Support ByteStreams in bytes starts-with and bytes ends-with (#12887)

# Description
Restores `bytes starts-with` so that it is able to work with byte
streams once again. For parity/consistency, this PR also adds byte
stream support to `bytes ends-with`.

# User-Facing Changes
- `bytes ends-with` now supports byte streams.

# Tests + Formatting
Re-enabled tests for `bytes starts-with` and added tests for `bytes
ends-with`.
This commit is contained in:
Ian Manske 2024-05-16 23:59:08 +00:00 committed by GitHub
parent aec41f3df0
commit 6891267b53
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
5 changed files with 275 additions and 96 deletions

View File

@ -0,0 +1,120 @@
use nu_test_support::nu;
#[test]
fn basic_binary_end_with() {
let actual = nu!(r#"
"hello world" | into binary | bytes ends-with 0x[77 6f 72 6c 64]
"#);
assert_eq!(actual.out, "true");
}
#[test]
fn basic_string_fails() {
let actual = nu!(r#"
"hello world" | bytes ends-with 0x[77 6f 72 6c 64]
"#);
assert!(actual.err.contains("command doesn't support"));
assert_eq!(actual.out, "");
}
#[test]
fn short_stream_binary() {
let actual = nu!(r#"
nu --testbin repeater (0x[01]) 5 | bytes ends-with 0x[010101]
"#);
assert_eq!(actual.out, "true");
}
#[test]
fn short_stream_mismatch() {
let actual = nu!(r#"
nu --testbin repeater (0x[010203]) 5 | bytes ends-with 0x[010204]
"#);
assert_eq!(actual.out, "false");
}
#[test]
fn short_stream_binary_overflow() {
let actual = nu!(r#"
nu --testbin repeater (0x[01]) 5 | bytes ends-with 0x[010101010101]
"#);
assert_eq!(actual.out, "false");
}
#[test]
fn long_stream_binary() {
let actual = nu!(r#"
nu --testbin repeater (0x[01]) 32768 | bytes ends-with 0x[010101]
"#);
assert_eq!(actual.out, "true");
}
#[test]
fn long_stream_binary_overflow() {
// .. ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
let actual = nu!(r#"
nu --testbin repeater (0x[01]) 32768 | bytes ends-with (0..32768 | each {|| 0x[01] } | bytes collect)
"#);
assert_eq!(actual.out, "false");
}
#[test]
fn long_stream_binary_exact() {
// ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
let actual = nu!(r#"
nu --testbin repeater (0x[01020304]) 8192 | bytes ends-with (0..<8192 | each {|| 0x[01020304] } | bytes collect)
"#);
assert_eq!(actual.out, "true");
}
#[test]
fn long_stream_string_exact() {
// ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
let actual = nu!(r#"
nu --testbin repeater hell 8192 | bytes ends-with (0..<8192 | each {|| "hell" | into binary } | bytes collect)
"#);
assert_eq!(actual.out, "true");
}
#[test]
fn long_stream_mixed_exact() {
// ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
let actual = nu!(r#"
let binseg = (0..<2048 | each {|| 0x[003d9fbf] } | bytes collect)
let strseg = (0..<2048 | each {|| "hell" | into binary } | bytes collect)
nu --testbin repeat_bytes 003d9fbf 2048 68656c6c 2048 | bytes ends-with (bytes build $binseg $strseg)
"#);
assert_eq!(
actual.err, "",
"invocation failed. command line limit likely reached"
);
assert_eq!(actual.out, "true");
}
#[test]
fn long_stream_mixed_overflow() {
// ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
let actual = nu!(r#"
let binseg = (0..<2048 | each {|| 0x[003d9fbf] } | bytes collect)
let strseg = (0..<2048 | each {|| "hell" | into binary } | bytes collect)
nu --testbin repeat_bytes 003d9fbf 2048 68656c6c 2048 | bytes ends-with (bytes build 0x[01] $binseg $strseg)
"#);
assert_eq!(
actual.err, "",
"invocation failed. command line limit likely reached"
);
assert_eq!(actual.out, "false");
}

View File

@ -1 +1,2 @@
mod ends_with;
mod starts_with;

View File

@ -19,102 +19,102 @@ fn basic_string_fails() {
assert_eq!(actual.out, "");
}
// #[test]
// fn short_stream_binary() {
// let actual = nu!(r#"
// nu --testbin repeater (0x[01]) 5 | bytes starts-with 0x[010101]
// "#);
#[test]
fn short_stream_binary() {
let actual = nu!(r#"
nu --testbin repeater (0x[01]) 5 | bytes starts-with 0x[010101]
"#);
// assert_eq!(actual.out, "true");
// }
assert_eq!(actual.out, "true");
}
// #[test]
// fn short_stream_mismatch() {
// let actual = nu!(r#"
// nu --testbin repeater (0x[010203]) 5 | bytes starts-with 0x[010204]
// "#);
#[test]
fn short_stream_mismatch() {
let actual = nu!(r#"
nu --testbin repeater (0x[010203]) 5 | bytes starts-with 0x[010204]
"#);
// assert_eq!(actual.out, "false");
// }
assert_eq!(actual.out, "false");
}
// #[test]
// fn short_stream_binary_overflow() {
// let actual = nu!(r#"
// nu --testbin repeater (0x[01]) 5 | bytes starts-with 0x[010101010101]
// "#);
#[test]
fn short_stream_binary_overflow() {
let actual = nu!(r#"
nu --testbin repeater (0x[01]) 5 | bytes starts-with 0x[010101010101]
"#);
// assert_eq!(actual.out, "false");
// }
assert_eq!(actual.out, "false");
}
// #[test]
// fn long_stream_binary() {
// let actual = nu!(r#"
// nu --testbin repeater (0x[01]) 32768 | bytes starts-with 0x[010101]
// "#);
#[test]
fn long_stream_binary() {
let actual = nu!(r#"
nu --testbin repeater (0x[01]) 32768 | bytes starts-with 0x[010101]
"#);
// assert_eq!(actual.out, "true");
// }
assert_eq!(actual.out, "true");
}
// #[test]
// fn long_stream_binary_overflow() {
// // .. ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
// let actual = nu!(r#"
// nu --testbin repeater (0x[01]) 32768 | bytes starts-with (0..32768 | each {|| 0x[01] } | bytes collect)
// "#);
#[test]
fn long_stream_binary_overflow() {
// .. ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
let actual = nu!(r#"
nu --testbin repeater (0x[01]) 32768 | bytes starts-with (0..32768 | each {|| 0x[01] } | bytes collect)
"#);
// assert_eq!(actual.out, "false");
// }
assert_eq!(actual.out, "false");
}
// #[test]
// fn long_stream_binary_exact() {
// // ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
// let actual = nu!(r#"
// nu --testbin repeater (0x[01020304]) 8192 | bytes starts-with (0..<8192 | each {|| 0x[01020304] } | bytes collect)
// "#);
#[test]
fn long_stream_binary_exact() {
// ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
let actual = nu!(r#"
nu --testbin repeater (0x[01020304]) 8192 | bytes starts-with (0..<8192 | each {|| 0x[01020304] } | bytes collect)
"#);
// assert_eq!(actual.out, "true");
// }
assert_eq!(actual.out, "true");
}
// #[test]
// fn long_stream_string_exact() {
// // ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
// let actual = nu!(r#"
// nu --testbin repeater hell 8192 | bytes starts-with (0..<8192 | each {|| "hell" | into binary } | bytes collect)
// "#);
#[test]
fn long_stream_string_exact() {
// ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
let actual = nu!(r#"
nu --testbin repeater hell 8192 | bytes starts-with (0..<8192 | each {|| "hell" | into binary } | bytes collect)
"#);
// assert_eq!(actual.out, "true");
// }
assert_eq!(actual.out, "true");
}
// #[test]
// fn long_stream_mixed_exact() {
// // ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
// let actual = nu!(r#"
// let binseg = (0..<2048 | each {|| 0x[003d9fbf] } | bytes collect)
// let strseg = (0..<2048 | each {|| "hell" | into binary } | bytes collect)
#[test]
fn long_stream_mixed_exact() {
// ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
let actual = nu!(r#"
let binseg = (0..<2048 | each {|| 0x[003d9fbf] } | bytes collect)
let strseg = (0..<2048 | each {|| "hell" | into binary } | bytes collect)
// nu --testbin repeat_bytes 003d9fbf 2048 68656c6c 2048 | bytes starts-with (bytes build $binseg $strseg)
// "#);
nu --testbin repeat_bytes 003d9fbf 2048 68656c6c 2048 | bytes starts-with (bytes build $binseg $strseg)
"#);
// assert_eq!(
// actual.err, "",
// "invocation failed. command line limit likely reached"
// );
// assert_eq!(actual.out, "true");
// }
assert_eq!(
actual.err, "",
"invocation failed. command line limit likely reached"
);
assert_eq!(actual.out, "true");
}
// #[test]
// fn long_stream_mixed_overflow() {
// // ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
// let actual = nu!(r#"
// let binseg = (0..<2048 | each {|| 0x[003d9fbf] } | bytes collect)
// let strseg = (0..<2048 | each {|| "hell" | into binary } | bytes collect)
#[test]
fn long_stream_mixed_overflow() {
// ranges are inclusive..inclusive, so we don't need to +1 to check for an overflow
let actual = nu!(r#"
let binseg = (0..<2048 | each {|| 0x[003d9fbf] } | bytes collect)
let strseg = (0..<2048 | each {|| "hell" | into binary } | bytes collect)
// nu --testbin repeat_bytes 003d9fbf 2048 68656c6c 2048 | bytes starts-with (bytes build $binseg $strseg 0x[01])
// "#);
nu --testbin repeat_bytes 003d9fbf 2048 68656c6c 2048 | bytes starts-with (bytes build $binseg $strseg 0x[01])
"#);
// assert_eq!(
// actual.err, "",
// "invocation failed. command line limit likely reached"
// );
// assert_eq!(actual.out, "false");
// }
assert_eq!(
actual.err, "",
"invocation failed. command line limit likely reached"
);
assert_eq!(actual.out, "false");
}

View File

@ -1,5 +1,9 @@
use nu_cmd_base::input_handler::{operate, CmdArgument};
use nu_engine::command_prelude::*;
use std::{
collections::VecDeque,
io::{self, BufRead},
};
struct Arguments {
pattern: Vec<u8>,
@ -52,14 +56,54 @@ impl Command for BytesEndsWith {
call: &Call,
input: PipelineData,
) -> Result<PipelineData, ShellError> {
let head = call.head;
let pattern: Vec<u8> = call.req(engine_state, stack, 0)?;
let cell_paths: Vec<CellPath> = call.rest(engine_state, stack, 1)?;
let cell_paths = (!cell_paths.is_empty()).then_some(cell_paths);
let arg = Arguments {
pattern,
cell_paths,
};
operate(ends_with, arg, input, call.head, engine_state.ctrlc.clone())
if let PipelineData::ByteStream(stream, ..) = input {
let span = stream.span();
if pattern.is_empty() {
return Ok(Value::bool(true, head).into_pipeline_data());
}
let Some(mut reader) = stream.reader() else {
return Ok(Value::bool(false, head).into_pipeline_data());
};
let cap = pattern.len();
let mut end = VecDeque::<u8>::with_capacity(cap);
loop {
let buf = match reader.fill_buf() {
Ok(&[]) => break,
Ok(buf) => buf,
Err(e) if e.kind() == io::ErrorKind::Interrupted => continue,
Err(e) => return Err(e.into_spanned(span).into()),
};
let len = buf.len();
if len >= cap {
end.clear();
end.extend(&buf[(len - cap)..])
} else {
let new_len = len + end.len();
if new_len > cap {
// The `drain` below will panic if `(new_len - cap) > end.len()`.
// But this cannot happen since we know `len < cap` (as checked above):
// (len + end.len() - cap) > end.len()
// => (len - cap) > 0
// => len > cap
end.drain(..(new_len - cap));
}
end.extend(buf);
}
reader.consume(len);
}
Ok(Value::bool(end == pattern, head).into_pipeline_data())
} else {
let arg = Arguments {
pattern,
cell_paths,
};
operate(ends_with, arg, input, head, engine_state.ctrlc.clone())
}
}
fn examples(&self) -> Vec<Example> {

View File

@ -1,5 +1,6 @@
use nu_cmd_base::input_handler::{operate, CmdArgument};
use nu_engine::command_prelude::*;
use std::io::Read;
struct Arguments {
pattern: Vec<u8>,
@ -53,20 +54,33 @@ impl Command for BytesStartsWith {
call: &Call,
input: PipelineData,
) -> Result<PipelineData, ShellError> {
let head = call.head;
let pattern: Vec<u8> = call.req(engine_state, stack, 0)?;
let cell_paths: Vec<CellPath> = call.rest(engine_state, stack, 1)?;
let cell_paths = (!cell_paths.is_empty()).then_some(cell_paths);
let arg = Arguments {
pattern,
cell_paths,
};
operate(
starts_with,
arg,
input,
call.head,
engine_state.ctrlc.clone(),
)
if let PipelineData::ByteStream(stream, ..) = input {
let span = stream.span();
if pattern.is_empty() {
return Ok(Value::bool(true, head).into_pipeline_data());
}
let Some(reader) = stream.reader() else {
return Ok(Value::bool(false, head).into_pipeline_data());
};
let mut start = Vec::with_capacity(pattern.len());
reader
.take(pattern.len() as u64)
.read_to_end(&mut start)
.err_span(span)?;
Ok(Value::bool(start == pattern, head).into_pipeline_data())
} else {
let arg = Arguments {
pattern,
cell_paths,
};
operate(starts_with, arg, input, head, engine_state.ctrlc.clone())
}
}
fn examples(&self) -> Vec<Example> {