refactor: split complex matchers, scanners and solver setup into helpers

Pure extraction, no behaviour change: same order of side effects, same
budget charges, same early returns and error values. Clears
cognitive-complexity findings on glob_inner, the PDF test ops scanner,
snap_ends, the loops inquiry run and build_budgeted (S3776).

crates/constraints/src/lib.rs

@@ -358,6 +358,56 @@ fn components<'a>(cons: &[&'a Constraint]) -> Vec<Vec<&'a Constraint>> {
     groups.into_values().collect()
 }
 
+/// The solver weight of each unknown of `model`: kept handles weigh more, preferred-to-move ones
+/// less, angles and ratios by size², and contact parameters move freely.
+fn solver_weights(model: &Model, opts: &SolveOptions) -> Vec<f64> {
+    let mut weights = vec![1.0; model.x0.len()];
+    for s in &model.slots {
+        let w = if opts.keep.contains(&s.handle) {
+            1e4
+        } else if opts.prefer_move.contains(&s.handle) {
+            1e-2
+        } else {
+            1.0
+        };
+        for j in s.off..s.off + s.len {
+            if let Some(slot) = weights.get_mut(j) {
+                *slot = w;
+            }
+        }
+        for (j, f) in slot_scales(model, s) {
+            if let Some(slot) = weights.get_mut(j) {
+                *slot *= f;
+            }
+        }
+    }
+    // Contact parameters move freely.
+    let geo = model.slots.iter().map(|s| s.off + s.len).max().unwrap_or(0);
+    for w in weights.iter_mut().skip(geo) {
+        *w = 1e-3;
+    }
+    weights
+}
+
+/// Per-unknown factors for the angles and ratios of one slot: size², so that the movement of
+/// points is what is minimised.
+fn slot_scales(model: &Model, s: &model::Slot) -> Vec<(usize, f64)> {
+    let sq = |j: usize| model.x0.get(j).map_or(1.0, |r| (r * r).max(1e-6));
+    match s.shape {
+        model::Shape::Arc => vec![(s.off + 3, sq(s.off + 2)), (s.off + 4, sq(s.off + 2))],
+        model::Shape::Ellipse { arc } => {
+            let a2 = (sq(s.off + 2) + sq(s.off + 3)).max(1e-6);
+            let mut v = vec![(s.off + 4, a2)];
+            if arc {
+                v.extend([(s.off + 5, a2), (s.off + 6, a2)]);
+            }
+            v
+        }
+        model::Shape::Insert { .. } => vec![(s.off + 2, model.rigid_extent2(s.handle).max(1.0))],
+        _ => Vec::new(),
+    }
+}
+
 /// Build the component systems. Constraints that cannot be interpreted are returned as skipped.
 pub(crate) fn build(d: &Drawing, cons: &[&Constraint], opts: &SolveOptions) -> Result<(Vec<Comp>, Vec<(u32, String)>), SolveError> {
     let mut members_left = model::MAX_RIGID_WORK;
@@ -396,46 +446,7 @@ pub(crate) fn build_budgeted(
         if built.is_empty() {
             continue;
         }
-        let mut weights = vec![1.0; model.x0.len()];
+        let weights = solver_weights(&model, opts);
-        for s in &model.slots {
-            let w = if opts.keep.contains(&s.handle) {
-                1e4
-            } else if opts.prefer_move.contains(&s.handle) {
-                1e-2
-            } else {
-                1.0
-            };
-            for j in s.off..s.off + s.len {
-                if let Some(slot) = weights.get_mut(j) {
-                    *slot = w;
-                }
-            }
-            // Angles and ratios: weight by size² so the movement of points is what is minimised.
-            let sq = |j: usize| model.x0.get(j).map_or(1.0, |r| (r * r).max(1e-6));
-            let scaled: Vec<(usize, f64)> = match s.shape {
-                model::Shape::Arc => vec![(s.off + 3, sq(s.off + 2)), (s.off + 4, sq(s.off + 2))],
-                model::Shape::Ellipse { arc } => {
-                    let a2 = (sq(s.off + 2) + sq(s.off + 3)).max(1e-6);
-                    let mut v = vec![(s.off + 4, a2)];
-                    if arc {
-                        v.extend([(s.off + 5, a2), (s.off + 6, a2)]);
-                    }
-                    v
-                }
-                model::Shape::Insert { .. } => vec![(s.off + 2, model.rigid_extent2(s.handle).max(1.0))],
-                _ => Vec::new(),
-            };
-            for (j, f) in scaled {
-                if let Some(slot) = weights.get_mut(j) {
-                    *slot *= f;
-                }
-            }
-        }
-        // Contact parameters move freely.
-        let geo = model.slots.iter().map(|s| s.off + s.len).max().unwrap_or(0);
-        for w in weights.iter_mut().skip(geo) {
-            *w = 1e-3;
-        }
         let held = opts.hold.iter().filter_map(|r| model.lock_indices(r)).flatten().collect();
         let tol = scale_tol(&model.x0);
         comps.push(Comp { model, built, weights, held, tol });

crates/engine/src/cmd/inquiry/loops.rs

@@ -117,21 +117,7 @@ pub fn run(s: &mut Session, p: &Value) -> Result<Value> {
 
     // Endpoints closer than the tolerance meet. Transitive components of three or more ends are
     // branch junctions; a component of exactly two is a simple joint.
-    let mut parent: Vec<usize> = (0..n).collect();
+    let mut parent = merge_ends(&ends, tol);
-    for i in 0..n {
-        let Some(a) = ends.get(i) else { continue };
-        for j in (i + 1)..n {
-            let Some(b) = ends.get(j) else { continue };
-            if a.at.dist(b.at) <= tol {
-                let (ra, rb) = (find(&mut parent, i), find(&mut parent, j));
-                if ra != rb
-                    && let Some(slot) = parent.get_mut(ra)
-                {
-                    *slot = rb;
-                }
-            }
-        }
-    }
     let mut groups: BTreeMap<usize, Vec<usize>> = BTreeMap::new();
     for i in 0..n {
         let root = find(&mut parent, i);
@@ -140,54 +126,8 @@ pub fn run(s: &mut Session, p: &Value) -> Result<Value> {
 
     // Junctions are reported and their pieces left out of the chains: not resolved.
     let mut excluded = vec![false; pieces.len()];
-    let mut junctions: Vec<Value> = Vec::new();
+    let (joints, junctions) = sort_groups(&groups, &ends, &pieces, &mut excluded);
-    let mut joints: Vec<(usize, usize)> = Vec::new();
+    let link = joint_links(joints, &ends, &pieces, &excluded);
-    for group in groups.values() {
-        match group.as_slice() {
-            &[a, b] => joints.push((a, b)),
-            [_, _, ..] => {
-                let scale = group.len() as f64;
-                let mut at = Vec2::ZERO;
-                let mut members: Vec<Handle> = Vec::new();
-                for &e in group {
-                    let Some(end) = ends.get(e) else { continue };
-                    at += end.at / scale;
-                    let Some(piece) = pieces.get(end.piece) else { continue };
-                    if !members.contains(&piece.handle) {
-                        members.push(piece.handle);
-                    }
-                    if let Some(slot) = excluded.get_mut(end.piece) {
-                        *slot = true;
-                    }
-                }
-                junctions.push(json!({
-                    "at": [at.x, at.y],
-                    "ends": group.len(),
-                    "members": members.iter().map(|h| h.hex()).collect::<Vec<_>>(),
-                }));
-            }
-            _ => {}
-        }
-    }
-
-    // A joint links the two ends for the walk. A joint into an excluded piece is not a link, and
-    // a line whose own ends meet within a huge tolerance is a spur, not a zero-area loop.
-    let mut link: Vec<Option<usize>> = vec![None; n];
-    for (a, b) in joints {
-        let (Some(ea), Some(eb)) = (ends.get(a), ends.get(b)) else { continue };
-        if ea.piece == eb.piece && pieces.get(ea.piece).is_some_and(|p| p.arc.is_none()) {
-            continue;
-        }
-        if excluded.get(ea.piece).copied().unwrap_or(true) || excluded.get(eb.piece).copied().unwrap_or(true) {
-            continue;
-        }
-        if let Some(slot) = link.get_mut(a) {
-            *slot = Some(b);
-        }
-        if let Some(slot) = link.get_mut(b) {
-            *slot = Some(a);
-        }
-    }
 
     let mut visited = vec![false; pieces.len()];
     let mut loops: Vec<Value> = Vec::new();
@@ -225,6 +165,84 @@ pub fn run(s: &mut Session, p: &Value) -> Result<Value> {
 }
 
 /// A report with nothing analyzed: an empty drawing, or one past the pair cap.
+/// Union-find over the ends: ends closer than `tol` share a root. Returns each end's parent.
+fn merge_ends(ends: &[End], tol: f64) -> Vec<usize> {
+    let n = ends.len();
+    let mut parent: Vec<usize> = (0..n).collect();
+    for i in 0..n {
+        let Some(a) = ends.get(i) else { continue };
+        for j in (i + 1)..n {
+            let Some(b) = ends.get(j) else { continue };
+            if a.at.dist(b.at) <= tol {
+                let (ra, rb) = (find(&mut parent, i), find(&mut parent, j));
+                if ra != rb
+                    && let Some(slot) = parent.get_mut(ra)
+                {
+                    *slot = rb;
+                }
+            }
+        }
+    }
+    parent
+}
+
+/// Each group of two ends is a joint (end indices); each of three or more is a junction, reported
+/// with its pieces marked `excluded`.
+fn sort_groups(groups: &BTreeMap<usize, Vec<usize>>, ends: &[End], pieces: &[Piece], excluded: &mut [bool]) -> (Vec<(usize, usize)>, Vec<Value>) {
+    let mut joints: Vec<(usize, usize)> = Vec::new();
+    let mut junctions: Vec<Value> = Vec::new();
+    for group in groups.values() {
+        match group.as_slice() {
+            &[a, b] => joints.push((a, b)),
+            [_, _, ..] => {
+                let scale = group.len() as f64;
+                let mut at = Vec2::ZERO;
+                let mut members: Vec<Handle> = Vec::new();
+                for &e in group {
+                    let Some(end) = ends.get(e) else { continue };
+                    at += end.at / scale;
+                    let Some(piece) = pieces.get(end.piece) else { continue };
+                    if !members.contains(&piece.handle) {
+                        members.push(piece.handle);
+                    }
+                    if let Some(slot) = excluded.get_mut(end.piece) {
+                        *slot = true;
+                    }
+                }
+                junctions.push(json!({
+                    "at": [at.x, at.y],
+                    "ends": group.len(),
+                    "members": members.iter().map(|h| h.hex()).collect::<Vec<_>>(),
+                }));
+            }
+            _ => {}
+        }
+    }
+    (joints, junctions)
+}
+
+/// The link each end takes from the joints. A joint into an excluded piece is not a link, and
+/// a line whose own ends meet within a huge tolerance is a spur, not a zero-area loop.
+fn joint_links(joints: Vec<(usize, usize)>, ends: &[End], pieces: &[Piece], excluded: &[bool]) -> Vec<Option<usize>> {
+    let mut link: Vec<Option<usize>> = vec![None; ends.len()];
+    for (a, b) in joints {
+        let (Some(ea), Some(eb)) = (ends.get(a), ends.get(b)) else { continue };
+        if ea.piece == eb.piece && pieces.get(ea.piece).is_some_and(|p| p.arc.is_none()) {
+            continue;
+        }
+        if excluded.get(ea.piece).copied().unwrap_or(true) || excluded.get(eb.piece).copied().unwrap_or(true) {
+            continue;
+        }
+        if let Some(slot) = link.get_mut(a) {
+            *slot = Some(b);
+        }
+        if let Some(slot) = link.get_mut(b) {
+            *slot = Some(a);
+        }
+    }
+    link
+}
+
 fn empty_report(offset: usize, tol: f64, pair_tests: usize, pair_tests_needed: usize, pair_limit: usize, capped: bool, skipped: Skipped) -> Value {
     json!({
         "schema": 1,

crates/engine/src/lisp/builtins.rs

@@ -785,6 +785,48 @@ fn glob_cached(
     Ok(matched)
 }
 
+/// For the bracket class opening at `pat[p]`, whether `text[t]` is in it: the index of its closing
+/// `]` when it is, `None` when it is not or the class is never closed.
+fn class_end(text: &[char], pat: &[char], p: usize, t: usize, budget: &mut u64) -> Result<Option<usize>, ()> {
+    let mut close = p;
+    while let Some(c) = pat.get(close) {
+        charge_match(budget)?;
+        if *c == ']' {
+            break;
+        }
+        close += 1;
+    }
+    if close == pat.len() {
+        return Ok(None);
+    }
+    let Some(&c) = text.get(t) else { return Ok(None) };
+    let set = &pat[p + 1..close];
+    let (neg, set) = match set.split_first() {
+        Some(('~', r)) if !r.is_empty() => (true, r),
+        _ => (false, set),
+    };
+    let mut hit = false;
+    let mut i = 0;
+    while i < set.len() {
+        charge_match(budget)?;
+        if set.get(i + 1) == Some(&'-') && i + 2 < set.len() {
+            if set[i] <= c && c <= set[i + 2] {
+                hit = true;
+            }
+            i += 3;
+        } else {
+            if set[i] == c {
+                hit = true;
+            }
+            i += 1;
+        }
+    }
+    if hit == neg {
+        return Ok(None);
+    }
+    Ok(Some(close))
+}
+
 /// Evaluate wildcard tokens and backtracking branches under shared work and recursion limits.
 fn glob_inner(
     text: &[char],
@@ -822,42 +864,7 @@ fn glob_inner(
                 p += 2;
             }
             '[' => {
-                let mut close = p;
+                let Some(close) = class_end(text, pat, p, t, budget)? else { return Ok(false) };
-                while let Some(c) = pat.get(close) {
-                    charge_match(budget)?;
-                    if *c == ']' {
-                        break;
-                    }
-                    close += 1;
-                }
-                if close == pat.len() {
-                    return Ok(false);
-                }
-                let Some(&c) = text.get(t) else { return Ok(false) };
-                let set = &pat[p + 1..close];
-                let (neg, set) = match set.split_first() {
-                    Some(('~', r)) if !r.is_empty() => (true, r),
-                    _ => (false, set),
-                };
-                let mut hit = false;
-                let mut i = 0;
-                while i < set.len() {
-                    charge_match(budget)?;
-                    if set.get(i + 1) == Some(&'-') && i + 2 < set.len() {
-                        if set[i] <= c && c <= set[i + 2] {
-                            hit = true;
-                        }
-                        i += 3;
-                    } else {
-                        if set[i] == c {
-                            hit = true;
-                        }
-                        i += 1;
-                    }
-                }
-                if hit == neg {
-                    return Ok(false);
-                }
                 t += 1;
                 p = close + 1;
             }

crates/geom/src/region.rs

@@ -447,6 +447,44 @@ fn islands(g: &Graph, outer: &[usize], p: Vec2) -> Vec<Island> {
     out
 }
 
+/// The point of the nearest other curve within `snap` of the end `e` of curve `i`, as (curve,
+/// segment, parameter, point). `None` when the end touches a curve, or none is near. `Err` when
+/// the budget of segment checks runs out.
+fn nearest_curve(polys: &[Vec<Vec2>], i: usize, e: Vec2, snap: f64, tol: f64, budget: &mut usize) -> Result<Option<(usize, usize, f64, Vec2)>, ()> {
+    let mut best: Option<(f64, usize, usize, f64, Vec2)> = None;
+    let mut touching = false;
+    for (j, other) in polys.iter().enumerate() {
+        if j == i {
+            continue;
+        }
+        for (k, w) in other.windows(2).enumerate() {
+            if *budget == 0 {
+                return Err(());
+            }
+            *budget -= 1;
+            let [a, b] = w else { continue };
+            if !(a.is_finite() && b.is_finite())
+                || e.x < a.x.min(b.x) - snap
+                || e.x > a.x.max(b.x) + snap
+                || e.y < a.y.min(b.y) - snap
+                || e.y > a.y.max(b.y) + snap
+            {
+                continue;
+            }
+            let line = Line::new(*a, *b);
+            let t = line.param_of(e).clamp(0.0, 1.0);
+            let q = line.at(t);
+            let d = q.dist(e);
+            if d <= tol {
+                touching = true;
+            } else if d <= snap && best.is_none_or(|(bd, ..)| d < bd) {
+                best = Some((d, j, k, t, q));
+            }
+        }
+    }
+    Ok(best.filter(|_| !touching).map(|(_, j, k, t, q)| (j, k, t, q)))
+}
+
 /// The curves with each open end that stops within `snap` of another curve moved onto that
 /// curve (which gains a vertex there), so the two meet exactly. A curve that is flattened
 /// into chords (an ellipse, a spline) lies up to the flattening tolerance off the true curve,
@@ -474,38 +512,8 @@ pub fn snap_ends(polys: &[Vec<Vec2>], snap: f64, tol: f64) -> Vec<Vec<Vec2>> {
             if !e.is_finite() {
                 continue;
             }
-            let mut best: Option<(f64, usize, usize, f64, Vec2)> = None;
+            let Ok(near) = nearest_curve(polys, i, e, snap, tol, &mut budget) else { break 'ends };
-            let mut touching = false;
+            let Some((j, k, t, q)) = near else { continue };
-            for (j, other) in polys.iter().enumerate() {
-                if j == i {
-                    continue;
-                }
-                for (k, w) in other.windows(2).enumerate() {
-                    if budget == 0 {
-                        break 'ends;
-                    }
-                    budget -= 1;
-                    let [a, b] = w else { continue };
-                    if !(a.is_finite() && b.is_finite())
-                        || e.x < a.x.min(b.x) - snap
-                        || e.x > a.x.max(b.x) + snap
-                        || e.y < a.y.min(b.y) - snap
-                        || e.y > a.y.max(b.y) + snap
-                    {
-                        continue;
-                    }
-                    let line = Line::new(*a, *b);
-                    let t = line.param_of(e).clamp(0.0, 1.0);
-                    let q = line.at(t);
-                    let d = q.dist(e);
-                    if d <= tol {
-                        touching = true;
-                    } else if d <= snap && best.is_none_or(|(bd, ..)| d < bd) {
-                        best = Some((d, j, k, t, q));
-                    }
-                }
-            }
-            let Some((_, j, k, t, q)) = best.filter(|_| !touching) else { continue };
             let Some(slot) = out.get_mut(i).and_then(|v| if at_end { v.last_mut() } else { v.first_mut() }) else { continue };
             *slot = q;
             let ends = polys.get(j).and_then(|o| o.get(k).zip(o.get(k + 1)));

crates/io/tests/support/pdf_read.rs

@@ -235,6 +235,46 @@ pub struct Op {
     pub names: Vec<String>,
 }
 
+/// The index just past the string literal opening at `b[i]`, honouring escapes and nesting.
+fn skip_string(b: &[char], mut i: usize) -> usize {
+    let mut depth = 0;
+    while i < b.len() {
+        match b[i] {
+            '\\' => i += 1,
+            '(' => depth += 1,
+            ')' => {
+                depth -= 1;
+                if depth == 0 {
+                    break;
+                }
+            }
+            _ => {}
+        }
+        i += 1;
+    }
+    i + 1
+}
+
+/// The index just past the inline dictionary opening at `b[i]` (`<<`).
+fn skip_dict(b: &[char], mut i: usize) -> usize {
+    let mut depth = 0;
+    while i < b.len() {
+        if b[i] == '<' && b.get(i + 1) == Some(&'<') {
+            depth += 1;
+            i += 1;
+        } else if b[i] == '>' && b.get(i + 1) == Some(&'>') {
+            depth -= 1;
+            i += 1;
+            if depth == 0 {
+                i += 1;
+                break;
+            }
+        }
+        i += 1;
+    }
+    i
+}
+
 /// The operators of content text in order, each with its numeric operands (other operands
 /// are skipped).
 pub fn ops(content: &str) -> Vec<Op> {
@@ -253,39 +293,10 @@ pub fn ops(content: &str) -> Vec<Op> {
                 i += 1;
             }
         } else if c == '(' {
-            let mut depth = 0;
+            i = skip_string(&b, i);
-            while i < b.len() {
-                match b[i] {
-                    '\\' => i += 1,
-                    '(' => depth += 1,
-                    ')' => {
-                        depth -= 1;
-                        if depth == 0 {
-                            break;
-                        }
-                    }
-                    _ => {}
-                }
-                i += 1;
-            }
-            i += 1;
         } else if c == '<' && b.get(i + 1) == Some(&'<') {
             // An inline dictionary (marked-content properties): skip to its end.
-            let mut depth = 0;
+            i = skip_dict(&b, i);
-            while i < b.len() {
-                if b[i] == '<' && b.get(i + 1) == Some(&'<') {
-                    depth += 1;
-                    i += 1;
-                } else if b[i] == '>' && b.get(i + 1) == Some(&'>') {
-                    depth -= 1;
-                    i += 1;
-                    if depth == 0 {
-                        i += 1;
-                        break;
-                    }
-                }
-                i += 1;
-            }
         } else if c == '<' {
             while i < b.len() && b[i] != '>' {
                 i += 1;