retex_engine/

image.rs

//! Image resources: `\saveimageresource`, `\useimageresource`,
//! `\lastsavedimageresourceindex` and `\lastsavedimageresourcepages`.
//!
//! Behaviour established black-box against LuaTeX 1.22:
//!
//! * `\saveimageresource` takes, in any order, `width`, `height`, `depth`
//!   (dimensions), `attr {…}`, `named {…}`, `page n`, `colorspace n`,
//!   `userpassword {…}`, `ownerpassword {…}` and a page box keyword
//!   (`mediabox`, `cropbox`, `bleedbox`, `trimbox`, `artbox`), then the file
//!   name in braces.
//! * A resource's index (`\lastsavedimageresourceindex`,
//!   `\lastsavedboxresourceindex`, the number `\useimageresource` and
//!   `\useboxresource` take) is the PDF object number reserved for it, so
//!   `\the\lastsavedboxresourceindex\space 0 R` refers to the form (pgf's
//!   soft masks rely on this). `pdf.getobjtype` reports `ximage` or `xform`
//!   for it.
//! * `\useimageresource [width…] [height…] [depth…] n` appends a rule node
//!   of subtype 2 (shown as `\image(h+d)xw`) in any mode.
//! * Sizes given to either command are completed from the image's natural
//!   aspect ratio: a width alone sets height = width × h₀ / w₀; a height
//!   (plus depth) alone sets width = (height + depth) × w₀ / h₀.
//! * The image is written to the PDF when it is first shipped out, or at
//!   once with `\immediate`. Its file name, as found, is printed as
//!   `<path>` then: at once with `\immediate`, otherwise at the end of the
//!   page (after the page's `\write`s), in the order of first use.

use std::path::PathBuf;

use retex_image::{Info, PageBox};

use crate::Engine;
use crate::nodes::NodeRef;

/// Rule subtype of image nodes.
pub const IMAGE_RULE: u16 = 2;

/// A saved image resource.
#[derive(Clone, Debug)]
pub struct ImageResource {
    /// The file as found.
    pub path: PathBuf,
    /// The name as given (for messages).
    pub name: String,
    /// What the file says about itself.
    pub info: Info,
    /// The page (PDF), 1-based.
    pub page: u32,
    /// The page box (PDF).
    pub page_box: PageBox,
    /// Extra entries for the image dictionary (`attr {…}`).
    pub attr: Vec<u8>,
    /// Size as saved: width, height, depth.
    pub size: (i32, i32, i32),
    /// Natural size: width, height.
    pub natural: (i32, i32),
    /// The PDF object number reserved for the image.
    pub obj: i32,
    /// Whether it has been written (and its name printed).
    pub written: bool,
}

/// Explicit dimensions of `\saveimageresource`/`\useimageresource`.
#[derive(Clone, Copy, Debug, Default)]
struct Dims {
    width: Option<i32>,
    height: Option<i32>,
    depth: Option<i32>,
}

/// `a × b / c` rounded, without overflow.
fn scale(a: i32, b: i32, c: i32) -> i32 {
    if c == 0 {
        return 0;
    }
    let (a, b, c) = (i64::from(a), i64::from(b), i64::from(c));
    ((a * b + c / 2) / c) as i32
}

/// Completes `d` from the natural size `(w0, h0)`.
fn complete(d: Dims, (w0, h0): (i32, i32)) -> (i32, i32, i32) {
    let depth = d.depth.unwrap_or(0);
    match (d.width, d.height, d.depth) {
        (None, None, None) => (w0, h0, 0),
        (Some(w), Some(h), _) => (w, h, depth),
        (Some(w), None, _) => (w, scale(w, h0, w0) - depth, depth),
        (None, Some(h), _) => (scale(h + depth, w0, h0), h, depth),
        (None, None, Some(dp)) => (w0, h0 - dp, dp),
    }
}

impl Engine {
    /// Scans `width`/`height`/`depth` keywords.
    fn scan_image_dims(&mut self, d: &mut Dims) -> bool {
        if self.scan_keyword("width") {
            self.scan_normal_dimen();
            d.width = Some(self.cur_val);
        } else if self.scan_keyword("height") {
            self.scan_normal_dimen();
            d.height = Some(self.cur_val);
        } else if self.scan_keyword("depth") {
            self.scan_normal_dimen();
            d.depth = Some(self.cur_val);
        } else {
            return false;
        }
        true
    }

    /// `\saveimageresource` (`immediate`: write the image at once).
    pub fn save_image_resource(&mut self, immediate: bool) {
        self.work_begin(crate::profile::Work::Images);
        self.save_image_resource_inner(immediate);
        self.work_end();
    }

    fn save_image_resource_inner(&mut self, immediate: bool) {
        let mut dims = Dims::default();
        let mut attr = Vec::new();
        let mut page = 1u32;
        let mut page_box = PageBox::Crop;
        loop {
            if self.scan_image_dims(&mut dims) {
                continue;
            }
            if self.scan_keyword("attr") {
                let toks = self.scan_toks_expanded();
                attr = self.print_to_string(|e| e.show_token_list(&toks, 0, toks.len(), i64::MAX));
            } else if self.scan_keyword("named") {
                // Named destinations of the PDF are not supported; LuaTeX
                // takes the page by name. Read and ignore it.
                let _ = self.scan_toks_expanded();
            } else if self.scan_keyword("page") {
                self.scan_int();
                page = self.cur_val.max(1) as u32;
            } else if self.scan_keyword("colorspace") {
                self.scan_int();
            } else if self.scan_keyword("userpassword") || self.scan_keyword("ownerpassword") {
                let _ = self.scan_toks_expanded();
            } else if self.scan_keyword("mediabox") {
                page_box = PageBox::Media;
            } else if self.scan_keyword("cropbox") {
                page_box = PageBox::Crop;
            } else if self.scan_keyword("bleedbox") {
                page_box = PageBox::Bleed;
            } else if self.scan_keyword("trimbox") {
                page_box = PageBox::Trim;
            } else if self.scan_keyword("artbox") {
                page_box = PageBox::Art;
            } else {
                break;
            }
        }
        let toks = self.scan_toks_expanded();
        let name_bytes =
            self.print_to_string(|e| e.show_token_list(&toks, 0, toks.len(), i64::MAX));
        let name = String::from_utf8_lossy(&name_bytes).into_owned();
        let found = self.find_input_file(&name, retex_kpath::FileFormat::Tex);
        if found.is_some() {
            self.set_svg_font_finder();
        }
        let info = found
            .as_ref()
            .map(|p| retex_image::probe_file(p, page, page_box));
        if let Some(p) = &found
            && !self.files.watched.contains(p)
        {
            self.files.watched.push(p.clone());
        }
        let (path, info) = match (found, info) {
            (Some(p), Some(Ok(i))) => (p, i),
            (Some(p), Some(Err(e))) if e.to_string().contains("does not exist") => {
                // Observed: a page that the PDF file does not have stops
                // LuaTeX, after its version warning (found by fuzzing,
                // \includegraphics[page=3]).
                if let Ok(i) = retex_image::probe_file(&p, 1, page_box) {
                    self.image_version_check(&name, &i);
                }
                let category = format!("file {name}) (pdf inclusion");
                let msg = format!("required page '{page}' does not exist");
                if self.eqtb.int(crate::primitives_gen::assign_int::OUTPUTMODE) > 0 {
                    self.pdf_fatal_in(&category, &msg);
                }
                self.luatex_fatal(&category, &msg);
            }
            (_, Some(Err(e))) => {
                self.image_error(&name, &e.to_string());
                return;
            }
            _ => {
                self.image_error(&name, "file not found");
                return;
            }
        };
        if !self.image_version_check(&name, &info) {
            return;
        }
        for (cat, msg) in &info.warnings {
            self.luatex_warning(&format!("file {name}) ({cat}"), msg);
        }

        let natural = retex_image::natural_size(&info);
        let size = complete(dims, natural);
        let obj = self.pdf_alloc_obj(crate::pdfext::ObjKind::Ximage);
        self.last_image_pages = info.pages as i32;
        self.images.push(ImageResource {
            path,
            name,
            info,
            page,
            page_box,
            attr,
            size,
            natural,
            obj,
            written: false,
        });
        self.last_image_index = obj;
        if immediate && let Some(mut be) = self.backend.take() {
            be.write_image(self, obj as u32);
            self.backend = Some(be);
            self.mark_image_written(obj as u32);
            self.print_image_name(obj as u32);
            self.pdf.stream_written = true;
        }
    }

    /// LuaTeX compares the file's header version with the output's: a
    /// warning, or an error with `\pdfvariable inclusionerrorlevel` > 0
    /// (then `false`).
    fn image_version_check(&mut self, name: &str, info: &retex_image::Info) -> bool {
        if let Some(found) = info.pdf_version {
            let int = |i: u32| self.eqtb.int(i).max(0) as u32;
            let allowed = (
                int(crate::backend::MAJOR_VERSION),
                int(crate::backend::MINOR_VERSION),
            );
            if let Some(msg) = retex_image::pdf::version_message(found, allowed) {
                if self.eqtb.int(145) > 0 {
                    self.image_error(name, &msg);
                    return false;
                }
                self.luatex_warning(&format!("file {name}) (pdf inclusion"), &msg);
            }
        }
        true
    }

    fn image_error(&mut self, name: &str, why: &str) {
        self.print_err("Image file `");
        self.printer.print_str(name);
        self.printer.print_str("': ");
        self.printer.print_str(why);
        self.help(&["The image could not be read, so nothing was saved."]);
        self.error();
    }

    /// `\useimageresource`: appends an image node.
    pub fn use_image_resource(&mut self) {
        let mut dims = Dims::default();
        while self.scan_image_dims(&mut dims) {}
        self.scan_int();
        let n = self.cur_val;
        let Some(img) = self.image(n as u32) else {
            self.print_err("Invalid image resource index ");
            self.printer.print_int(i64::from(n));
            self.help(&["\\useimageresource needs a number from \\lastsavedimageresourceindex."]);
            self.error();
            return;
        };
        let (w, h, d) = if dims.width.is_none() && dims.height.is_none() && dims.depth.is_none() {
            img.size
        } else {
            complete(dims, img.natural)
        };
        let p = self.nodes.new_rule(w, h, d);
        self.nodes.get_mut(p).subtype = IMAGE_RULE;
        if self.eqtb.attrs_in_use {
            let a = self.cur_attr();
            self.nodes.set_attr(p, a);
        }
        if let crate::nodes::NodeData::Rule { index, .. } = &mut self.nodes.get_mut(p).data {
            *index = n;
        }
        self.tail_append(p);
    }

    /// Whether node `p` is an image.
    pub fn is_image(&self, p: NodeRef) -> bool {
        let n = self.nodes.get(p);
        n.subtype == IMAGE_RULE && matches!(n.data, crate::nodes::NodeData::Rule { .. })
    }

    /// The image resource with index (object number) `n`.
    pub fn image(&self, n: u32) -> Option<&ImageResource> {
        // Object numbers grow, so the list is sorted by them.
        let i = self
            .images
            .binary_search_by_key(&(n as i32), |r| r.obj)
            .ok()?;
        self.images.get(i)
    }

    /// Records that image `n` is written; true the first time.
    pub fn mark_image_written(&mut self, n: u32) -> bool {
        let Ok(i) = self.images.binary_search_by_key(&(n as i32), |r| r.obj) else {
            return false;
        };
        !std::mem::replace(&mut self.images[i].written, true)
    }

    /// Prints `<path>` for image `n`, as when it is written.
    pub fn print_image_name(&mut self, n: u32) {
        let Some(path) = self.image(n).map(|i| i.path.to_string_lossy().into_owned()) else {
            return;
        };
        // Printed like a C string (log lines may reach 80 characters).
        let mut text = b"<".to_vec();
        text.extend_from_slice(path.as_bytes());
        text.push(b'>');
        self.printer.print_lua_bytes(&text);
    }

    /// The resource number of image `n` (`/Im<k>`: the k-th image saved).
    pub fn image_name(&self, n: u32) -> u32 {
        self.images
            .binary_search_by_key(&(n as i32), |r| r.obj)
            .map_or(n, |i| i as u32 + 1)
    }

    /// The box resource with index (object number) `n` and its resource
    /// number (`/Fm<k>`: the k-th box resource saved).
    pub fn box_resource(&self, n: u32) -> Option<(&BoxResource, u32)> {
        let i = self.box_resource_slot(n)?;
        Some((&self.box_resources[i], i as u32 + 1))
    }

    /// The position in [`Engine::box_resources`] of the box resource with
    /// index (object number) `n`.
    pub fn box_resource_slot(&self, n: u32) -> Option<usize> {
        self.box_resources
            .binary_search_by_key(&(n as i32), |r| r.obj)
            .ok()
    }
}

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

    #[test]
    fn completing_sizes() {
        let nat = (26312704, 19734528); // 401.5pt x 301.125pt
        let d = |w, h, dp| Dims {
            width: w,
            height: h,
            depth: dp,
        };
        // LuaTeX: width 2cm → 56.9055pt x 42.67912pt.
        assert_eq!(
            complete(d(Some(3729359), None, None), nat),
            (3729359, 2797019, 0)
        );
        // height 5pt depth 1pt → 8pt wide.
        assert_eq!(
            complete(d(None, Some(327680), Some(65536)), nat),
            (524288, 327680, 65536)
        );
    }
}

/// Rule subtype of box-resource nodes (`\useboxresource`).
pub const BOX_RESOURCE_RULE: u16 = 1;

/// A saved box resource (`\saveboxresource`): a TeX box written as a PDF
/// form XObject when first used (or at once with `\immediate`).
#[derive(Clone, Debug)]
pub struct BoxResource {
    /// The box (taken from its register).
    pub list: NodeRef,
    /// Width, height and depth of the box.
    pub size: (i32, i32, i32),
    /// Extra entries for the form dictionary (`attr {…}`).
    pub attr: Vec<u8>,
    /// Extra entries for the form's resources (`resources {…}`).
    pub resources: Vec<u8>,
    /// Whether the form has been written.
    pub written: bool,
    /// The PDF object number reserved for the form (also its index).
    pub obj: i32,
}

impl Engine {
    /// `\saveboxresource [attr {…}] [resources {…}] <register>`
    /// (`immediate`: write the form at once).
    pub fn save_box_resource(&mut self, immediate: bool) {
        let mut attr = Vec::new();
        let mut resources = Vec::new();
        loop {
            if self.scan_keyword("attr") {
                let toks = self.scan_toks_expanded();
                attr = self.print_to_string(|e| e.show_token_list(&toks, 0, toks.len(), i64::MAX));
            } else if self.scan_keyword("resources") {
                let toks = self.scan_toks_expanded();
                resources =
                    self.print_to_string(|e| e.show_token_list(&toks, 0, toks.len(), i64::MAX));
            } else if self.scan_keyword("type") {
                self.scan_int();
            } else if self.scan_keyword("margin") {
                self.scan_normal_dimen();
            } else {
                break;
            }
        }
        self.scan_register_num();
        let n = self.cur_val as usize;
        let list = self.eqtb.boxes[n].val;
        if list.is_null() {
            self.print_err("Invalid \\box");
            self.help(&["\\saveboxresource needs a box register that is not void."]);
            self.error();
            return;
        }
        self.eqtb.boxes[n].val = NodeRef::NULL;
        let obj = self.pdf_alloc_obj(crate::pdfext::ObjKind::Xform);
        let size = self
            .nodes
            .box_data(list)
            .map_or((0, 0, 0), |b| (b.width, b.height, b.depth));
        self.box_resources.push(BoxResource {
            list,
            size,
            attr,
            resources,
            written: false,
            obj,
        });
        self.last_box_index = obj;
        if immediate && let Some(be) = self.backend.take() {
            let be = self.write_box_resource(be, obj as u32);
            self.pdf.stream_written = true;
            self.backend = Some(be);
        }
    }

    /// `\useboxresource [width…] [height…] [depth…] n`.
    pub fn use_box_resource(&mut self) {
        let mut dims = Dims::default();
        while self.scan_image_dims(&mut dims) {}
        self.scan_int();
        let n = self.cur_val;
        let Some(res) = u32::try_from(n)
            .ok()
            .and_then(|n| self.box_resource_slot(n))
            .map(|i| &self.box_resources[i])
        else {
            self.print_err("Invalid box resource index ");
            self.printer.print_int(i64::from(n));
            self.help(&["\\useboxresource needs a number from \\lastsavedboxresourceindex."]);
            self.error();
            return;
        };
        let (w0, h0, d0) = res.size;
        // A given dimension scales the others in proportion.
        let (w, h, d) = match (dims.width, dims.height, dims.depth) {
            (None, None, None) => (w0, h0, d0),
            (Some(w), Some(h), Some(d)) => (w, h, d),
            (Some(w), h, d) => (
                w,
                h.unwrap_or_else(|| scale(h0, w, w0)),
                d.unwrap_or_else(|| scale(d0, w, w0)),
            ),
            (None, Some(h), d) => (scale(w0, h, h0), h, d.unwrap_or_else(|| scale(d0, h, h0))),
            (None, None, Some(d)) => (scale(w0, d, d0), scale(h0, d, d0), d),
        };
        let p = self.nodes.new_rule(w, h, d);
        self.nodes.get_mut(p).subtype = BOX_RESOURCE_RULE;
        if self.eqtb.attrs_in_use {
            let a = self.cur_attr();
            self.nodes.set_attr(p, a);
        }
        if let crate::nodes::NodeData::Rule { index, .. } = &mut self.nodes.get_mut(p).data {
            *index = n;
        }
        self.tail_append(p);
    }
}

/// What the default output mode adds to package files, before their final
/// `\endinput`: the file name, a text that identifies the file, and the
/// addition. `--compat=lualatex` reads the files as they are.
///
/// * graphicx's driver `luatex.def`: `.svg` files are included as PDF
///   pages (`retex-image` converts them), and found after the other
///   extensions.
/// * The `svg` package: `\includesvg` includes the SVG file itself with
///   `\includegraphics` and the package's size and angle options, without
///   Inkscape; `inkscapelatex` is ignored (the SVG file's own text is
///   drawn).
const PACKAGE_ADDITIONS: &[(&str, &[u8], &[u8])] = &[
    (
        "luatex.def",
        b"\\def\\Gin@extensions",
        b"\\@namedef{Gin@rule@.svg}#1{{pdf}{.svg}{#1}}%
\\@namedef{Gin@rule@.SVG}#1{{pdf}{.SVG}{#1}}%
\\edef\\Gin@extensions{\\Gin@extensions,.svg,.SVG}%
",
    ),
    (
        "svg.sty",
        b"\\newcommand*\\includesvg",
        b"\\renewcommand*\\includesvg[2][]{%
  \\begingroup
    \\svg@deprecated@param
    \\svg@local@param@set{#1}%
    \\@svg@ink@latexfalse
    \\svg@get@path{#2}{}%
    \\if@svg@file@found
      \\svg@@input{\\svg@file@base.\\svg@file@ext}%
    \\else
      \\svg@file@missing[\\svg@file@ext]{\\svg@file@base}{}%
    \\fi
  \\endgroup
}%
",
    ),
];

impl Engine {
    /// The text of an input file, with its [`PACKAGE_ADDITIONS`] in the
    /// default output mode.
    pub(crate) fn extend_package(&self, path: &std::path::Path, data: Vec<u8>) -> Vec<u8> {
        if self.diagnostics.is_none() {
            return data;
        }
        let Some(&(_, _, addition)) = PACKAGE_ADDITIONS.iter().find(|(name, mark, _)| {
            path.file_name() == Some(std::ffi::OsStr::new(name))
                && data.windows(mark.len()).any(|w| w == *mark)
        }) else {
            return data;
        };
        let Some(at) = data
            .windows(b"\\endinput".len())
            .rposition(|w| w == b"\\endinput")
            .filter(|&i| i == 0 || data[i - 1] == b'\n')
        else {
            return data;
        };
        [&data[..at], addition, &data[at..]].concat()
    }
}

impl Engine {
    /// Before the first image: SVG text finds its fonts by family name in
    /// the font name database (the one of luaotfload's `name:` lookups,
    /// scanned when first needed).
    fn set_svg_font_finder(&mut self) {
        if self.svg_fonts_set {
            return;
        }
        self.svg_fonts_set = true;
        let Some(k) = &self.kpse else {
            return;
        };
        let mut k = k.borrow_mut();
        let dirs = retex_font::loader::name_dirs(&mut k);
        let cache = k.cache_dir().map(std::path::Path::to_path_buf);
        retex_image::svg::set_font_finder(std::sync::Arc::new(move |family| {
            retex_font::names::NameDatabase::shared(dirs.clone(), cache.as_deref())
                .family_files(family)
                .into_iter()
                .map(std::path::Path::to_path_buf)
                .collect()
        }));
    }
}