standard-mdbooks/epub2mdbook
standard-mdbooks/epub2mdbook
1
0
Fork 0
mirror of https://github.com/cyborg42/epub2mdbook.git synced 2026-02-10 16:24:50 -05:00
Code Issues Projects Releases Packages Wiki Activity Actions

deps: update

Browse source
This commit is contained in:
Maverick Liu 2026-01-06 19:05:13 +08:00
parent 2bf293c4d7
commit 6fbd50bb1b
4 changed files with 278 additions and 1912 deletions
Download patch file Download diff file Expand all files Collapse all files

2098
Cargo.lock generated
View file

File diff suppressed because it is too large Load diff

16
Cargo.toml
View file

@ -1,6 +1,6 @@
[package] [package]
name = "epub2mdbook" name = "epub2mdbook"
version = "0.15.0" version = "0.16.0"
edition = "2024" edition = "2024"
description = "A tool to convert EPUB files to MDBook format" description = "A tool to convert EPUB files to MDBook format"
authors = ["Maverick Liu <maverick.liu42@gmail.com>"] authors = ["Maverick Liu <maverick.liu42@gmail.com>"]
@ -10,10 +10,10 @@ keywords = ["epub", "mdbook", "converter", "ebook"]
categories = ["command-line-utilities", "text-processing"] categories = ["command-line-utilities", "text-processing"]
[dependencies] [dependencies]
clap = { version = "4.5.30", features = ["derive"] } clap = { version = "4.5.54", features = ["derive"] }
epub = "2.1.1" epub = "2.1.5"
htmd = "0.1.6" htmd = "0.5.0"
mdbook = "0.4.45" mdbook-core = "0.5.2"
regex = "1.11.1" regex = "1.12.2"
thiserror = "2.0.11" thiserror = "2.0.17"
toml = "0.8.20" toml = "0.9.10"

71
src/lib.rs
View file

@ -2,7 +2,7 @@ pub mod error;
use epub::doc::{EpubDoc, NavPoint}; use epub::doc::{EpubDoc, NavPoint};
use error::Error; use error::Error;
use mdbook::config::BookConfig; use mdbook_core::config::BookConfig;
use regex::{Captures, Regex}; use regex::{Captures, Regex};
use std::collections::HashMap; use std::collections::HashMap;
use std::ffi::OsStr; use std::ffi::OsStr;
@ -15,20 +15,21 @@ use std::{fs, io};
/// ///
/// # Arguments /// # Arguments
/// ///
/// * `epub_path` - The path to the EPUB file /// * `epub_path` - Path to the input EPUB file
/// * `output_dir` - The path to the output directory /// * `output_dir` - Path to the output directory
/// * `with_file_name` - Whether to use the file name as the output directory /// * `create_subdir` - If `true`, creates a subdirectory named after the EPUB file
/// (e.g., `output_dir/book_name/`). If `false`, outputs directly to `output_dir`.
pub fn convert_epub_to_mdbook( pub fn convert_epub_to_mdbook(
epub_path: impl AsRef<Path>, epub_path: impl AsRef<Path>,
output_dir: impl AsRef<Path>, output_dir: impl AsRef<Path>,
with_file_name: bool, create_subdir: bool,
) -> Result<(), Error> { ) -> Result<(), Error> {
let epub_path = epub_path.as_ref(); let epub_path = epub_path.as_ref();
if !epub_path.is_file() { if !epub_path.is_file() {
return Err(Error::NotAFile(epub_path.display().to_string())); return Err(Error::NotAFile(epub_path.display().to_string()));
} }
let mut output_dir = output_dir.as_ref().to_owned(); let mut output_dir = output_dir.as_ref().to_owned();
if with_file_name { if create_subdir {
let book_name = epub_path let book_name = epub_path
.with_extension("") .with_extension("")
.file_name() .file_name()
@ -80,7 +81,7 @@ fn epub_nav_to_md(
pub fn generate_summary_md<R: Read + Seek>( pub fn generate_summary_md<R: Read + Seek>(
epub_doc: &EpubDoc<R>, epub_doc: &EpubDoc<R>,
) -> (String, HashMap<PathBuf, PathBuf>) { ) -> (String, HashMap<PathBuf, PathBuf>) {
let title = epub_doc.metadata.get("title").and_then(|v| v.first()); let title = epub_doc.get_title();
let mut summary_md = if let Some(title) = title { let mut summary_md = if let Some(title) = title {
format!("# {}\n\n", title) format!("# {}\n\n", title)
} else { } else {
@ -89,8 +90,10 @@ pub fn generate_summary_md<R: Read + Seek>(
let html_to_md = epub_doc let html_to_md = epub_doc
.resources .resources
.iter() .iter()
.filter(|(_, (_, mime))| ["application/xhtml+xml", "text/html"].contains(&&**mime)) .filter(|(_, resource)| {
.map(|(_, (path, _))| (path.clone(), path.with_extension("md"))) ["application/xhtml+xml", "text/html"].contains(&resource.mime.as_str())
})
.map(|(_, resource)| (resource.path.clone(), resource.path.with_extension("md")))
.collect::<HashMap<PathBuf, PathBuf>>(); .collect::<HashMap<PathBuf, PathBuf>>();
for nav in &epub_doc.toc { for nav in &epub_doc.toc {
if let Some(md) = epub_nav_to_md(nav, 0, &html_to_md) { if let Some(md) = epub_nav_to_md(nav, 0, &html_to_md) {
@ -110,12 +113,13 @@ fn extract_chapters_and_resources<R: Read + Seek>(
.filter_map(|(k, v)| Some((k.file_name()?, v.file_name()?))) .filter_map(|(k, v)| Some((k.file_name()?, v.file_name()?)))
.collect::<HashMap<_, _>>(); .collect::<HashMap<_, _>>();
let src_dir = output_dir.as_ref().join("src"); let src_dir = output_dir.as_ref().join("src");
for (_, (path, _)) in epub_doc.resources.clone().into_iter() { for (_, resource) in epub_doc.resources.clone() {
let mut content = match epub_doc.get_resource_by_path(&path) { let path = &resource.path;
let mut content = match epub_doc.get_resource_by_path(path) {
Some(content) => content, Some(content) => content,
None => continue, // unreachable None => continue, // unreachable
}; };
let target_path = if let Some(md_path) = html_to_md.get(&path) { let target_path = if let Some(md_path) = html_to_md.get(path) {
// html file, convert to md // html file, convert to md
let html = String::from_utf8(content.clone())?; let html = String::from_utf8(content.clone())?;
let markdown = htmd::convert(&html)?; let markdown = htmd::convert(&html)?;
@ -127,7 +131,7 @@ fn extract_chapters_and_resources<R: Read + Seek>(
} }
} else { } else {
// other file, just copy // other file, just copy
src_dir.join(&path) src_dir.join(path)
}; };
// write to target path // write to target path
if let Some(parent) = target_path.parent() { if let Some(parent) = target_path.parent() {
@ -139,7 +143,7 @@ fn extract_chapters_and_resources<R: Read + Seek>(
} }
/// Capture the `{link}` without `#`, eg: /// Capture the `{link}` without `#`, eg:
/// ``` /// ```text
/// [ABC]({abc.html}#xxx) /// [ABC]({abc.html}#xxx)
/// [ABC]({abc.html}) /// [ABC]({abc.html})
/// ``` /// ```
@ -147,7 +151,7 @@ static LINK: LazyLock<Regex> = LazyLock::new(|| {
Regex::new(r#"\[[^\]]+\]\((?P<link>[^#)]+)(#[^)]+)?\)"#).expect("unreachable") Regex::new(r#"\[[^\]]+\]\((?P<link>[^#)]+)(#[^)]+)?\)"#).expect("unreachable")
}); });
/// Match the URL link, eg: /// Match the URL link, eg:
/// ``` /// ```text
/// https://www.example.com\ /// https://www.example.com\
/// ``` /// ```
static URL_LINK: LazyLock<Regex> = static URL_LINK: LazyLock<Regex> =
@ -183,29 +187,26 @@ fn write_book_toml<R: Read + Seek>(
output_dir: impl AsRef<Path>, output_dir: impl AsRef<Path>,
) -> io::Result<()> { ) -> io::Result<()> {
let output_dir = output_dir.as_ref(); let output_dir = output_dir.as_ref();
let title = epub_doc let title = epub_doc.get_title();
let authors = epub_doc
.metadata .metadata
.get("title") .iter()
.and_then(|v| v.first().cloned()); .filter(|m| m.property == "creator")
let authors = epub_doc.metadata.get("creator").cloned().unwrap_or(vec![]); .map(|m| m.value.clone())
.collect::<Vec<_>>();
let description = epub_doc let description = epub_doc
.metadata .mdata("description")
.get("description") .map(|m| htmd::convert(&m.value).expect("unreachable"));
.and_then(|v| v.first().cloned())
.map(|s| htmd::convert(&s).expect("unreachable"));
let lang = epub_doc let lang = epub_doc
.metadata .mdata("language")
.get("lang") .or_else(|| epub_doc.mdata("lang"))
.and_then(|v| v.first().cloned()); .map(|m| m.value.clone());
let config = BookConfig { let mut config = BookConfig::default();
title, config.title = title;
authors, config.authors = authors;
description, config.description = description;
src: PathBuf::from("src"), config.src = PathBuf::from("src");
multilingual: false, config.language = lang;
language: lang,
text_direction: None,
};
let toml_content = format!("[book]\n{}", toml::to_string(&config).expect("unreachable")); let toml_content = format!("[book]\n{}", toml::to_string(&config).expect("unreachable"));
fs::write(output_dir.join("book.toml"), toml_content)?; fs::write(output_dir.join("book.toml"), toml_content)?;
Ok(()) Ok(())

5
src/main.rs
View file

@ -10,11 +10,14 @@ struct Args {
/// The path to the output directory /// The path to the output directory
#[clap(short, long, default_value = ".")] #[clap(short, long, default_value = ".")]
output_dir: PathBuf, output_dir: PathBuf,
/// Output directly to the output directory without creating a subdirectory named after the book
#[clap(short, long)]
flat: bool,
} }
fn main() -> Result<(), Error> { fn main() -> Result<(), Error> {
let args = Args::parse(); let args = Args::parse();
convert_epub_to_mdbook(args.input_epub, args.output_dir, true)?; convert_epub_to_mdbook(args.input_epub, args.output_dir, !args.flat)?;
println!("Conversion completed successfully!"); println!("Conversion completed successfully!");
Ok(()) Ok(())
} }