ZacharyZhang-NY
b7fca0146d
ElyDoc, ElyTable, ElyShow (LibreOfficeKit), ElyPDF (MuPDF), ElyTree (Git), ElyMark (Milkdown and CodeMirror in CEF). ElyMail and ElyAgent follow.
106 lines
3.4 KiB
Rust
106 lines
3.4 KiB
Rust
//! Removal and cleaning into new files (PDF-09, PDF-10).
|
|
|
|
use std::{io, path::Path};
|
|
|
|
use serde_json::{Value, json};
|
|
|
|
use crate::client::{Doc, Worker};
|
|
use crate::proto::tag;
|
|
|
|
/// A region to remove, in page points.
|
|
#[derive(Clone, Debug, PartialEq)]
|
|
pub struct Region {
|
|
pub page: usize,
|
|
pub rect: [f32; 4],
|
|
}
|
|
|
|
/// The written file's check: zeros, and one version.
|
|
#[derive(Clone, Debug, PartialEq, Eq)]
|
|
pub struct Redacted {
|
|
pub regions: usize,
|
|
/// Characters still inside a region.
|
|
pub text: usize,
|
|
/// Pixels inside a region that are not black.
|
|
pub light: usize,
|
|
/// Drawings still inside a region.
|
|
pub objects: usize,
|
|
pub versions: usize,
|
|
}
|
|
|
|
/// What cleaning reaches, or what is left after.
|
|
#[derive(Clone, Debug, Default, PartialEq, Eq)]
|
|
pub struct Reach {
|
|
/// Info entries, plus one for an XMP stream.
|
|
pub meta: usize,
|
|
pub annots: usize,
|
|
pub files: usize,
|
|
/// Pages with hidden text or layers.
|
|
pub hidden: usize,
|
|
pub scripts: usize,
|
|
pub hidden_annots: usize,
|
|
}
|
|
|
|
/// What to clean.
|
|
#[derive(Clone, Copy, Debug, Default, PartialEq, Eq)]
|
|
pub struct Cleaning {
|
|
pub meta: bool,
|
|
pub annots: bool,
|
|
pub files: bool,
|
|
/// Invisible text, scripts and hidden annotations.
|
|
pub hidden: bool,
|
|
}
|
|
|
|
impl Cleaning {
|
|
fn mask(self) -> u64 {
|
|
u64::from(self.meta) | u64::from(self.annots) << 1 | u64::from(self.files) << 2 | u64::from(self.hidden) << 3
|
|
}
|
|
}
|
|
|
|
fn count(value: &Value, key: &str) -> io::Result<usize> {
|
|
let n = value[key].as_u64().ok_or_else(|| io::Error::other(format!("a bad {key} in {value}")))?;
|
|
usize::try_from(n).map_err(io::Error::other)
|
|
}
|
|
|
|
fn reach(bytes: &[u8]) -> io::Result<Reach> {
|
|
let value: Value = serde_json::from_slice(bytes).map_err(io::Error::other)?;
|
|
Ok(Reach {
|
|
meta: count(&value, "meta")?,
|
|
annots: count(&value, "annots")?,
|
|
files: count(&value, "files")?,
|
|
hidden: count(&value, "hidden")?,
|
|
scripts: count(&value, "scripts")?,
|
|
hidden_annots: count(&value, "hidden_annots")?,
|
|
})
|
|
}
|
|
|
|
fn utf8(path: &Path) -> io::Result<&str> {
|
|
path.to_str().ok_or_else(|| io::Error::other("a path that is not UTF-8"))
|
|
}
|
|
|
|
impl Worker {
|
|
/// Writes `out` without `regions`; an unclean result is an error.
|
|
pub fn redact(&mut self, doc: Doc, regions: &[Region], out: &Path) -> io::Result<Redacted> {
|
|
let spec: Vec<Value> = regions.iter().map(|r| json!({ "page": r.page, "rect": r.rect })).collect();
|
|
let ask = self.on(&doc)?.str(&Value::Array(spec).to_string()).str(utf8(out)?);
|
|
let value: Value = serde_json::from_slice(&self.request(tag::REDACT, &ask.bytes())?).map_err(io::Error::other)?;
|
|
Ok(Redacted {
|
|
regions: count(&value, "regions")?,
|
|
text: count(&value, "text")?,
|
|
light: count(&value, "light")?,
|
|
objects: count(&value, "objects")?,
|
|
versions: count(&value, "versions")?,
|
|
})
|
|
}
|
|
|
|
/// Counts what cleaning reaches; `scratch` holds a passing copy.
|
|
pub fn clean_scan(&mut self, doc: Doc, scratch: &Path) -> io::Result<Reach> {
|
|
reach(&self.request(tag::CLEAN_SCAN, &self.whole(&doc)?.str(utf8(scratch)?).bytes())?)
|
|
}
|
|
|
|
/// Writes `out` cleaned; what is left in it.
|
|
pub fn clean(&mut self, doc: Doc, what: Cleaning, out: &Path) -> io::Result<Reach> {
|
|
let ask = self.whole(&doc)?.u64(what.mask()).str(utf8(out)?);
|
|
reach(&self.request(tag::CLEAN, &ask.bytes())?)
|
|
}
|
|
}
|