ztb_recode/src/main.rs

130 lines
3.6 KiB
Rust

extern crate encoding;
extern crate textstream;
extern crate zip;
use encoding::all::{KOI8_R, UTF_8};
use encoding::Encoding;
use encoding::{DecoderTrap, EncoderTrap};
use std::io::{Read, Seek, Write};
use std::path::PathBuf;
use std::{fs, io};
use textstream::TextReader;
use zip::write::FileOptions;
use zip::ZipWriter;
const INPUT_FILENAME: &str = "baza.zip";
const OUTPUT_FILENAME: &str = "baza_utf8.zip";
struct OutFileDescr<'a> {
name: String,
options: &'a FileOptions,
encoding: &'a dyn Encoding,
trap: EncoderTrap,
}
impl<'a> OutFileDescr<'a> {
fn new(
name: String,
options: &'a FileOptions,
enc: &'a dyn Encoding,
trap: EncoderTrap,
) -> OutFileDescr<'a> {
OutFileDescr {
name,
options,
encoding: enc,
trap,
}
}
}
fn read_file(file: impl Read, enc: &dyn Encoding) -> Result<String, Box<dyn std::error::Error>> {
let buf = io::BufReader::new(file);
let mut reader = TextReader::new(buf, enc, DecoderTrap::Ignore);
let mut result = String::new();
reader.read_to_end(&mut result).or(Err("decode error"))?;
Ok(result)
}
fn write_file<T: Seek + Write>(
arc: &mut ZipWriter<T>,
f: OutFileDescr,
data: String,
) -> Result<(), Box<dyn std::error::Error>> {
arc.start_file(f.name, *f.options)?;
let data = f.encoding.encode(data.as_str(), f.trap)?;
arc.write_all(&data)?;
Ok(())
}
fn process_files(files: &[PathBuf]) {
if files.is_empty() {
return;
}
let enc_input = KOI8_R;
let enc_output = UTF_8;
let options = zip::write::FileOptions::default()
.compression_method(zip::CompressionMethod::Deflated)
.compression_level(Some(9));
let zip_file = fs::File::open(INPUT_FILENAME).unwrap();
let zip_reader = io::BufReader::new(zip_file);
let mut archive = zip::ZipArchive::new(zip_reader).unwrap();
let mut outfile = fs::File::create(OUTPUT_FILENAME).unwrap();
let mut zip_writer = ZipWriter::new(&mut outfile);
files.iter().for_each(|name| {
let name_str = name.to_str().unwrap();
// read string from file in input zip
let file = archive.by_name(name_str).unwrap();
let data = read_file(file, enc_input).unwrap();
// write string to file in output zip
let out_file = OutFileDescr::new(
name_str.to_string(),
&options,
enc_output,
EncoderTrap::Ignore,
);
write_file(&mut zip_writer, out_file, data).unwrap();
});
zip_writer.finish().unwrap();
}
fn main() -> Result<(), Box<dyn std::error::Error>> {
// open archive just to list files
let zip_file = fs::File::open(INPUT_FILENAME)?;
let zip_reader = io::BufReader::new(zip_file);
let mut archive = zip::ZipArchive::new(zip_reader)?;
let mut source_files: Vec<PathBuf> = (0..archive.len())
.map(|i| archive.by_index(i).unwrap().mangled_name())
.filter(|name| {
// skip files without "txt" extension
match name.extension() {
Some(ext) => match ext.to_str() {
Some(ext_str) => ext_str.eq_ignore_ascii_case("txt"),
_ => false, // extension is not valid unicode or not txt
},
_ => false, // no extension in filename
}
})
.collect();
drop(archive);
println!("processing {} files...", source_files.len());
source_files.sort();
let source_files = source_files;
process_files(&source_files);
println!("done");
Ok(())
}