From 4640585fbdf72df993dbed46799844aa78996cce Mon Sep 17 00:00:00 2001 From: Martin Haug Date: Sat, 4 Jun 2022 12:57:45 +0200 Subject: First iteration of outline items --- src/export/pdf.rs | 190 +++++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 189 insertions(+), 1 deletion(-) (limited to 'src/export') diff --git a/src/export/pdf.rs b/src/export/pdf.rs index aa7acd41..843e6f37 100644 --- a/src/export/pdf.rs +++ b/src/export/pdf.rs @@ -16,7 +16,7 @@ use ttf_parser::{name_id, GlyphId, Tag}; use super::subset::subset; use crate::font::{find_name, FaceId, FontStore}; -use crate::frame::{Destination, Element, Frame, Group, Text}; +use crate::frame::{Destination, Element, Frame, Group, Role, Text}; use crate::geom::{ self, Color, Dir, Em, Geometry, Length, Numeric, Paint, Point, Ratio, Shape, Size, Stroke, Transform, @@ -313,6 +313,8 @@ impl<'a> PdfExporter<'a> { } let mut languages = HashMap::new(); + let mut heading_tree: Vec = vec![]; + for (page, page_id) in self.pages.into_iter().zip(page_refs.iter()) { let content_id = self.alloc.bump(); @@ -356,6 +358,20 @@ impl<'a> PdfExporter<'a> { .or_insert_with(|| count); } + for heading in page.headings.into_iter() { + if let Some(last) = heading_tree.pop() { + let new = last.clone().insert(heading.clone(), *page_id, 1); + if let Some(new) = new { + heading_tree.push(new); + } else { + heading_tree.push(last); + heading_tree.push(HeadingNode::Leaf(heading, *page_id)) + } + } else { + heading_tree.push(HeadingNode::Leaf(heading, *page_id)) + } + } + self.writer .stream(content_id, &deflate(&page.content.finish())) .filter(Filter::FlateDecode); @@ -388,6 +404,39 @@ impl<'a> PdfExporter<'a> { resources.finish(); pages.finish(); + // Build the heading tree. + let outline_root_id = self.alloc.bump(); + + let start_ref = self.alloc.bump(); + let mut current_ref = start_ref; + let mut prev_ref = None; + + for (i, node) in heading_tree.iter().enumerate() { + let next = write_outline_item( + &mut self.writer, + node, + current_ref, + prev_ref, + i == heading_tree.len() - 1, + outline_root_id, + ); + prev_ref = Some(current_ref); + current_ref = next; + } + + + self.alloc = Ref::new( + start_ref.get() + + heading_tree.iter().map(HeadingNode::len).sum::() as i32, + ); + + if let Some(prev_ref) = prev_ref { + let mut outline_root = self.writer.outline(outline_root_id); + outline_root.first(start_ref); + outline_root.last(prev_ref); + outline_root.count(heading_tree.len() as i32); + } + let lang = languages .into_iter() .max_by(|(_, v1), (_, v2)| v1.cmp(v2)) @@ -405,6 +454,11 @@ impl<'a> PdfExporter<'a> { catalog.pages(page_tree_ref); catalog.viewer_preferences().direction(dir); + + if !heading_tree.is_empty() { + catalog.outlines(outline_root_id); + } + if let Some(lang) = lang { catalog.lang(TextStr(lang.as_str())); } @@ -426,6 +480,7 @@ struct PageExporter<'a> { links: Vec<(Destination, Rect)>, state: State, saves: Vec, + headings: Vec, } /// Data for an exported page. @@ -434,6 +489,7 @@ struct Page { content: Content, links: Vec<(Destination, Rect)>, languages: HashMap, + headings: Vec, } /// A simulated graphics state used to deduplicate graphics state changes and @@ -448,6 +504,64 @@ struct State { stroke_space: Option>, } +/// A heading that can later be linked in the outline panel. +#[derive(Debug, Clone)] +struct Heading { + content: String, + level: usize, + position: Point, +} + +#[derive(Debug, Clone)] +enum HeadingNode { + Leaf(Heading, Ref), + Branch(Heading, Ref, Vec), +} + +impl HeadingNode { + fn heading(&self) -> &Heading { + match self { + HeadingNode::Leaf(h, _) => h, + HeadingNode::Branch(h, _, _) => h, + } + } + + fn reference(&self) -> Ref { + match self { + HeadingNode::Leaf(_, r) => *r, + HeadingNode::Branch(_, r, _) => *r, + } + } + + fn len(&self) -> usize { + match self { + HeadingNode::Leaf(_, _) => 1, + HeadingNode::Branch(_, _, children) => { + 1 + children.iter().map(|c| c.len()).sum::() + } + } + } + + fn insert(self, other: Heading, page: Ref, level: usize) -> Option { + if level >= other.level { + return None; + } + + let mut node = match self { + HeadingNode::Leaf(h, r) => (h, r, vec![]), + HeadingNode::Branch(h, r, v) if level + 1 == other.level => (h, r, v), + HeadingNode::Branch(h, r, mut v) => { + let new = v.pop().unwrap().insert(other, page, level + 1).unwrap(); + v.push(new); + return Some(HeadingNode::Branch(h, r, v)); + } + }; + + node.2.push(HeadingNode::Leaf(other, page)); + Some(HeadingNode::Branch(node.0, node.1, node.2)) + } +} + impl<'a> PageExporter<'a> { fn new(exporter: &'a mut PdfExporter) -> Self { Self { @@ -461,6 +575,7 @@ impl<'a> PageExporter<'a> { links: vec![], state: State::default(), saves: vec![], + headings: vec![], } } @@ -481,10 +596,22 @@ impl<'a> PageExporter<'a> { content: self.content, links: self.links, languages: self.languages, + headings: self.headings, } } fn write_frame(&mut self, frame: &Frame) { + if let Some(Role::Heading(level)) = frame.role() { + self.headings.push(Heading { + position: Point::new( + self.state.transform.tx, + self.state.transform.ty + Length::pt(3.0), + ), + content: frame.inner_text().to_string(), + level, + }) + } + for &(pos, ref element) in &frame.elements { let x = pos.x.to_f32(); let y = pos.y.to_f32(); @@ -815,6 +942,67 @@ fn encode_image(img: &RasterImage) -> ImageResult<(Vec, Filter, bool)> { }) } +fn write_outline_item( + writer: &mut PdfWriter, + node: &HeadingNode, + current_ref: Ref, + prev_ref: Option, + is_last: bool, + parent_ref: Ref, +) -> Ref { + let mut outline = writer.outline_item(current_ref); + let next = Ref::new(current_ref.get() + node.len() as i32); + outline.parent(parent_ref); + + if !is_last { + outline.next(next); + } + + if let Some(prev_ref) = prev_ref { + outline.prev(prev_ref); + } + + if let HeadingNode::Branch(_, _, children) = node { + let current_child = Ref::new(current_ref.get() + 1); + if children.len() > 0 { + outline.first(current_child); + outline.last(Ref::new(next.get() - 1)); + } + + outline.count(-1 * children.len() as i32); + } + + let heading = node.heading(); + outline.title(TextStr(&heading.content)); + outline.dest_direct().page(node.reference()).xyz( + heading.position.x.to_f32(), + heading.position.y.to_f32(), + None, + ); + + outline.finish(); + + if let HeadingNode::Branch(_, _, children) = node { + let mut current_child = Ref::new(current_ref.get() + 1); + let mut prev_ref = None; + + for (i, child) in children.iter().enumerate() { + write_outline_item( + writer, + child, + current_child, + prev_ref, + i == children.len() - 1, + current_ref, + ); + prev_ref = Some(current_child); + current_child = Ref::new(current_child.get() + 1); + } + } + + next +} + /// Encode an image's alpha channel if present. fn encode_alpha(img: &RasterImage) -> (Vec, Filter) { let pixels: Vec<_> = img.buf.pixels().map(|(_, _, Rgba([_, _, _, a]))| a).collect(); -- cgit v1.2.3 From 55dce19f4925d6505a6382f4584a2532ce67bed2 Mon Sep 17 00:00:00 2001 From: Martin Haug Date: Wed, 8 Jun 2022 16:23:32 +0200 Subject: Restructure PDF structure writing --- src/export/pdf.rs | 345 ++++++++++++++++++++++++++---------------------------- 1 file changed, 165 insertions(+), 180 deletions(-) (limited to 'src/export') diff --git a/src/export/pdf.rs b/src/export/pdf.rs index 843e6f37..b5768755 100644 --- a/src/export/pdf.rs +++ b/src/export/pdf.rs @@ -22,6 +22,7 @@ use crate::geom::{ Stroke, Transform, }; use crate::image::{Image, ImageId, ImageStore, RasterImage}; +use crate::library::prelude::EcoString; use crate::library::text::Lang; use crate::Context; @@ -74,7 +75,8 @@ impl<'a> PdfExporter<'a> { self.build_pages(frames); self.write_fonts(); self.write_images(); - self.write_structure() + self.write_structure(); + self.writer.finish() } fn build_pages(&mut self, frames: &[Arc]) { @@ -299,7 +301,7 @@ impl<'a> PdfExporter<'a> { } } - fn write_structure(mut self) -> Vec { + fn write_structure(&mut self) { // The root page tree. let page_tree_ref = self.alloc.bump(); @@ -315,70 +317,94 @@ impl<'a> PdfExporter<'a> { let mut languages = HashMap::new(); let mut heading_tree: Vec = vec![]; - for (page, page_id) in self.pages.into_iter().zip(page_refs.iter()) { - let content_id = self.alloc.bump(); - - let mut page_writer = self.writer.page(*page_id); - page_writer.parent(page_tree_ref); - - let w = page.size.x.to_f32(); - let h = page.size.y.to_f32(); - page_writer.media_box(Rect::new(0.0, 0.0, w, h)); - page_writer.contents(content_id); - - let mut annotations = page_writer.annotations(); - for (dest, rect) in page.links { - let mut link = annotations.push(); - link.subtype(AnnotationType::Link).rect(rect); - match dest { - Destination::Url(uri) => { - link.action() - .action_type(ActionType::Uri) - .uri(Str(uri.as_str().as_bytes())); - } - Destination::Internal(loc) => { - let index = loc.page - 1; - let height = page_heights[index]; - link.action() - .action_type(ActionType::GoTo) - .destination_direct() - .page(page_refs[index]) - .xyz(loc.pos.x.to_f32(), height - loc.pos.y.to_f32(), None); - } + for (page, page_id) in + std::mem::take(&mut self.pages).into_iter().zip(page_refs.iter()) + { + self.write_page( + page, + *page_id, + &page_refs, + page_tree_ref, + &mut languages, + &mut heading_tree, + &mut page_heights, + ); + } + + self.write_page_tree(&page_refs, page_tree_ref); + self.write_catalog(page_tree_ref, &languages, &heading_tree); + } + + fn write_page( + &mut self, + page: Page, + page_id: Ref, + page_refs: &[Ref], + page_tree_ref: Ref, + languages: &mut HashMap, + heading_tree: &mut Vec, + page_heights: &mut Vec, + ) { + let content_id = self.alloc.bump(); + + let mut page_writer = self.writer.page(page_id); + page_writer.parent(page_tree_ref); + + let w = page.size.x.to_f32(); + let h = page.size.y.to_f32(); + page_writer.media_box(Rect::new(0.0, 0.0, w, h)); + page_writer.contents(content_id); + + let mut annotations = page_writer.annotations(); + for (dest, rect) in page.links { + let mut link = annotations.push(); + link.subtype(AnnotationType::Link).rect(rect); + match dest { + Destination::Url(uri) => { + link.action() + .action_type(ActionType::Uri) + .uri(Str(uri.as_str().as_bytes())); + } + Destination::Internal(loc) => { + let index = loc.page - 1; + let height = page_heights[index]; + link.action() + .action_type(ActionType::GoTo) + .destination_direct() + .page(page_refs[index]) + .xyz(loc.pos.x.to_f32(), height - loc.pos.y.to_f32(), None); } } + } - annotations.finish(); - page_writer.finish(); + annotations.finish(); + page_writer.finish(); - for (lang, count) in page.languages { - languages - .entry(lang) - .and_modify(|x| *x += count) - .or_insert_with(|| count); - } + for (lang, count) in page.languages { + languages + .entry(lang) + .and_modify(|x| *x += count) + .or_insert_with(|| count); + } - for heading in page.headings.into_iter() { - if let Some(last) = heading_tree.pop() { - let new = last.clone().insert(heading.clone(), *page_id, 1); - if let Some(new) = new { - heading_tree.push(new); - } else { - heading_tree.push(last); - heading_tree.push(HeadingNode::Leaf(heading, *page_id)) - } - } else { - heading_tree.push(HeadingNode::Leaf(heading, *page_id)) + for heading in page.headings.into_iter() { + if let Some(last) = heading_tree.last_mut() { + if !last.insert(heading.clone(), page_id, 1) { + heading_tree.push(HeadingNode::leaf(heading, page_id)) } + } else { + heading_tree.push(HeadingNode::leaf(heading, page_id)) } - - self.writer - .stream(content_id, &deflate(&page.content.finish())) - .filter(Filter::FlateDecode); } + self.writer + .stream(content_id, &deflate(&page.content.finish())) + .filter(Filter::FlateDecode); + } + + fn write_page_tree(&mut self, page_refs: &[Ref], page_tree_ref: Ref) { let mut pages = self.writer.pages(page_tree_ref); - pages.count(page_refs.len() as i32).kids(page_refs); + pages.count(page_refs.len() as i32).kids(page_refs.iter().copied()); let mut resources = pages.resources(); let mut spaces = resources.color_spaces(); @@ -403,37 +429,32 @@ impl<'a> PdfExporter<'a> { images.finish(); resources.finish(); pages.finish(); + } - // Build the heading tree. + fn write_catalog( + &mut self, + page_tree_ref: Ref, + languages: &HashMap, + heading_tree: &Vec, + ) { + // Build the outline tree. let outline_root_id = self.alloc.bump(); - let start_ref = self.alloc.bump(); - let mut current_ref = start_ref; - let mut prev_ref = None; + let outline_start_ref = self.alloc; for (i, node) in heading_tree.iter().enumerate() { - let next = write_outline_item( - &mut self.writer, + self.write_outline_item( node, - current_ref, - prev_ref, - i == heading_tree.len() - 1, + i == 0, + i + 1 == heading_tree.len(), outline_root_id, ); - prev_ref = Some(current_ref); - current_ref = next; } - - self.alloc = Ref::new( - start_ref.get() - + heading_tree.iter().map(HeadingNode::len).sum::() as i32, - ); - - if let Some(prev_ref) = prev_ref { + if !heading_tree.is_empty() { let mut outline_root = self.writer.outline(outline_root_id); - outline_root.first(start_ref); - outline_root.last(prev_ref); + outline_root.first(outline_start_ref); + outline_root.last(Ref::new(self.alloc.get() - 1)); outline_root.count(heading_tree.len() as i32); } @@ -442,7 +463,7 @@ impl<'a> PdfExporter<'a> { .max_by(|(_, v1), (_, v2)| v1.cmp(v2)) .map(|(k, _)| k); - let dir = if lang.map(Lang::dir) == Some(Dir::RTL) { + let dir = if lang.copied().map(Lang::dir) == Some(Dir::RTL) { Direction::R2L } else { Direction::L2R @@ -454,7 +475,6 @@ impl<'a> PdfExporter<'a> { catalog.pages(page_tree_ref); catalog.viewer_preferences().direction(dir); - if !heading_tree.is_empty() { catalog.outlines(outline_root_id); } @@ -464,7 +484,51 @@ impl<'a> PdfExporter<'a> { } catalog.finish(); - self.writer.finish() + } + + fn write_outline_item( + &mut self, + node: &HeadingNode, + is_first: bool, + is_last: bool, + parent_ref: Ref, + ) { + let id = self.alloc.bump(); + let next = Ref::new(id.get() + node.len() as i32); + + let mut outline = self.writer.outline_item(id); + outline.parent(parent_ref); + + if !is_last { + outline.next(next); + } + + if !is_first { + outline.prev(Ref::new(id.get() - 1)); + } + + if !node.children.is_empty() { + let current_child = Ref::new(id.get() + 1); + outline.first(current_child); + outline.last(Ref::new(next.get() - 1)); + + outline.count(-1 * node.children.len() as i32); + } + + outline.title(TextStr(&node.heading.content)); + outline.dest_direct().page(node.page).xyz( + node.heading.position.x.to_f32(), + (node.heading.position.y + Length::pt(3.0)).to_f32(), + None, + ); + + outline.finish(); + + if !node.children.is_empty() { + for (i, child) in node.children.iter().enumerate() { + self.write_outline_item(child, i == 0, i + 1 == node.children.len(), id); + } + } } } @@ -507,58 +571,43 @@ struct State { /// A heading that can later be linked in the outline panel. #[derive(Debug, Clone)] struct Heading { - content: String, + content: EcoString, level: usize, position: Point, } #[derive(Debug, Clone)] -enum HeadingNode { - Leaf(Heading, Ref), - Branch(Heading, Ref, Vec), +struct HeadingNode { + heading: Heading, + page: Ref, + children: Vec, } impl HeadingNode { - fn heading(&self) -> &Heading { - match self { - HeadingNode::Leaf(h, _) => h, - HeadingNode::Branch(h, _, _) => h, - } - } - - fn reference(&self) -> Ref { - match self { - HeadingNode::Leaf(_, r) => *r, - HeadingNode::Branch(_, r, _) => *r, - } + fn leaf(heading: Heading, page: Ref) -> Self { + HeadingNode { heading, page, children: Vec::new() } } fn len(&self) -> usize { - match self { - HeadingNode::Leaf(_, _) => 1, - HeadingNode::Branch(_, _, children) => { - 1 + children.iter().map(|c| c.len()).sum::() - } - } + 1 + self.children.iter().map(|c| c.len()).sum::() } - fn insert(self, other: Heading, page: Ref, level: usize) -> Option { + fn insert(&mut self, other: Heading, page: Ref, level: usize) -> bool { if level >= other.level { - return None; + return false; } - let mut node = match self { - HeadingNode::Leaf(h, r) => (h, r, vec![]), - HeadingNode::Branch(h, r, v) if level + 1 == other.level => (h, r, v), - HeadingNode::Branch(h, r, mut v) => { - let new = v.pop().unwrap().insert(other, page, level + 1).unwrap(); - v.push(new); - return Some(HeadingNode::Branch(h, r, v)); - } - }; + if !self.children.is_empty() && level + 1 > other.level { + return self.children.last_mut().unwrap().insert(other, page, level + 1); + } + + self.children.push(HeadingNode { + heading: other, + page, + children: Vec::new(), + }); - node.2.push(HeadingNode::Leaf(other, page)); - Some(HeadingNode::Branch(node.0, node.1, node.2)) + true } } @@ -603,11 +652,8 @@ impl<'a> PageExporter<'a> { fn write_frame(&mut self, frame: &Frame) { if let Some(Role::Heading(level)) = frame.role() { self.headings.push(Heading { - position: Point::new( - self.state.transform.tx, - self.state.transform.ty + Length::pt(3.0), - ), - content: frame.inner_text().to_string(), + position: Point::new(self.state.transform.tx, self.state.transform.ty), + content: frame.inner_text(), level, }) } @@ -942,67 +988,6 @@ fn encode_image(img: &RasterImage) -> ImageResult<(Vec, Filter, bool)> { }) } -fn write_outline_item( - writer: &mut PdfWriter, - node: &HeadingNode, - current_ref: Ref, - prev_ref: Option, - is_last: bool, - parent_ref: Ref, -) -> Ref { - let mut outline = writer.outline_item(current_ref); - let next = Ref::new(current_ref.get() + node.len() as i32); - outline.parent(parent_ref); - - if !is_last { - outline.next(next); - } - - if let Some(prev_ref) = prev_ref { - outline.prev(prev_ref); - } - - if let HeadingNode::Branch(_, _, children) = node { - let current_child = Ref::new(current_ref.get() + 1); - if children.len() > 0 { - outline.first(current_child); - outline.last(Ref::new(next.get() - 1)); - } - - outline.count(-1 * children.len() as i32); - } - - let heading = node.heading(); - outline.title(TextStr(&heading.content)); - outline.dest_direct().page(node.reference()).xyz( - heading.position.x.to_f32(), - heading.position.y.to_f32(), - None, - ); - - outline.finish(); - - if let HeadingNode::Branch(_, _, children) = node { - let mut current_child = Ref::new(current_ref.get() + 1); - let mut prev_ref = None; - - for (i, child) in children.iter().enumerate() { - write_outline_item( - writer, - child, - current_child, - prev_ref, - i == children.len() - 1, - current_ref, - ); - prev_ref = Some(current_child); - current_child = Ref::new(current_child.get() + 1); - } - } - - next -} - /// Encode an image's alpha channel if present. fn encode_alpha(img: &RasterImage) -> (Vec, Filter) { let pixels: Vec<_> = img.buf.pixels().map(|(_, _, Rgba([_, _, _, a]))| a).collect(); -- cgit v1.2.3 From 9dca4c2f7833055edd3c1682e98dcc3f86b7e31b Mon Sep 17 00:00:00 2001 From: Martin Haug Date: Wed, 8 Jun 2022 17:05:07 +0200 Subject: Preallocate Page Refs --- src/export/pdf.rs | 108 +++++++++++++++++++++++++----------------------------- 1 file changed, 49 insertions(+), 59 deletions(-) (limited to 'src/export') diff --git a/src/export/pdf.rs b/src/export/pdf.rs index b5768755..f8144ad9 100644 --- a/src/export/pdf.rs +++ b/src/export/pdf.rs @@ -53,6 +53,8 @@ struct PdfExporter<'a> { glyph_sets: HashMap>, image_map: Remapper, image_refs: Vec, + page_refs: Vec, + heading_tree: Vec, } impl<'a> PdfExporter<'a> { @@ -68,6 +70,8 @@ impl<'a> PdfExporter<'a> { glyph_sets: HashMap::new(), image_map: Remapper::new(), image_refs: vec![], + page_refs: vec![], + heading_tree: vec![], } } @@ -81,7 +85,9 @@ impl<'a> PdfExporter<'a> { fn build_pages(&mut self, frames: &[Arc]) { for frame in frames { - let page = PageExporter::new(self).export(frame); + let page_id = self.alloc.bump(); + self.page_refs.push(page_id); + let page = PageExporter::new(self, page_id).export(frame); self.pages.push(page); } } @@ -306,43 +312,34 @@ impl<'a> PdfExporter<'a> { let page_tree_ref = self.alloc.bump(); // The page objects (non-root nodes in the page tree). - let mut page_refs = vec![]; - let mut page_heights = vec![]; - for page in &self.pages { - let page_id = self.alloc.bump(); - page_refs.push(page_id); - page_heights.push(page.size.y.to_f32()); - } + let mut page_heights = + self.pages.iter().map(|page| page.size.y.to_f32()).collect(); let mut languages = HashMap::new(); - let mut heading_tree: Vec = vec![]; - for (page, page_id) in - std::mem::take(&mut self.pages).into_iter().zip(page_refs.iter()) + for (page, page_id) in std::mem::take(&mut self.pages) + .into_iter() + .zip(self.page_refs.clone().iter()) { self.write_page( page, *page_id, - &page_refs, page_tree_ref, &mut languages, - &mut heading_tree, &mut page_heights, ); } - self.write_page_tree(&page_refs, page_tree_ref); - self.write_catalog(page_tree_ref, &languages, &heading_tree); + self.write_page_tree(page_tree_ref); + self.write_catalog(page_tree_ref, &languages); } fn write_page( &mut self, page: Page, page_id: Ref, - page_refs: &[Ref], page_tree_ref: Ref, languages: &mut HashMap, - heading_tree: &mut Vec, page_heights: &mut Vec, ) { let content_id = self.alloc.bump(); @@ -371,7 +368,7 @@ impl<'a> PdfExporter<'a> { link.action() .action_type(ActionType::GoTo) .destination_direct() - .page(page_refs[index]) + .page(self.page_refs[index]) .xyz(loc.pos.x.to_f32(), height - loc.pos.y.to_f32(), None); } } @@ -387,24 +384,16 @@ impl<'a> PdfExporter<'a> { .or_insert_with(|| count); } - for heading in page.headings.into_iter() { - if let Some(last) = heading_tree.last_mut() { - if !last.insert(heading.clone(), page_id, 1) { - heading_tree.push(HeadingNode::leaf(heading, page_id)) - } - } else { - heading_tree.push(HeadingNode::leaf(heading, page_id)) - } - } - self.writer .stream(content_id, &deflate(&page.content.finish())) .filter(Filter::FlateDecode); } - fn write_page_tree(&mut self, page_refs: &[Ref], page_tree_ref: Ref) { + fn write_page_tree(&mut self, page_tree_ref: Ref) { let mut pages = self.writer.pages(page_tree_ref); - pages.count(page_refs.len() as i32).kids(page_refs.iter().copied()); + pages + .count(self.page_refs.len() as i32) + .kids(self.page_refs.iter().copied()); let mut resources = pages.resources(); let mut spaces = resources.color_spaces(); @@ -431,31 +420,26 @@ impl<'a> PdfExporter<'a> { pages.finish(); } - fn write_catalog( - &mut self, - page_tree_ref: Ref, - languages: &HashMap, - heading_tree: &Vec, - ) { + fn write_catalog(&mut self, page_tree_ref: Ref, languages: &HashMap) { // Build the outline tree. let outline_root_id = self.alloc.bump(); let outline_start_ref = self.alloc; - for (i, node) in heading_tree.iter().enumerate() { + for (i, node) in std::mem::take(&mut self.heading_tree).iter().enumerate() { self.write_outline_item( node, i == 0, - i + 1 == heading_tree.len(), + i + 1 == self.heading_tree.len(), outline_root_id, ); } - if !heading_tree.is_empty() { + if !self.heading_tree.is_empty() { let mut outline_root = self.writer.outline(outline_root_id); outline_root.first(outline_start_ref); outline_root.last(Ref::new(self.alloc.get() - 1)); - outline_root.count(heading_tree.len() as i32); + outline_root.count(self.heading_tree.len() as i32); } let lang = languages @@ -475,7 +459,7 @@ impl<'a> PdfExporter<'a> { catalog.pages(page_tree_ref); catalog.viewer_preferences().direction(dir); - if !heading_tree.is_empty() { + if !self.heading_tree.is_empty() { catalog.outlines(outline_root_id); } @@ -516,7 +500,7 @@ impl<'a> PdfExporter<'a> { } outline.title(TextStr(&node.heading.content)); - outline.dest_direct().page(node.page).xyz( + outline.dest_direct().page(node.heading.page).xyz( node.heading.position.x.to_f32(), (node.heading.position.y + Length::pt(3.0)).to_f32(), None, @@ -538,13 +522,14 @@ struct PageExporter<'a> { font_map: &'a mut Remapper, image_map: &'a mut Remapper, glyphs: &'a mut HashMap>, + heading_tree: &'a mut Vec, + page_ref: Ref, languages: HashMap, bottom: f32, content: Content, links: Vec<(Destination, Rect)>, state: State, saves: Vec, - headings: Vec, } /// Data for an exported page. @@ -553,7 +538,6 @@ struct Page { content: Content, links: Vec<(Destination, Rect)>, languages: HashMap, - headings: Vec, } /// A simulated graphics state used to deduplicate graphics state changes and @@ -574,57 +558,55 @@ struct Heading { content: EcoString, level: usize, position: Point, + page: Ref, } #[derive(Debug, Clone)] struct HeadingNode { heading: Heading, - page: Ref, children: Vec, } impl HeadingNode { - fn leaf(heading: Heading, page: Ref) -> Self { - HeadingNode { heading, page, children: Vec::new() } + fn leaf(heading: Heading) -> Self { + HeadingNode { heading, children: Vec::new() } } fn len(&self) -> usize { 1 + self.children.iter().map(|c| c.len()).sum::() } - fn insert(&mut self, other: Heading, page: Ref, level: usize) -> bool { + fn insert(&mut self, other: Heading, level: usize) -> bool { if level >= other.level { return false; } if !self.children.is_empty() && level + 1 > other.level { - return self.children.last_mut().unwrap().insert(other, page, level + 1); + return self.children.last_mut().unwrap().insert(other, level + 1); } - self.children.push(HeadingNode { - heading: other, - page, - children: Vec::new(), - }); + self.children + .push(HeadingNode { heading: other, children: Vec::new() }); true } } impl<'a> PageExporter<'a> { - fn new(exporter: &'a mut PdfExporter) -> Self { + fn new(exporter: &'a mut PdfExporter, page_ref: Ref) -> Self { Self { fonts: exporter.fonts, font_map: &mut exporter.face_map, image_map: &mut exporter.image_map, glyphs: &mut exporter.glyph_sets, + heading_tree: &mut exporter.heading_tree, + page_ref, languages: HashMap::new(), bottom: 0.0, content: Content::new(), links: vec![], state: State::default(), saves: vec![], - headings: vec![], } } @@ -645,17 +627,25 @@ impl<'a> PageExporter<'a> { content: self.content, links: self.links, languages: self.languages, - headings: self.headings, } } fn write_frame(&mut self, frame: &Frame) { if let Some(Role::Heading(level)) = frame.role() { - self.headings.push(Heading { + let heading = Heading { position: Point::new(self.state.transform.tx, self.state.transform.ty), content: frame.inner_text(), + page: self.page_ref, level, - }) + }; + + if let Some(last) = self.heading_tree.last_mut() { + if !last.insert(heading.clone(), 1) { + self.heading_tree.push(HeadingNode::leaf(heading)) + } + } else { + self.heading_tree.push(HeadingNode::leaf(heading)) + } } for &(pos, ref element) in &frame.elements { -- cgit v1.2.3 From 72d3f3fffabe6872eb7839585bea925b89aac6a4 Mon Sep 17 00:00:00 2001 From: Martin Haug Date: Wed, 8 Jun 2022 18:43:00 +0200 Subject: CR: Whoever said orange is the new pink was seriously disturbed. --- src/export/pdf.rs | 226 ++++++++++++++++++++++++------------------------------ 1 file changed, 100 insertions(+), 126 deletions(-) (limited to 'src/export') diff --git a/src/export/pdf.rs b/src/export/pdf.rs index f8144ad9..b8fc1e39 100644 --- a/src/export/pdf.rs +++ b/src/export/pdf.rs @@ -43,34 +43,42 @@ const SRGB_GRAY: Name<'static> = Name(b"srgbgray"); /// An exporter for a whole PDF document. struct PdfExporter<'a> { + writer: PdfWriter, fonts: &'a FontStore, images: &'a ImageStore, - writer: PdfWriter, - alloc: Ref, pages: Vec, - face_map: Remapper, + page_heights: Vec, + alloc: Ref, + page_tree_ref: Ref, face_refs: Vec, - glyph_sets: HashMap>, - image_map: Remapper, image_refs: Vec, page_refs: Vec, + face_map: Remapper, + image_map: Remapper, + glyph_sets: HashMap>, + languages: HashMap, heading_tree: Vec, } impl<'a> PdfExporter<'a> { fn new(ctx: &'a Context) -> Self { + let mut alloc = Ref::new(1); + let page_tree_ref = alloc.bump(); Self { + writer: PdfWriter::new(), fonts: &ctx.fonts, images: &ctx.images, - writer: PdfWriter::new(), - alloc: Ref::new(1), pages: vec![], - face_map: Remapper::new(), + page_heights: vec![], + alloc, + page_tree_ref, + page_refs: vec![], face_refs: vec![], - glyph_sets: HashMap::new(), - image_map: Remapper::new(), image_refs: vec![], - page_refs: vec![], + face_map: Remapper::new(), + image_map: Remapper::new(), + glyph_sets: HashMap::new(), + languages: HashMap::new(), heading_tree: vec![], } } @@ -79,7 +87,15 @@ impl<'a> PdfExporter<'a> { self.build_pages(frames); self.write_fonts(); self.write_images(); - self.write_structure(); + + // The root page tree. + for page in std::mem::take(&mut self.pages).into_iter() { + self.write_page(page); + } + + self.write_page_tree(); + self.write_catalog(); + self.writer.finish() } @@ -88,6 +104,7 @@ impl<'a> PdfExporter<'a> { let page_id = self.alloc.bump(); self.page_refs.push(page_id); let page = PageExporter::new(self, page_id).export(frame); + self.page_heights.push(page.size.y.to_f32()); self.pages.push(page); } } @@ -307,45 +324,11 @@ impl<'a> PdfExporter<'a> { } } - fn write_structure(&mut self) { - // The root page tree. - let page_tree_ref = self.alloc.bump(); - - // The page objects (non-root nodes in the page tree). - let mut page_heights = - self.pages.iter().map(|page| page.size.y.to_f32()).collect(); - - let mut languages = HashMap::new(); - - for (page, page_id) in std::mem::take(&mut self.pages) - .into_iter() - .zip(self.page_refs.clone().iter()) - { - self.write_page( - page, - *page_id, - page_tree_ref, - &mut languages, - &mut page_heights, - ); - } - - self.write_page_tree(page_tree_ref); - self.write_catalog(page_tree_ref, &languages); - } - - fn write_page( - &mut self, - page: Page, - page_id: Ref, - page_tree_ref: Ref, - languages: &mut HashMap, - page_heights: &mut Vec, - ) { + fn write_page(&mut self, page: Page) { let content_id = self.alloc.bump(); - let mut page_writer = self.writer.page(page_id); - page_writer.parent(page_tree_ref); + let mut page_writer = self.writer.page(page.id); + page_writer.parent(self.page_tree_ref); let w = page.size.x.to_f32(); let h = page.size.y.to_f32(); @@ -364,7 +347,7 @@ impl<'a> PdfExporter<'a> { } Destination::Internal(loc) => { let index = loc.page - 1; - let height = page_heights[index]; + let height = self.page_heights[index]; link.action() .action_type(ActionType::GoTo) .destination_direct() @@ -377,20 +360,13 @@ impl<'a> PdfExporter<'a> { annotations.finish(); page_writer.finish(); - for (lang, count) in page.languages { - languages - .entry(lang) - .and_modify(|x| *x += count) - .or_insert_with(|| count); - } - self.writer .stream(content_id, &deflate(&page.content.finish())) .filter(Filter::FlateDecode); } - fn write_page_tree(&mut self, page_tree_ref: Ref) { - let mut pages = self.writer.pages(page_tree_ref); + fn write_page_tree(&mut self) { + let mut pages = self.writer.pages(self.page_tree_ref); pages .count(self.page_refs.len() as i32) .kids(self.page_refs.iter().copied()); @@ -420,34 +396,36 @@ impl<'a> PdfExporter<'a> { pages.finish(); } - fn write_catalog(&mut self, page_tree_ref: Ref, languages: &HashMap) { + fn write_catalog(&mut self) { // Build the outline tree. - let outline_root_id = self.alloc.bump(); - + let outline_root_id = (!self.heading_tree.is_empty()).then(|| self.alloc.bump()); let outline_start_ref = self.alloc; + let len = self.heading_tree.len(); + let mut prev_ref = None; for (i, node) in std::mem::take(&mut self.heading_tree).iter().enumerate() { - self.write_outline_item( + prev_ref = Some(self.write_outline_item( node, - i == 0, - i + 1 == self.heading_tree.len(), - outline_root_id, - ); + outline_root_id.unwrap(), + prev_ref, + i + 1 == len, + )); } - if !self.heading_tree.is_empty() { + if let Some(outline_root_id) = outline_root_id { let mut outline_root = self.writer.outline(outline_root_id); outline_root.first(outline_start_ref); outline_root.last(Ref::new(self.alloc.get() - 1)); outline_root.count(self.heading_tree.len() as i32); } - let lang = languages - .into_iter() - .max_by(|(_, v1), (_, v2)| v1.cmp(v2)) - .map(|(k, _)| k); + let lang = self + .languages + .iter() + .max_by_key(|(&lang, &count)| (count, lang)) + .map(|(&k, _)| k); - let dir = if lang.copied().map(Lang::dir) == Some(Dir::RTL) { + let dir = if lang.map(Lang::dir) == Some(Dir::RTL) { Direction::R2L } else { Direction::L2R @@ -456,10 +434,10 @@ impl<'a> PdfExporter<'a> { // Write the document information, catalog and wrap it up! self.writer.document_info(self.alloc.bump()).creator(TextStr("Typst")); let mut catalog = self.writer.catalog(self.alloc.bump()); - catalog.pages(page_tree_ref); + catalog.pages(self.page_tree_ref); catalog.viewer_preferences().direction(dir); - if !self.heading_tree.is_empty() { + if let Some(outline_root_id) = outline_root_id { catalog.outlines(outline_root_id); } @@ -473,29 +451,28 @@ impl<'a> PdfExporter<'a> { fn write_outline_item( &mut self, node: &HeadingNode, - is_first: bool, - is_last: bool, parent_ref: Ref, - ) { + prev_ref: Option, + is_last: bool, + ) -> Ref { let id = self.alloc.bump(); - let next = Ref::new(id.get() + node.len() as i32); + let next_ref = Ref::new(id.get() + node.len() as i32); let mut outline = self.writer.outline_item(id); outline.parent(parent_ref); if !is_last { - outline.next(next); + outline.next(next_ref); } - if !is_first { - outline.prev(Ref::new(id.get() - 1)); + if let Some(prev_rev) = prev_ref { + outline.prev(prev_rev); } if !node.children.is_empty() { let current_child = Ref::new(id.get() + 1); outline.first(current_child); - outline.last(Ref::new(next.get() - 1)); - + outline.last(Ref::new(next_ref.get() - 1)); outline.count(-1 * node.children.len() as i32); } @@ -508,36 +485,37 @@ impl<'a> PdfExporter<'a> { outline.finish(); - if !node.children.is_empty() { - for (i, child) in node.children.iter().enumerate() { - self.write_outline_item(child, i == 0, i + 1 == node.children.len(), id); - } + let mut prev_ref = None; + for (i, child) in node.children.iter().enumerate() { + prev_ref = Some(self.write_outline_item( + child, + id, + prev_ref, + i + 1 == node.children.len(), + )); } + + id } } /// An exporter for the contents of a single PDF page. -struct PageExporter<'a> { - fonts: &'a FontStore, - font_map: &'a mut Remapper, - image_map: &'a mut Remapper, - glyphs: &'a mut HashMap>, - heading_tree: &'a mut Vec, +struct PageExporter<'a, 'b> { + exporter: &'a mut PdfExporter<'b>, page_ref: Ref, - languages: HashMap, - bottom: f32, content: Content, - links: Vec<(Destination, Rect)>, state: State, saves: Vec, + bottom: f32, + links: Vec<(Destination, Rect)>, } /// Data for an exported page. struct Page { + id: Ref, size: Size, content: Content, links: Vec<(Destination, Rect)>, - languages: HashMap, } /// A simulated graphics state used to deduplicate graphics state changes and @@ -573,7 +551,7 @@ impl HeadingNode { } fn len(&self) -> usize { - 1 + self.children.iter().map(|c| c.len()).sum::() + 1 + self.children.iter().map(Self::len).sum::() } fn insert(&mut self, other: Heading, level: usize) -> bool { @@ -581,32 +559,27 @@ impl HeadingNode { return false; } - if !self.children.is_empty() && level + 1 > other.level { - return self.children.last_mut().unwrap().insert(other, level + 1); + if let Some(child) = self.children.last_mut() { + if child.insert(other.clone(), level + 1) { + return true; + } } - self.children - .push(HeadingNode { heading: other, children: Vec::new() }); - + self.children.push(Self::leaf(other)); true } } -impl<'a> PageExporter<'a> { - fn new(exporter: &'a mut PdfExporter, page_ref: Ref) -> Self { +impl<'a, 'b> PageExporter<'a, 'b> { + fn new(exporter: &'a mut PdfExporter<'b>, page_ref: Ref) -> Self { Self { - fonts: exporter.fonts, - font_map: &mut exporter.face_map, - image_map: &mut exporter.image_map, - glyphs: &mut exporter.glyph_sets, - heading_tree: &mut exporter.heading_tree, + exporter, page_ref, - languages: HashMap::new(), - bottom: 0.0, content: Content::new(), - links: vec![], state: State::default(), saves: vec![], + bottom: 0.0, + links: vec![], } } @@ -625,8 +598,8 @@ impl<'a> PageExporter<'a> { Page { size: frame.size, content: self.content, + id: self.page_ref, links: self.links, - languages: self.languages, } } @@ -634,17 +607,17 @@ impl<'a> PageExporter<'a> { if let Some(Role::Heading(level)) = frame.role() { let heading = Heading { position: Point::new(self.state.transform.tx, self.state.transform.ty), - content: frame.inner_text(), + content: frame.text(), page: self.page_ref, level, }; - if let Some(last) = self.heading_tree.last_mut() { + if let Some(last) = self.exporter.heading_tree.last_mut() { if !last.insert(heading.clone(), 1) { - self.heading_tree.push(HeadingNode::leaf(heading)) + self.exporter.heading_tree.push(HeadingNode::leaf(heading)) } } else { - self.heading_tree.push(HeadingNode::leaf(heading)) + self.exporter.heading_tree.push(HeadingNode::leaf(heading)) } } @@ -684,13 +657,14 @@ impl<'a> PageExporter<'a> { } fn write_text(&mut self, x: f32, y: f32, text: &Text) { - *self.languages.entry(text.lang).or_insert(0) += text.glyphs.len(); - self.glyphs + *self.exporter.languages.entry(text.lang).or_insert(0) += text.glyphs.len(); + self.exporter + .glyph_sets .entry(text.face_id) .or_default() .extend(text.glyphs.iter().map(|g| g.id)); - let face = self.fonts.get(text.face_id); + let face = self.exporter.fonts.get(text.face_id); self.set_fill(text.fill); self.set_font(text.face_id, text.size); @@ -804,8 +778,8 @@ impl<'a> PageExporter<'a> { } fn write_image(&mut self, x: f32, y: f32, id: ImageId, size: Size) { - self.image_map.insert(id); - let name = format_eco!("Im{}", self.image_map.map(id)); + self.exporter.image_map.insert(id); + let name = format_eco!("Im{}", self.exporter.image_map.map(id)); let w = size.x.to_f32(); let h = size.y.to_f32(); self.content.save_state(); @@ -868,8 +842,8 @@ impl<'a> PageExporter<'a> { fn set_font(&mut self, face_id: FaceId, size: Length) { if self.state.font != Some((face_id, size)) { - self.font_map.insert(face_id); - let name = format_eco!("F{}", self.font_map.map(face_id)); + self.exporter.face_map.insert(face_id); + let name = format_eco!("F{}", self.exporter.face_map.map(face_id)); self.content.set_font(Name(name.as_bytes()), size.to_f32()); self.state.font = Some((face_id, size)); } -- cgit v1.2.3