
}
/// Notes pages and handouts print on the chosen paper, landscape included, with every slide box
+1 -1
View File
@@ -53,7 +53,7 @@ pub use extent::{Region, TILE, layout_extent, pasteboard_region, slide_extent};
pub use images::{decode as decode_image, gif_frame};
pub use morph_path::PathMorph;
pub use morph_text::TextMorph;
pub use placed::{Placed, PlacedLink, PlacedShape, PlacedText, place_master_texts, place_slide, text_outline};
pub use placed::{Placed, PlacedLink, PlacedShape, PlacedText, place_master_texts, place_slide, place_slide_with, text_outline};
pub use table::cell_body as table_cell_body;
pub use table::cell_rects as table_cell_rects;
pub use table::row_heights as table_row_heights;
+18 -12
View File
@@ -52,10 +52,16 @@ pub struct Placed {
/// Text and links of slide `index` (master and layout graphics first, as rendered).
pub fn place_slide(pres: &Presentation, index: usize) -> Placed {
place_slide_with(pres, index, &crate::RenderOpts::default())
}
/// [`place_slide`] with the render options the slide was drawn with, so date fields take the same
/// date and time as the raster (`opts.now`).
pub fn place_slide_with(pres: &Presentation, index: usize, opts: &crate::RenderOpts) -> Placed {
let mut out = Placed::default();
let Some(slide) = pres.slides.get(index) else { return out };
let Some(rctx) = Ctx::for_slide(pres, slide) else { return out };
let fields = SlideFields::new(pres.slide_number(index), pres.header_footer.footer_text.clone(), &crate::RenderOpts::default());
let fields = SlideFields::new(pres.slide_number(index), pres.header_footer.footer_text.clone(), opts);
let (show_layout, show_master) = resolve::show_master_shapes(slide, rctx.layout);
if show_master {
let mctx = Ctx::for_master(pres, rctx.master);
@@ -164,20 +170,20 @@ fn place_text(out: &mut Placed, rctx: &Ctx, fields: &SlideFields, owner: ShapeId
}
// Field runs carry their saved text (`‹#›`); the layout drew their values, so the body that
// goes with it (character offsets, text layer) must hold the same values.
let mut body = body.clone();
for para in &mut body.paragraphs {
let src = para.clone();
for run in &mut para.runs {
if let RunKind::Field { field } = &run.kind {
let lang = resolve::run(rctx, s, &src, &run.props).lang;
if let Some(v) = fields.field_in(field, lang.as_deref()) {
run.text = v;
run.kind = RunKind::Text;
}
let mut placed = body.clone();
for (pi, para) in body.paragraphs.iter().enumerate() {
for (ri, run) in para.runs.iter().enumerate() {
let RunKind::Field { field } = &run.kind else { continue };
let lang = resolve::run(rctx, s, para, &run.props).lang;
if let Some(v) = fields.field_in(field, lang.as_deref())
&& let Some(out_run) = placed.paragraphs.get_mut(pi).and_then(|p| p.runs.get_mut(ri))
{
out_run.text = v;
out_run.kind = RunKind::Text;
}
}
}
out.texts.push(PlacedText { layout, transform: m, body });
out.texts.push(PlacedText { layout, transform: m, body: placed });
}
/// The hyperlink of the model run containing character `at` of paragraph `para`.
+9
View File
@@ -2607,3 +2607,12 @@ fn placed_text_holds_the_field_value_not_the_saved_placeholder() {
// The layout was drawn from the same value, so the offsets agree with the body.
assert_eq!(placed.texts[0].layout.runs.iter().map(|r| r.chars.1).max(), Some(1));
}
#[test]
fn placed_text_takes_its_date_from_the_given_render_options() {
use deckcraft_text::datetime::DateTime;
let p = deck_with(vec![field_shape(Some("datetime1"), "1/1/2000")]);
let at = |now: DateTime| place_slide_with(&p, 0, &RenderOpts { now: Some(now), ..Default::default() });
assert_eq!(at(DateTime::new(2026, 10, 9, 20, 5, 7)).texts[0].body.text(), "10/9/2026");
assert_eq!(at(DateTime::new(2027, 3, 4, 0, 0, 0)).texts[0].body.text(), "3/4/2027");
}