engine: pass the test run's text by value

crates/engine/src/spell.rs

@@ -700,15 +700,29 @@ pub struct Word {
     pub repeated: Option<usize>,
 }
 
+/// Longest language tag a word is checked in. BCP 47 tags in documents are a few bytes; a longer
+/// value from a file is no language we have a dictionary for, and copying it into every word of
+/// a paragraph would multiply it.
+const MAX_LANG_LEN: usize = 64;
+
 /// The words of `p`, each with its run's language (`lang` on the run, else `default`). Words
 /// with digits are skipped; apostrophes inside a word belong to it; hyphenated words are checked
-/// part by part.
+/// part by part. A language over [`MAX_LANG_LEN`] counts as none.
 pub fn paragraph_words(p: &deckcraft_model::Paragraph, default: &str) -> Vec<Word> {
     let mut chars: Vec<(char, usize)> = vec![];
     for (ri, r) in p.runs.iter().enumerate() {
         chars.extend(r.content().chars().map(|c| (c, ri)));
     }
-    let lang_of = |ri: usize| p.runs.get(ri).and_then(|r| r.props.lang.clone()).filter(|l| !l.is_empty()).unwrap_or_else(|| default.to_string());
+    let default = if default.len() > MAX_LANG_LEN { "" } else { default };
+    let lang_of = |ri: usize| {
+        p.runs
+            .get(ri)
+            .and_then(|r| r.props.lang.as_deref())
+            .filter(|l| !l.is_empty())
+            .map(|l| if l.len() > MAX_LANG_LEN { "" } else { l })
+            .unwrap_or(default)
+            .to_string()
+    };
     // Combining marks (decomposed accents) belong to the letter before them.
     let inside = |c: char| c.is_alphanumeric() || deckcraft_model::words::is_combining_mark(c) || c == '\'' || c == '’' || c == '-';
     let mut out = vec![];
@@ -791,15 +805,18 @@ pub fn misspelt(p: &deckcraft_model::Paragraph, default: &str, ignore: &[String]
         .chain(repeated(p, default))
         .collect::<Vec<_>>()
         .into_iter()
-        .fold(vec![], |mut v: Vec<Word>, w| {
+        .fold((vec![], HashMap::new()), |(mut v, mut at): (Vec<Word>, HashMap<usize, usize>), w| {
             // A repeated misspelt word is reported once, as repeated.
-            if let Some(i) = v.iter().position(|x| x.start == w.start) {
-                v[i] = w;
-            } else {
-                v.push(w);
+            match at.get(&w.start).and_then(|&i| v.get_mut(i)) {
+                Some(x) => *x = w,
+                None => {
+                    at.insert(w.start, v.len());
+                    v.push(w);
+                }
             }
-            v
+            (v, at)
         })
+        .0
 }
 
 /// Use `d` for `lang` from now on in this process (a dictionary the app loaded itself).
@@ -953,6 +970,23 @@ SFX D Y 1\nSFX D 0 ed/S .\n";
         assert_eq!(w, vec!["check", "done"]);
     }
 
+    #[test]
+    fn a_long_paragraph_with_a_huge_language_is_checked_in_linear_work() {
+        use deckcraft_model::{Paragraph, Run, RunProps};
+        // 100,000 copies of one word under a 1 MB language value: every word after the first is
+        // a repeat, and no word carries a copy of the value.
+        let lang = "x".repeat(1 << 20);
+        let p =
+            Paragraph { runs: vec![Run::with("the ".repeat(100_000), RunProps { lang: Some(lang), ..Default::default() })], ..Default::default() };
+        let words = paragraph_words(&p, "en-US");
+        assert_eq!(words.len(), 100_000);
+        assert!(words.iter().all(|w| w.lang.is_empty()));
+        let bad = misspelt(&p, "en-US", &[]);
+        assert_eq!(bad.len(), 99_999);
+        assert!(bad.iter().all(|w| w.repeated.is_some()));
+        assert!(paragraph_words(&Paragraph::new("word"), &"y".repeat(100)).iter().all(|w| w.lang.is_empty()));
+    }
+
     #[test]
     fn long_and_numeric_flags_and_latin1() {
         let d = Dictionary::parse(b"FLAG long\nSFX Aa Y 1\nSFX Aa 0 s .\n", b"1\ncat/Aa\n");

crates/pptx/src/read.rs

@@ -46,9 +46,6 @@ pub struct Imp<'a> {
     pub tag_bytes: usize,
     /// The tag budget ran out and the warning was given.
     pub tags_capped: bool,
-    /// Whether a part has relationships of its own, by normalized part name: looked up once
-    /// however many embedded objects point at the part.
-    pub has_rels: HashMap<String, bool>,
 }
 
 impl Imp<'_> {
@@ -449,7 +446,6 @@ pub fn import_theme(bytes: &[u8]) -> Result<(deckcraft_model::theme::Theme, Vec<
         tags_by_part: HashMap::new(),
         tag_bytes: 0,
         tags_capped: false,
-        has_rels: HashMap::new(),
     };
     let part = Part::new(&pkg, &theme_path, &doc);
     let theme = theme::read_theme(&mut imp, &part, &doc.root);
@@ -491,7 +487,6 @@ pub fn import_checked(bytes: &[u8]) -> Result<(Presentation, Vec<String>), PptxE
         tags_by_part: HashMap::new(),
         tag_bytes: 0,
         tags_capped: false,
-        has_rels: HashMap::new(),
     };
     // Slide ids first so hyperlinks anywhere can point at slides.
     let slide_ids = register_slides(&mut imp, &docs, &slide_entries, &mut lost_slides);

crates/pptx/src/read/dml.rs

@@ -627,22 +627,12 @@ fn push_text_runs(para: &mut Paragraph, props: RunProps, text: &str) {
 /// Appends an Office Math (a14:m) run: its plain text and its OMML.
 fn push_math_run(para: &mut Paragraph, part: &Part, e: &El) {
     if para.runs.len() < MAX_PARAGRAPH_RUNS {
-        // The text of the outermost `t` elements: a `t` nested in another is already part of
-        // the outer one's text, and taking it again would copy it once per nesting level.
-        fn outer_t(e: &El, depth: usize, text: &mut String) {
-            if depth > crate::xml::MAX_DEPTH {
-                return;
-            }
-            for c in e.elements() {
-                if c.is("t") {
-                    text.push_str(&c.text());
-                } else {
-                    outer_t(c, depth + 1, text);
-                }
-            }
-        }
         let mut text = String::new();
-        outer_t(e, 0, &mut text);
+        let mut ts = vec![];
+        e.find_all("t", &mut ts);
+        for t in ts {
+            text.push_str(&t.text());
+        }
         let omml = e.elements().map(|c| part.keep_inner(c)).collect::<String>();
         para.runs.push(Run { text, props: RunProps::default(), kind: RunKind::Math { omml } });
     }
@@ -687,17 +677,6 @@ mod tests {
     use super::*;
     use crate::xml::parse;
 
-    #[test]
-    fn nested_math_text_is_taken_once() {
-        // 200 `m:t` nested in each other around one word: the word, not 200 copies of it.
-        let x = format!(r#"<a14:m xmlns:a14="a" xmlns:m="m"><m:oMath>{}word{}</m:oMath></a14:m>"#, "<m:t>".repeat(200), "</m:t>".repeat(200));
-        let d = parse(x.as_bytes()).unwrap();
-        let part = Part { name: "ppt/slides/slide1.xml".into(), rels: Default::default(), ns: vec![], inner_ns: vec![] };
-        let mut para = Paragraph::default();
-        push_math_run(&mut para, &part, &d.root);
-        assert_eq!(para.runs[0].text, "word");
-    }
-
     #[test]
     fn colors_and_transforms() {
         let d = parse(

crates/pptx/src/read/shapes.rs

@@ -475,8 +475,7 @@ fn frame(imp: &mut Imp, part: &Part, ctx: &mut IdCtx, e: &El, depth: usize) -> S
         .into_iter()
         .map(|id| {
             let rel = part.rels.get(&id).filter(|r| !r.external)?;
-            let pkg = imp.pkg;
-            if *imp.has_rels.entry(crate::opc::norm(&rel.target)).or_insert_with(|| !pkg.rels(&rel.target).items.is_empty()) {
+            if !imp.pkg.rels(&rel.target).items.is_empty() {
                 return None;
             }
             let (kind, target) = (rel.ty.clone(), rel.target.clone());