Rust preserves-tools

This commit is contained in:
Tony Garnock-Jones 2021-08-04 15:42:07 +02:00
parent 3078396487
commit b24aca8f0f
3 changed files with 294 additions and 0 deletions

View File

@ -2,4 +2,5 @@
members = [
"preserves",
"preserves-schema",
"preserves-tools",
]

View File

@ -0,0 +1,16 @@
[package]
name = "preserves-tools"
version = "0.1.0"
authors = ["Tony Garnock-Jones <tonyg@leastfixedpoint.com>"]
edition = "2018"
description = "Command-line utilities for working with Preserves documents."
homepage = "https://preserves.gitlab.io/"
repository = "https://gitlab.com/preserves/preserves"
license = "Apache-2.0"
[dependencies]
preserves = { path = "../preserves", version = "0.16.0" }
bytes = "1.0"
clap = "3.0.0-beta.2"
clap_generate = "3.0.0-beta.2"

View File

@ -0,0 +1,277 @@
use bytes::Buf;
use bytes::BufMut;
use bytes::BytesMut;
use clap::ArgEnum;
use clap::Clap;
use clap::IntoApp;
use clap_generate::{generate, generators};
use preserves::value::IOBinarySource;
use preserves::value::IOValue;
use preserves::value::IOValueDomainCodec;
use preserves::value::PackedReader;
use preserves::value::PackedWriter;
use preserves::value::Reader;
use preserves::value::TextReader;
use preserves::value::TextWriter;
use preserves::value::ViaCodec;
use preserves::value::Writer;
use std::io;
// #[derive(ArgEnum, Clone, Debug)]
// enum Encoding {
// None,
// Base64,
// Hex,
// }
#[derive(ArgEnum, Clone, Debug)]
enum InputFormat {
AutoDetect,
Text,
Binary,
}
#[derive(ArgEnum, Clone, Debug)]
enum OutputFormat {
Text,
Binary,
}
#[derive(ArgEnum, Clone, Debug)]
enum CompletionDialect {
Bash,
Zsh,
PowerShell,
Fish,
Elvish,
}
#[derive(ArgEnum, Clone, Copy, Debug)]
enum Boolish {
#[clap(alias = "no", alias = "n", alias = "off", alias = "0", alias = "false")]
Disabled,
#[clap(alias = "yes", alias = "y", alias = "on", alias = "1", alias = "true")]
Enabled,
}
impl From<Boolish> for bool {
fn from(b: Boolish) -> Self {
matches!(b, Boolish::Enabled)
}
}
#[derive(Clap, Clone, Debug)]
struct Convert {
// #[clap(long, arg_enum, default_value = "none")]
// input_encoding: Encoding,
// #[clap(long, arg_enum, default_value = "none")]
// output_encoding: Encoding,
#[clap(long, short, arg_enum, default_value = "auto-detect")]
input_format: InputFormat,
#[clap(long, short, arg_enum, default_value = "text")]
output_format: OutputFormat,
#[clap(long)]
limit: Option<usize>,
#[clap(long, arg_enum, value_name = "on/off", default_value = "on")]
indent: Boolish,
#[clap(long, arg_enum, value_name = "on/off", default_value = "on")]
annotations: Boolish,
}
#[derive(Clap, Clone, Debug)]
enum Subcommand {
Convert(Convert),
Completions {
#[clap(arg_enum, value_name = "dialect")]
dialect: CompletionDialect,
}
}
#[derive(Clap, Clone, Debug)]
#[clap(version)]
struct CommandLine {
#[clap(subcommand)]
command: Subcommand,
}
fn main() -> io::Result<()> {
let args = CommandLine::parse();
match args.command {
Subcommand::Completions { dialect } => {
let mut app = CommandLine::into_app();
match dialect {
CompletionDialect::Bash => generate::<generators::Bash, _>(&mut app, "preserves-tool", &mut io::stdout()),
CompletionDialect::Zsh => generate::<generators::Zsh, _>(&mut app, "preserves-tool", &mut io::stdout()),
CompletionDialect::PowerShell => generate::<generators::PowerShell, _>(&mut app, "preserves-tool", &mut io::stdout()),
CompletionDialect::Fish => generate::<generators::Fish, _>(&mut app, "preserves-tool", &mut io::stdout()),
CompletionDialect::Elvish => generate::<generators::Elvish, _>(&mut app, "preserves-tool", &mut io::stdout()),
}
},
Subcommand::Convert(c) => convert(c)?,
}
Ok(())
}
struct RollingBuffer<R: io::Read> {
r: R,
discarded: usize,
pos: usize,
buf: BytesMut,
}
impl<R: io::Read> RollingBuffer<R> {
fn new(r: R) -> Self {
RollingBuffer {
r,
discarded: 0,
pos: 0,
buf: BytesMut::new(),
}
}
fn read_more(&mut self) -> io::Result<usize> {
let mut buf = [0; 8192];
let n = self.r.read(&mut buf)?;
self.buf.put(&buf[.. n]);
Ok(n)
}
fn peek_buf(&mut self) -> io::Result<&[u8]> {
if self.rhs() == self.pos {
let _ = self.read_more()?;
}
return Ok(&self.buf[self.pos - self.discarded ..]);
}
fn rhs(&self) -> usize {
self.buf.remaining() + self.discarded
}
fn discard(&mut self, count: usize) {
self.buf.advance(count);
self.discarded += count;
}
fn discard_to_pos(&mut self) {
self.discard(self.pos - self.discarded)
}
}
impl<R: io::Read> io::Seek for RollingBuffer<R> {
fn seek(&mut self, offset: io::SeekFrom) -> io::Result<u64> {
let new_position = match offset {
io::SeekFrom::Current(delta) => if delta >= 0 {
self.pos + delta as usize
} else {
self.pos - (-delta) as usize
},
io::SeekFrom::End(_) => Err(io::Error::new(
io::ErrorKind::Unsupported, "Cannot seek wrt end on open-ended stream"))?,
io::SeekFrom::Start(new_position) => new_position as usize,
};
if new_position > self.rhs() {
Err(io::Error::new(
io::ErrorKind::InvalidInput, "Attempt to seek beyond end of buffer"))?;
}
if new_position < self.discarded {
Err(io::Error::new(
io::ErrorKind::InvalidInput, "Attempt to seek before start of buffer"))?;
}
self.pos = new_position;
Ok(new_position as u64)
}
fn stream_position(&mut self) -> io::Result<u64> {
Ok(self.pos as u64)
}
}
impl<R: io::Read> io::Read for RollingBuffer<R> {
fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> {
let i = self.pos - self.discarded;
loop {
let n = std::cmp::min(self.buf.remaining() - i , buf.len());
if n == 0 {
if self.read_more()? == 0 {
return Ok(0);
}
continue;
}
&buf[.. n].copy_from_slice(&self.buf[i .. i + n]);
self.pos += n;
return Ok(n);
}
}
}
fn convert(c: Convert) -> io::Result<()> {
let mut r = RollingBuffer::new(io::stdin());
let mut count = 0;
let mut w: Box<dyn FnMut(&IOValue) -> io::Result<()>> = match c.output_format {
OutputFormat::Text => {
let mut t = TextWriter::new(io::stdout());
if c.indent.into() {
t.indentation = 2;
}
Box::new(move |v| {
t.write(&mut IOValueDomainCodec, v)?;
println!();
Ok(())
})
}
OutputFormat::Binary => {
let mut p = PackedWriter::new(io::stdout());
Box::new(move |v| p.write(&mut IOValueDomainCodec, v))
}
};
loop {
let is_text = {
let peek_buf = r.peek_buf()?;
if peek_buf.is_empty() {
return Ok(());
}
peek_buf[0] < 128
};
let maybe_value: Option<IOValue> = if is_text {
match c.input_format {
InputFormat::AutoDetect | InputFormat::Text => (),
InputFormat::Binary => return Err(io::Error::new(
io::ErrorKind::InvalidData, "Expected binary input, saw text input")),
}
TextReader::new(&mut IOBinarySource::new(&mut r), ViaCodec::new(IOValueDomainCodec))
.next(c.annotations.into())?
} else {
match c.input_format {
InputFormat::AutoDetect | InputFormat::Binary => (),
InputFormat::Text => return Err(io::Error::new(
io::ErrorKind::InvalidData, "Expected text input, saw binary input")),
}
PackedReader::new(&mut IOBinarySource::new(&mut r), IOValueDomainCodec)
.next(c.annotations.into())?
};
match maybe_value {
None => return Ok(()),
Some(value) => {
w(&value)?;
r.discard_to_pos();
count += 1;
if let Some(limit) = c.limit {
if count >= limit {
return Ok(());
}
}
}
}
}
}