Merge pull request #1492 from switch87/fix/1484-compact-anchor-heavy-saves

Native save: documents whose JSON is large are written compact

contributors/people.toml

@@ -89,3 +89,7 @@ sync_github_name = false
 
 [people.Nael-Nathanael]
 sync_github_name = true
+
+[people.switch87]
+real_name = "Gert Pellin"
+sync_github_name = false

crates/format/src/lib.rs

@@ -243,13 +243,32 @@ impl SaveOptions {
         Ok(())
     }
 
-    /// The defaults for saving `doc` to disk: pretty when small (see [`PRETTY_MAX_OBJECTS`]).
+    /// The defaults for saving `doc` to disk: pretty when small (see [`PRETTY_MAX_OBJECTS`] and
+    /// [`PRETTY_MAX_BYTES`]).
     pub fn for_doc(doc: &Document) -> Self {
         let objects: usize = doc.layers.iter().map(|l| l.count()).sum();
-        Self { pretty: objects <= PRETTY_MAX_OBJECTS, ..Self::default() }
+        Self { pretty: objects <= PRETTY_MAX_OBJECTS && compact_size_at_most(doc, PRETTY_MAX_BYTES), ..Self::default() }
     }
 }
 
+/// Is `doc` at most `limit` bytes as compact JSON? Serializing stops as soon as it is past the
+/// limit, so a large document costs no more than `limit` bytes of work.
+fn compact_size_at_most(doc: &Document, limit: usize) -> bool {
+    struct Capped {
+        left: usize,
+    }
+    impl Write for Capped {
+        fn write(&mut self, buf: &[u8]) -> std::io::Result<usize> {
+            self.left = self.left.checked_sub(buf.len()).ok_or_else(|| std::io::Error::other("over the limit"))?;
+            Ok(buf.len())
+        }
+        fn flush(&mut self) -> std::io::Result<()> {
+            Ok(())
+        }
+    }
+    serde_json::to_writer(Capped { left: limit }, doc).is_ok()
+}
+
 /// Serialize a document (pretty = human-diffable) in the current version. Editing-mode working
 /// copies are left out ([`Document::without_edit_modes`]).
 pub fn save(doc: &Document, pretty: bool) -> Vec<u8> {
@@ -260,6 +279,11 @@ pub fn save(doc: &Document, pretty: bool) -> Vec<u8> {
 /// compact, where indentation would multiply the file size and slow opening down.
 pub const PRETTY_MAX_OBJECTS: usize = 2000;
 
+/// Documents whose compact JSON is larger than this are saved compact even with few objects:
+/// pretty-printing spreads every path anchor over several lines, so traced or generated art (in
+/// layers, symbols or patterns alike) would grow about tenfold.
+pub const PRETTY_MAX_BYTES: usize = 1 << 20;
+
 /// Serialize for saving to disk: pretty when small, compact when large.
 pub fn save_file(doc: &Document) -> Vec<u8> {
     save_with(doc, &SaveOptions::for_doc(doc)).unwrap_or_default()

crates/format/tests/save_options.rs

@@ -8,9 +8,11 @@ use std::sync::Arc;
 
 use serde_json::Value;
 use vectorcraft_doc::links::LinkInfo;
-use vectorcraft_doc::{Document, ImageBlob, ImageObject, Node, NodeKind};
+use vectorcraft_doc::{Document, ImageBlob, ImageObject, Node, NodeKind, Symbol};
-use vectorcraft_format::{ICC_MIME, SaveOptions, load, load_file, pdf_content, save_with};
+use vectorcraft_format::{ICC_MIME, PRETTY_MAX_BYTES, PRETTY_MAX_OBJECTS, SaveOptions, load, load_file, pdf_content, save_file, save_with};
-use vectorcraft_geom::Affine;
+use vectorcraft_geom::path::{PathData, SubPath};
+use vectorcraft_geom::{Affine, Point};
+use vectorcraft_testkit::fixtures::DocBuilder;
 
 fn json_of(bytes: &[u8]) -> Value {
     serde_json::from_slice(bytes).unwrap()
@@ -82,3 +84,66 @@ fn embedded_profiles_and_pdf_round_trip() {
     let damaged = String::from_utf8(plain).unwrap().replacen("AAECA/o=", "not base64!", 1);
     assert!(load_file(damaged.as_bytes()).unwrap().profiles.is_empty());
 }
+
+/// A document of `paths` open paths with `anchors` anchors each.
+fn paths_doc(paths: usize, anchors: usize) -> Document {
+    let mut b = DocBuilder::new(1000.0, 1000.0);
+    for i in 0..paths {
+        let points: Vec<Point> = (0..anchors).map(|j| Point::new(j as f64, i as f64 + (j % 2) as f64)).collect();
+        let node = b.path_node(PathData::single(SubPath::polyline(&points, false)), Default::default());
+        b.add(node);
+    }
+    b.doc
+}
+
+fn is_pretty(bytes: &[u8]) -> bool {
+    bytes.windows(3).any(|w| w == b"\n  ")
+}
+
+#[test]
+fn small_documents_are_saved_pretty() {
+    let d = paths_doc(10, 10);
+    assert!(SaveOptions::for_doc(&d).pretty);
+    assert!(is_pretty(&save_file(&d)));
+}
+
+#[test]
+fn many_objects_are_saved_compact() {
+    let d = paths_doc(PRETTY_MAX_OBJECTS + 1, 2);
+    assert!(!SaveOptions::for_doc(&d).pretty);
+}
+
+/// Anchors per path that make a document of `paths` paths about `bytes` of compact JSON.
+fn anchors_for(paths: usize, bytes: usize) -> usize {
+    let per_anchor = serde_json::to_vec(&paths_doc(1, 1000)).unwrap().len() / 1000;
+    bytes / per_anchor / paths
+}
+
+#[test]
+fn few_objects_with_many_anchors_are_saved_compact() {
+    let d = paths_doc(3, anchors_for(3, 2 * PRETTY_MAX_BYTES));
+    assert!(!SaveOptions::for_doc(&d).pretty, "3 objects, but about {} bytes", 2 * PRETTY_MAX_BYTES);
+    let bytes = save_file(&d);
+    assert!(!is_pretty(&bytes));
+    assert_eq!(load(&bytes).unwrap().layers[0].count(), d.layers[0].count());
+}
+
+#[test]
+fn documents_under_the_size_limit_stay_pretty() {
+    let d = paths_doc(2, anchors_for(2, PRETTY_MAX_BYTES / 2));
+    assert!(SaveOptions::for_doc(&d).pretty);
+}
+
+#[test]
+fn art_outside_the_layers_counts_too() {
+    // Generated art kept in a symbol: one instance-sized document, a symbol full of anchors.
+    let mut d = paths_doc(1, 4);
+    let mut art = paths_doc(1, anchors_for(1, 2 * PRETTY_MAX_BYTES));
+    let path = Arc::clone(&art.layers[0]);
+    art.layers.clear();
+    d.symbols.push(Symbol { name: "Leaves".into(), art: path });
+    assert!(!SaveOptions::for_doc(&d).pretty);
+    let bytes = save_file(&d);
+    assert!(!is_pretty(&bytes));
+    assert_eq!(load(&bytes).unwrap().symbols.len(), 1);
+}