oxideav-pdf 0.1.2

//! PDF content-stream operator parser — inverse of [`crate::operators`].
//!
//! Walks the operator stream emitted by a per-page Contents object and
//! reconstructs the [`oxideav_core::vector::Group`] tree that the
//! writer originally walked. The mapping is the same one the writer
//! uses, run in reverse:
//!
//! | PDF operator         | Vector IR                                |
//! |----------------------|------------------------------------------|
//! | `q` / `Q`            | enter / leave a child [`Group`]          |
//! | `cm`                 | concat into the current group's transform |
//! | `m` / `l` / `c` / `h`| [`PathCommand::MoveTo`] / `LineTo` / `CubicCurveTo` / `Close` |
//! | `v` / `y`            | shorthand cubic — lifted to a full `c`   |
//! | `re`                 | rectangle subpath (m + 3*l + h)          |
//! | `f` / `f*`           | fill (NonZero / EvenOdd)                 |
//! | `S`                  | stroke                                   |
//! | `B` / `B*`           | fill + stroke                            |
//! | `b` / `b*`           | close + fill + stroke                    |
//! | `n`                  | no-op paint (consume current path)       |
//! | `W` / `W*`           | clip — assigns to the current group's `clip` |
//! | `rg` / `RG`          | fill / stroke colour (DeviceRGB)         |
//! | `g` / `G`            | grayscale fill / stroke (round-3 maps to RGB triplet) |
//! | `k` / `K`            | DeviceCMYK fill / stroke — converted to RGB per §10.3.5 |
//! | `w` / `J` / `j` / `M`| stroke width / cap / join / miter limit  |
//! | `d`                  | dash array + offset                      |
//! | `gs` / `cs` / `CS` / `scn` / `SCN` | resource references — round-3 records names but doesn't resolve |
//!
//! The parser does not reach into the page's `/Resources` dict for
//! gradient / pattern lookups in this commit — those land in round-3
//! commit 5 (the top-level walker that has the resolved Document).
//! For now, a `/Pat0 scn` pair produces a black solid fill (matches
//! the writer's "unknown-paint fallback" behaviour, so the
//! roundtrip remains semantically conservative).
//!
//! Text-showing operators (`BT` / `ET` / `Tj` / `TJ`) are skipped
//! silently — text rendering is round-4+.

use std::str;

use oxideav_core::vector::{
    DashPattern, FillRule, Group, LineCap, LineJoin, Node, Paint, Path, PathCommand, PathNode,
    Point, Rgba, Stroke, Transform2D,
};

use crate::error::PdfError;

/// Parse a content-stream byte sequence into a single [`Group`]
/// containing every shape painted by the stream. Nested `q`/`Q`
/// brackets become nested `Node::Group` children. The returned root
/// group has identity transform; per-`q` transforms live on the
/// child groups.
pub fn parse_content_stream(input: &[u8]) -> Result<Group, PdfError> {
    let mut state = State::new();
    state.parse(input)?;
    Ok(state.finish())
}

// ───────────────────────── parser state ─────────────────────────

/// Per-graphics-state tracker. Pushed on `q`, popped on `Q`. The
/// active state is `stack.last_mut().unwrap()`; the always-present
/// root frame collects whatever the input emits before any explicit
/// `q`/`Q`.
struct State {
    /// Argument stack — operands are pushed as the parser scans
    /// numbers, names, arrays; an operator keyword consumes them.
    operands: Vec<Operand>,
    /// Group stack mirroring PDF's graphics-state stack.
    stack: Vec<Frame>,
    /// Current path being built (the most recent `m`/`l`/`c`/`re`
    /// sequence). `None` after a paint operator commits it.
    current_path: Option<Path>,
    /// Tracking for the current path's last endpoint — needed to
    /// handle the shorthand cubics `v` (use current pt as c1) and
    /// `y` (use end pt as c2).
    current_point: Point,
    /// Last set fill / stroke paint state. Reset on each `q`/`Q`
    /// (PDF graphics state) since `q` saves the entire colour /
    /// stroke state and `Q` restores it.
    fill_paint: Option<Paint>,
    stroke_paint: Option<Paint>,
    stroke_width: f32,
    line_cap: LineCap,
    line_join: LineJoin,
    miter_limit: f32,
    dash: Option<DashPattern>,
}

struct Frame {
    /// Transform applied to this group via `cm` operators since `q`.
    transform: Transform2D,
    /// Children accumulated while this `q` is active.
    children: Vec<Node>,
    /// Clip path, if a `W`/`W*` was issued.
    clip: Option<Path>,
}

#[derive(Clone, Debug)]
enum Operand {
    Number(f32),
    Array(Vec<f32>),
    /// Name operand — the inner string is unused by the round-3
    /// commit-4 dispatcher (resource references like `/Pat0 scn`
    /// land in round-3 commit 5, when the page's `/Resources` dict
    /// is available). The variant is kept so the operand stack stays
    /// well-typed; future commits switch the inner string from `_`
    /// to a proper lookup key.
    #[allow(dead_code)]
    Name(String),
}

impl State {
    fn new() -> Self {
        Self {
            operands: Vec::new(),
            stack: vec![Frame::new()],
            current_path: None,
            current_point: Point::default(),
            fill_paint: None,
            stroke_paint: None,
            stroke_width: 1.0,
            line_cap: LineCap::Butt,
            line_join: LineJoin::Miter,
            miter_limit: 10.0,
            dash: None,
        }
    }

    fn finish(mut self) -> Group {
        // Unwind any unmatched `q` frames by promoting them in order
        // — the input was malformed but we'd rather salvage what we
        // can than refuse the whole document.
        while self.stack.len() > 1 {
            self.pop_q();
        }
        let root = self.stack.pop().expect("root frame present");
        Group {
            transform: root.transform,
            opacity: 1.0,
            clip: root.clip,
            children: root.children,
            ..Group::default()
        }
    }

    fn current(&mut self) -> &mut Frame {
        self.stack.last_mut().expect("at least the root frame")
    }

    fn push_q(&mut self) {
        self.stack.push(Frame::new());
    }

    fn pop_q(&mut self) {
        // Only pop if we have more than the root frame — otherwise
        // ignore the unbalanced `Q` per the writer's "permissive
        // recovery" stance.
        if self.stack.len() <= 1 {
            return;
        }
        let frame = self.stack.pop().unwrap();
        // Translate the frame into a Node::Group child of its parent
        // — but skip empty groups (just `q Q` with nothing in
        // between is a no-op for the IR).
        if frame.is_effectively_empty() {
            return;
        }
        let g = Group {
            transform: frame.transform,
            opacity: 1.0,
            clip: frame.clip,
            children: frame.children,
            ..Group::default()
        };
        self.current().children.push(Node::Group(g));
    }

    /// Handle one keyword (operator). Operands have already been
    /// pushed to `self.operands`.
    fn dispatch(&mut self, op: &[u8]) -> Result<(), PdfError> {
        match op {
            // Graphics state -------------------------------------
            b"q" => {
                self.push_q();
            }
            b"Q" => {
                self.pop_q();
            }
            b"cm" => {
                let nums = self.take_numbers(6)?;
                let t = Transform2D {
                    a: nums[0],
                    b: nums[1],
                    c: nums[2],
                    d: nums[3],
                    e: nums[4],
                    f: nums[5],
                };
                let frame = self.current();
                frame.transform = compose(frame.transform, t);
            }
            b"gs" => {
                // /GSx gs — references an ExtGState. Round-3 doesn't
                // resolve resources, so we drop the operand. (The
                // top-level walker — round-3 commit 5 — will look up
                // the gradient / opacity from /Resources.)
                self.operands.clear();
            }

            // Path construction ----------------------------------
            b"m" => {
                let p = self.take_point()?;
                let path = self.path_mut();
                path.commands.push(PathCommand::MoveTo(p));
                self.current_point = p;
            }
            b"l" => {
                let p = self.take_point()?;
                let path = self.path_mut();
                path.commands.push(PathCommand::LineTo(p));
                self.current_point = p;
            }
            b"c" => {
                let nums = self.take_numbers(6)?;
                let c1 = Point::new(nums[0], nums[1]);
                let c2 = Point::new(nums[2], nums[3]);
                let end = Point::new(nums[4], nums[5]);
                let path = self.path_mut();
                path.commands
                    .push(PathCommand::CubicCurveTo { c1, c2, end });
                self.current_point = end;
            }
            b"v" => {
                // Shorthand cubic: c1 = current point.
                let nums = self.take_numbers(4)?;
                let c1 = self.current_point;
                let c2 = Point::new(nums[0], nums[1]);
                let end = Point::new(nums[2], nums[3]);
                let path = self.path_mut();
                path.commands
                    .push(PathCommand::CubicCurveTo { c1, c2, end });
                self.current_point = end;
            }
            b"y" => {
                // Shorthand cubic: c2 = end.
                let nums = self.take_numbers(4)?;
                let c1 = Point::new(nums[0], nums[1]);
                let end = Point::new(nums[2], nums[3]);
                let c2 = end;
                let path = self.path_mut();
                path.commands
                    .push(PathCommand::CubicCurveTo { c1, c2, end });
                self.current_point = end;
            }
            b"re" => {
                // x y w h re — a rectangle as a closed subpath.
                let nums = self.take_numbers(4)?;
                let (x, y, w, h) = (nums[0], nums[1], nums[2], nums[3]);
                let path = self.path_mut();
                path.commands.push(PathCommand::MoveTo(Point::new(x, y)));
                path.commands
                    .push(PathCommand::LineTo(Point::new(x + w, y)));
                path.commands
                    .push(PathCommand::LineTo(Point::new(x + w, y + h)));
                path.commands
                    .push(PathCommand::LineTo(Point::new(x, y + h)));
                path.commands.push(PathCommand::Close);
                self.current_point = Point::new(x, y);
            }
            b"h" => {
                let path = self.path_mut();
                path.commands.push(PathCommand::Close);
            }

            // Painting -------------------------------------------
            b"f" | b"F" => self.commit_path(true, false, FillRule::NonZero),
            b"f*" => self.commit_path(true, false, FillRule::EvenOdd),
            b"S" => self.commit_path(false, true, FillRule::NonZero),
            b"s" => {
                // s = h + S — implicit close before stroke.
                if let Some(p) = &mut self.current_path {
                    p.commands.push(PathCommand::Close);
                }
                self.commit_path(false, true, FillRule::NonZero);
            }
            b"B" => self.commit_path(true, true, FillRule::NonZero),
            b"B*" => self.commit_path(true, true, FillRule::EvenOdd),
            b"b" => {
                if let Some(p) = &mut self.current_path {
                    p.commands.push(PathCommand::Close);
                }
                self.commit_path(true, true, FillRule::NonZero);
            }
            b"b*" => {
                if let Some(p) = &mut self.current_path {
                    p.commands.push(PathCommand::Close);
                }
                self.commit_path(true, true, FillRule::EvenOdd);
            }
            b"n" => {
                // No-op paint — drop the current path.
                self.current_path = None;
                self.operands.clear();
            }

            // Clip --------------------------------------------------
            b"W" | b"W*" => {
                // The clip operator consumes the current path as the
                // clip region — but in PDF the clip is committed by
                // the next paint operator, conventionally `n`. We
                // record it onto the current frame here; if the
                // upcoming paint is `n` it'll just discard the path
                // (which we've already moved into `clip`).
                if let Some(p) = self.current_path.take() {
                    self.current().clip = Some(p);
                }
                self.operands.clear();
            }

            // Colour ----------------------------------------------
            b"rg" => {
                let nums = self.take_numbers(3)?;
                self.fill_paint = Some(Paint::Solid(rgb_from_unit(nums[0], nums[1], nums[2])));
            }
            b"RG" => {
                let nums = self.take_numbers(3)?;
                self.stroke_paint = Some(Paint::Solid(rgb_from_unit(nums[0], nums[1], nums[2])));
            }
            b"g" => {
                let nums = self.take_numbers(1)?;
                self.fill_paint = Some(Paint::Solid(rgb_from_unit(nums[0], nums[0], nums[0])));
            }
            b"G" => {
                let nums = self.take_numbers(1)?;
                self.stroke_paint = Some(Paint::Solid(rgb_from_unit(nums[0], nums[0], nums[0])));
            }
            b"k" | b"K" => {
                // DeviceCMYK fill (`k`) / stroke (`K`). The IR carries
                // only RGB, so convert per ISO 32000-1 §10.3.5
                // (DeviceCMYK → DeviceRGB): a simple operation that does
                // not involve black generation or undercolour removal.
                let nums = self.take_numbers(4)?;
                let p = Some(Paint::Solid(rgb_from_cmyk(
                    nums[0], nums[1], nums[2], nums[3],
                )));
                if op == b"K" {
                    self.stroke_paint = p;
                } else {
                    self.fill_paint = p;
                }
            }
            b"sc" | b"scn" => {
                // Pattern / multi-component colour — round-3 fallback.
                self.fill_paint = self
                    .fill_paint
                    .clone()
                    .or(Some(Paint::Solid(Rgba::opaque(0, 0, 0))));
                self.operands.clear();
            }
            b"SC" | b"SCN" => {
                self.stroke_paint = self
                    .stroke_paint
                    .clone()
                    .or(Some(Paint::Solid(Rgba::opaque(0, 0, 0))));
                self.operands.clear();
            }
            b"cs" | b"CS" => {
                // Colour-space switch — operand is /Name. Drop it.
                self.operands.clear();
            }

            // Stroke style -----------------------------------------
            b"w" => {
                let nums = self.take_numbers(1)?;
                self.stroke_width = nums[0];
            }
            b"J" => {
                let nums = self.take_numbers(1)?;
                self.line_cap = match nums[0] as i32 {
                    0 => LineCap::Butt,
                    1 => LineCap::Round,
                    2 => LineCap::Square,
                    _ => LineCap::Butt,
                };
            }
            b"j" => {
                let nums = self.take_numbers(1)?;
                self.line_join = match nums[0] as i32 {
                    0 => LineJoin::Miter,
                    1 => LineJoin::Round,
                    2 => LineJoin::Bevel,
                    _ => LineJoin::Miter,
                };
            }
            b"M" => {
                let nums = self.take_numbers(1)?;
                self.miter_limit = nums[0];
            }
            b"d" => {
                // [array] offset d
                if self.operands.len() < 2 {
                    self.operands.clear();
                    return Ok(());
                }
                let offset = match self.operands.pop().unwrap() {
                    Operand::Number(n) => n,
                    _ => 0.0,
                };
                let array = match self.operands.pop().unwrap() {
                    Operand::Array(v) => v,
                    _ => Vec::new(),
                };
                self.dash = if array.is_empty() {
                    None
                } else {
                    Some(DashPattern { array, offset })
                };
                self.operands.clear();
            }

            // Text — silently skip the operands. Round-4+.
            b"BT" | b"ET" | b"Tj" | b"TJ" | b"Tf" | b"Tc" | b"Tw" | b"Tz" | b"TL" | b"Tr"
            | b"Ts" | b"Td" | b"TD" | b"Tm" | b"T*" | b"'" | b"\"" => {
                self.operands.clear();
            }

            // XObject paint ----------------------------------------
            b"Do" => {
                // /Imx Do — paint an image XObject. Round-3 doesn't
                // resolve XObject images yet (round-4+), drop.
                self.operands.clear();
            }

            // Marked-content + everything else ---------------------
            _ => {
                self.operands.clear();
            }
        }
        Ok(())
    }

    fn commit_path(&mut self, fill: bool, stroke: bool, rule: FillRule) {
        let Some(path) = self.current_path.take() else {
            self.operands.clear();
            return;
        };
        let fill_paint = if fill {
            self.fill_paint
                .clone()
                .or(Some(Paint::Solid(Rgba::opaque(0, 0, 0))))
        } else {
            None
        };
        let stroke_obj = if stroke {
            Some(Stroke {
                width: self.stroke_width,
                paint: self
                    .stroke_paint
                    .clone()
                    .unwrap_or(Paint::Solid(Rgba::opaque(0, 0, 0))),
                cap: self.line_cap,
                join: self.line_join,
                miter_limit: self.miter_limit,
                dash: self.dash.clone(),
            })
        } else {
            None
        };
        let node = Node::Path(PathNode {
            path,
            fill: fill_paint,
            stroke: stroke_obj,
            fill_rule: rule,
        });
        self.current().children.push(node);
        self.operands.clear();
    }

    fn path_mut(&mut self) -> &mut Path {
        if self.current_path.is_none() {
            self.current_path = Some(Path::new());
        }
        self.current_path.as_mut().unwrap()
    }

    fn take_numbers(&mut self, n: usize) -> Result<Vec<f32>, PdfError> {
        if self.operands.len() < n {
            return Err(PdfError::other(format!(
                "PDF content parser: operator needed {n} numeric operands, got {}",
                self.operands.len()
            )));
        }
        let split = self.operands.len() - n;
        let tail: Vec<Operand> = self.operands.drain(split..).collect();
        let mut out = Vec::with_capacity(n);
        for op in tail {
            match op {
                Operand::Number(f) => out.push(f),
                other => {
                    return Err(PdfError::other(format!(
                        "PDF content parser: expected numeric operand, got {other:?}"
                    )));
                }
            }
        }
        Ok(out)
    }

    fn take_point(&mut self) -> Result<Point, PdfError> {
        let nums = self.take_numbers(2)?;
        Ok(Point::new(nums[0], nums[1]))
    }

    fn parse(&mut self, input: &[u8]) -> Result<(), PdfError> {
        let mut i = 0;
        while i < input.len() {
            let b = input[i];
            if is_whitespace(b) {
                i += 1;
                continue;
            }
            if b == b'%' {
                // Comment to end of line.
                while i < input.len() && input[i] != b'\n' && input[i] != b'\r' {
                    i += 1;
                }
                continue;
            }
            if b == b'(' {
                // Skip a literal string operand (text-show argument
                // for `Tj` etc). Track depth + escape.
                let (end, _bytes) = read_literal_string(input, i)?;
                i = end;
                continue;
            }
            if b == b'<' && input.get(i + 1) != Some(&b'<') {
                // Hex string — also a text-show operand.
                let end = read_hex_string(input, i)?;
                i = end;
                continue;
            }
            if b == b'[' {
                // Array operand — for the dash array `[5 3] 0 d`.
                let (end, nums) = read_number_array(input, i);
                self.operands.push(Operand::Array(nums));
                i = end;
                continue;
            }
            if b == b'/' {
                // Name operand.
                let mut end = i + 1;
                while end < input.len() && !is_whitespace(input[end]) && !is_delimiter(input[end]) {
                    end += 1;
                }
                // We don't bother decoding #xx in content-stream
                // names — round-3 callers never produce such names.
                let name = String::from_utf8_lossy(&input[i + 1..end]).into_owned();
                self.operands.push(Operand::Name(name));
                i = end;
                continue;
            }
            if matches!(b, b'+' | b'-' | b'.' | b'0'..=b'9') {
                // Number operand.
                let mut end = i;
                if matches!(input[end], b'+' | b'-') {
                    end += 1;
                }
                let mut saw_digit = false;
                let mut saw_dot = false;
                while end < input.len() {
                    let c = input[end];
                    if c.is_ascii_digit() {
                        end += 1;
                        saw_digit = true;
                    } else if c == b'.' && !saw_dot {
                        end += 1;
                        saw_dot = true;
                    } else {
                        break;
                    }
                }
                if !saw_digit {
                    // Bare sign / dot — fall through to keyword
                    // handling.
                    let kw_end = scan_keyword_end(input, i);
                    let kw = &input[i..kw_end];
                    self.dispatch(kw)?;
                    i = kw_end;
                    continue;
                }
                let s = str::from_utf8(&input[i..end]).map_err(|_| {
                    PdfError::other(format!("PDF content parser: non-UTF-8 number at byte {i}"))
                })?;
                let f: f32 = s.parse().map_err(|_| {
                    PdfError::other(format!(
                        "PDF content parser: invalid number `{s}` at byte {i}"
                    ))
                })?;
                self.operands.push(Operand::Number(f));
                i = end;
                continue;
            }
            // Anything else is a keyword (operator).
            let kw_end = scan_keyword_end(input, i);
            if kw_end == i {
                // Unrecognised single byte — skip to avoid infinite
                // loop.
                i += 1;
                continue;
            }
            let kw = &input[i..kw_end];
            self.dispatch(kw)?;
            i = kw_end;
        }
        Ok(())
    }
}

impl Frame {
    fn new() -> Self {
        Self {
            transform: Transform2D::identity(),
            children: Vec::new(),
            clip: None,
        }
    }

    fn is_effectively_empty(&self) -> bool {
        self.children.is_empty() && self.clip.is_none() && self.transform.is_identity()
    }
}

// ───────────────────────── helpers ─────────────────────────

fn is_whitespace(b: u8) -> bool {
    matches!(b, 0x00 | b'\t' | b'\n' | 0x0C | b'\r' | b' ')
}

fn is_delimiter(b: u8) -> bool {
    matches!(
        b,
        b'(' | b')' | b'<' | b'>' | b'[' | b']' | b'{' | b'}' | b'/' | b'%'
    )
}

fn rgb_from_unit(r: f32, g: f32, b: f32) -> Rgba {
    Rgba::opaque(unit_to_byte(r), unit_to_byte(g), unit_to_byte(b))
}

/// Convert a DeviceCMYK colour value to DeviceRGB per ISO 32000-1
/// §10.3.5 ("Conversion from DeviceCMYK to DeviceRGB"):
///
/// ```text
/// red   = 1.0 − min(1.0, cyan    + black)
/// green = 1.0 − min(1.0, magenta + black)
/// blue  = 1.0 − min(1.0, yellow  + black)
/// ```
///
/// The black component is added to each of the other components, which
/// are then converted to their complementary colours by subtracting
/// each from 1.0. No black generation or undercolour removal is
/// involved. Components are clamped into 0.0..=1.0 first so an
/// out-of-range operand cannot escape the 1.0 ceiling (§10.3.4 NOTE 4
/// applies the same nearest-valid-value substitution without error).
fn rgb_from_cmyk(cyan: f32, magenta: f32, yellow: f32, black: f32) -> Rgba {
    let c = cyan.clamp(0.0, 1.0);
    let m = magenta.clamp(0.0, 1.0);
    let y = yellow.clamp(0.0, 1.0);
    let k = black.clamp(0.0, 1.0);
    let red = 1.0 - (c + k).min(1.0);
    let green = 1.0 - (m + k).min(1.0);
    let blue = 1.0 - (y + k).min(1.0);
    rgb_from_unit(red, green, blue)
}

fn unit_to_byte(f: f32) -> u8 {
    (f.clamp(0.0, 1.0) * 255.0).round() as u8
}

fn compose(a: Transform2D, b: Transform2D) -> Transform2D {
    // PDF `cm` post-concatenates: new CTM = b * old CTM. In the
    // SVG/IR convention, group.transform applies to the children
    // *before* any parent transform — so when we encounter a `cm`
    // inside a frame whose existing transform is `a`, the resulting
    // group transform is `a * b`.
    Transform2D {
        a: a.a * b.a + a.c * b.b,
        b: a.b * b.a + a.d * b.b,
        c: a.a * b.c + a.c * b.d,
        d: a.b * b.c + a.d * b.d,
        e: a.a * b.e + a.c * b.f + a.e,
        f: a.b * b.e + a.d * b.f + a.f,
    }
}

fn scan_keyword_end(input: &[u8], start: usize) -> usize {
    let mut end = start;
    while end < input.len() && !is_whitespace(input[end]) && !is_delimiter(input[end]) {
        end += 1;
    }
    end
}

fn read_literal_string(input: &[u8], start: usize) -> Result<(usize, Vec<u8>), PdfError> {
    let mut end = start + 1;
    let mut depth = 1u32;
    let mut decoded = Vec::new();
    while end < input.len() {
        let b = input[end];
        if b == b'\\' {
            end += 1;
            if end < input.len() {
                end += 1;
            }
            continue;
        }
        if b == b'(' {
            depth += 1;
        }
        if b == b')' {
            depth -= 1;
            if depth == 0 {
                end += 1;
                return Ok((end, decoded));
            }
        }
        decoded.push(b);
        end += 1;
    }
    Err(PdfError::other(
        "PDF content parser: unterminated literal string",
    ))
}

fn read_hex_string(input: &[u8], start: usize) -> Result<usize, PdfError> {
    let mut end = start + 1;
    while end < input.len() {
        if input[end] == b'>' {
            return Ok(end + 1);
        }
        end += 1;
    }
    Err(PdfError::other(
        "PDF content parser: unterminated hex string",
    ))
}

fn read_number_array(input: &[u8], start: usize) -> (usize, Vec<f32>) {
    let mut end = start + 1;
    let mut nums = Vec::new();
    while end < input.len() && input[end] != b']' {
        if is_whitespace(input[end]) {
            end += 1;
            continue;
        }
        if matches!(input[end], b'+' | b'-' | b'.' | b'0'..=b'9') {
            let nstart = end;
            if matches!(input[end], b'+' | b'-') {
                end += 1;
            }
            let mut saw_dot = false;
            while end < input.len()
                && (input[end].is_ascii_digit() || (input[end] == b'.' && !saw_dot))
            {
                if input[end] == b'.' {
                    saw_dot = true;
                }
                end += 1;
            }
            if let Ok(s) = str::from_utf8(&input[nstart..end]) {
                if let Ok(f) = s.parse::<f32>() {
                    nums.push(f);
                }
            }
        } else {
            end += 1;
        }
    }
    if end < input.len() {
        end += 1;
    } // skip `]`
    (end, nums)
}

#[cfg(test)]
mod tests {
    use super::*;

    fn parse(input: &[u8]) -> Group {
        parse_content_stream(input).unwrap()
    }

    #[test]
    fn empty_content_yields_empty_group() {
        let g = parse(b"");
        assert!(g.children.is_empty());
        assert!(g.clip.is_none());
    }

    #[test]
    fn rect_fill_round_trips() {
        // The writer would emit something like:
        //   q 1 0 0 rg 10 10 m 110 10 l 110 60 l 10 60 l h f Q
        let bytes = b"q 1 0 0 rg 10 10 m 110 10 l 110 60 l 10 60 l h f Q\n";
        let root = parse(bytes);
        // One child group containing the path.
        assert_eq!(root.children.len(), 1);
        let Node::Group(g) = &root.children[0] else {
            panic!("expected group")
        };
        assert_eq!(g.children.len(), 1);
        let Node::Path(pn) = &g.children[0] else {
            panic!("expected path")
        };
        // 4 verts + close = 5 commands.
        assert_eq!(pn.path.commands.len(), 5);
        assert!(matches!(pn.path.commands[0], PathCommand::MoveTo(p) if (p.x - 10.0).abs() < 1e-3));
        assert!(matches!(pn.path.commands[4], PathCommand::Close));
        assert_eq!(pn.fill_rule, FillRule::NonZero);
        // Fill is solid red.
        match &pn.fill {
            Some(Paint::Solid(r)) => assert_eq!((r.r, r.g, r.b), (255, 0, 0)),
            other => panic!("unexpected fill: {other:?}"),
        }
        assert!(pn.stroke.is_none());
    }

    #[test]
    fn nested_q_groups_are_promoted_to_node_groups() {
        let bytes = b"q q 1 0 0 1 5 5 cm 0 0 m 10 10 l S Q Q\n";
        let root = parse(bytes);
        assert_eq!(root.children.len(), 1);
        let Node::Group(outer) = &root.children[0] else {
            panic!()
        };
        assert_eq!(outer.children.len(), 1);
        let Node::Group(inner) = &outer.children[0] else {
            panic!()
        };
        // Inner group has the cm transform.
        assert!(!inner.transform.is_identity());
        assert_eq!(inner.children.len(), 1);
    }

    #[test]
    fn rectangle_operator_re_expands_to_subpath() {
        // 10 20 30 40 re → subpath of M(10,20), L(40,20), L(40,60), L(10,60), h
        let bytes = b"q 0.5 0.5 0.5 rg 10 20 30 40 re f Q\n";
        let root = parse(bytes);
        let Node::Group(g) = &root.children[0] else {
            panic!()
        };
        let Node::Path(p) = &g.children[0] else {
            panic!()
        };
        assert_eq!(p.path.commands.len(), 5);
        assert!(
            matches!(p.path.commands[0], PathCommand::MoveTo(pp) if pp.x == 10.0 && pp.y == 20.0)
        );
        assert!(
            matches!(p.path.commands[1], PathCommand::LineTo(pp) if pp.x == 40.0 && pp.y == 20.0)
        );
        assert!(
            matches!(p.path.commands[2], PathCommand::LineTo(pp) if pp.x == 40.0 && pp.y == 60.0)
        );
        assert!(
            matches!(p.path.commands[3], PathCommand::LineTo(pp) if pp.x == 10.0 && pp.y == 60.0)
        );
        assert!(matches!(p.path.commands[4], PathCommand::Close));
    }

    #[test]
    fn cubic_curve_roundtrips() {
        let bytes = b"q 0 0 m 1 1 2 1 3 0 c S Q\n";
        let root = parse(bytes);
        let Node::Group(g) = &root.children[0] else {
            panic!()
        };
        let Node::Path(p) = &g.children[0] else {
            panic!()
        };
        assert!(matches!(
            p.path.commands[1],
            PathCommand::CubicCurveTo { c1, c2, end }
                if c1.x == 1.0 && c1.y == 1.0 && c2.x == 2.0 && c2.y == 1.0 && end.x == 3.0 && end.y == 0.0
        ));
    }

    #[test]
    fn fill_rule_evenodd_recognised() {
        let bytes = b"q 0 0 m 10 0 l 10 10 l h f* Q\n";
        let root = parse(bytes);
        let Node::Group(g) = &root.children[0] else {
            panic!()
        };
        let Node::Path(p) = &g.children[0] else {
            panic!()
        };
        assert_eq!(p.fill_rule, FillRule::EvenOdd);
    }

    #[test]
    fn cm_translate_lands_on_group_transform() {
        let bytes = b"q 1 0 0 1 100 200 cm 0 0 m 5 5 l S Q\n";
        let root = parse(bytes);
        let Node::Group(g) = &root.children[0] else {
            panic!()
        };
        assert!((g.transform.e - 100.0).abs() < 1e-3);
        assert!((g.transform.f - 200.0).abs() < 1e-3);
    }

    #[test]
    fn stroke_style_w_j_m_d_recorded() {
        let bytes = b"q 2.5 w 1 J 2 j 8 M [5 3] 1 d 0 0 0 RG 0 0 m 10 10 l S Q\n";
        let root = parse(bytes);
        let Node::Group(g) = &root.children[0] else {
            panic!()
        };
        let Node::Path(p) = &g.children[0] else {
            panic!()
        };
        let s = p.stroke.as_ref().expect("stroke set");
        assert!((s.width - 2.5).abs() < 1e-3);
        assert!(matches!(s.cap, LineCap::Round));
        assert!(matches!(s.join, LineJoin::Bevel));
        assert!((s.miter_limit - 8.0).abs() < 1e-3);
        let dash = s.dash.as_ref().expect("dash set");
        assert_eq!(dash.array, vec![5.0, 3.0]);
        assert!((dash.offset - 1.0).abs() < 1e-3);
    }

    #[test]
    fn clip_w_assigns_to_group_clip() {
        // Clip operator: build a path, hit `W`, then `n` to consume.
        let bytes =
            b"q 10 10 m 50 10 l 50 50 l 10 50 l h W n 0 0 0 rg 20 20 m 30 20 l 30 30 l h f Q\n";
        let root = parse(bytes);
        let Node::Group(g) = &root.children[0] else {
            panic!()
        };
        assert!(g.clip.is_some());
        // The triangle painted afterwards lives as a child node.
        assert_eq!(g.children.len(), 1);
    }

    /// §10.3.5 fundamental cases: pure inks convert to their RGB
    /// complements, and pure black yields RGB black.
    #[test]
    fn cmyk_pure_inks_convert_per_10_3_5() {
        // cyan=1 → red=1−min(1,1+0)=0, green=blue=1 → (0,255,255).
        assert_eq!(rgb_from_cmyk(1.0, 0.0, 0.0, 0.0), Rgba::opaque(0, 255, 255));
        // magenta=1 → (255,0,255).
        assert_eq!(rgb_from_cmyk(0.0, 1.0, 0.0, 0.0), Rgba::opaque(255, 0, 255));
        // yellow=1 → (255,255,0).
        assert_eq!(rgb_from_cmyk(0.0, 0.0, 1.0, 0.0), Rgba::opaque(255, 255, 0));
        // black=1 → every channel 1−min(1,0+1)=0 → (0,0,0).
        assert_eq!(rgb_from_cmyk(0.0, 0.0, 0.0, 1.0), Rgba::opaque(0, 0, 0));
        // all zero → white.
        assert_eq!(
            rgb_from_cmyk(0.0, 0.0, 0.0, 0.0),
            Rgba::opaque(255, 255, 255)
        );
    }

    /// The `min(1.0, comp + black)` ceiling caps the sum so an ink
    /// plus black never wraps past full saturation.
    #[test]
    fn cmyk_component_plus_black_clamps_at_one() {
        // cyan=0.7 black=0.7 → red=1−min(1,1.4)=0; green/blue=1−0.7=0.3.
        let r = rgb_from_cmyk(0.7, 0.0, 0.0, 0.7);
        assert_eq!(r.r, 0);
        assert_eq!(r.g, (0.3f32 * 255.0).round() as u8);
        assert_eq!(r.b, (0.3f32 * 255.0).round() as u8);
    }

    /// Out-of-range operands are clamped before the formula (§10.3.4
    /// NOTE 4 nearest-valid-value substitution).
    #[test]
    fn cmyk_out_of_range_operands_clamp() {
        // Negative and >1 operands behave as 0.0 / 1.0.
        assert_eq!(
            rgb_from_cmyk(-0.5, 2.0, 0.0, 0.0),
            rgb_from_cmyk(0.0, 1.0, 0.0, 0.0)
        );
    }

    /// End-to-end through the content parser: `k` sets the fill paint,
    /// `K` sets the stroke paint, both via the §10.3.5 conversion.
    #[test]
    fn k_and_upper_k_operators_apply_cmyk_conversion() {
        // Fill = pure cyan (0,255,255); stroke = pure magenta (255,0,255).
        let bytes = b"q 1 0 0 0 k 0 1 0 0 K 0 0 m 10 10 l 10 0 l h B Q\n";
        let root = parse(bytes);
        let Node::Group(g) = &root.children[0] else {
            panic!("expected group")
        };
        let Node::Path(p) = &g.children[0] else {
            panic!("expected path")
        };
        match &p.fill {
            Some(Paint::Solid(c)) => assert_eq!((c.r, c.g, c.b), (0, 255, 255)),
            other => panic!("unexpected fill: {other:?}"),
        }
        let s = p.stroke.as_ref().expect("stroke set");
        match &s.paint {
            Paint::Solid(c) => assert_eq!((c.r, c.g, c.b), (255, 0, 255)),
            other => panic!("unexpected stroke paint: {other:?}"),
        }
    }
}