Decode images in parallel

This commit is contained in:
Marc Plano-Lesay 2024-10-11 16:04:27 +11:00
parent c968d9086d
commit 04f03b4efc

View file

@ -1,4 +1,5 @@
use clap::{Parser, ValueHint};
use image::DynamicImage;
use log::info;
use pdf_writer::{Content, Filter, Finish, Name, Pdf, Rect, Ref};
use rayon::prelude::*;
@ -56,20 +57,37 @@ struct ImageFile {
pub data: Vec<u8>,
}
struct DecodedImageFile {
pub name: String,
pub data: Vec<u8>,
pub image: DynamicImage,
}
impl From<&ImageFile> for DecodedImageFile {
fn from(value: &ImageFile) -> Self {
let image = image::load_from_memory(&value.data).unwrap();
Self {
name: value.name.clone(),
data: value.data.clone(),
image,
}
}
}
fn convert_cbz(cbz_path: &Path, output_dir: &Path) -> Result<(), Box<dyn std::error::Error>> {
info!("Converting {:?}", cbz_path);
let a4 = Rect::new(0.0, 0.0, 595.0, 842.0);
let mut zip = ZipArchive::new(File::open(cbz_path)?)?;
let mut images = Vec::new();
let mut files = Vec::new();
for i in 0..zip.len() {
let mut file = zip.by_index(i)?;
let mut image_data = Vec::new();
let name = file.enclosed_name().expect("Failed to read file name");
if name.extension() == Some(OsStr::new("jpg")) {
file.read_to_end(&mut image_data)?;
images.push(ImageFile {
files.push(ImageFile {
name: name
.file_name()
.expect("Failed to read file name")
@ -80,6 +98,11 @@ fn convert_cbz(cbz_path: &Path, output_dir: &Path) -> Result<(), Box<dyn std::er
}
}
let mut images = Vec::new();
files
.par_iter()
.map(|f| DecodedImageFile::from(f))
.collect_into_vec(&mut images);
images.par_sort_by_key(|img| img.name.clone());
let mut pdf = Pdf::new();
@ -98,7 +121,6 @@ fn convert_cbz(cbz_path: &Path, output_dir: &Path) -> Result<(), Box<dyn std::er
pages.push(page_id);
let mut page = pdf.page(page_id);
let image_name = Name(b"Im1");
let dynamic = image::load_from_memory(&image.data).unwrap();
page.media_box(a4);
page.parent(page_tree_id);
@ -108,8 +130,8 @@ fn convert_cbz(cbz_path: &Path, output_dir: &Path) -> Result<(), Box<dyn std::er
let mut pdf_image = pdf.image_xobject(image_id, &image.data);
pdf_image.filter(Filter::DctDecode);
pdf_image.width(dynamic.width() as i32);
pdf_image.height(dynamic.height() as i32);
pdf_image.width(image.image.width() as i32);
pdf_image.height(image.image.height() as i32);
pdf_image.color_space().device_rgb();
pdf_image.bits_per_component(8);
pdf_image.finish();