mirror of
https://github.com/storytold/vectorcraft.git
synced 2026-10-26 18:56:55 -06:00
The engine built the PDF options without a creation date, so the writer dated every PDF and PDF-compatible .ai file with the time it was written. A document created on 2020-01-01 (File › File Info shows the date) exported as a PDF with today's date in /CreationDate and xmp:CreateDate. SVG export with the metadata option already writes the document's own date. The options now carry the document's File Info created date and the time of writing as the modification date. A document without a creation date is dated with the time of writing, as before, so PDF/X files still get their required date. The PDF writer writes one date for both, so the export then sets /ModDate, xmp:ModifyDate and the PDF/A history dates to the time of writing; /CreationDate and xmp:CreateDate keep the document's date. The page information mark and the embedded editing data's modification date keep the time of writing. Docs: docs/mcp.md (File Info), docs/target-app-parity.md.
176 lines
8.5 KiB
Rust
176 lines
8.5 KiB
Rust
//! What the export does to the file once it is written, in this order: the pages' thumbnails
|
||
//! (General › Embed page thumbnails), the seal of the editing data, fast web view (a linearised
|
||
//! file, [`crate::linearize`]) and encryption ([`crate::encrypt`]), which a linearised file goes
|
||
//! through between its two steps so that it stays linearised. Also here: what makes a file say it
|
||
//! is PDF 1.3 ([`declare_pdf13`]), for PDF 1.3 and PDF/X-1a and PDF/X-3 files, the file's
|
||
//! modification date ([`modification_date`]) and the pages' user units ([`user_units`]), which the
|
||
//! export writes in that order before the thumbnails.
|
||
|
||
use std::borrow::Cow;
|
||
|
||
use crate::encrypt::Obj;
|
||
use crate::linearize::{self, Parsed};
|
||
use crate::patch::Patch;
|
||
use crate::{PdfError, PdfOptions, PdfSettings};
|
||
|
||
/// Longest side of a page thumbnail, in pixels.
|
||
pub const THUMBNAIL_SIZE: u32 = 106;
|
||
|
||
/// A page thumbnail: 8-bit RGB pixels, row by row, at most [`THUMBNAIL_SIZE`] pixels a side.
|
||
#[derive(Clone, Debug, Default, PartialEq)]
|
||
pub struct Thumbnail {
|
||
pub width: u32,
|
||
pub height: u32,
|
||
pub rgb: Vec<u8>,
|
||
}
|
||
|
||
impl Thumbnail {
|
||
/// The image XObject of a page's `/Thumb` entry; `gray`: in DeviceGray (for PDF/X files, whose
|
||
/// device colours must suit the printing condition, an RGB thumbnail wouldn't).
|
||
fn object(&self, gray: bool) -> Result<Vec<u8>, PdfError> {
|
||
let (w, h) = (self.width, self.height);
|
||
let sized = (1..=THUMBNAIL_SIZE).contains(&w) && (1..=THUMBNAIL_SIZE).contains(&h);
|
||
if !sized || self.rgb.len() as u64 != u64::from(w) * u64::from(h) * 3 {
|
||
return Err(PdfError::BadSetting(format!("a page thumbnail must be RGB pixels, 1 to {THUMBNAIL_SIZE} a side (got {w} × {h})")));
|
||
}
|
||
let (space, pixels) = if gray {
|
||
// Rec. 601 luma.
|
||
let luma = |p: &[u8; 3]| ((299 * u32::from(p[0]) + 587 * u32::from(p[1]) + 114 * u32::from(p[2]) + 500) / 1000) as u8;
|
||
("DeviceGray", Cow::Owned(self.rgb.as_chunks::<3>().0.iter().map(luma).collect()))
|
||
} else {
|
||
("DeviceRGB", Cow::Borrowed(self.rgb.as_slice()))
|
||
};
|
||
let data = crate::output::deflate(&pixels).map_err(|e| PdfError::Write(format!("page thumbnail: {e}")))?;
|
||
let mut out = format!("<</Width {w}/Height {h}/ColorSpace/{space}/BitsPerComponent 8/Filter/FlateDecode/Length {}>>\nstream\n", data.len())
|
||
.into_bytes();
|
||
out.extend_from_slice(&data);
|
||
out.extend_from_slice(b"\nendstream");
|
||
Ok(out)
|
||
}
|
||
}
|
||
|
||
/// `pdf` with `thumbs` as the thumbnails of its pages, in page order (in grey when `gray`).
|
||
fn thumbnails(pdf: Vec<u8>, thumbs: &[Thumbnail], gray: bool) -> Result<Vec<u8>, PdfError> {
|
||
let failed = |why: &str| PdfError::Write(format!("can't add page thumbnails: {why}"));
|
||
let Parsed { xref, pages, .. } = linearize::parse(&pdf).map_err(failed)?;
|
||
let mut patch = Patch::new(&xref);
|
||
for (page, thumb) in pages.iter().zip(thumbs) {
|
||
let (dict, _) = xref.dict(&pdf, *page).ok_or_else(|| failed("a page can't be read"))?;
|
||
let n = patch.add_object(thumb.object(gray)?);
|
||
patch.replace(dict, dict, format!("/Thumb {n} 0 R").into_bytes());
|
||
}
|
||
patch.apply(&pdf, &xref)
|
||
}
|
||
|
||
/// `pdf` with `units` as the `UserUnit` entries of its pages, in page order (a unit of 1 writes
|
||
/// none); unchanged when every unit is 1.
|
||
pub(crate) fn user_units(pdf: Vec<u8>, units: &[f64]) -> Result<Vec<u8>, PdfError> {
|
||
if units.iter().all(|u| *u == 1.0) {
|
||
return Ok(pdf);
|
||
}
|
||
let failed = |why: &str| PdfError::Write(format!("can't write the pages' user units: {why}"));
|
||
let Parsed { xref, pages, .. } = linearize::parse(&pdf).map_err(failed)?;
|
||
let mut patch = Patch::new(&xref);
|
||
for (page, unit) in pages.iter().zip(units).filter(|(_, u)| **u != 1.0) {
|
||
let (dict, _) = xref.dict(&pdf, *page).ok_or_else(|| failed("a page can't be read"))?;
|
||
patch.replace(dict, dict, format!("/UserUnit {unit}").into_bytes());
|
||
}
|
||
patch.apply(&pdf, &xref)
|
||
}
|
||
|
||
/// `pdf` dated as modified at `modified` (Unix seconds, UTC), its creation dates kept: the
|
||
/// document information's `/ModDate` (a PDF date, ISO 32000-1 §7.9.4) and every `xmp:ModifyDate`,
|
||
/// `xmp:MetadataDate` and `stEvt:when` of the catalog's XMP metadata (ISO 8601 dates), in UTC as
|
||
/// the PDF writer writes the creation date. The writer gives `/ModDate`, `xmp:ModifyDate` and
|
||
/// `stEvt:when` the creation date.
|
||
pub(crate) fn modification_date(pdf: Vec<u8>, modified: i64) -> Result<Vec<u8>, PdfError> {
|
||
let failed = |why: &str| PdfError::Write(format!("can't write the file's modification date: {why}"));
|
||
let Parsed { xref, objects, trailer, root, .. } = linearize::parse(&pdf).map_err(failed)?;
|
||
let [y, mo, d, h, mi, s] = vectorcraft_doc::metadata::civil(modified);
|
||
let mut patch = Patch::new(&xref);
|
||
if let Some(&Obj::Ref(n, _)) = trailer.get(b"Info")
|
||
&& let Some(Obj::Str { start, end, .. }) = objects.get(&n).and_then(|o| o.value.get(b"ModDate"))
|
||
{
|
||
patch.replace(*start, *end, format!("(D:{y:04}{mo:02}{d:02}{h:02}{mi:02}{s:02}Z)").into_bytes());
|
||
}
|
||
let metadata = match objects.get(&root).and_then(|c| c.value.get(b"Metadata")) {
|
||
Some(&Obj::Ref(n, _)) => objects.get(&n),
|
||
_ => None,
|
||
};
|
||
if let Some(o) = metadata
|
||
&& let Some(&Obj::Int { value, start, end }) = o.value.get(b"Length")
|
||
{
|
||
let date = format!("{y:04}-{mo:02}-{d:02}T{h:02}:{mi:02}:{s:02}+00:00");
|
||
let xmp = pdf.get(..o.endobj).unwrap_or_default();
|
||
let mut grown = 0;
|
||
for tag in ["xmp:ModifyDate", "xmp:MetadataDate", "stEvt:when"] {
|
||
let (open, close) = (format!("<{tag}>"), format!("</{tag}>"));
|
||
let mut from = o.value_span.1;
|
||
while let Some(at) = crate::lab_spot::find(xmp, open.as_bytes(), from).map(|i| i + open.len())
|
||
&& let Some(to) = crate::lab_spot::find(xmp, close.as_bytes(), at)
|
||
{
|
||
grown += date.len() as i64 - (to - at) as i64;
|
||
patch.replace(at, to, date.clone().into_bytes());
|
||
from = to + close.len();
|
||
}
|
||
}
|
||
if grown != 0 {
|
||
patch.replace(start, end, (value + grown).to_string().into_bytes());
|
||
}
|
||
}
|
||
patch.apply(&pdf, &xref)
|
||
}
|
||
|
||
/// Linearise `pdf`, encrypting it on the way as `set` says.
|
||
fn linearised(pdf: &[u8], set: &PdfSettings) -> Result<Vec<u8>, PdfError> {
|
||
let (prepared, plan) = linearize::prepare(pdf)?;
|
||
let (encrypted, mut cipher) = crate::encrypt::protect_keeping(prepared, set)?;
|
||
linearize::write(&encrypted, &plan, cipher.as_mut())
|
||
}
|
||
|
||
/// The file the export wrote (`pdf`), finished as `opts` say; `editing`: it carries the editing
|
||
/// data, to seal once the pages are final. What can't be done comes back in `warnings`.
|
||
pub(crate) fn finish(pdf: Vec<u8>, opts: &PdfOptions, editing: bool, warnings: &mut Vec<String>) -> Result<Vec<u8>, PdfError> {
|
||
let set = &opts.settings;
|
||
let pdf = match (set.thumbnails, opts.thumbnails.is_empty()) {
|
||
(true, false) => thumbnails(pdf, &opts.thumbnails, set.standard.is_pdfx())?,
|
||
(true, true) => {
|
||
warnings.push("page thumbnails need the pages drawn, which weren't given: none are embedded".into());
|
||
pdf
|
||
}
|
||
(false, _) => pdf,
|
||
};
|
||
let pdf = if editing { crate::editing::seal(pdf)? } else { pdf };
|
||
if !set.fast_web_view {
|
||
return crate::encrypt::protect(pdf, set);
|
||
}
|
||
match linearised(&pdf, set) {
|
||
Ok(out) => Ok(out),
|
||
Err(e) => {
|
||
warnings.push(format!("the file isn't optimized for fast web view: {e}"));
|
||
crate::encrypt::protect(pdf, set)
|
||
}
|
||
}
|
||
}
|
||
|
||
/// Make a file written with the PDF 1.4 settings say it is PDF 1.3: its header and the version
|
||
/// its XMP metadata gives, rewritten in place.
|
||
pub(crate) fn declare_pdf13(pdf: &mut [u8]) {
|
||
if let Some(v) = pdf.get_mut(..8).filter(|h| h.starts_with(b"%PDF-1.4")) {
|
||
v.copy_from_slice(b"%PDF-1.3");
|
||
}
|
||
if let Some(digit) = metadata_version(pdf).and_then(|at| pdf.get_mut(at)) {
|
||
*digit = b'3';
|
||
}
|
||
}
|
||
|
||
/// Where the minor version digit is in the `1.4` of the XMP metadata stream of `pdf`'s catalog.
|
||
fn metadata_version(pdf: &[u8]) -> Option<usize> {
|
||
const VERSION: &[u8] = b"<pdf:PDFVersion>1.4<";
|
||
let p = linearize::parse(pdf).ok()?;
|
||
let Some(&Obj::Ref(n, _)) = p.objects.get(&p.root)?.value.get(b"Metadata") else { return None };
|
||
let o = p.objects.get(&n)?;
|
||
let at = crate::lab_spot::find(pdf.get(..o.endobj)?, VERSION, o.value_span.1)?;
|
||
Some(at + VERSION.len() - 2)
|
||
}
|