Support font encoder

Author: vidy

Cargo.toml

@@ -9,15 +9,20 @@ description = "PDF text extraction"
 
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
+
+[dependencies.pdf]
+default-features=false 
+features = ["cache", "dump"]
+git = "https://github.com/pdf-rs/pdf"
+
 [dependencies]
-pdf = { git = "https://github.com/pdf-rs/pdf", features = ["cache"] }
-pdf_render = { git = "https://github.com/pdf-rs/pdf_render" }
-font = { git = "https://github.com/pdf-rs/font" }
+pdf_render= { git = "https://github.com/videni/pdf_render_with_vello", branch="vello_wip"}
 itertools = "*"
 log = "*"
 ordered-float = "*"
 serde = { version = "*", features = ["derive"] }
 unicode-normalization = "0.1.19"
+font = { git = "https://github.com/pdf-rs/font", branch = "vello", features=['cff']}
 
 pathfinder_geometry = { git = "https://github.com/servo/pathfinder" }
 pathfinder_color = { git = "https://github.com/servo/pathfinder" }

src/lib.rs

@@ -2,18 +2,18 @@ use std::collections::HashSet;
 
 use entry::Flow;
 use pdf::{backend::Backend, object::{Page, Resolve}, PdfError};
-use pdf_render::{tracer::{TraceCache, Tracer, DrawItem}, Fill, render_pattern, render_page, FillMode};
+use pdf_render::{tracer::{TraceCache, Tracer, DrawItem}, Fill, render_pattern, render_page, FillMode, font::OutlineBuilder};
 
 mod tree;
 mod util;
 mod text;
 pub mod entry;
 
 pub fn run<B: Backend>(file: &pdf::file::CachedFile<B>, page: &Page, resolve: &impl Resolve) -> Result<Flow, PdfError> {
-    let cache = TraceCache::new();
+    let mut cache = TraceCache::new(OutlineBuilder::default());
 
     let mut clip_paths = vec![];
-    let mut tracer = Tracer::new(&cache, &mut clip_paths);
+    let mut tracer = Tracer::new(&mut cache, &mut clip_paths);
 
     render_page(&mut tracer, resolve, &page, Default::default())?;
 
@@ -68,7 +68,7 @@ pub fn run<B: Backend>(file: &pdf::file::CachedFile<B>, page: &Page, resolve: &i
                 continue;
             }
         };
-        let mut pat_tracer = Tracer::new(&cache, &mut clip_paths);
+        let mut pat_tracer = Tracer::new(&mut cache, &mut clip_paths);
 
         render_pattern(&mut pat_tracer, &*pattern, resolve)?;
         let pat_items = pat_tracer.finish();

src/text.rs

@@ -1,10 +1,11 @@
+use font::Encoder;
 use pathfinder_geometry::vector::Vector2F;
 use pdf_render::TextSpan;
 use itertools::{Itertools};
 use unicode_normalization::UnicodeNormalization;
 use crate::{util::avg, entry::Word, util::Rect};
 
-pub fn concat_text<'a>(out: &mut String, items: impl Iterator<Item=&'a TextSpan> + Clone) -> Vec<Word> {
+pub fn concat_text<'a, E: Encoder + 'a>(out: &mut String, items: impl Iterator<Item=&'a TextSpan<E>> + Clone) -> Vec<Word> {
     let mut words = vec![];
     let gaps = items.clone()
         .flat_map(|s| {

src/tree.rs

@@ -16,8 +16,9 @@ use crate::util::{is_number, avg, CellContent};
 use crate::text::{concat_text};
 use std::mem::take;
 use table::Table;
+use font::{Encoder, Glyph};
 
-pub fn build(spans: &[TextSpan], bbox: RectF, lines: &[[f32; 4]]) -> Node {
+pub fn build<E: Encoder>(spans: &[TextSpan<E>], bbox: RectF, lines: &[[f32; 4]]) -> Node {
     if spans.len() == 0 {
         return Node::singleton(&[]);
     }
@@ -180,7 +181,7 @@ impl Span {
     }
 }
 
-pub fn split2(boxes: &mut [(RectF, usize)], spans: &[TextSpan], lines_info: &Lines) -> Node {
+pub fn split2<E: Encoder>(boxes: &mut [(RectF, usize)], spans: &[TextSpan<E>], lines_info: &Lines) -> Node {
     use std::mem::replace;
 
     #[derive(Debug)]
@@ -383,7 +384,7 @@ pub enum NodeTag {
     Complex,
 }
 
-pub fn items(mut flow: &mut Flow, spans: &[TextSpan], node: &Node, x_anchor: f32) {
+pub fn items<E: Encoder>(mut flow: &mut Flow, spans: &[TextSpan<E>], node: &Node, x_anchor: f32) {
     match *node {
         Node::Final { ref indices } => {
             if indices.len() > 0 {
@@ -534,10 +535,10 @@ pub fn items(mut flow: &mut Flow, spans: &[TextSpan], node: &Node, x_anchor: f32
 }
 
 
-pub fn render(w: &mut String, spans: &[TextSpan], node: &Node, bbox: RectF) {
+pub fn render<E: Encoder>(w: &mut String, spans: &[TextSpan<E>], node: &Node, bbox: RectF) {
     _render(w, spans, node, bbox, 0)
 }
-fn _render(w: &mut String, spans: &[TextSpan], node: &Node, bbox: RectF, level: usize) {
+fn _render<E: Encoder>(w: &mut String, spans: &[TextSpan<E>], node: &Node, bbox: RectF, level: usize) {
     use std::fmt::Write;
 
     match *node {
@@ -596,7 +597,7 @@ fn _render(w: &mut String, spans: &[TextSpan], node: &Node, bbox: RectF, level:
     }
 }
 
-fn split(boxes: &mut [(RectF, usize)], spans: &[TextSpan], lines: &Lines) -> Node {
+fn split<E: Encoder>(boxes: &mut [(RectF, usize)], spans: &[TextSpan<E>], lines: &Lines) -> Node {
     let num_boxes = boxes.len();
     if num_boxes < 2 {
         return Node::singleton(boxes);
@@ -925,13 +926,13 @@ impl TriCount {
         }
     }
 }
-fn classify<'a>(spans: impl Iterator<Item=&'a TextSpan>) -> Class {
+fn classify<'a, E: Encoder + 'a>(spans: impl Iterator<Item=&'a TextSpan<E>>) -> Class {
     use pdf_render::FontEntry;
 
     let mut bold = TriCount::new();
     let mut numeric = TriCount::new();
     let mut uniform = TriCount::new();
-    let mut first_font: *const FontEntry = std::ptr::null();
+    let mut first_font: *const FontEntry<E> = std::ptr::null();
 
     for s in spans {
         numeric.add(is_number(&s.text));