Bug 1489792 - Part 2: Revendor dependencies. r=ato

This commit is contained in:
Bastien Orivel 2018-09-08 20:25:45 +02:00 committed by Andreas Tolfsen
parent 7ef3fa6f4a
commit 9ffad5d4b8
18 changed files with 3540 additions and 1491 deletions

View File

@ -1 +1 @@
{"files":{"Cargo.toml":"6f4dcaea19fa469551f91670c673d9e240fee726ea78c0ab46ba7b779b3a0d31","LICENSE-APACHE":"a60eea817514531668d7e00765731449fe14d059d3249e0bc93b36de45f759f2","LICENSE-MIT":"0dd882e53de11566d50f8e8e2d5a651bcf3fabee4987d70f306233cf39094ba7","README.md":"ac936d8d62e3aab7294e077901010ed0daf99b5f1df0569ffe197859955338c1","benches/benchmarks.rs":"9a7167a38f6f42be4ac32e723aeedcdfd8740a90775c461170a4203809e768f3","examples/make_tables.rs":"3c80f2a8cdb204168cc1b60f8904d544b2da067b9e6a7b40ade5fb4a994b4175","src/lib.rs":"cc226627c531a6d84c95f7b41dc024a26bb0b2341554b4b937baac4b591adf77","src/line_wrap.rs":"3a8f8fb96e3f63d80b87a94e8d4a235ad4e18bd3ee6413c2fd9dcd70354177b6","src/tables.rs":"378743892907cde87c1a92e6afee2df36ce590311e61381b2cc0404b3e018039","src/tests.rs":"61a54ba78255fd360f8b39ca47cefae2e5cb568b525ca1bef7e92dc127a0a485","tests/tests.rs":"80b9b3a44517baaf9bf584cf02fe521594f7993270dffeed3d1e743cf617206c"},"package":"96434f987501f0ed4eb336a411e0631ecd1afa11574fe148587adc4ff96143c9"}
{"files":{"Cargo.toml":"e6e873cf85b808ef7df811221f90dec19bfc9efc9b40237f325dfda8fc64fbfc","LICENSE-APACHE":"a60eea817514531668d7e00765731449fe14d059d3249e0bc93b36de45f759f2","LICENSE-MIT":"0dd882e53de11566d50f8e8e2d5a651bcf3fabee4987d70f306233cf39094ba7","README.md":"9941b171f4885557b0dfaac9e6bba9d0d18df6143f9501cf4605ddb7d2fdffe4","RELEASE-NOTES.md":"d9a4dfe7499c69247f3216f681df0dca7cc217b6f7d38816f3b943f66544e220","benches/benchmarks.rs":"bfc990dfa1716acf5bbd7189b3d7438145a9441833d6e4a796d2ce4428e55e14","examples/make_tables.rs":"e7c3e665874faa10ff1906420eb5182d1b725bcd30ff0d6da719bb368683e197","src/chunked_encoder.rs":"3d92a8241ed6d1a402ad1c1a8c7aa9c736b62cecc4071f44bc29b1b7cc70488c","src/decode.rs":"968c9c13746338ea307d3ff0b528df7a4fa20c7a1071c6adc1fb23c573741b57","src/display.rs":"d8ec9fa3f2b03640b026bbdfd817cd76454e1bb9b6ae53bd793334cd0a4ac62e","src/encode.rs":"c38c64582105ea698c5d1ec1596f5f1d99cfc1fb894226b79c5c8e96f73e6541","src/lib.rs":"4aeed1db6057b03532b20c34dbb822c1edcb0b5a2c21cdd50e7ac6b06858843a","src/line_wrap.rs":"fd136457629b803417bd25eb191fd3665bb57b8726fc4d74e7325e499e4e88af","src/tables.rs":"10b09997ed9765dc46c2d1f0df21b915d88c4f47a292b5c7df5c1851e724fb03","src/tests.rs":"cf0ddeb01b2149041b7ea99a881b6f149b1630110c5c2eedee6b1f44e873ea6c","tests/decode.rs":"c06a7595a0ac708961a180d4a8393f6d377b56028f75b29e6e34fd8451d84226","tests/encode.rs":"2eb3c8f81a4f807be69d1dfb0857625130017b0fa7037f7168f434c886cc851b","tests/helpers.rs":"a76015e4a4e8f98213bdbaa592cd9574ccdc95a28e1b1f835a2753e09fa6037f","tests/tests.rs":"13929406ba13504ff49d1a16a2578a87c560834fba0eaad78edecf0793261b0e"},"package":"85415d2594767338a74a30c1d370b2f3262ec1b4ed2d7bba5b3faf4de40467d9"}

View File

@ -1,22 +1,32 @@
# THIS FILE IS AUTOMATICALLY GENERATED BY CARGO
#
# When uploading crates to the registry Cargo will automatically
# "normalize" Cargo.toml files for maximal compatibility
# with all versions of Cargo and also rewrite `path` dependencies
# to registry (e.g. crates.io) dependencies
#
# If you believe there's an error in this file please file an
# issue against the rust-lang/cargo repository. If you're
# editing this file be aware that the upstream Cargo.toml
# will likely look very different (and much more reasonable)
[package]
name = "base64"
version = "0.6.0"
version = "0.9.2"
authors = ["Alice Maz <alice@alicemaz.com>", "Marshall Pierce <marshall@mpierce.org>"]
description = "encodes and decodes base64 as bytes or utf8"
repository = "https://github.com/alicemaz/rust-base64"
documentation = "https://github.com/alicemaz/rust-base64/blob/master/README.md"
readme = "README.md"
keywords = ["base64", "utf8", "encode", "decode"]
categories = ["encoding"]
license = "MIT/Apache-2.0"
[dependencies]
byteorder = "1.0.0"
safemem = "0.2.0"
[dev-dependencies]
rand = "=0.3.15"
repository = "https://github.com/alicemaz/rust-base64"
[profile.bench]
# Uncomment when using `perf record`
#debug = true
debug = true
[dependencies.byteorder]
version = "1.1.0"
[dependencies.safemem]
version = "0.2.0"
[dev-dependencies.rand]
version = "0.4"

View File

@ -1,5 +1,6 @@
[base64](https://crates.io/crates/base64)
===
[![Docs](https://docs.rs/base64/badge.svg)](https://docs.rs/base64)
It's base64. What more could anyone want?
@ -22,23 +23,7 @@ fn main() {
}
```
API
---
base64 exposes six functions:
```rust
fn encode<T: ?Sized + AsRef<[u8]>>(&T) -> String;
fn decode<T: ?Sized + AsRef<[u8]>>(&T) -> Result<Vec<u8>, DecodeError>;
fn encode_config<T: ?Sized + AsRef<[u8]>>(&T, Config) -> String;
fn encode_config_buf<T: ?Sized + AsRef<[u8]>>(&T, Config, &mut String);
fn decode_config<T: ?Sized + AsRef<[u8]>>(&T, Config) -> Result<Vec<u8>, DecodeError>;
fn decode_config_buf<T: ?Sized + AsRef<[u8]>>(&T, Config, &mut Vec<u8>) -> Result<(), DecodeError>;
```
`STANDARD`, `URL_SAFE`, `URL_SAFE_NO_PAD`, and `MIME` configuation structs are provided for convenience. `encode` and `decode` are convenience wrappers for the `_config` functions called with the `STANDARD` config, and they are themselves wrappers of the `_buf` functions that allocate on the user's behalf. Encode produces valid padding absent a config that states otherwise; decode produces the same output for valid or omitted padding in all cases, but errors on invalid (superfluous) padding. Whitespace in the input to decode is an error for all modes except `MIME`, which disregards it ("whitespace" according to POSIX-locale `isspace`, meaning \n \r \f \t \v and space).
`Config` exposes a constructor to allow custom combinations of character set, output padding, input whitespace permissiveness, linewrapping, and line ending character(s). The vast majority of usecases should be covered by the four provided, however.
See the [docs](https://docs.rs/base64) for all the details.
Purpose
---
@ -63,14 +48,7 @@ cargo run --example make_tables > src/tables.rs.tmp && mv src/tables.rs.tmp src/
Profiling
---
On Linux, you can use [perf](https://perf.wiki.kernel.org/index.php/Main_Page) for profiling. First, enable debug symbols in Cargo.toml. Don't commit this change, though, since it's usually not what you want (and costs some performance):
```
[profile.release]
debug = true
```
Then compile the benchmarks. (Just re-run them and ^C once the benchmarks start running; all that's needed is to recompile them.)
On Linux, you can use [perf](https://perf.wiki.kernel.org/index.php/Main_Page) for profiling. Then compile the benchmarks with `rustup nightly run cargo bench --no-run`.
Run the benchmark binary with `perf` (shown here filtering to one particular benchmark, which will make the results easier to read). `perf` is only available to the root user on most systems as it fiddles with event counters in your CPU, so use `sudo`. We need to run the actual benchmark binary, hence the path into `target`. You can see the actual full path with `rustup run nightly cargo bench -v`; it will print out the commands it runs. If you use the exact path that `bench` outputs, make sure you get the one that's for the benchmarks, not the tests. You may also want to `cargo clean` so you have only one `benchmarks-` binary (they tend to accumulate).
@ -113,10 +91,10 @@ Fuzzing
This uses [cargo-fuzz](https://github.com/rust-fuzz/cargo-fuzz). See `fuzz/fuzzers` for the available fuzzing scripts. To run, use an invocation like these:
```
rustup run nightly cargo fuzz run roundtrip
rustup run nightly cargo fuzz run roundtrip_no_pad
rustup run nightly cargo fuzz run roundtrip_mime -- -max_len=10240
rustup run nightly cargo fuzz run roundtrip_random_config -- -max_len=10240
cargo +nightly fuzz run roundtrip
cargo +nightly fuzz run roundtrip_no_pad
cargo +nightly fuzz run roundtrip_mime -- -max_len=10240
cargo +nightly fuzz run roundtrip_random_config -- -max_len=10240
```

View File

@ -0,0 +1,58 @@
# 0.9.2
- Derive `Clone` for `DecodeError`.
# 0.9.1
- Add support for `crypt(3)`'s base64 variant.
# 0.9.0
- `decode_config_slice` function for no-allocation decoding, analogous to `encode_config_slice`
- Decode performance optimization
# 0.8.0
- `encode_config_slice` function for no-allocation encoding
# 0.7.0
- `STANDARD_NO_PAD` config
- `Base64Display` heap-free wrapper for use in format strings, etc
# 0.6.0
- Decode performance improvements
- Use `unsafe` in fewer places
- Added fuzzers
# 0.5.2
- Avoid usize overflow when calculating length
- Better line wrapping performance
# 0.5.1
- Temporarily disable line wrapping
- Add Apache 2.0 license
# 0.5.0
- MIME support, including configurable line endings and line wrapping
- Removed `decode_ws`
- Renamed `Base64Error` to `DecodeError`
# 0.4.1
- Allow decoding a `AsRef<[u8]>` instead of just a `&str`
# 0.4.0
- Configurable padding
- Encode performance improvements
# 0.3.0
- Added encode/decode functions that do not allocate their own storage
- Decode performance improvements
- Extraneous padding bytes are no longer ignored. Now, an error will be returned.

View File

@ -1,13 +1,15 @@
#![feature(test)]
extern crate base64;
extern crate test;
extern crate rand;
extern crate test;
use base64::{decode, decode_config_buf, encode, encode_config_buf, Config, MIME, STANDARD};
use base64::display;
use base64::{decode, decode_config_buf, decode_config_slice, encode, encode_config_buf,
encode_config_slice, Config, MIME, STANDARD};
use test::Bencher;
use rand::Rng;
use test::Bencher;
#[bench]
fn encode_3b(b: &mut Bencher) {
@ -19,16 +21,31 @@ fn encode_3b_reuse_buf(b: &mut Bencher) {
do_encode_bench_reuse_buf(b, 3, STANDARD)
}
#[bench]
fn encode_3b_slice(b: &mut Bencher) {
do_encode_bench_slice(b, 3, STANDARD)
}
#[bench]
fn encode_50b(b: &mut Bencher) {
do_encode_bench(b, 50)
}
#[bench]
fn encode_50b_display(b: &mut Bencher) {
do_encode_bench_display(b, 50)
}
#[bench]
fn encode_50b_reuse_buf(b: &mut Bencher) {
do_encode_bench_reuse_buf(b, 50, STANDARD)
}
#[bench]
fn encode_50b_slice(b: &mut Bencher) {
do_encode_bench_slice(b, 50, STANDARD)
}
#[bench]
fn encode_100b(b: &mut Bencher) {
do_encode_bench(b, 100)
@ -59,11 +76,21 @@ fn encode_3kib(b: &mut Bencher) {
do_encode_bench(b, 3 * 1024)
}
#[bench]
fn encode_3kib_display(b: &mut Bencher) {
do_encode_bench_display(b, 3 * 1024)
}
#[bench]
fn encode_3kib_reuse_buf(b: &mut Bencher) {
do_encode_bench_reuse_buf(b, 3 * 1024, STANDARD)
}
#[bench]
fn encode_3kib_slice(b: &mut Bencher) {
do_encode_bench_slice(b, 3 * 1024, STANDARD)
}
#[bench]
fn encode_3kib_reuse_buf_mime(b: &mut Bencher) {
do_encode_bench_reuse_buf(b, 3 * 1024, MIME)
@ -74,11 +101,21 @@ fn encode_3mib(b: &mut Bencher) {
do_encode_bench(b, 3 * 1024 * 1024)
}
#[bench]
fn encode_3mib_display(b: &mut Bencher) {
do_encode_bench_display(b, 3 * 1024 * 1024)
}
#[bench]
fn encode_3mib_reuse_buf(b: &mut Bencher) {
do_encode_bench_reuse_buf(b, 3 * 1024 * 1024, STANDARD)
}
#[bench]
fn encode_3mib_slice(b: &mut Bencher) {
do_encode_bench_slice(b, 3 * 1024 * 1024, STANDARD)
}
#[bench]
fn encode_10mib(b: &mut Bencher) {
do_encode_bench(b, 10 * 1024 * 1024)
@ -99,6 +136,11 @@ fn encode_30mib_reuse_buf(b: &mut Bencher) {
do_encode_bench_reuse_buf(b, 30 * 1024 * 1024, STANDARD)
}
#[bench]
fn encode_30mib_slice(b: &mut Bencher) {
do_encode_bench_slice(b, 30 * 1024 * 1024, STANDARD)
}
#[bench]
fn decode_3b(b: &mut Bencher) {
do_decode_bench(b, 3)
@ -109,6 +151,11 @@ fn decode_3b_reuse_buf(b: &mut Bencher) {
do_decode_bench_reuse_buf(b, 3)
}
#[bench]
fn decode_3b_slice(b: &mut Bencher) {
do_decode_bench_slice(b, 3)
}
#[bench]
fn decode_50b(b: &mut Bencher) {
do_decode_bench(b, 50)
@ -119,6 +166,11 @@ fn decode_50b_reuse_buf(b: &mut Bencher) {
do_decode_bench_reuse_buf(b, 50)
}
#[bench]
fn decode_50b_slice(b: &mut Bencher) {
do_decode_bench_slice(b, 50)
}
#[bench]
fn decode_100b(b: &mut Bencher) {
do_decode_bench(b, 100)
@ -149,6 +201,11 @@ fn decode_3kib_reuse_buf(b: &mut Bencher) {
do_decode_bench_reuse_buf(b, 3 * 1024)
}
#[bench]
fn decode_3kib_slice(b: &mut Bencher) {
do_decode_bench_slice(b, 3 * 1024)
}
#[bench]
fn decode_3mib(b: &mut Bencher) {
do_decode_bench(b, 3 * 1024 * 1024)
@ -159,6 +216,11 @@ fn decode_3mib_reuse_buf(b: &mut Bencher) {
do_decode_bench_reuse_buf(b, 3 * 1024 * 1024)
}
#[bench]
fn decode_3mib_slice(b: &mut Bencher) {
do_decode_bench_slice(b, 3 * 1024 * 1024)
}
#[bench]
fn decode_10mib(b: &mut Bencher) {
do_decode_bench(b, 10 * 1024 * 1024)
@ -179,6 +241,11 @@ fn decode_30mib_reuse_buf(b: &mut Bencher) {
do_decode_bench_reuse_buf(b, 30 * 1024 * 1024)
}
#[bench]
fn decode_30mib_slice(b: &mut Bencher) {
do_decode_bench_slice(b, 30 * 1024 * 1024)
}
fn do_decode_bench(b: &mut Bencher, size: usize) {
let mut v: Vec<u8> = Vec::with_capacity(size * 3 / 4);
fill(&mut v);
@ -205,6 +272,20 @@ fn do_decode_bench_reuse_buf(b: &mut Bencher, size: usize) {
});
}
fn do_decode_bench_slice(b: &mut Bencher, size: usize) {
let mut v: Vec<u8> = Vec::with_capacity(size * 3 / 4);
fill(&mut v);
let encoded = encode(&v);
let mut buf = Vec::new();
buf.resize(size, 0);
b.bytes = encoded.len() as u64;
b.iter(|| {
decode_config_slice(&encoded, STANDARD, &mut buf).unwrap();
test::black_box(&buf);
});
}
fn do_encode_bench(b: &mut Bencher, size: usize) {
let mut v: Vec<u8> = Vec::with_capacity(size);
fill(&mut v);
@ -216,6 +297,17 @@ fn do_encode_bench(b: &mut Bencher, size: usize) {
});
}
fn do_encode_bench_display(b: &mut Bencher, size: usize) {
let mut v: Vec<u8> = Vec::with_capacity(size);
fill(&mut v);
b.bytes = v.len() as u64;
b.iter(|| {
let e = format!("{}", display::Base64Display::standard(&v));
test::black_box(&e);
});
}
fn do_encode_bench_reuse_buf(b: &mut Bencher, size: usize, config: Config) {
let mut v: Vec<u8> = Vec::with_capacity(size);
fill(&mut v);
@ -229,6 +321,20 @@ fn do_encode_bench_reuse_buf(b: &mut Bencher, size: usize, config: Config) {
});
}
fn do_encode_bench_slice(b: &mut Bencher, size: usize, config: Config) {
let mut v: Vec<u8> = Vec::with_capacity(size);
fill(&mut v);
let mut buf = Vec::new();
b.bytes = v.len() as u64;
// conservative estimate of encoded size
buf.resize(size * 2, 0);
b.iter(|| {
encode_config_slice(&v, config, &mut buf);
});
}
fn fill(v: &mut Vec<u8>) {
let cap = v.capacity();
// weak randomness is plenty; we just want to not be completely friendly to the branch predictor

View File

@ -32,17 +32,40 @@ fn main() {
print_encode_table(&url_alphabet, "URL_SAFE_ENCODE", 0);
print_decode_table(&url_alphabet, "URL_SAFE_DECODE", 0);
// ./0123456789
let crypt_alphabet: Vec<u8> = (b'.'..(b'9'+1))
// A-Z
.chain(b'A'..(b'Z'+1))
// a-z
.chain(b'a'..(b'z'+1))
.collect();
print_encode_table(&crypt_alphabet, "CRYPT_ENCODE", 0);
print_decode_table(&crypt_alphabet, "CRYPT_DECODE", 0);
}
fn print_encode_table(alphabet: &[u8], const_name: &str, indent_depth: usize) {
println!("{:width$}pub const {}: &'static [u8; 64] = &[", "", const_name, width=indent_depth);
println!("#[cfg_attr(rustfmt, rustfmt_skip)]");
println!(
"{:width$}pub const {}: &'static [u8; 64] = &[",
"",
const_name,
width = indent_depth
);
for (i, b) in alphabet.iter().enumerate() {
println!("{:width$}{}, // input {} (0x{:X}) => '{}' (0x{:X})", "",
b, i, i, String::from_utf8(vec!(*b as u8)).unwrap(), b, width=indent_depth + 4);
println!(
"{:width$}{}, // input {} (0x{:X}) => '{}' (0x{:X})",
"",
b,
i,
i,
String::from_utf8(vec![*b as u8]).unwrap(),
b,
width = indent_depth + 4
);
}
println!("{:width$}];", "", width=indent_depth);
println!("{:width$}];", "", width = indent_depth);
}
fn print_decode_table(alphabet: &[u8], const_name: &str, indent_depth: usize) {
@ -55,19 +78,39 @@ fn print_decode_table(alphabet: &[u8], const_name: &str, indent_depth: usize) {
let _ = input_to_morsel.insert(*ascii_byte, morsel as u8);
}
println!("{:width$}pub const {}: &'static [u8; 256] = &[", "", const_name, width=indent_depth);
println!("#[cfg_attr(rustfmt, rustfmt_skip)]");
println!(
"{:width$}pub const {}: &'static [u8; 256] = &[",
"",
const_name,
width = indent_depth
);
for ascii_byte in 0..256 {
let (value, comment) = match input_to_morsel.get(&(ascii_byte as u8)) {
None => ("INVALID_VALUE".to_string(),
format!("input {} (0x{:X})", ascii_byte, ascii_byte)),
Some(v) => (format!("{}", *v),
format!("input {} (0x{:X} char '{}') => {} (0x{:X})",
ascii_byte,
ascii_byte,
String::from_utf8(vec!(ascii_byte as u8)).unwrap(), *v, *v))
None => (
"INVALID_VALUE".to_string(),
format!("input {} (0x{:X})", ascii_byte, ascii_byte),
),
Some(v) => (
format!("{}", *v),
format!(
"input {} (0x{:X} char '{}') => {} (0x{:X})",
ascii_byte,
ascii_byte,
String::from_utf8(vec![ascii_byte as u8]).unwrap(),
*v,
*v
),
),
};
println!("{:width$}{}, // {}", "", value, comment, width=indent_depth + 4);
println!(
"{:width$}{}, // {}",
"",
value,
comment,
width = indent_depth + 4
);
}
println!("{:width$}];", "", width=indent_depth);
println!("{:width$}];", "", width = indent_depth);
}

View File

@ -0,0 +1,509 @@
use encode::{add_padding, encode_to_slice};
use line_wrap::line_wrap;
use std::cmp;
use {Config, LineEnding, LineWrap};
/// The output mechanism for ChunkedEncoder's encoded bytes.
pub trait Sink {
type Error;
/// Handle a chunk of encoded base64 data (as UTF-8 bytes)
fn write_encoded_bytes(&mut self, encoded: &[u8]) -> Result<(), Self::Error>;
}
#[derive(Debug, PartialEq)]
pub enum ChunkedEncoderError {
/// If wrapping is configured, the line length must be a multiple of 4, and must not be absurdly
/// large (see BUF_SIZE).
InvalidLineLength,
}
const BUF_SIZE: usize = 1024;
/// A base64 encoder that emits encoded bytes in chunks without heap allocation.
pub struct ChunkedEncoder {
config: Config,
max_input_chunk_len: usize,
}
impl ChunkedEncoder {
pub fn new(config: Config) -> Result<ChunkedEncoder, ChunkedEncoderError> {
Ok(ChunkedEncoder {
config,
max_input_chunk_len: max_input_length(BUF_SIZE, &config)?,
})
}
pub fn encode<S: Sink>(&self, bytes: &[u8], sink: &mut S) -> Result<(), S::Error> {
let mut encode_buf: [u8; BUF_SIZE] = [0; BUF_SIZE];
let encode_table = self.config.char_set.encode_table();
let mut input_index = 0;
while input_index < bytes.len() {
// either the full input chunk size, or it's the last iteration
let input_chunk_len = cmp::min(self.max_input_chunk_len, bytes.len() - input_index);
let chunk = &bytes[input_index..(input_index + input_chunk_len)];
let mut b64_bytes_written = encode_to_slice(chunk, &mut encode_buf, encode_table);
input_index += input_chunk_len;
let more_input_left = input_index < bytes.len();
if self.config.pad && !more_input_left {
// no more input, add padding if needed. Buffer will have room because
// max_input_length leaves room for it.
b64_bytes_written += add_padding(bytes.len(), &mut encode_buf[b64_bytes_written..]);
}
let line_ending_bytes = match self.config.line_wrap {
LineWrap::NoWrap => 0,
LineWrap::Wrap(line_len, line_ending) => {
let initial_line_ending_bytes =
line_wrap(&mut encode_buf, b64_bytes_written, line_len, line_ending);
if more_input_left {
assert_eq!(input_chunk_len, self.max_input_chunk_len);
// If there are more bytes of input, then we know we didn't just do the
// last chunk. line_wrap() doesn't put an ending after the last line, so we
// append one more line ending here. Since the chunk just encoded was not
// the last one, it was multiple of the line length (max_input_chunk_len),
// and therefore we can just put the line ending bytes at the end of the
// contents of the buffer.
match line_ending {
LineEnding::LF => {
encode_buf[b64_bytes_written + initial_line_ending_bytes] = b'\n';
initial_line_ending_bytes + 1
}
LineEnding::CRLF => {
encode_buf[b64_bytes_written + initial_line_ending_bytes] = b'\r';
encode_buf[b64_bytes_written + initial_line_ending_bytes + 1] =
b'\n';
initial_line_ending_bytes + 2
}
}
} else {
initial_line_ending_bytes
}
}
};
let total_bytes_written = b64_bytes_written + line_ending_bytes;
sink.write_encoded_bytes(&encode_buf[0..total_bytes_written])?;
}
Ok(())
}
}
/// Calculate the longest input that can be encoded for the given output buffer size.
///
/// If config requires line wrap, the calculated input length will be the maximum number of input
/// lines that can fit in the output buffer after each line has had its line ending appended.
///
/// If the config requires padding, two bytes of buffer space will be set aside so that the last
/// chunk of input can be encoded safely.
///
/// The input length will always be a multiple of 3 so that no encoding state has to be carried over
/// between chunks.
///
/// If the configured line length is not divisible by 4 (and therefore would require carrying
/// encoder state between chunks), or if the line length is too big for the buffer, an error will be
/// returned.
///
/// Note that the last overall line of input should *not* have an ending appended, but this will
/// conservatively calculate space as if it should because encoding is done in chunks, and all the
/// chunks before the last one will need a line ending after the last encoded line in that chunk.
fn max_input_length(encoded_buf_len: usize, config: &Config) -> Result<usize, ChunkedEncoderError> {
let effective_buf_len = if config.pad {
// make room for padding
encoded_buf_len
.checked_sub(2)
.expect("Don't use a tiny buffer")
} else {
encoded_buf_len
};
match config.line_wrap {
// No wrapping, no padding, so just normal base64 expansion.
LineWrap::NoWrap => Ok((effective_buf_len / 4) * 3),
LineWrap::Wrap(line_len, line_ending) => {
// To avoid complicated encode buffer shuffling, only allow line lengths that are
// multiples of 4 (which map to input lengths that are multiples of 3).
// line_len is never 0.
if line_len % 4 != 0 {
return Err(ChunkedEncoderError::InvalidLineLength);
}
let single_encoded_full_line_with_ending_len = line_len
.checked_add(line_ending.len())
.expect("Encoded line length with ending exceeds usize");
// max number of complete lines with endings that will fit in buffer
let num_encoded_wrapped_lines_in_buffer =
effective_buf_len / single_encoded_full_line_with_ending_len;
if num_encoded_wrapped_lines_in_buffer == 0 {
// line + ending is longer than can fit into encode buffer; give up
Err(ChunkedEncoderError::InvalidLineLength)
} else {
let input_len_for_line_len = (line_len / 4) * 3;
let input_len = input_len_for_line_len
.checked_mul(num_encoded_wrapped_lines_in_buffer)
.expect("Max input size exceeds usize");
assert!(input_len % 3 == 0 && input_len > 1);
Ok(input_len)
}
}
}
}
#[cfg(test)]
pub mod tests {
extern crate rand;
use super::*;
use tests::random_config;
use *;
use std::str;
use self::rand::distributions::{IndependentSample, Range};
use self::rand::Rng;
#[test]
fn chunked_encode_empty() {
assert_eq!("", chunked_encode_str(&[], STANDARD));
}
#[test]
fn chunked_encode_intermediate_fast_loop() {
// > 8 bytes input, will enter the pretty fast loop
assert_eq!(
"Zm9vYmFyYmF6cXV4",
chunked_encode_str(b"foobarbazqux", STANDARD)
);
}
#[test]
fn chunked_encode_fast_loop() {
// > 32 bytes input, will enter the uber fast loop
assert_eq!(
"Zm9vYmFyYmF6cXV4cXV1eGNvcmdlZ3JhdWx0Z2FycGx5eg==",
chunked_encode_str(b"foobarbazquxquuxcorgegraultgarplyz", STANDARD)
);
}
#[test]
fn chunked_encode_slow_loop_only() {
// < 8 bytes input, slow loop only
assert_eq!("Zm9vYmFy", chunked_encode_str(b"foobar", STANDARD));
}
#[test]
fn chunked_encode_line_wrap_padding() {
// < 8 bytes input, slow loop only
let config = config_wrap(true, 4, LineEnding::LF);
assert_eq!(
"Zm9v\nYmFy\nZm9v\nYmFy\nZg==",
chunked_encode_str(b"foobarfoobarf", config)
);
}
#[test]
fn chunked_encode_longer_than_one_buffer_adds_final_line_wrap_lf() {
// longest line len possible
let config = config_wrap(false, 1020, LineEnding::LF);
let input = vec![0xFF; 768];
let encoded = chunked_encode_str(&input, config);
// got a line wrap
assert_eq!(1024 + 1, encoded.len());
for &b in encoded.as_bytes()[0..1020].iter() {
// ascii /
assert_eq!(47, b);
}
assert_eq!(10, encoded.as_bytes()[1020]);
for &b in encoded.as_bytes()[1021..].iter() {
// ascii /
assert_eq!(47, b);
}
}
#[test]
fn chunked_encode_longer_than_one_buffer_adds_final_line_wrap_crlf() {
// longest line len possible
let config = config_wrap(false, 1020, LineEnding::CRLF);
let input = vec![0xFF; 768];
let encoded = chunked_encode_str(&input, config);
// got a line wrap
assert_eq!(1024 + 2, encoded.len());
for &b in encoded.as_bytes()[0..1020].iter() {
// ascii /
assert_eq!(47, b);
}
assert_eq!(13, encoded.as_bytes()[1020]);
assert_eq!(10, encoded.as_bytes()[1021]);
for &b in encoded.as_bytes()[1022..].iter() {
// ascii /
assert_eq!(47, b);
}
}
#[test]
fn chunked_encode_matches_normal_encode_random_string_sink() {
let helper = StringSinkTestHelper;
chunked_encode_matches_normal_encode_random(&helper);
}
#[test]
fn max_input_length_no_wrap_no_pad() {
let config = config_no_wrap(false);
assert_eq!(768, max_input_length(1024, &config).unwrap());
}
#[test]
fn max_input_length_no_wrap_with_pad_decrements_one_triple() {
let config = config_no_wrap(true);
assert_eq!(765, max_input_length(1024, &config).unwrap());
}
#[test]
fn max_input_length_no_wrap_with_pad_one_byte_short() {
let config = config_no_wrap(true);
assert_eq!(765, max_input_length(1025, &config).unwrap());
}
#[test]
fn max_input_length_no_wrap_with_pad_fits_exactly() {
let config = config_no_wrap(true);
assert_eq!(768, max_input_length(1026, &config).unwrap());
}
#[test]
fn max_input_length_wrap_with_lf_fits_exactly_no_pad() {
// 10 * (72 + 1) = 730. 54 input bytes = 72 encoded bytes, + 1 for LF.
let config = config_wrap(false, 72, LineEnding::LF);
assert_eq!(540, max_input_length(730, &config).unwrap());
}
#[test]
fn max_input_length_wrap_with_lf_fits_one_spare_byte_no_pad() {
// 10 * (72 + 1) = 730. 54 input bytes = 72 encoded bytes, + 1 for LF.
let config = config_wrap(false, 72, LineEnding::LF);
assert_eq!(540, max_input_length(731, &config).unwrap());
}
#[test]
fn max_input_length_wrap_with_lf_size_one_byte_short_of_another_line_no_pad() {
// 10 * (72 + 1) = 730. 54 input bytes = 72 encoded bytes, + 1 for LF.
// 73 * 11 = 803
let config = config_wrap(false, 72, LineEnding::LF);
assert_eq!(540, max_input_length(802, &config).unwrap());
}
#[test]
fn max_input_length_wrap_with_lf_size_another_line_no_pad() {
// 10 * (72 + 1) = 730. 54 input bytes = 72 encoded bytes, + 1 for LF.
// 73 * 11 = 803
let config = config_wrap(false, 72, LineEnding::LF);
assert_eq!(594, max_input_length(803, &config).unwrap());
}
#[test]
fn max_input_length_wrap_with_lf_one_byte_short_with_pad() {
// one fewer input line
let config = config_wrap(true, 72, LineEnding::LF);
assert_eq!(486, max_input_length(731, &config).unwrap());
}
#[test]
fn max_input_length_wrap_with_lf_fits_exactly_with_pad() {
// 10 * (72 + 1) = 730. 54 input bytes = 72 encoded bytes, + 1 for LF.
let config = config_wrap(true, 72, LineEnding::LF);
assert_eq!(540, max_input_length(732, &config).unwrap());
}
#[test]
fn max_input_length_wrap_line_len_wont_fit_one_line_lf() {
// 300 bytes is 400 encoded, + 1 for LF
let config = config_wrap(false, 400, LineEnding::LF);
assert_eq!(
ChunkedEncoderError::InvalidLineLength,
max_input_length(400, &config).unwrap_err()
);
}
#[test]
fn max_input_length_wrap_line_len_just_fits_one_line_lf() {
// 300 bytes is 400 encoded, + 1 for LF
let config = Config::new(
CharacterSet::Standard,
false,
false,
LineWrap::Wrap(400, LineEnding::LF),
);
assert_eq!(300, max_input_length(401, &config).unwrap());
}
#[test]
fn max_input_length_wrap_with_crlf_fits_exactly_no_pad() {
// 10 * (72 + 2) = 740. 54 input bytes = 72 encoded bytes, + 2 for CRLF.
let config = config_wrap(false, 72, LineEnding::CRLF);
assert_eq!(540, max_input_length(740, &config).unwrap());
}
#[test]
fn max_input_length_wrap_with_crlf_fits_one_spare_byte_no_pad() {
// 10 * (72 + 2) = 740. 54 input bytes = 72 encoded bytes, + 2 for CRLF.
let config = config_wrap(false, 72, LineEnding::CRLF);
assert_eq!(540, max_input_length(741, &config).unwrap());
}
#[test]
fn max_input_length_wrap_with_crlf_size_one_byte_short_of_another_line_no_pad() {
// 10 * (72 + 2) = 740. 54 input bytes = 72 encoded bytes, + 2 for CRLF.
// 74 * 11 = 814
let config = config_wrap(false, 72, LineEnding::CRLF);
assert_eq!(540, max_input_length(813, &config).unwrap());
}
#[test]
fn max_input_length_wrap_with_crlf_size_another_line_no_pad() {
// 10 * (72 + 2) = 740. 54 input bytes = 72 encoded bytes, + 2 for CRLF.
// 74 * 11 = 814
let config = config_wrap(false, 72, LineEnding::CRLF);
assert_eq!(594, max_input_length(814, &config).unwrap());
}
#[test]
fn max_input_length_wrap_line_len_not_multiple_of_4_rejected() {
let config = config_wrap(false, 41, LineEnding::LF);
assert_eq!(
ChunkedEncoderError::InvalidLineLength,
max_input_length(400, &config).unwrap_err()
);
}
pub fn chunked_encode_matches_normal_encode_random<S: SinkTestHelper>(sink_test_helper: &S) {
let mut input_buf: Vec<u8> = Vec::new();
let mut output_buf = String::new();
let mut rng = rand::weak_rng();
let line_len_range = Range::new(1, 1020);
let input_len_range = Range::new(1, 10_000);
for _ in 0..5_000 {
input_buf.clear();
output_buf.clear();
let buf_len = input_len_range.ind_sample(&mut rng);
for _ in 0..buf_len {
input_buf.push(rng.gen());
}
let config = random_config_for_chunked_encoder(&mut rng, &line_len_range);
let chunk_encoded_string = sink_test_helper.encode_to_string(config, &input_buf);
encode_config_buf(&input_buf, config, &mut output_buf);
assert_eq!(
output_buf, chunk_encoded_string,
"input len={}, config: pad={}, wrap={:?}",
buf_len, config.pad, config.line_wrap
);
}
}
fn chunked_encode_str(bytes: &[u8], config: Config) -> String {
let mut sink = StringSink::new();
{
let encoder = ChunkedEncoder::new(config).unwrap();
encoder.encode(bytes, &mut sink).unwrap();
}
return sink.string;
}
fn random_config_for_chunked_encoder<R: Rng>(
rng: &mut R,
line_len_range: &Range<usize>,
) -> Config {
loop {
let config = random_config(rng, line_len_range);
// only use a config with line_len that is divisible by 4
match config.line_wrap {
LineWrap::NoWrap => return config,
LineWrap::Wrap(line_len, _) => if line_len % 4 == 0 {
return config;
},
}
}
}
fn config_no_wrap(pad: bool) -> Config {
Config::new(CharacterSet::Standard, pad, false, LineWrap::NoWrap)
}
fn config_wrap(pad: bool, line_len: usize, line_ending: LineEnding) -> Config {
Config::new(
CharacterSet::Standard,
pad,
false,
LineWrap::Wrap(line_len, line_ending),
)
}
// An abstraction around sinks so that we can have tests that easily to any sink implementation
pub trait SinkTestHelper {
fn encode_to_string(&self, config: Config, bytes: &[u8]) -> String;
}
// A really simple sink that just appends to a string for testing
struct StringSink {
string: String,
}
impl StringSink {
fn new() -> StringSink {
StringSink {
string: String::new(),
}
}
}
impl Sink for StringSink {
type Error = ();
fn write_encoded_bytes(&mut self, s: &[u8]) -> Result<(), Self::Error> {
self.string.push_str(str::from_utf8(s).unwrap());
Ok(())
}
}
struct StringSinkTestHelper;
impl SinkTestHelper for StringSinkTestHelper {
fn encode_to_string(&self, config: Config, bytes: &[u8]) -> String {
let encoder = ChunkedEncoder::new(config).unwrap();
let mut sink = StringSink::new();
encoder.encode(bytes, &mut sink).unwrap();
sink.string
}
}
}

710
third_party/rust/base64/src/decode.rs vendored Normal file
View File

@ -0,0 +1,710 @@
use byteorder::{BigEndian, ByteOrder};
use {tables, CharacterSet, Config, STANDARD};
use std::{error, fmt, str};
// decode logic operates on chunks of 8 input bytes without padding
const INPUT_CHUNK_LEN: usize = 8;
const DECODED_CHUNK_LEN: usize = 6;
// we read a u64 and write a u64, but a u64 of input only yields 6 bytes of output, so the last
// 2 bytes of any output u64 should not be counted as written to (but must be available in a
// slice).
const DECODED_CHUNK_SUFFIX: usize = 2;
// how many u64's of input to handle at a time
const CHUNKS_PER_FAST_LOOP_BLOCK: usize = 4;
const INPUT_BLOCK_LEN: usize = CHUNKS_PER_FAST_LOOP_BLOCK * INPUT_CHUNK_LEN;
// includes the trailing 2 bytes for the final u64 write
const DECODED_BLOCK_LEN: usize =
CHUNKS_PER_FAST_LOOP_BLOCK * DECODED_CHUNK_LEN + DECODED_CHUNK_SUFFIX;
/// Errors that can occur while decoding.
#[derive(Clone, Debug, PartialEq, Eq)]
pub enum DecodeError {
/// An invalid byte was found in the input. The offset and offending byte are provided.
InvalidByte(usize, u8),
/// The length of the input is invalid.
InvalidLength,
}
impl fmt::Display for DecodeError {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match *self {
DecodeError::InvalidByte(index, byte) => {
write!(f, "Invalid byte {}, offset {}.", byte, index)
}
DecodeError::InvalidLength => write!(f, "Encoded text cannot have a 6-bit remainder."),
}
}
}
impl error::Error for DecodeError {
fn description(&self) -> &str {
match *self {
DecodeError::InvalidByte(_, _) => "invalid byte",
DecodeError::InvalidLength => "invalid length",
}
}
fn cause(&self) -> Option<&error::Error> {
None
}
}
///Decode from string reference as octets.
///Returns a Result containing a Vec<u8>.
///Convenience `decode_config(input, base64::STANDARD);`.
///
///# Example
///
///```rust
///extern crate base64;
///
///fn main() {
/// let bytes = base64::decode("aGVsbG8gd29ybGQ=").unwrap();
/// println!("{:?}", bytes);
///}
///```
pub fn decode<T: ?Sized + AsRef<[u8]>>(input: &T) -> Result<Vec<u8>, DecodeError> {
decode_config(input, STANDARD)
}
///Decode from string reference as octets.
///Returns a Result containing a Vec<u8>.
///
///# Example
///
///```rust
///extern crate base64;
///
///fn main() {
/// let bytes = base64::decode_config("aGVsbG8gd29ybGR+Cg==", base64::STANDARD).unwrap();
/// println!("{:?}", bytes);
///
/// let bytes_url = base64::decode_config("aGVsbG8gaW50ZXJuZXR-Cg==", base64::URL_SAFE).unwrap();
/// println!("{:?}", bytes_url);
///}
///```
pub fn decode_config<T: ?Sized + AsRef<[u8]>>(
input: &T,
config: Config,
) -> Result<Vec<u8>, DecodeError> {
let mut buffer = Vec::<u8>::with_capacity(input.as_ref().len() * 4 / 3);
decode_config_buf(input, config, &mut buffer).map(|_| buffer)
}
///Decode from string reference as octets.
///Writes into the supplied buffer to avoid allocation.
///Returns a Result containing an empty tuple, aka ().
///
///# Example
///
///```rust
///extern crate base64;
///
///fn main() {
/// let mut buffer = Vec::<u8>::new();
/// base64::decode_config_buf("aGVsbG8gd29ybGR+Cg==", base64::STANDARD, &mut buffer).unwrap();
/// println!("{:?}", buffer);
///
/// buffer.clear();
///
/// base64::decode_config_buf("aGVsbG8gaW50ZXJuZXR-Cg==", base64::URL_SAFE, &mut buffer)
/// .unwrap();
/// println!("{:?}", buffer);
///}
///```
pub fn decode_config_buf<T: ?Sized + AsRef<[u8]>>(
input: &T,
config: Config,
buffer: &mut Vec<u8>,
) -> Result<(), DecodeError> {
let input_copy;
let input_bytes = if config.strip_whitespace {
input_copy = copy_without_whitespace(input.as_ref());
input_copy.as_ref()
} else {
input.as_ref()
};
let starting_output_len = buffer.len();
let num_chunks = num_chunks(input_bytes);
let decoded_len_estimate = num_chunks
.checked_mul(DECODED_CHUNK_LEN)
.and_then(|p| p.checked_add(starting_output_len))
.expect("Overflow when calculating output buffer length");
buffer.resize(decoded_len_estimate, 0);
let bytes_written;
{
let buffer_slice = &mut buffer.as_mut_slice()[starting_output_len..];
bytes_written = decode_helper(input_bytes, num_chunks, &config.char_set, buffer_slice)?;
}
buffer.truncate(starting_output_len + bytes_written);
Ok(())
}
/// Decode the input into the provided output slice.
///
/// This will not write any bytes past exactly what is decoded (no stray garbage bytes at the end).
///
/// If you don't know ahead of time what the decoded length should be, size your buffer with a
/// conservative estimate for the decoded length of an input: 3 bytes of output for every 4 bytes of
/// input, rounded up, or in other words `(input_len + 3) / 4 * 3`.
///
/// If the slice is not large enough, this will panic.
pub fn decode_config_slice<T: ?Sized + AsRef<[u8]>>(
input: &T,
config: Config,
output: &mut [u8],
) -> Result<usize, DecodeError> {
let input_copy;
let input_bytes = if config.strip_whitespace {
input_copy = copy_without_whitespace(input.as_ref());
input_copy.as_ref()
} else {
input.as_ref()
};
decode_helper(
input_bytes,
num_chunks(input_bytes),
&config.char_set,
output,
)
}
/// Return the number of input chunks (including a possibly partial final chunk) in the input
fn num_chunks(input: &[u8]) -> usize {
input
.len()
.checked_add(INPUT_CHUNK_LEN - 1)
.expect("Overflow when calculating number of chunks in input") / INPUT_CHUNK_LEN
}
fn copy_without_whitespace(input: &[u8]) -> Vec<u8> {
let mut input_copy = Vec::<u8>::with_capacity(input.len());
input_copy.extend(input.iter().filter(|b| !b" \n\t\r\x0b\x0c".contains(b)));
input_copy
}
/// Helper to avoid duplicating num_chunks calculation, which is costly on short inputs.
/// Returns the number of bytes written, or an error.
// We're on the fragile edge of compiler heuristics here. If this is not inlined, slow. If this is
// inlined(always), a different slow. plain ol' inline makes the benchmarks happiest at the moment,
// but this is fragile and the best setting changes with only minor code modifications.
#[inline]
fn decode_helper(
input: &[u8],
num_chunks: usize,
char_set: &CharacterSet,
output: &mut [u8],
) -> Result<usize, DecodeError> {
let decode_table = char_set.decode_table();
let remainder_len = input.len() % INPUT_CHUNK_LEN;
// Because the fast decode loop writes in groups of 8 bytes (unrolled to
// CHUNKS_PER_FAST_LOOP_BLOCK times 8 bytes, where possible) and outputs 8 bytes at a time (of
// which only 6 are valid data), we need to be sure that we stop using the fast decode loop
// soon enough that there will always be 2 more bytes of valid data written after that loop.
let trailing_bytes_to_skip = match remainder_len {
// if input is a multiple of the chunk size, ignore the last chunk as it may have padding,
// and the fast decode logic cannot handle padding
0 => INPUT_CHUNK_LEN,
// 1 and 5 trailing bytes are illegal: can't decode 6 bits of input into a byte
1 | 5 => return Err(DecodeError::InvalidLength),
// This will decode to one output byte, which isn't enough to overwrite the 2 extra bytes
// written by the fast decode loop. So, we have to ignore both these 2 bytes and the
// previous chunk.
2 => INPUT_CHUNK_LEN + 2,
// If this is 3 unpadded chars, then it would actually decode to 2 bytes. However, if this
// is an erroneous 2 chars + 1 pad char that would decode to 1 byte, then it should fail
// with an error, not panic from going past the bounds of the output slice, so we let it
// use stage 3 + 4.
3 => INPUT_CHUNK_LEN + 3,
// This can also decode to one output byte because it may be 2 input chars + 2 padding
// chars, which would decode to 1 byte.
4 => INPUT_CHUNK_LEN + 4,
// Everything else is a legal decode len (given that we don't require padding), and will
// decode to at least 2 bytes of output.
_ => remainder_len,
};
// rounded up to include partial chunks
let mut remaining_chunks = num_chunks;
let mut input_index = 0;
let mut output_index = 0;
{
let length_of_fast_decode_chunks = input.len().saturating_sub(trailing_bytes_to_skip);
// Fast loop, stage 1
// manual unroll to CHUNKS_PER_FAST_LOOP_BLOCK of u64s to amortize slice bounds checks
if let Some(max_start_index) = length_of_fast_decode_chunks.checked_sub(INPUT_BLOCK_LEN) {
while input_index <= max_start_index {
let input_slice = &input[input_index..(input_index + INPUT_BLOCK_LEN)];
let output_slice = &mut output[output_index..(output_index + DECODED_BLOCK_LEN)];
decode_chunk(
&input_slice[0..],
input_index,
decode_table,
&mut output_slice[0..],
)?;
decode_chunk(
&input_slice[8..],
input_index + 8,
decode_table,
&mut output_slice[6..],
)?;
decode_chunk(
&input_slice[16..],
input_index + 16,
decode_table,
&mut output_slice[12..],
)?;
decode_chunk(
&input_slice[24..],
input_index + 24,
decode_table,
&mut output_slice[18..],
)?;
input_index += INPUT_BLOCK_LEN;
output_index += DECODED_BLOCK_LEN - DECODED_CHUNK_SUFFIX;
remaining_chunks -= CHUNKS_PER_FAST_LOOP_BLOCK;
}
}
// Fast loop, stage 2 (aka still pretty fast loop)
// 8 bytes at a time for whatever we didn't do in stage 1.
if let Some(max_start_index) = length_of_fast_decode_chunks.checked_sub(INPUT_CHUNK_LEN) {
while input_index < max_start_index {
decode_chunk(
&input[input_index..(input_index + INPUT_CHUNK_LEN)],
input_index,
decode_table,
&mut output
[output_index..(output_index + DECODED_CHUNK_LEN + DECODED_CHUNK_SUFFIX)],
)?;
output_index += DECODED_CHUNK_LEN;
input_index += INPUT_CHUNK_LEN;
remaining_chunks -= 1;
}
}
}
// Stage 3
// If input length was such that a chunk had to be deferred until after the fast loop
// because decoding it would have produced 2 trailing bytes that wouldn't then be
// overwritten, we decode that chunk here. This way is slower but doesn't write the 2
// trailing bytes.
// However, we still need to avoid the last chunk (partial or complete) because it could
// have padding, so we always do 1 fewer to avoid the last chunk.
for _ in 1..remaining_chunks {
decode_chunk_precise(
&input[input_index..],
input_index,
decode_table,
&mut output[output_index..(output_index + DECODED_CHUNK_LEN)],
)?;
input_index += INPUT_CHUNK_LEN;
output_index += DECODED_CHUNK_LEN;
}
// Stage 4
// Finally, decode any leftovers that aren't a complete input block of 8 bytes.
// Use a u64 as a stack-resident 8 byte buffer.
let mut leftover_bits: u64 = 0;
let mut morsels_in_leftover = 0;
let mut padding_bytes = 0;
let mut first_padding_index: usize = 0;
let start_of_leftovers = input_index;
for (i, b) in input[start_of_leftovers..].iter().enumerate() {
// '=' padding
if *b == 0x3D {
// There can be bad padding in a few ways:
// 1 - Padding with non-padding characters after it
// 2 - Padding after zero or one non-padding characters before it
// in the current quad.
// 3 - More than two characters of padding. If 3 or 4 padding chars
// are in the same quad, that implies it will be caught by #2.
// If it spreads from one quad to another, it will be caught by
// #2 in the second quad.
if i % 4 < 2 {
// Check for case #2.
let bad_padding_index = start_of_leftovers + if padding_bytes > 0 {
// If we've already seen padding, report the first padding index.
// This is to be consistent with the faster logic above: it will report an
// error on the first padding character (since it doesn't expect to see
// anything but actual encoded data).
first_padding_index
} else {
// haven't seen padding before, just use where we are now
i
};
return Err(DecodeError::InvalidByte(bad_padding_index, *b));
}
if padding_bytes == 0 {
first_padding_index = i;
}
padding_bytes += 1;
continue;
}
// Check for case #1.
// To make '=' handling consistent with the main loop, don't allow
// non-suffix '=' in trailing chunk either. Report error as first
// erroneous padding.
if padding_bytes > 0 {
return Err(DecodeError::InvalidByte(
start_of_leftovers + first_padding_index,
0x3D,
));
}
// can use up to 8 * 6 = 48 bits of the u64, if last chunk has no padding.
// To minimize shifts, pack the leftovers from left to right.
let shift = 64 - (morsels_in_leftover + 1) * 6;
// tables are all 256 elements, lookup with a u8 index always succeeds
let morsel = decode_table[*b as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(start_of_leftovers + i, *b));
}
leftover_bits |= (morsel as u64) << shift;
morsels_in_leftover += 1;
}
let leftover_bits_ready_to_append = match morsels_in_leftover {
0 => 0,
2 => 8,
3 => 16,
4 => 24,
6 => 32,
7 => 40,
8 => 48,
_ => unreachable!(
"Impossible: must only have 0 to 8 input bytes in last chunk, with no invalid lengths"
),
};
let mut leftover_bits_appended_to_buf = 0;
while leftover_bits_appended_to_buf < leftover_bits_ready_to_append {
// `as` simply truncates the higher bits, which is what we want here
let selected_bits = (leftover_bits >> (56 - leftover_bits_appended_to_buf)) as u8;
output[output_index] = selected_bits;
output_index += 1;
leftover_bits_appended_to_buf += 8;
}
Ok(output_index)
}
/// Decode 8 bytes of input into 6 bytes of output. 8 bytes of output will be written, but only the
/// first 6 of those contain meaningful data.
///
/// `input` is the bytes to decode, of which the first 8 bytes will be processed.
/// `index_at_start_of_input` is the offset in the overall input (used for reporting errors
/// accurately)
/// `decode_table` is the lookup table for the particular base64 alphabet.
/// `output` will have its first 8 bytes overwritten, of which only the first 6 are valid decoded
/// data.
// yes, really inline (worth 30-50% speedup)
#[inline(always)]
fn decode_chunk(
input: &[u8],
index_at_start_of_input: usize,
decode_table: &[u8; 256],
output: &mut [u8],
) -> Result<(), DecodeError> {
let mut accum: u64;
let morsel = decode_table[input[0] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(index_at_start_of_input, input[0]));
}
accum = (morsel as u64) << 58;
let morsel = decode_table[input[1] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(
index_at_start_of_input + 1,
input[1],
));
}
accum |= (morsel as u64) << 52;
let morsel = decode_table[input[2] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(
index_at_start_of_input + 2,
input[2],
));
}
accum |= (morsel as u64) << 46;
let morsel = decode_table[input[3] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(
index_at_start_of_input + 3,
input[3],
));
}
accum |= (morsel as u64) << 40;
let morsel = decode_table[input[4] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(
index_at_start_of_input + 4,
input[4],
));
}
accum |= (morsel as u64) << 34;
let morsel = decode_table[input[5] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(
index_at_start_of_input + 5,
input[5],
));
}
accum |= (morsel as u64) << 28;
let morsel = decode_table[input[6] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(
index_at_start_of_input + 6,
input[6],
));
}
accum |= (morsel as u64) << 22;
let morsel = decode_table[input[7] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(
index_at_start_of_input + 7,
input[7],
));
}
accum |= (morsel as u64) << 16;
BigEndian::write_u64(output, accum);
Ok(())
}
/// Decode an 8-byte chunk, but only write the 6 bytes actually decoded instead of including 2
/// trailing garbage bytes.
#[inline]
fn decode_chunk_precise(
input: &[u8],
index_at_start_of_input: usize,
decode_table: &[u8; 256],
output: &mut [u8],
) -> Result<(), DecodeError> {
let mut tmp_buf = [0_u8; 8];
decode_chunk(
input,
index_at_start_of_input,
decode_table,
&mut tmp_buf[..],
)?;
output[0..6].copy_from_slice(&tmp_buf[0..6]);
Ok(())
}
#[cfg(test)]
mod tests {
extern crate rand;
use super::*;
use encode::encode_config_buf;
use tests::{assert_encode_sanity, random_config};
use self::rand::distributions::{IndependentSample, Range};
use self::rand::Rng;
#[test]
fn decode_chunk_precise_writes_only_6_bytes() {
let input = b"Zm9vYmFy"; // "foobar"
let mut output = [0_u8, 1, 2, 3, 4, 5, 6, 7];
decode_chunk_precise(&input[..], 0, tables::STANDARD_DECODE, &mut output).unwrap();
assert_eq!(&vec![b'f', b'o', b'o', b'b', b'a', b'r', 6, 7], &output);
}
#[test]
fn decode_chunk_writes_8_bytes() {
let input = b"Zm9vYmFy"; // "foobar"
let mut output = [0_u8, 1, 2, 3, 4, 5, 6, 7];
decode_chunk(&input[..], 0, tables::STANDARD_DECODE, &mut output).unwrap();
assert_eq!(&vec![b'f', b'o', b'o', b'b', b'a', b'r', 0, 0], &output);
}
#[test]
fn decode_into_nonempty_vec_doesnt_clobber_existing_prefix() {
let mut orig_data = Vec::new();
let mut encoded_data = String::new();
let mut decoded_with_prefix = Vec::new();
let mut decoded_without_prefix = Vec::new();
let mut prefix = Vec::new();
let prefix_len_range = Range::new(0, 1000);
let input_len_range = Range::new(0, 1000);
let line_len_range = Range::new(1, 1000);
let mut rng = rand::weak_rng();
for _ in 0..10_000 {
orig_data.clear();
encoded_data.clear();
decoded_with_prefix.clear();
decoded_without_prefix.clear();
prefix.clear();
let input_len = input_len_range.ind_sample(&mut rng);
for _ in 0..input_len {
orig_data.push(rng.gen());
}
let config = random_config(&mut rng, &line_len_range);
encode_config_buf(&orig_data, config, &mut encoded_data);
assert_encode_sanity(&encoded_data, &config, input_len);
let prefix_len = prefix_len_range.ind_sample(&mut rng);
// fill the buf with a prefix
for _ in 0..prefix_len {
prefix.push(rng.gen());
}
decoded_with_prefix.resize(prefix_len, 0);
decoded_with_prefix.copy_from_slice(&prefix);
// decode into the non-empty buf
decode_config_buf(&encoded_data, config, &mut decoded_with_prefix).unwrap();
// also decode into the empty buf
decode_config_buf(&encoded_data, config, &mut decoded_without_prefix).unwrap();
assert_eq!(
prefix_len + decoded_without_prefix.len(),
decoded_with_prefix.len()
);
assert_eq!(orig_data, decoded_without_prefix);
// append plain decode onto prefix
prefix.append(&mut decoded_without_prefix);
assert_eq!(prefix, decoded_with_prefix);
}
}
#[test]
fn decode_into_slice_doesnt_clobber_existing_prefix_or_suffix() {
let mut orig_data = Vec::new();
let mut encoded_data = String::new();
let mut decode_buf = Vec::new();
let mut decode_buf_copy: Vec<u8> = Vec::new();
let input_len_range = Range::new(0, 1000);
let line_len_range = Range::new(1, 1000);
let mut rng = rand::weak_rng();
for _ in 0..10_000 {
orig_data.clear();
encoded_data.clear();
decode_buf.clear();
decode_buf_copy.clear();
let input_len = input_len_range.ind_sample(&mut rng);
for _ in 0..input_len {
orig_data.push(rng.gen());
}
let config = random_config(&mut rng, &line_len_range);
encode_config_buf(&orig_data, config, &mut encoded_data);
assert_encode_sanity(&encoded_data, &config, input_len);
// fill the buffer with random garbage, long enough to have some room before and after
for _ in 0..5000 {
decode_buf.push(rng.gen());
}
// keep a copy for later comparison
decode_buf_copy.extend(decode_buf.iter());
let offset = 1000;
// decode into the non-empty buf
let decode_bytes_written =
decode_config_slice(&encoded_data, config, &mut decode_buf[offset..]).unwrap();
assert_eq!(orig_data.len(), decode_bytes_written);
assert_eq!(
orig_data,
&decode_buf[offset..(offset + decode_bytes_written)]
);
assert_eq!(&decode_buf_copy[0..offset], &decode_buf[0..offset]);
assert_eq!(
&decode_buf_copy[offset + decode_bytes_written..],
&decode_buf[offset + decode_bytes_written..]
);
}
}
#[test]
fn decode_into_slice_fits_in_precisely_sized_slice() {
let mut orig_data = Vec::new();
let mut encoded_data = String::new();
let mut decode_buf = Vec::new();
let input_len_range = Range::new(0, 1000);
let line_len_range = Range::new(1, 1000);
let mut rng = rand::weak_rng();
for _ in 0..10_000 {
orig_data.clear();
encoded_data.clear();
decode_buf.clear();
let input_len = input_len_range.ind_sample(&mut rng);
for _ in 0..input_len {
orig_data.push(rng.gen());
}
let config = random_config(&mut rng, &line_len_range);
encode_config_buf(&orig_data, config, &mut encoded_data);
assert_encode_sanity(&encoded_data, &config, input_len);
decode_buf.resize(input_len, 0);
// decode into the non-empty buf
let decode_bytes_written =
decode_config_slice(&encoded_data, config, &mut decode_buf[..]).unwrap();
assert_eq!(orig_data.len(), decode_bytes_written);
assert_eq!(orig_data, decode_buf);
}
}
}

112
third_party/rust/base64/src/display.rs vendored Normal file
View File

@ -0,0 +1,112 @@
//! Enables base64'd output anywhere you might use a `Display` implementation, like a format string.
//!
//! ```
//! use base64::display::Base64Display;
//!
//! let data = vec![0x0, 0x1, 0x2, 0x3];
//! let wrapper = Base64Display::standard(&data);
//!
//! assert_eq!("base64: AAECAw==", format!("base64: {}", wrapper));
//! ```
use super::chunked_encoder::{ChunkedEncoder, ChunkedEncoderError};
use super::Config;
use std::fmt::{Display, Formatter};
use std::{fmt, str};
// I'm not convinced that we should expose ChunkedEncoder or its error type since it's just an
// implementation detail, so use a different error type.
/// Errors that can occur initializing a Base64Display.
#[derive(Debug, PartialEq)]
pub enum DisplayError {
/// If wrapping is configured, the line length must be a multiple of 4, and must not be absurdly
/// large (currently capped at 1024, subject to change).
InvalidLineLength,
}
/// A convenience wrapper for base64'ing bytes into a format string without heap allocation.
pub struct Base64Display<'a> {
bytes: &'a [u8],
chunked_encoder: ChunkedEncoder,
}
impl<'a> Base64Display<'a> {
/// Create a `Base64Display` with the provided config.
pub fn with_config(bytes: &[u8], config: Config) -> Result<Base64Display, DisplayError> {
ChunkedEncoder::new(config)
.map(|c| Base64Display {
bytes,
chunked_encoder: c,
})
.map_err(|e| match e {
ChunkedEncoderError::InvalidLineLength => DisplayError::InvalidLineLength,
})
}
/// Convenience method for creating a `Base64Display` with the `STANDARD` configuration.
pub fn standard(bytes: &[u8]) -> Base64Display {
Base64Display::with_config(bytes, super::STANDARD).expect("STANDARD is valid")
}
/// Convenience method for creating a `Base64Display` with the `URL_SAFE` configuration.
pub fn url_safe(bytes: &[u8]) -> Base64Display {
Base64Display::with_config(bytes, super::URL_SAFE).expect("URL_SAFE is valid")
}
}
impl<'a> Display for Base64Display<'a> {
fn fmt(&self, formatter: &mut Formatter) -> Result<(), fmt::Error> {
let mut sink = FormatterSink { f: formatter };
self.chunked_encoder.encode(self.bytes, &mut sink)
}
}
struct FormatterSink<'a, 'b: 'a> {
f: &'a mut Formatter<'b>,
}
impl<'a, 'b: 'a> super::chunked_encoder::Sink for FormatterSink<'a, 'b> {
type Error = fmt::Error;
fn write_encoded_bytes(&mut self, encoded: &[u8]) -> Result<(), Self::Error> {
// Avoid unsafe. If max performance is needed, write your own display wrapper that uses
// unsafe here to gain about 10-15%.
self.f
.write_str(str::from_utf8(encoded).expect("base64 data was not utf8"))
}
}
#[cfg(test)]
mod tests {
use super::super::chunked_encoder::tests::{chunked_encode_matches_normal_encode_random,
SinkTestHelper};
use super::super::*;
use super::*;
#[test]
fn basic_display() {
assert_eq!(
"~$Zm9vYmFy#*",
format!("~${}#*", Base64Display::standard("foobar".as_bytes()))
);
assert_eq!(
"~$Zm9vYmFyZg==#*",
format!("~${}#*", Base64Display::standard("foobarf".as_bytes()))
);
}
#[test]
fn display_encode_matches_normal_encode() {
let helper = DisplaySinkTestHelper;
chunked_encode_matches_normal_encode_random(&helper);
}
struct DisplaySinkTestHelper;
impl SinkTestHelper for DisplaySinkTestHelper {
fn encode_to_string(&self, config: Config, bytes: &[u8]) -> String {
format!("{}", Base64Display::with_config(bytes, config).unwrap())
}
}
}

761
third_party/rust/base64/src/encode.rs vendored Normal file
View File

@ -0,0 +1,761 @@
use byteorder::{BigEndian, ByteOrder};
use {line_wrap, line_wrap_parameters, Config, LineWrap, STANDARD};
///Encode arbitrary octets as base64.
///Returns a String.
///Convenience for `encode_config(input, base64::STANDARD);`.
///
///# Example
///
///```rust
///extern crate base64;
///
///fn main() {
/// let b64 = base64::encode(b"hello world");
/// println!("{}", b64);
///}
///```
pub fn encode<T: ?Sized + AsRef<[u8]>>(input: &T) -> String {
encode_config(input, STANDARD)
}
///Encode arbitrary octets as base64.
///Returns a String.
///
///# Example
///
///```rust
///extern crate base64;
///
///fn main() {
/// let b64 = base64::encode_config(b"hello world~", base64::STANDARD);
/// println!("{}", b64);
///
/// let b64_url = base64::encode_config(b"hello internet~", base64::URL_SAFE);
/// println!("{}", b64_url);
///}
///```
pub fn encode_config<T: ?Sized + AsRef<[u8]>>(input: &T, config: Config) -> String {
let mut buf = match encoded_size(input.as_ref().len(), &config) {
Some(n) => String::with_capacity(n),
None => panic!("integer overflow when calculating buffer size"),
};
encode_config_buf(input, config, &mut buf);
buf
}
///Encode arbitrary octets as base64.
///Writes into the supplied output buffer, which will grow the buffer if needed.
///
///# Example
///
///```rust
///extern crate base64;
///
///fn main() {
/// let mut buf = String::new();
/// base64::encode_config_buf(b"hello world~", base64::STANDARD, &mut buf);
/// println!("{}", buf);
///
/// buf.clear();
/// base64::encode_config_buf(b"hello internet~", base64::URL_SAFE, &mut buf);
/// println!("{}", buf);
///}
///```
pub fn encode_config_buf<T: ?Sized + AsRef<[u8]>>(input: &T, config: Config, buf: &mut String) {
let input_bytes = input.as_ref();
let encoded_size = encoded_size(input_bytes.len(), &config)
.expect("usize overflow when calculating buffer size");
let orig_buf_len = buf.len();
// we're only going to insert valid utf8
let buf_bytes;
unsafe {
buf_bytes = buf.as_mut_vec();
}
buf_bytes.resize(
orig_buf_len
.checked_add(encoded_size)
.expect("usize overflow when calculating expanded buffer size"),
0,
);
let mut b64_output = &mut buf_bytes[orig_buf_len..];
encode_with_padding_line_wrap(&input_bytes, &config, encoded_size, &mut b64_output);
}
/// Encode arbitrary octets as base64.
/// Writes into the supplied output buffer.
///
/// This is useful if you wish to avoid allocation entirely (e.g. encoding into a stack-resident
/// or statically-allocated buffer).
///
/// # Panics
///
/// If `output` is too small to hold the encoded version of `input`, a panic will result.
///
/// # Example
///
/// ```rust
/// extern crate base64;
///
/// fn main() {
/// let s = b"hello internet!";
/// let mut buf = Vec::new();
/// // make sure we'll have a slice big enough for base64 + padding
/// buf.resize(s.len() * 4 / 3 + 4, 0);
///
/// let bytes_written = base64::encode_config_slice(s,
/// base64::STANDARD, &mut buf);
///
/// // shorten our vec down to just what was written
/// buf.resize(bytes_written, 0);
///
/// assert_eq!(s, base64::decode(&buf).unwrap().as_slice());
/// }
/// ```
pub fn encode_config_slice<T: ?Sized + AsRef<[u8]>>(
input: &T,
config: Config,
output: &mut [u8],
) -> usize {
let input_bytes = input.as_ref();
let encoded_size = encoded_size(input_bytes.len(), &config)
.expect("usize overflow when calculating buffer size");
let mut b64_output = &mut output[0..encoded_size];
encode_with_padding_line_wrap(&input_bytes, &config, encoded_size, &mut b64_output);
encoded_size
}
/// B64-encode, pad, and line wrap (if configured).
///
/// This helper exists to avoid recalculating encoded_size, which is relatively expensive on short
/// inputs.
///
/// `encoded_size` is the encoded size calculated for `input`.
///
/// `output` must be of size `encoded_size`.
///
/// All bytes in `output` will be written to since it is exactly the size of the output.
pub fn encode_with_padding_line_wrap(
input: &[u8],
config: &Config,
encoded_size: usize,
output: &mut [u8],
) {
debug_assert_eq!(encoded_size, output.len());
let b64_bytes_written = encode_to_slice(input, output, config.char_set.encode_table());
let padding_bytes = if config.pad {
add_padding(input.len(), &mut output[b64_bytes_written..])
} else {
0
};
let encoded_bytes = b64_bytes_written
.checked_add(padding_bytes)
.expect("usize overflow when calculating b64 length");
let line_ending_bytes = if let LineWrap::Wrap(line_len, line_end) = config.line_wrap {
line_wrap(output, encoded_bytes, line_len, line_end)
} else {
0
};
debug_assert_eq!(encoded_size, encoded_bytes + line_ending_bytes);
}
/// Encode input bytes to utf8 base64 bytes. Does not pad or line wrap.
/// `output` must be long enough to hold the encoded `input` without padding or line wrapping.
/// Returns the number of bytes written.
#[inline]
pub fn encode_to_slice(input: &[u8], output: &mut [u8], encode_table: &[u8; 64]) -> usize {
let mut input_index: usize = 0;
const BLOCKS_PER_FAST_LOOP: usize = 4;
const LOW_SIX_BITS: u64 = 0x3F;
// we read 8 bytes at a time (u64) but only actually consume 6 of those bytes. Thus, we need
// 2 trailing bytes to be available to read..
let last_fast_index = input.len().saturating_sub(BLOCKS_PER_FAST_LOOP * 6 + 2);
let mut output_index = 0;
if last_fast_index > 0 {
while input_index <= last_fast_index {
// Major performance wins from letting the optimizer do the bounds check once, mostly
// on the output side
let input_chunk = &input[input_index..(input_index + (BLOCKS_PER_FAST_LOOP * 6 + 2))];
let output_chunk = &mut output[output_index..(output_index + BLOCKS_PER_FAST_LOOP * 8)];
// Hand-unrolling for 32 vs 16 or 8 bytes produces yields performance about equivalent
// to unsafe pointer code on a Xeon E5-1650v3. 64 byte unrolling was slightly better for
// large inputs but significantly worse for 50-byte input, unsurprisingly. I suspect
// that it's a not uncommon use case to encode smallish chunks of data (e.g. a 64-byte
// SHA-512 digest), so it would be nice if that fit in the unrolled loop at least once.
// Plus, single-digit percentage performance differences might well be quite different
// on different hardware.
let input_u64 = BigEndian::read_u64(&input_chunk[0..]);
output_chunk[0] = encode_table[((input_u64 >> 58) & LOW_SIX_BITS) as usize];
output_chunk[1] = encode_table[((input_u64 >> 52) & LOW_SIX_BITS) as usize];
output_chunk[2] = encode_table[((input_u64 >> 46) & LOW_SIX_BITS) as usize];
output_chunk[3] = encode_table[((input_u64 >> 40) & LOW_SIX_BITS) as usize];
output_chunk[4] = encode_table[((input_u64 >> 34) & LOW_SIX_BITS) as usize];
output_chunk[5] = encode_table[((input_u64 >> 28) & LOW_SIX_BITS) as usize];
output_chunk[6] = encode_table[((input_u64 >> 22) & LOW_SIX_BITS) as usize];
output_chunk[7] = encode_table[((input_u64 >> 16) & LOW_SIX_BITS) as usize];
let input_u64 = BigEndian::read_u64(&input_chunk[6..]);
output_chunk[8] = encode_table[((input_u64 >> 58) & LOW_SIX_BITS) as usize];
output_chunk[9] = encode_table[((input_u64 >> 52) & LOW_SIX_BITS) as usize];
output_chunk[10] = encode_table[((input_u64 >> 46) & LOW_SIX_BITS) as usize];
output_chunk[11] = encode_table[((input_u64 >> 40) & LOW_SIX_BITS) as usize];
output_chunk[12] = encode_table[((input_u64 >> 34) & LOW_SIX_BITS) as usize];
output_chunk[13] = encode_table[((input_u64 >> 28) & LOW_SIX_BITS) as usize];
output_chunk[14] = encode_table[((input_u64 >> 22) & LOW_SIX_BITS) as usize];
output_chunk[15] = encode_table[((input_u64 >> 16) & LOW_SIX_BITS) as usize];
let input_u64 = BigEndian::read_u64(&input_chunk[12..]);
output_chunk[16] = encode_table[((input_u64 >> 58) & LOW_SIX_BITS) as usize];
output_chunk[17] = encode_table[((input_u64 >> 52) & LOW_SIX_BITS) as usize];
output_chunk[18] = encode_table[((input_u64 >> 46) & LOW_SIX_BITS) as usize];
output_chunk[19] = encode_table[((input_u64 >> 40) & LOW_SIX_BITS) as usize];
output_chunk[20] = encode_table[((input_u64 >> 34) & LOW_SIX_BITS) as usize];
output_chunk[21] = encode_table[((input_u64 >> 28) & LOW_SIX_BITS) as usize];
output_chunk[22] = encode_table[((input_u64 >> 22) & LOW_SIX_BITS) as usize];
output_chunk[23] = encode_table[((input_u64 >> 16) & LOW_SIX_BITS) as usize];
let input_u64 = BigEndian::read_u64(&input_chunk[18..]);
output_chunk[24] = encode_table[((input_u64 >> 58) & LOW_SIX_BITS) as usize];
output_chunk[25] = encode_table[((input_u64 >> 52) & LOW_SIX_BITS) as usize];
output_chunk[26] = encode_table[((input_u64 >> 46) & LOW_SIX_BITS) as usize];
output_chunk[27] = encode_table[((input_u64 >> 40) & LOW_SIX_BITS) as usize];
output_chunk[28] = encode_table[((input_u64 >> 34) & LOW_SIX_BITS) as usize];
output_chunk[29] = encode_table[((input_u64 >> 28) & LOW_SIX_BITS) as usize];
output_chunk[30] = encode_table[((input_u64 >> 22) & LOW_SIX_BITS) as usize];
output_chunk[31] = encode_table[((input_u64 >> 16) & LOW_SIX_BITS) as usize];
output_index += BLOCKS_PER_FAST_LOOP * 8;
input_index += BLOCKS_PER_FAST_LOOP * 6;
}
}
// Encode what's left after the fast loop.
const LOW_SIX_BITS_U8: u8 = 0x3F;
let rem = input.len() % 3;
let start_of_rem = input.len() - rem;
// start at the first index not handled by fast loop, which may be 0.
while input_index < start_of_rem {
let input_chunk = &input[input_index..(input_index + 3)];
let output_chunk = &mut output[output_index..(output_index + 4)];
output_chunk[0] = encode_table[(input_chunk[0] >> 2) as usize];
output_chunk[1] =
encode_table[((input_chunk[0] << 4 | input_chunk[1] >> 4) & LOW_SIX_BITS_U8) as usize];
output_chunk[2] =
encode_table[((input_chunk[1] << 2 | input_chunk[2] >> 6) & LOW_SIX_BITS_U8) as usize];
output_chunk[3] = encode_table[(input_chunk[2] & LOW_SIX_BITS_U8) as usize];
input_index += 3;
output_index += 4;
}
if rem == 2 {
output[output_index] = encode_table[(input[start_of_rem] >> 2) as usize];
output[output_index + 1] =
encode_table[((input[start_of_rem] << 4 | input[start_of_rem + 1] >> 4)
& LOW_SIX_BITS_U8) as usize];
output[output_index + 2] =
encode_table[((input[start_of_rem + 1] << 2) & LOW_SIX_BITS_U8) as usize];
output_index += 3;
} else if rem == 1 {
output[output_index] = encode_table[(input[start_of_rem] >> 2) as usize];
output[output_index + 1] =
encode_table[((input[start_of_rem] << 4) & LOW_SIX_BITS_U8) as usize];
output_index += 2;
}
output_index
}
/// calculate the base64 encoded string size, including padding and line wraps if appropriate
pub fn encoded_size(bytes_len: usize, config: &Config) -> Option<usize> {
let rem = bytes_len % 3;
let complete_input_chunks = bytes_len / 3;
let complete_chunk_output = complete_input_chunks.checked_mul(4);
let encoded_len_no_wrap = if rem > 0 {
if config.pad {
complete_chunk_output.and_then(|c| c.checked_add(4))
} else {
let encoded_rem = match rem {
1 => 2,
2 => 3,
_ => unreachable!("Impossible remainder"),
};
complete_chunk_output.and_then(|c| c.checked_add(encoded_rem))
}
} else {
complete_chunk_output
};
encoded_len_no_wrap.map(|e| match config.line_wrap {
LineWrap::NoWrap => e,
LineWrap::Wrap(line_len, line_ending) => {
line_wrap_parameters(e, line_len, line_ending).total_len
}
})
}
/// Write padding characters.
/// `output` is the slice where padding should be written, of length at least 2.
///
/// Returns the number of padding bytes written.
pub fn add_padding(input_len: usize, output: &mut [u8]) -> usize {
let rem = input_len % 3;
let mut bytes_written = 0;
for _ in 0..((3 - rem) % 3) {
output[bytes_written] = b'=';
bytes_written += 1;
}
bytes_written
}
#[cfg(test)]
mod tests {
extern crate rand;
use super::*;
use decode::decode_config_buf;
use tests::{assert_encode_sanity, random_config};
use {CharacterSet, Config, LineEnding, LineWrap, MIME, STANDARD, URL_SAFE_NO_PAD};
use self::rand::distributions::{IndependentSample, Range};
use self::rand::Rng;
use std;
use std::str;
#[test]
fn encoded_size_correct_standard() {
assert_encoded_length(0, 0, STANDARD);
assert_encoded_length(1, 4, STANDARD);
assert_encoded_length(2, 4, STANDARD);
assert_encoded_length(3, 4, STANDARD);
assert_encoded_length(4, 8, STANDARD);
assert_encoded_length(5, 8, STANDARD);
assert_encoded_length(6, 8, STANDARD);
assert_encoded_length(7, 12, STANDARD);
assert_encoded_length(8, 12, STANDARD);
assert_encoded_length(9, 12, STANDARD);
assert_encoded_length(54, 72, STANDARD);
assert_encoded_length(55, 76, STANDARD);
assert_encoded_length(56, 76, STANDARD);
assert_encoded_length(57, 76, STANDARD);
assert_encoded_length(58, 80, STANDARD);
}
#[test]
fn encoded_size_correct_no_pad_no_wrap() {
assert_encoded_length(0, 0, URL_SAFE_NO_PAD);
assert_encoded_length(1, 2, URL_SAFE_NO_PAD);
assert_encoded_length(2, 3, URL_SAFE_NO_PAD);
assert_encoded_length(3, 4, URL_SAFE_NO_PAD);
assert_encoded_length(4, 6, URL_SAFE_NO_PAD);
assert_encoded_length(5, 7, URL_SAFE_NO_PAD);
assert_encoded_length(6, 8, URL_SAFE_NO_PAD);
assert_encoded_length(7, 10, URL_SAFE_NO_PAD);
assert_encoded_length(8, 11, URL_SAFE_NO_PAD);
assert_encoded_length(9, 12, URL_SAFE_NO_PAD);
assert_encoded_length(54, 72, URL_SAFE_NO_PAD);
assert_encoded_length(55, 74, URL_SAFE_NO_PAD);
assert_encoded_length(56, 75, URL_SAFE_NO_PAD);
assert_encoded_length(57, 76, URL_SAFE_NO_PAD);
assert_encoded_length(58, 78, URL_SAFE_NO_PAD);
}
#[test]
fn encoded_size_correct_mime() {
assert_encoded_length(0, 0, MIME);
assert_encoded_length(1, 4, MIME);
assert_encoded_length(2, 4, MIME);
assert_encoded_length(3, 4, MIME);
assert_encoded_length(4, 8, MIME);
assert_encoded_length(5, 8, MIME);
assert_encoded_length(6, 8, MIME);
assert_encoded_length(7, 12, MIME);
assert_encoded_length(8, 12, MIME);
assert_encoded_length(9, 12, MIME);
assert_encoded_length(54, 72, MIME);
assert_encoded_length(55, 76, MIME);
assert_encoded_length(56, 76, MIME);
assert_encoded_length(57, 76, MIME);
assert_encoded_length(58, 82, MIME);
assert_encoded_length(59, 82, MIME);
assert_encoded_length(60, 82, MIME);
}
#[test]
fn encoded_size_correct_lf_pad() {
let config = Config::new(
CharacterSet::Standard,
true,
false,
LineWrap::Wrap(76, LineEnding::LF),
);
assert_encoded_length(0, 0, config);
assert_encoded_length(1, 4, config);
assert_encoded_length(2, 4, config);
assert_encoded_length(3, 4, config);
assert_encoded_length(4, 8, config);
assert_encoded_length(5, 8, config);
assert_encoded_length(6, 8, config);
assert_encoded_length(7, 12, config);
assert_encoded_length(8, 12, config);
assert_encoded_length(9, 12, config);
assert_encoded_length(54, 72, config);
assert_encoded_length(55, 76, config);
assert_encoded_length(56, 76, config);
assert_encoded_length(57, 76, config);
// one fewer than MIME
assert_encoded_length(58, 81, config);
assert_encoded_length(59, 81, config);
assert_encoded_length(60, 81, config);
}
#[test]
fn encoded_size_overflow() {
assert_eq!(None, encoded_size(std::usize::MAX, &STANDARD));
}
#[test]
fn encode_config_buf_into_nonempty_buffer_doesnt_clobber_prefix() {
let mut orig_data = Vec::new();
let mut prefix = String::new();
let mut encoded_data_no_prefix = String::new();
let mut encoded_data_with_prefix = String::new();
let mut decoded = Vec::new();
let prefix_len_range = Range::new(0, 1000);
let input_len_range = Range::new(0, 1000);
let line_len_range = Range::new(1, 1000);
let mut rng = rand::weak_rng();
for _ in 0..10_000 {
orig_data.clear();
prefix.clear();
encoded_data_no_prefix.clear();
encoded_data_with_prefix.clear();
decoded.clear();
let input_len = input_len_range.ind_sample(&mut rng);
for _ in 0..input_len {
orig_data.push(rng.gen());
}
let prefix_len = prefix_len_range.ind_sample(&mut rng);
for _ in 0..prefix_len {
// getting convenient random single-byte printable chars that aren't base64 is
// annoying
prefix.push('#');
}
encoded_data_with_prefix.push_str(&prefix);
let config = random_config(&mut rng, &line_len_range);
encode_config_buf(&orig_data, config, &mut encoded_data_no_prefix);
encode_config_buf(&orig_data, config, &mut encoded_data_with_prefix);
assert_eq!(
encoded_data_no_prefix.len() + prefix_len,
encoded_data_with_prefix.len()
);
assert_encode_sanity(&encoded_data_no_prefix, &config, input_len);
assert_encode_sanity(&encoded_data_with_prefix[prefix_len..], &config, input_len);
// append plain encode onto prefix
prefix.push_str(&mut encoded_data_no_prefix);
assert_eq!(prefix, encoded_data_with_prefix);
decode_config_buf(&encoded_data_no_prefix, config, &mut decoded).unwrap();
assert_eq!(orig_data, decoded);
}
}
#[test]
fn encode_config_slice_into_nonempty_buffer_doesnt_clobber_suffix() {
let mut orig_data = Vec::new();
let mut encoded_data = Vec::new();
let mut encoded_data_original_state = Vec::new();
let mut decoded = Vec::new();
let input_len_range = Range::new(0, 1000);
let line_len_range = Range::new(1, 1000);
let mut rng = rand::weak_rng();
for _ in 0..10_000 {
orig_data.clear();
encoded_data.clear();
encoded_data_original_state.clear();
decoded.clear();
let input_len = input_len_range.ind_sample(&mut rng);
for _ in 0..input_len {
orig_data.push(rng.gen());
}
// plenty of existing garbage in the encoded buffer
for _ in 0..10 * input_len {
encoded_data.push(rng.gen());
}
encoded_data_original_state.extend_from_slice(&encoded_data);
let config = random_config(&mut rng, &line_len_range);
let encoded_size = encoded_size(input_len, &config).unwrap();
assert_eq!(
encoded_size,
encode_config_slice(&orig_data, config, &mut encoded_data)
);
assert_encode_sanity(
std::str::from_utf8(&encoded_data[0..encoded_size]).unwrap(),
&config,
input_len,
);
assert_eq!(
&encoded_data[encoded_size..],
&encoded_data_original_state[encoded_size..]
);
decode_config_buf(&encoded_data[0..encoded_size], config, &mut decoded).unwrap();
assert_eq!(orig_data, decoded);
}
}
#[test]
fn encode_config_slice_fits_into_precisely_sized_slice() {
let mut orig_data = Vec::new();
let mut encoded_data = Vec::new();
let mut decoded = Vec::new();
let input_len_range = Range::new(0, 1000);
let line_len_range = Range::new(1, 1000);
let mut rng = rand::weak_rng();
for _ in 0..10_000 {
orig_data.clear();
encoded_data.clear();
decoded.clear();
let input_len = input_len_range.ind_sample(&mut rng);
for _ in 0..input_len {
orig_data.push(rng.gen());
}
let config = random_config(&mut rng, &line_len_range);
let encoded_size = encoded_size(input_len, &config).unwrap();
encoded_data.resize(encoded_size, 0);
assert_eq!(
encoded_size,
encode_config_slice(&orig_data, config, &mut encoded_data)
);
assert_encode_sanity(
std::str::from_utf8(&encoded_data[0..encoded_size]).unwrap(),
&config,
input_len,
);
decode_config_buf(&encoded_data[0..encoded_size], config, &mut decoded).unwrap();
assert_eq!(orig_data, decoded);
}
}
#[test]
fn encode_to_slice_random_valid_utf8() {
let mut input = Vec::new();
let mut output = Vec::new();
let input_len_range = Range::new(0, 1000);
let line_len_range = Range::new(1, 1000);
let mut rng = rand::weak_rng();
for _ in 0..10_000 {
input.clear();
output.clear();
let input_len = input_len_range.ind_sample(&mut rng);
for _ in 0..input_len {
input.push(rng.gen());
}
let config = random_config(&mut rng, &line_len_range);
// fill up the output buffer with garbage
let encoded_size = encoded_size(input_len, &config).unwrap();
for _ in 0..encoded_size {
output.push(rng.gen());
}
let orig_output_buf = output.to_vec();
let bytes_written =
encode_to_slice(&input, &mut output, config.char_set.encode_table());
// make sure the part beyond bytes_written is the same garbage it was before
assert_eq!(orig_output_buf[bytes_written..], output[bytes_written..]);
// make sure the encoded bytes are UTF-8
let _ = str::from_utf8(&output[0..bytes_written]).unwrap();
}
}
#[test]
fn encode_with_padding_line_wrap_random_valid_utf8() {
let mut input = Vec::new();
let mut output = Vec::new();
let input_len_range = Range::new(0, 1000);
let line_len_range = Range::new(1, 1000);
let mut rng = rand::weak_rng();
for _ in 0..10_000 {
input.clear();
output.clear();
let input_len = input_len_range.ind_sample(&mut rng);
for _ in 0..input_len {
input.push(rng.gen());
}
let config = random_config(&mut rng, &line_len_range);
// fill up the output buffer with garbage
let encoded_size = encoded_size(input_len, &config).unwrap();
for _ in 0..encoded_size + 1000 {
output.push(rng.gen());
}
let orig_output_buf = output.to_vec();
encode_with_padding_line_wrap(
&input,
&config,
encoded_size,
&mut output[0..encoded_size],
);
// make sure the part beyond b64 is the same garbage it was before
assert_eq!(orig_output_buf[encoded_size..], output[encoded_size..]);
// make sure the encoded bytes are UTF-8
let _ = str::from_utf8(&output[0..encoded_size]).unwrap();
}
}
#[test]
fn add_padding_random_valid_utf8() {
let mut output = Vec::new();
let mut rng = rand::weak_rng();
// cover our bases for length % 3
for input_len in 0..10 {
output.clear();
// fill output with random
for _ in 0..10 {
output.push(rng.gen());
}
let orig_output_buf = output.to_vec();
let bytes_written = add_padding(input_len, &mut output);
// make sure the part beyond bytes_written is the same garbage it was before
assert_eq!(orig_output_buf[bytes_written..], output[bytes_written..]);
// make sure the encoded bytes are UTF-8
let _ = str::from_utf8(&output[0..bytes_written]).unwrap();
}
}
fn assert_encoded_length(input_len: usize, encoded_len: usize, config: Config) {
assert_eq!(encoded_len, encoded_size(input_len, &config).unwrap());
let mut bytes: Vec<u8> = Vec::new();
let mut rng = rand::weak_rng();
for _ in 0..input_len {
bytes.push(rng.gen());
}
let encoded = encode_config(&bytes, config);
assert_encode_sanity(&encoded, &config, input_len);
assert_eq!(encoded_len, encoded.len());
}
}

View File

@ -1,13 +1,82 @@
//! # Configs
//!
//! There isn't just one type of Base64; that would be too simple. You need to choose a character
//! set (standard or URL-safe), padding suffix (yes/no), and line wrap (line length, line ending).
//! The `Config` struct encapsulates this info. There are some common configs included: `STANDARD`,
//! `MIME`, etc. You can also make your own `Config` if needed.
//!
//! The functions that don't have `config` in the name (e.g. `encode()` and `decode()`) use the
//! `STANDARD` config .
//!
//! The functions that write to a slice (the ones that end in `_slice`) are generally the fastest
//! because they don't need to resize anything. If it fits in your workflow and you care about
//! performance, keep using the same buffer (growing as need be) and use the `_slice` methods for
//! the best performance.
//!
//! # Encoding
//!
//! Several different encoding functions are available to you depending on your desire for
//! convenience vs performance.
//!
//! | Function | Output | Allocates |
//! | ----------------------- | ---------------------------- | ------------------------------ |
//! | `encode` | Returns a new `String` | Always |
//! | `encode_config` | Returns a new `String` | Always |
//! | `encode_config_buf` | Appends to provided `String` | Only if `String` needs to grow |
//! | `encode_config_slice` | Writes to provided `&[u8]` | Never |
//!
//! All of the encoding functions that take a `Config` will pad, line wrap, etc, as per the config.
//!
//! # Decoding
//!
//! Just as for encoding, there are different decoding functions available.
//!
//! Note that all decode functions that take a config will allocate a copy of the input if you
//! specify a config that requires whitespace to be stripped. If you care about speed, don't use
//! formats that line wrap and then require whitespace stripping.
//!
//! | Function | Output | Allocates |
//! | ----------------------- | ----------------------------- | ------------------------------ |
//! | `decode` | Returns a new `Vec<u8>` | Always |
//! | `decode_config` | Returns a new `Vec<u8>` | Always |
//! | `decode_config_buf` | Appends to provided `Vec<u8>` | Only if `Vec` needs to grow |
//! | `decode_config_slice` | Writes to provided `&[u8]` | Never |
//!
//! Unlike encoding, where all possible input is valid, decoding can fail (see `DecodeError`).
//!
//! Input can be invalid because it has invalid characters or invalid padding. (No padding at all is
//! valid, but excess padding is not.)
//!
//! Whitespace in the input is invalid unless `strip_whitespace` is enabled in the `Config` used.
//!
//! # Panics
//!
//! If length calculations result in overflowing `usize`, a panic will result.
//!
//! The `_slice` flavors of encode or decode will panic if the provided output slice is too small,
#![deny(
missing_docs, trivial_casts, trivial_numeric_casts, unused_extern_crates, unused_import_braces,
unused_results, variant_size_differences, warnings
)]
extern crate byteorder;
use std::{fmt, error, str};
use byteorder::{BigEndian, ByteOrder};
mod chunked_encoder;
pub mod display;
mod line_wrap;
mod tables;
mod line_wrap;
use line_wrap::{line_wrap_parameters, line_wrap};
use line_wrap::{line_wrap, line_wrap_parameters};
mod encode;
pub use encode::{encode, encode_config, encode_config_buf, encode_config_slice};
mod decode;
pub use decode::{decode, decode_config, decode_config_buf, decode_config_slice, DecodeError};
#[cfg(test)]
mod tests;
/// Available encoding character sets
#[derive(Clone, Copy, Debug)]
@ -15,28 +84,35 @@ pub enum CharacterSet {
/// The standard character set (uses `+` and `/`)
Standard,
/// The URL safe character set (uses `-` and `_`)
UrlSafe
UrlSafe,
/// The `crypt(3)` character set (uses `./0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz`)
Crypt,
}
impl CharacterSet {
fn encode_table(&self) -> &'static [u8; 64] {
match *self {
CharacterSet::Standard => tables::STANDARD_ENCODE,
CharacterSet::UrlSafe => tables::URL_SAFE_ENCODE
CharacterSet::UrlSafe => tables::URL_SAFE_ENCODE,
CharacterSet::Crypt => tables::CRYPT_ENCODE,
}
}
fn decode_table(&self) -> &'static [u8; 256] {
match *self {
CharacterSet::Standard => tables::STANDARD_DECODE,
CharacterSet::UrlSafe => tables::URL_SAFE_DECODE
CharacterSet::UrlSafe => tables::URL_SAFE_DECODE,
CharacterSet::Crypt => tables::CRYPT_DECODE,
}
}
}
/// Line ending used in optional line wrapping.
#[derive(Clone, Copy, Debug)]
pub enum LineEnding {
/// Unix-style \n
LF,
/// Windows-style \r\n
CRLF,
}
@ -44,16 +120,18 @@ impl LineEnding {
fn len(&self) -> usize {
match *self {
LineEnding::LF => 1,
LineEnding::CRLF => 2
LineEnding::CRLF => 2,
}
}
}
/// Line wrap configuration.
#[derive(Clone, Copy, Debug)]
pub enum LineWrap {
/// Don't wrap.
NoWrap,
// wrap length is always > 0
Wrap(usize, LineEnding)
/// Wrap lines with the specified length and line ending. The length must be > 0.
Wrap(usize, LineEnding),
}
/// Contains configuration parameters for base64 encoding
@ -63,653 +141,80 @@ pub struct Config {
char_set: CharacterSet,
/// True to pad output with `=` characters
pad: bool,
/// Remove whitespace before decoding, at the cost of an allocation
/// Remove whitespace before decoding, at the cost of an allocation. Whitespace is defined
/// according to POSIX-locale `isspace`, meaning \n \r \f \t \v and space.
strip_whitespace: bool,
/// ADT signifying whether to linewrap output, and if so by how many characters and with what ending
/// ADT signifying whether to linewrap output, and if so by how many characters and with what
/// ending
line_wrap: LineWrap,
}
impl Config {
pub fn new(char_set: CharacterSet,
pad: bool,
strip_whitespace: bool,
input_line_wrap: LineWrap) -> Config {
let line_wrap = match input_line_wrap {
/// Create a new `Config`.
pub fn new(
char_set: CharacterSet,
pad: bool,
strip_whitespace: bool,
input_line_wrap: LineWrap,
) -> Config {
let line_wrap = match input_line_wrap {
LineWrap::Wrap(0, _) => LineWrap::NoWrap,
_ => input_line_wrap,
};
Config {
char_set: char_set,
pad: pad,
strip_whitespace: strip_whitespace,
line_wrap: line_wrap,
char_set,
pad,
strip_whitespace,
line_wrap,
}
}
}
pub static STANDARD: Config = Config {
/// Standard character set with padding.
pub const STANDARD: Config = Config {
char_set: CharacterSet::Standard,
pad: true,
strip_whitespace: false,
line_wrap: LineWrap::NoWrap,
};
pub static MIME: Config = Config {
/// Standard character set without padding.
pub const STANDARD_NO_PAD: Config = Config {
char_set: CharacterSet::Standard,
pad: false,
strip_whitespace: false,
line_wrap: LineWrap::NoWrap,
};
/// As per standards for MIME encoded messages
pub const MIME: Config = Config {
char_set: CharacterSet::Standard,
pad: true,
strip_whitespace: true,
line_wrap: LineWrap::Wrap(76, LineEnding::CRLF),
};
pub static URL_SAFE: Config = Config {
/// URL-safe character set with padding
pub const URL_SAFE: Config = Config {
char_set: CharacterSet::UrlSafe,
pad: true,
strip_whitespace: false,
line_wrap: LineWrap::NoWrap,
};
pub static URL_SAFE_NO_PAD: Config = Config {
/// URL-safe character set without padding
pub const URL_SAFE_NO_PAD: Config = Config {
char_set: CharacterSet::UrlSafe,
pad: false,
strip_whitespace: false,
line_wrap: LineWrap::NoWrap,
};
#[derive(Debug, PartialEq, Eq)]
pub enum DecodeError {
InvalidByte(usize, u8),
InvalidLength,
}
impl fmt::Display for DecodeError {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
match *self {
DecodeError::InvalidByte(index, byte) =>
write!(f, "Invalid byte {}, offset {}.", byte, index),
DecodeError::InvalidLength =>
write!(f, "Encoded text cannot have a 6-bit remainder.")
}
}
}
impl error::Error for DecodeError {
fn description(&self) -> &str {
match *self {
DecodeError::InvalidByte(_, _) => "invalid byte",
DecodeError::InvalidLength => "invalid length"
}
}
fn cause(&self) -> Option<&error::Error> {
None
}
}
///Encode arbitrary octets as base64.
///Returns a String.
///Convenience for `encode_config(input, base64::STANDARD);`.
///
///# Example
///
///```rust
///extern crate base64;
///
///fn main() {
/// let b64 = base64::encode(b"hello world");
/// println!("{}", b64);
///}
///```
pub fn encode<T: ?Sized + AsRef<[u8]>>(input: &T) -> String {
encode_config(input, STANDARD)
}
///Decode from string reference as octets.
///Returns a Result containing a Vec<u8>.
///Convenience `decode_config(input, base64::STANDARD);`.
///
///# Example
///
///```rust
///extern crate base64;
///
///fn main() {
/// let bytes = base64::decode("aGVsbG8gd29ybGQ=").unwrap();
/// println!("{:?}", bytes);
///}
///```
pub fn decode<T: ?Sized + AsRef<[u8]>>(input: &T) -> Result<Vec<u8>, DecodeError> {
decode_config(input, STANDARD)
}
///Encode arbitrary octets as base64.
///Returns a String.
///
///# Example
///
///```rust
///extern crate base64;
///
///fn main() {
/// let b64 = base64::encode_config(b"hello world~", base64::STANDARD);
/// println!("{}", b64);
///
/// let b64_url = base64::encode_config(b"hello internet~", base64::URL_SAFE);
/// println!("{}", b64_url);
///}
///```
pub fn encode_config<T: ?Sized + AsRef<[u8]>>(input: &T, config: Config) -> String {
let mut buf = match encoded_size(input.as_ref().len(), &config) {
Some(n) => String::with_capacity(n),
None => panic!("integer overflow when calculating buffer size")
};
encode_config_buf(input, config, &mut buf);
buf
}
/// calculate the base64 encoded string size, including padding
fn encoded_size(bytes_len: usize, config: &Config) -> Option<usize> {
let rem = bytes_len % 3;
let complete_input_chunks = bytes_len / 3;
let complete_chunk_output = complete_input_chunks.checked_mul(4);
let encoded_len_no_wrap = if rem > 0 {
if config.pad {
complete_chunk_output.and_then(|c| c.checked_add(4))
} else {
let encoded_rem = match rem {
1 => 2,
2 => 3,
_ => panic!("Impossible remainder")
};
complete_chunk_output.and_then(|c| c.checked_add(encoded_rem))
}
} else {
complete_chunk_output
};
encoded_len_no_wrap.map(|e| {
match config.line_wrap {
LineWrap::NoWrap => e,
LineWrap::Wrap(line_len, line_ending) => {
line_wrap_parameters(e, line_len, line_ending).total_len
}
}
})
}
///Encode arbitrary octets as base64.
///Writes into the supplied buffer to avoid allocations.
///
///# Example
///
///```rust
///extern crate base64;
///
///fn main() {
/// let mut buf = String::new();
/// base64::encode_config_buf(b"hello world~", base64::STANDARD, &mut buf);
/// println!("{}", buf);
///
/// buf.clear();
/// base64::encode_config_buf(b"hello internet~", base64::URL_SAFE, &mut buf);
/// println!("{}", buf);
///}
///```
pub fn encode_config_buf<T: ?Sized + AsRef<[u8]>>(input: &T, config: Config, buf: &mut String) {
let input_bytes = input.as_ref();
let encoded_size = encoded_size(input_bytes.len(), &config)
.expect("usize overflow when calculating buffer size");
let orig_buf_len = buf.len();
// we're only going to insert valid utf8
let mut buf_bytes;
unsafe {
buf_bytes = buf.as_mut_vec();
}
buf_bytes.resize(orig_buf_len.checked_add(encoded_size)
.expect("usize overflow when calculating expanded buffer size"), 0);
let mut b64_output = &mut buf_bytes[orig_buf_len..];
let encoded_bytes = encode_with_padding(input_bytes, b64_output, config.char_set.encode_table(),
config.pad);
if let LineWrap::Wrap(line_len, line_end) = config.line_wrap {
line_wrap(b64_output, encoded_bytes, line_len, line_end);
}
}
/// Encode input bytes and pad if configured.
/// `output` must be long enough to hold the encoded `input` with padding.
/// Returns the number of bytes written.
fn encode_with_padding(input: &[u8], output: &mut [u8], encode_table: &[u8; 64], pad: bool) -> usize {
let b64_bytes_written = encode_to_slice(input, output, encode_table);
let padding_bytes = if pad {
add_padding(input.len(), &mut output[b64_bytes_written..])
} else {
0
};
b64_bytes_written.checked_add(padding_bytes)
.expect("usize overflow when calculating b64 length")
}
/// Encode input bytes to utf8 base64 bytes. Does not pad or line wrap.
/// `output` must be long enough to hold the encoded `input` without padding or line wrapping.
/// Returns the number of bytes written.
#[inline]
fn encode_to_slice(input: &[u8], output: &mut [u8], encode_table: &[u8; 64]) -> usize {
let mut input_index: usize = 0;
const BLOCKS_PER_FAST_LOOP: usize = 4;
const LOW_SIX_BITS: u64 = 0x3F;
// we read 8 bytes at a time (u64) but only actually consume 6 of those bytes. Thus, we need
// 2 trailing bytes to be available to read..
let last_fast_index = input.len().saturating_sub(BLOCKS_PER_FAST_LOOP * 6 + 2);
let mut output_index = 0;
if last_fast_index > 0 {
while input_index <= last_fast_index {
// Major performance wins from letting the optimizer do the bounds check once, mostly
// on the output side
let input_chunk = &input[input_index..(input_index + (BLOCKS_PER_FAST_LOOP * 6 + 2))];
let mut output_chunk = &mut output[output_index..(output_index + BLOCKS_PER_FAST_LOOP * 8)];
// Hand-unrolling for 32 vs 16 or 8 bytes produces yields performance about equivalent
// to unsafe pointer code on a Xeon E5-1650v3. 64 byte unrolling was slightly better for
// large inputs but significantly worse for 50-byte input, unsurprisingly. I suspect
// that it's a not uncommon use case to encode smallish chunks of data (e.g. a 64-byte
// SHA-512 digest), so it would be nice if that fit in the unrolled loop at least once.
// Plus, single-digit percentage performance differences might well be quite different
// on different hardware.
let input_u64 = BigEndian::read_u64(&input_chunk[0..]);
output_chunk[0] = encode_table[((input_u64 >> 58) & LOW_SIX_BITS) as usize];
output_chunk[1] = encode_table[((input_u64 >> 52) & LOW_SIX_BITS) as usize];
output_chunk[2] = encode_table[((input_u64 >> 46) & LOW_SIX_BITS) as usize];
output_chunk[3] = encode_table[((input_u64 >> 40) & LOW_SIX_BITS) as usize];
output_chunk[4] = encode_table[((input_u64 >> 34) & LOW_SIX_BITS) as usize];
output_chunk[5] = encode_table[((input_u64 >> 28) & LOW_SIX_BITS) as usize];
output_chunk[6] = encode_table[((input_u64 >> 22) & LOW_SIX_BITS) as usize];
output_chunk[7] = encode_table[((input_u64 >> 16) & LOW_SIX_BITS) as usize];
let input_u64 = BigEndian::read_u64(&input_chunk[6..]);
output_chunk[8] = encode_table[((input_u64 >> 58) & LOW_SIX_BITS) as usize];
output_chunk[9] = encode_table[((input_u64 >> 52) & LOW_SIX_BITS) as usize];
output_chunk[10] = encode_table[((input_u64 >> 46) & LOW_SIX_BITS) as usize];
output_chunk[11] = encode_table[((input_u64 >> 40) & LOW_SIX_BITS) as usize];
output_chunk[12] = encode_table[((input_u64 >> 34) & LOW_SIX_BITS) as usize];
output_chunk[13] = encode_table[((input_u64 >> 28) & LOW_SIX_BITS) as usize];
output_chunk[14] = encode_table[((input_u64 >> 22) & LOW_SIX_BITS) as usize];
output_chunk[15] = encode_table[((input_u64 >> 16) & LOW_SIX_BITS) as usize];
let input_u64 = BigEndian::read_u64(&input_chunk[12..]);
output_chunk[16] = encode_table[((input_u64 >> 58) & LOW_SIX_BITS) as usize];
output_chunk[17] = encode_table[((input_u64 >> 52) & LOW_SIX_BITS) as usize];
output_chunk[18] = encode_table[((input_u64 >> 46) & LOW_SIX_BITS) as usize];
output_chunk[19] = encode_table[((input_u64 >> 40) & LOW_SIX_BITS) as usize];
output_chunk[20] = encode_table[((input_u64 >> 34) & LOW_SIX_BITS) as usize];
output_chunk[21] = encode_table[((input_u64 >> 28) & LOW_SIX_BITS) as usize];
output_chunk[22] = encode_table[((input_u64 >> 22) & LOW_SIX_BITS) as usize];
output_chunk[23] = encode_table[((input_u64 >> 16) & LOW_SIX_BITS) as usize];
let input_u64 = BigEndian::read_u64(&input_chunk[18..]);
output_chunk[24] = encode_table[((input_u64 >> 58) & LOW_SIX_BITS) as usize];
output_chunk[25] = encode_table[((input_u64 >> 52) & LOW_SIX_BITS) as usize];
output_chunk[26] = encode_table[((input_u64 >> 46) & LOW_SIX_BITS) as usize];
output_chunk[27] = encode_table[((input_u64 >> 40) & LOW_SIX_BITS) as usize];
output_chunk[28] = encode_table[((input_u64 >> 34) & LOW_SIX_BITS) as usize];
output_chunk[29] = encode_table[((input_u64 >> 28) & LOW_SIX_BITS) as usize];
output_chunk[30] = encode_table[((input_u64 >> 22) & LOW_SIX_BITS) as usize];
output_chunk[31] = encode_table[((input_u64 >> 16) & LOW_SIX_BITS) as usize];
output_index += BLOCKS_PER_FAST_LOOP * 8;
input_index += BLOCKS_PER_FAST_LOOP * 6;
}
}
// Encode what's left after the fast loop.
const LOW_SIX_BITS_U8: u8 = 0x3F;
let rem = input.len() % 3;
let start_of_rem = input.len() - rem;
// start at the first index not handled by fast loop, which may be 0.
while input_index < start_of_rem {
let input_chunk = &input[input_index..(input_index + 3)];
let mut output_chunk = &mut output[output_index..(output_index + 4)];
output_chunk[0] = encode_table[(input_chunk[0] >> 2) as usize];
output_chunk[1] = encode_table[((input_chunk[0] << 4 | input_chunk[1] >> 4) & LOW_SIX_BITS_U8) as usize];
output_chunk[2] = encode_table[((input_chunk[1] << 2 | input_chunk[2] >> 6) & LOW_SIX_BITS_U8) as usize];
output_chunk[3] = encode_table[(input_chunk[2] & LOW_SIX_BITS_U8) as usize];
input_index += 3;
output_index += 4;
}
if rem == 2 {
output[output_index] = encode_table[(input[start_of_rem] >> 2) as usize];
output[output_index + 1] = encode_table[((input[start_of_rem] << 4 | input[start_of_rem + 1] >> 4) & LOW_SIX_BITS_U8) as usize];
output[output_index + 2] = encode_table[((input[start_of_rem + 1] << 2) & LOW_SIX_BITS_U8) as usize];
output_index += 3;
} else if rem == 1 {
output[output_index] = encode_table[(input[start_of_rem] >> 2) as usize];
output[output_index + 1] = encode_table[((input[start_of_rem] << 4) & LOW_SIX_BITS_U8) as usize];
output_index += 2;
}
output_index
}
/// Write padding characters.
/// `output` is the slice where padding should be written, of length at least 2.
fn add_padding(input_len: usize, output: &mut[u8]) -> usize {
let rem = input_len % 3;
let mut bytes_written = 0;
for _ in 0..((3 - rem) % 3) {
output[bytes_written] = b'=';
bytes_written += 1;
}
bytes_written
}
///Decode from string reference as octets.
///Returns a Result containing a Vec<u8>.
///
///# Example
///
///```rust
///extern crate base64;
///
///fn main() {
/// let bytes = base64::decode_config("aGVsbG8gd29ybGR+Cg==", base64::STANDARD).unwrap();
/// println!("{:?}", bytes);
///
/// let bytes_url = base64::decode_config("aGVsbG8gaW50ZXJuZXR-Cg==", base64::URL_SAFE).unwrap();
/// println!("{:?}", bytes_url);
///}
///```
pub fn decode_config<T: ?Sized + AsRef<[u8]>>(input: &T, config: Config) -> Result<Vec<u8>, DecodeError> {
let mut buffer = Vec::<u8>::with_capacity(input.as_ref().len() * 4 / 3);
decode_config_buf(input, config, &mut buffer).map(|_| buffer)
}
///Decode from string reference as octets.
///Writes into the supplied buffer to avoid allocation.
///Returns a Result containing an empty tuple, aka ().
///
///# Example
///
///```rust
///extern crate base64;
///
///fn main() {
/// let mut buffer = Vec::<u8>::new();
/// base64::decode_config_buf("aGVsbG8gd29ybGR+Cg==", base64::STANDARD, &mut buffer).unwrap();
/// println!("{:?}", buffer);
///
/// buffer.clear();
///
/// base64::decode_config_buf("aGVsbG8gaW50ZXJuZXR-Cg==", base64::URL_SAFE, &mut buffer).unwrap();
/// println!("{:?}", buffer);
///}
///```
pub fn decode_config_buf<T: ?Sized + AsRef<[u8]>>(input: &T,
config: Config,
buffer: &mut Vec<u8>)
-> Result<(), DecodeError> {
let mut input_copy;
let input_bytes = if config.strip_whitespace {
input_copy = Vec::<u8>::with_capacity(input.as_ref().len());
input_copy.extend(input.as_ref().iter().filter(|b| !b" \n\t\r\x0b\x0c".contains(b)));
input_copy.as_ref()
} else {
input.as_ref()
};
let decode_table = &config.char_set.decode_table();
// decode logic operates on chunks of 8 input bytes without padding
const INPUT_CHUNK_LEN: usize = 8;
const DECODED_CHUNK_LEN: usize = 6;
// we read a u64 and write a u64, but a u64 of input only yields 6 bytes of output, so the last
// 2 bytes of any output u64 should not be counted as written to (but must be available in a
// slice).
const DECODED_CHUNK_SUFFIX: usize = 2;
let remainder_len = input_bytes.len() % INPUT_CHUNK_LEN;
let trailing_bytes_to_skip = if remainder_len == 0 {
// if input is a multiple of the chunk size, ignore the last chunk as it may have padding,
// and the fast decode logic cannot handle padding
INPUT_CHUNK_LEN
} else {
remainder_len
};
let length_of_full_chunks = input_bytes.len().saturating_sub(trailing_bytes_to_skip);
let starting_output_index = buffer.len();
// Resize to hold decoded output from fast loop. Need the extra two bytes because
// we write a full 8 bytes for the last 6-byte decoded chunk and then truncate off the last two.
let new_size = starting_output_index
.checked_add(length_of_full_chunks / INPUT_CHUNK_LEN * DECODED_CHUNK_LEN)
.and_then(|l| l.checked_add(DECODED_CHUNK_SUFFIX))
.expect("Overflow when calculating output buffer length");
buffer.resize(new_size, 0);
{
let mut output_index = 0;
let mut input_index = 0;
let buffer_slice = &mut buffer.as_mut_slice()[starting_output_index..];
// how many u64's of input to handle at a time
const CHUNKS_PER_FAST_LOOP_BLOCK: usize = 4;
const INPUT_BLOCK_LEN: usize = CHUNKS_PER_FAST_LOOP_BLOCK * INPUT_CHUNK_LEN;
// includes the trailing 2 bytes for the final u64 write
const DECODED_BLOCK_LEN: usize = CHUNKS_PER_FAST_LOOP_BLOCK * DECODED_CHUNK_LEN +
DECODED_CHUNK_SUFFIX;
// the start index of the last block of data that is big enough to use the unrolled loop
let last_block_start_index = length_of_full_chunks
.saturating_sub(INPUT_CHUNK_LEN * CHUNKS_PER_FAST_LOOP_BLOCK);
// manual unroll to CHUNKS_PER_FAST_LOOP_BLOCK of u64s to amortize slice bounds checks
if last_block_start_index > 0 {
while input_index <= last_block_start_index {
let input_slice = &input_bytes[input_index..(input_index + INPUT_BLOCK_LEN)];
let output_slice = &mut buffer_slice[output_index..(output_index + DECODED_BLOCK_LEN)];
decode_chunk(&input_slice[0..], input_index, decode_table, &mut output_slice[0..])?;
decode_chunk(&input_slice[8..], input_index + 8, decode_table, &mut output_slice[6..])?;
decode_chunk(&input_slice[16..], input_index + 16, decode_table, &mut output_slice[12..])?;
decode_chunk(&input_slice[24..], input_index + 24, decode_table, &mut output_slice[18..])?;
input_index += INPUT_BLOCK_LEN;
output_index += DECODED_BLOCK_LEN - DECODED_CHUNK_SUFFIX;
}
}
// still pretty fast loop: 8 bytes at a time for whatever we didn't do in the faster loop.
while input_index < length_of_full_chunks {
decode_chunk(&input_bytes[input_index..(input_index + 8)], input_index, decode_table,
&mut buffer_slice[output_index..(output_index + 8)])?;
output_index += DECODED_CHUNK_LEN;
input_index += INPUT_CHUNK_LEN;
}
}
// Truncate off the last two bytes from writing the last u64.
// Unconditional because we added on the extra 2 bytes in the resize before the loop,
// so it will never underflow.
let new_len = buffer.len() - DECODED_CHUNK_SUFFIX;
buffer.truncate(new_len);
// handle leftovers (at most 8 bytes, decoded to 6).
// Use a u64 as a stack-resident 8 byte buffer.
let mut leftover_bits: u64 = 0;
let mut morsels_in_leftover = 0;
let mut padding_bytes = 0;
let mut first_padding_index: usize = 0;
for (i, b) in input_bytes[length_of_full_chunks..].iter().enumerate() {
// '=' padding
if *b == 0x3D {
// There can be bad padding in a few ways:
// 1 - Padding with non-padding characters after it
// 2 - Padding after zero or one non-padding characters before it
// in the current quad.
// 3 - More than two characters of padding. If 3 or 4 padding chars
// are in the same quad, that implies it will be caught by #2.
// If it spreads from one quad to another, it will be caught by
// #2 in the second quad.
if i % 4 < 2 {
// Check for case #2.
let bad_padding_index = length_of_full_chunks + if padding_bytes > 0 {
// If we've already seen padding, report the first padding index.
// This is to be consistent with the faster logic above: it will report an error
// on the first padding character (since it doesn't expect to see anything but
// actual encoded data).
first_padding_index
} else {
// haven't seen padding before, just use where we are now
i
};
return Err(DecodeError::InvalidByte(bad_padding_index, *b));
}
if padding_bytes == 0 {
first_padding_index = i;
}
padding_bytes += 1;
continue;
}
// Check for case #1.
// To make '=' handling consistent with the main loop, don't allow
// non-suffix '=' in trailing chunk either. Report error as first
// erroneous padding.
if padding_bytes > 0 {
return Err(DecodeError::InvalidByte(
length_of_full_chunks + first_padding_index, 0x3D));
}
// can use up to 8 * 6 = 48 bits of the u64, if last chunk has no padding.
// To minimize shifts, pack the leftovers from left to right.
let shift = 64 - (morsels_in_leftover + 1) * 6;
// tables are all 256 elements, lookup with a u8 index always succeeds
let morsel = decode_table[*b as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(length_of_full_chunks + i, *b));
}
leftover_bits |= (morsel as u64) << shift;
morsels_in_leftover += 1;
}
let leftover_bits_ready_to_append = match morsels_in_leftover {
0 => 0,
1 => return Err(DecodeError::InvalidLength),
2 => 8,
3 => 16,
4 => 24,
5 => return Err(DecodeError::InvalidLength),
6 => 32,
7 => 40,
8 => 48,
_ => panic!("Impossible: must only have 0 to 4 input bytes in last quad")
};
let mut leftover_bits_appended_to_buf = 0;
while leftover_bits_appended_to_buf < leftover_bits_ready_to_append {
// `as` simply truncates the higher bits, which is what we want here
let selected_bits = (leftover_bits >> (56 - leftover_bits_appended_to_buf)) as u8;
buffer.push(selected_bits);
leftover_bits_appended_to_buf += 8;
};
Ok(())
}
// yes, really inline (worth 30-50% speedup)
#[inline(always)]
fn decode_chunk(input: &[u8], index_at_start_of_input: usize, decode_table: &[u8; 256],
output: &mut [u8]) -> Result<(), DecodeError> {
let mut accum: u64;
let morsel = decode_table[input[0] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(index_at_start_of_input, input[0]));
}
accum = (morsel as u64) << 58;
let morsel = decode_table[input[1] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(index_at_start_of_input + 1, input[1]));
}
accum |= (morsel as u64) << 52;
let morsel = decode_table[input[2] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(index_at_start_of_input + 2, input[2]));
}
accum |= (morsel as u64) << 46;
let morsel = decode_table[input[3] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(index_at_start_of_input + 3, input[3]));
}
accum |= (morsel as u64) << 40;
let morsel = decode_table[input[4] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(index_at_start_of_input + 4, input[4]));
}
accum |= (morsel as u64) << 34;
let morsel = decode_table[input[5] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(index_at_start_of_input + 5, input[5]));
}
accum |= (morsel as u64) << 28;
let morsel = decode_table[input[6] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(index_at_start_of_input + 6, input[6]));
}
accum |= (morsel as u64) << 22;
let morsel = decode_table[input[7] as usize];
if morsel == tables::INVALID_VALUE {
return Err(DecodeError::InvalidByte(index_at_start_of_input + 7, input[7]));
}
accum |= (morsel as u64) << 16;
BigEndian::write_u64(output, accum);
Ok(())
}
#[cfg(test)]
mod tests;
/// As per `crypt(3)` requirements
pub const CRYPT: Config = Config {
char_set: CharacterSet::Crypt,
pad: false,
strip_whitespace: false,
line_wrap: LineWrap::NoWrap,
};

View File

@ -2,8 +2,6 @@ extern crate safemem;
use super::*;
use std::str;
#[derive(Debug, PartialEq)]
pub struct LineWrapParameters {
// number of lines that need an ending
@ -15,13 +13,16 @@ pub struct LineWrapParameters {
// length of all lines, including endings for the ones that need them
pub total_len: usize,
// length of the line endings only
pub total_line_endings_len: usize
pub total_line_endings_len: usize,
}
/// Calculations about how many lines we'll get for a given line length, line ending, etc.
/// This assumes that the last line will not get an ending, even if it is the full line length.
pub fn line_wrap_parameters(input_len: usize, line_len: usize, line_ending: LineEnding)
-> LineWrapParameters {
pub fn line_wrap_parameters(
input_len: usize,
line_len: usize,
line_ending: LineEnding,
) -> LineWrapParameters {
let line_ending_len = line_ending.len();
if input_len <= line_len {
@ -31,7 +32,7 @@ pub fn line_wrap_parameters(input_len: usize, line_len: usize, line_ending: Line
last_line_len: input_len,
total_full_wrapped_lines_len: 0,
total_len: input_len,
total_line_endings_len: 0
total_line_endings_len: 0,
};
};
@ -50,16 +51,19 @@ pub fn line_wrap_parameters(input_len: usize, line_len: usize, line_ending: Line
// line wrapping out of the normal encode path and have it be a separate step. Then only users
// who need line wrapping would care about the possibility for error.
let single_full_line_with_ending_len = line_len.checked_add(line_ending_len)
let single_full_line_with_ending_len = line_len
.checked_add(line_ending_len)
.expect("Line length with ending exceeds usize");
// length of just the full lines with line endings
let total_full_wrapped_lines_len = num_lines_with_endings
.checked_mul(single_full_line_with_ending_len)
.expect("Full lines with endings length exceeds usize");
// all lines with appropriate endings, including the last line
let total_all_wrapped_len = total_full_wrapped_lines_len.checked_add(last_line_length)
let total_all_wrapped_len = total_full_wrapped_lines_len
.checked_add(last_line_length)
.expect("All lines with endings length exceeds usize");
let total_line_endings_len = num_lines_with_endings.checked_mul(line_ending_len)
let total_line_endings_len = num_lines_with_endings
.checked_mul(line_ending_len)
.expect("Total line endings length exceeds usize");
LineWrapParameters {
@ -67,11 +71,10 @@ pub fn line_wrap_parameters(input_len: usize, line_len: usize, line_ending: Line
last_line_len: last_line_length,
total_full_wrapped_lines_len: total_full_wrapped_lines_len,
total_len: total_all_wrapped_len,
total_line_endings_len: total_line_endings_len
total_line_endings_len: total_line_endings_len,
}
}
/// Insert line endings into the encoded base64 after each complete line (except the last line, even
/// if it is complete).
/// The provided buffer must be large enough to handle the increased size after endings are
@ -79,24 +82,38 @@ pub fn line_wrap_parameters(input_len: usize, line_len: usize, line_ending: Line
/// `input_len` is the length of the encoded data in `encoded_buf`.
/// `line_len` is the width without line ending characters.
/// Returns the number of line ending bytes added.
pub fn line_wrap(encoded_buf: &mut [u8], input_len: usize, line_len: usize, line_ending: LineEnding) -> usize {
pub fn line_wrap(
encoded_buf: &mut [u8],
input_len: usize,
line_len: usize,
line_ending: LineEnding,
) -> usize {
let line_wrap_params = line_wrap_parameters(input_len, line_len, line_ending);
// ptr.offset() is undefined if it wraps, and there is no checked_offset(). However, because
// we perform this check up front to make sure we have enough capacity, we know that none of
// the subsequent pointer operations (assuming they implement the desired behavior of course!)
// will overflow.
assert!(encoded_buf.len() >= line_wrap_params.total_len,
"Buffer must be able to hold encoded data after line wrapping");
assert!(
encoded_buf.len() >= line_wrap_params.total_len,
"Buffer must be able to hold encoded data after line wrapping"
);
// Move the last line, either partial or full, by itself as it does not have a line ending
// afterwards
let last_line_start = line_wrap_params.lines_with_endings.checked_mul(line_len)
let last_line_start = line_wrap_params
.lines_with_endings
.checked_mul(line_len)
.expect("Start of last line in input exceeds usize");
// last line starts immediately after all the wrapped full lines
let new_line_start = line_wrap_params.total_full_wrapped_lines_len;
safemem::copy_over(encoded_buf, last_line_start, new_line_start, line_wrap_params.last_line_len);
safemem::copy_over(
encoded_buf,
last_line_start,
new_line_start,
line_wrap_params.last_line_len,
);
let mut line_ending_bytes = 0;
@ -106,15 +123,18 @@ pub fn line_wrap(encoded_buf: &mut [u8], input_len: usize, line_len: usize, line
for line_num in 0..line_wrap_params.lines_with_endings {
// doesn't underflow because line_num < lines_with_endings
let lines_before_this_line = line_wrap_params.lines_with_endings - 1 - line_num;
let old_line_start = lines_before_this_line.checked_mul(line_len)
let old_line_start = lines_before_this_line
.checked_mul(line_len)
.expect("Old line start index exceeds usize");
let new_line_start = lines_before_this_line.checked_mul(line_ending_len)
let new_line_start = lines_before_this_line
.checked_mul(line_ending_len)
.and_then(|i| i.checked_add(old_line_start))
.expect("New line start index exceeds usize");
safemem::copy_over(encoded_buf, old_line_start, new_line_start, line_len);
let after_new_line = new_line_start.checked_add(line_len)
let after_new_line = new_line_start
.checked_add(line_len)
.expect("Line ending index exceeds usize");
match line_ending {
@ -124,8 +144,9 @@ pub fn line_wrap(encoded_buf: &mut [u8], input_len: usize, line_len: usize, line
}
LineEnding::CRLF => {
encoded_buf[after_new_line] = b'\r';
encoded_buf[after_new_line.checked_add(1).expect("Line ending index exceeds usize")]
= b'\n';
encoded_buf[after_new_line
.checked_add(1)
.expect("Line ending index exceeds usize")] = b'\n';
line_ending_bytes += 2;
}
}
@ -140,82 +161,96 @@ pub fn line_wrap(encoded_buf: &mut [u8], input_len: usize, line_len: usize, line
mod tests {
extern crate rand;
use super::super::*;
use super::*;
use self::rand::Rng;
use self::rand::distributions::{IndependentSample, Range};
use self::rand::Rng;
#[test]
fn line_params_perfect_multiple_of_line_length_lf() {
let params = line_wrap_parameters(100, 20, LineEnding::LF);
assert_eq!(LineWrapParameters {
lines_with_endings: 4,
last_line_len: 20,
total_full_wrapped_lines_len: 84,
total_len: 104,
total_line_endings_len: 4
}, params);
assert_eq!(
LineWrapParameters {
lines_with_endings: 4,
last_line_len: 20,
total_full_wrapped_lines_len: 84,
total_len: 104,
total_line_endings_len: 4,
},
params
);
}
#[test]
fn line_params_partial_last_line_crlf() {
let params = line_wrap_parameters(103, 20, LineEnding::CRLF);
assert_eq!(LineWrapParameters {
lines_with_endings: 5,
last_line_len: 3,
total_full_wrapped_lines_len: 110,
total_len: 113,
total_line_endings_len: 10
}, params);
assert_eq!(
LineWrapParameters {
lines_with_endings: 5,
last_line_len: 3,
total_full_wrapped_lines_len: 110,
total_len: 113,
total_line_endings_len: 10,
},
params
);
}
#[test]
fn line_params_line_len_1_crlf() {
let params = line_wrap_parameters(100, 1, LineEnding::CRLF);
assert_eq!(LineWrapParameters {
lines_with_endings: 99,
last_line_len: 1,
total_full_wrapped_lines_len: 99 * 3,
total_len: 99 * 3 + 1,
total_line_endings_len: 99 * 2
}, params);
assert_eq!(
LineWrapParameters {
lines_with_endings: 99,
last_line_len: 1,
total_full_wrapped_lines_len: 99 * 3,
total_len: 99 * 3 + 1,
total_line_endings_len: 99 * 2,
},
params
);
}
#[test]
fn line_params_line_len_longer_than_input_crlf() {
let params = line_wrap_parameters(100, 200, LineEnding::CRLF);
assert_eq!(LineWrapParameters {
lines_with_endings: 0,
last_line_len: 100,
total_full_wrapped_lines_len: 0,
total_len: 100,
total_line_endings_len: 0
}, params);
assert_eq!(
LineWrapParameters {
lines_with_endings: 0,
last_line_len: 100,
total_full_wrapped_lines_len: 0,
total_len: 100,
total_line_endings_len: 0,
},
params
);
}
#[test]
fn line_params_line_len_same_as_input_crlf() {
let params = line_wrap_parameters(100, 100, LineEnding::CRLF);
assert_eq!(LineWrapParameters {
lines_with_endings: 0,
last_line_len: 100,
total_full_wrapped_lines_len: 0,
total_len: 100,
total_line_endings_len: 0
}, params);
assert_eq!(
LineWrapParameters {
lines_with_endings: 0,
last_line_len: 100,
total_full_wrapped_lines_len: 0,
total_len: 100,
total_line_endings_len: 0,
},
params
);
}
#[test]
fn line_wrap_length_1_lf() {
let mut buf = vec![0x1, 0x2, 0x3, 0x4];
do_line_wrap(&mut buf, 1, LineEnding::LF);
assert_eq!(3, do_line_wrap(&mut buf, 1, LineEnding::LF));
assert_eq!(vec![0x1, 0xA, 0x2, 0xA, 0x3, 0xA, 0x4], buf);
}
@ -224,7 +259,7 @@ mod tests {
fn line_wrap_length_1_crlf() {
let mut buf = vec![0x1, 0x2, 0x3, 0x4];
do_line_wrap(&mut buf, 1, LineEnding::CRLF);
assert_eq!(6, do_line_wrap(&mut buf, 1, LineEnding::CRLF));
assert_eq!(vec![0x1, 0xD, 0xA, 0x2, 0xD, 0xA, 0x3, 0xD, 0xA, 0x4], buf);
}
@ -233,7 +268,7 @@ mod tests {
fn line_wrap_length_2_lf_full_lines() {
let mut buf = vec![0x1, 0x2, 0x3, 0x4];
do_line_wrap(&mut buf, 2, LineEnding::LF);
assert_eq!(1, do_line_wrap(&mut buf, 2, LineEnding::LF));
assert_eq!(vec![0x1, 0x2, 0xA, 0x3, 0x4], buf);
}
@ -242,7 +277,7 @@ mod tests {
fn line_wrap_length_2_crlf_full_lines() {
let mut buf = vec![0x1, 0x2, 0x3, 0x4];
do_line_wrap(&mut buf, 2, LineEnding::CRLF);
assert_eq!(2, do_line_wrap(&mut buf, 2, LineEnding::CRLF));
assert_eq!(vec![0x1, 0x2, 0xD, 0xA, 0x3, 0x4], buf);
}
@ -251,7 +286,7 @@ mod tests {
fn line_wrap_length_2_lf_partial_line() {
let mut buf = vec![0x1, 0x2, 0x3, 0x4, 0x5];
do_line_wrap(&mut buf, 2, LineEnding::LF);
assert_eq!(2, do_line_wrap(&mut buf, 2, LineEnding::LF));
assert_eq!(vec![0x1, 0x2, 0xA, 0x3, 0x4, 0xA, 0x5], buf);
}
@ -260,7 +295,7 @@ mod tests {
fn line_wrap_length_2_crlf_partial_line() {
let mut buf = vec![0x1, 0x2, 0x3, 0x4, 0x5];
do_line_wrap(&mut buf, 2, LineEnding::CRLF);
assert_eq!(4, do_line_wrap(&mut buf, 2, LineEnding::CRLF));
assert_eq!(vec![0x1, 0x2, 0xD, 0xA, 0x3, 0x4, 0xD, 0xA, 0x5], buf);
}
@ -299,7 +334,7 @@ mod tests {
let line_ending_offset = (line_ending_num + 1) * line_len;
for _ in 0..line_ending_len {
buf.remove(line_ending_offset);
let _ = buf.remove(line_ending_offset);
}
}
@ -331,17 +366,25 @@ mod tests {
// make sure line_wrap didn't touch anything beyond what it should
let start_of_untouched_data = orig_len + bytes_written;
assert_eq!(before_line_wrap[start_of_untouched_data..],
buf[start_of_untouched_data..]);
assert_eq!(
before_line_wrap[start_of_untouched_data..],
buf[start_of_untouched_data..]
);
// also make sure that line wrapping will fit into a slice no bigger than what it should
// need
let bytes_written_precise_fit =
line_wrap(&mut before_line_wrap[0..(params.total_len)], orig_len, line_len,
line_ending);
let bytes_written_precise_fit = line_wrap(
&mut before_line_wrap[0..(params.total_len)],
orig_len,
line_len,
line_ending,
);
assert_eq!(bytes_written, bytes_written_precise_fit);
assert_eq!(&buf[0..(params.total_len)], &before_line_wrap[0..(params.total_len)]);
assert_eq!(
&buf[0..(params.total_len)],
&before_line_wrap[0..(params.total_len)]
);
buf.truncate(orig_len + bytes_written);

View File

@ -1,4 +1,5 @@
pub const INVALID_VALUE: u8 = 255;
#[cfg_attr(rustfmt, rustfmt_skip)]
pub const STANDARD_ENCODE: &'static [u8; 64] = &[
65, // input 0 (0x0) => 'A' (0x41)
66, // input 1 (0x1) => 'B' (0x42)
@ -65,6 +66,7 @@ pub const STANDARD_ENCODE: &'static [u8; 64] = &[
43, // input 62 (0x3E) => '+' (0x2B)
47, // input 63 (0x3F) => '/' (0x2F)
];
#[cfg_attr(rustfmt, rustfmt_skip)]
pub const STANDARD_DECODE: &'static [u8; 256] = &[
INVALID_VALUE, // input 0 (0x0)
INVALID_VALUE, // input 1 (0x1)
@ -323,6 +325,7 @@ pub const STANDARD_DECODE: &'static [u8; 256] = &[
INVALID_VALUE, // input 254 (0xFE)
INVALID_VALUE, // input 255 (0xFF)
];
#[cfg_attr(rustfmt, rustfmt_skip)]
pub const URL_SAFE_ENCODE: &'static [u8; 64] = &[
65, // input 0 (0x0) => 'A' (0x41)
66, // input 1 (0x1) => 'B' (0x42)
@ -389,6 +392,7 @@ pub const URL_SAFE_ENCODE: &'static [u8; 64] = &[
45, // input 62 (0x3E) => '-' (0x2D)
95, // input 63 (0x3F) => '_' (0x5F)
];
#[cfg_attr(rustfmt, rustfmt_skip)]
pub const URL_SAFE_DECODE: &'static [u8; 256] = &[
INVALID_VALUE, // input 0 (0x0)
INVALID_VALUE, // input 1 (0x1)
@ -647,3 +651,329 @@ pub const URL_SAFE_DECODE: &'static [u8; 256] = &[
INVALID_VALUE, // input 254 (0xFE)
INVALID_VALUE, // input 255 (0xFF)
];
#[cfg_attr(rustfmt, rustfmt_skip)]
pub const CRYPT_ENCODE: &'static [u8; 64] = &[
46, // input 0 (0x0) => '.' (0x2E)
47, // input 1 (0x1) => '/' (0x2F)
48, // input 2 (0x2) => '0' (0x30)
49, // input 3 (0x3) => '1' (0x31)
50, // input 4 (0x4) => '2' (0x32)
51, // input 5 (0x5) => '3' (0x33)
52, // input 6 (0x6) => '4' (0x34)
53, // input 7 (0x7) => '5' (0x35)
54, // input 8 (0x8) => '6' (0x36)
55, // input 9 (0x9) => '7' (0x37)
56, // input 10 (0xA) => '8' (0x38)
57, // input 11 (0xB) => '9' (0x39)
65, // input 12 (0xC) => 'A' (0x41)
66, // input 13 (0xD) => 'B' (0x42)
67, // input 14 (0xE) => 'C' (0x43)
68, // input 15 (0xF) => 'D' (0x44)
69, // input 16 (0x10) => 'E' (0x45)
70, // input 17 (0x11) => 'F' (0x46)
71, // input 18 (0x12) => 'G' (0x47)
72, // input 19 (0x13) => 'H' (0x48)
73, // input 20 (0x14) => 'I' (0x49)
74, // input 21 (0x15) => 'J' (0x4A)
75, // input 22 (0x16) => 'K' (0x4B)
76, // input 23 (0x17) => 'L' (0x4C)
77, // input 24 (0x18) => 'M' (0x4D)
78, // input 25 (0x19) => 'N' (0x4E)
79, // input 26 (0x1A) => 'O' (0x4F)
80, // input 27 (0x1B) => 'P' (0x50)
81, // input 28 (0x1C) => 'Q' (0x51)
82, // input 29 (0x1D) => 'R' (0x52)
83, // input 30 (0x1E) => 'S' (0x53)
84, // input 31 (0x1F) => 'T' (0x54)
85, // input 32 (0x20) => 'U' (0x55)
86, // input 33 (0x21) => 'V' (0x56)
87, // input 34 (0x22) => 'W' (0x57)
88, // input 35 (0x23) => 'X' (0x58)
89, // input 36 (0x24) => 'Y' (0x59)
90, // input 37 (0x25) => 'Z' (0x5A)
97, // input 38 (0x26) => 'a' (0x61)
98, // input 39 (0x27) => 'b' (0x62)
99, // input 40 (0x28) => 'c' (0x63)
100, // input 41 (0x29) => 'd' (0x64)
101, // input 42 (0x2A) => 'e' (0x65)
102, // input 43 (0x2B) => 'f' (0x66)
103, // input 44 (0x2C) => 'g' (0x67)
104, // input 45 (0x2D) => 'h' (0x68)
105, // input 46 (0x2E) => 'i' (0x69)
106, // input 47 (0x2F) => 'j' (0x6A)
107, // input 48 (0x30) => 'k' (0x6B)
108, // input 49 (0x31) => 'l' (0x6C)
109, // input 50 (0x32) => 'm' (0x6D)
110, // input 51 (0x33) => 'n' (0x6E)
111, // input 52 (0x34) => 'o' (0x6F)
112, // input 53 (0x35) => 'p' (0x70)
113, // input 54 (0x36) => 'q' (0x71)
114, // input 55 (0x37) => 'r' (0x72)
115, // input 56 (0x38) => 's' (0x73)
116, // input 57 (0x39) => 't' (0x74)
117, // input 58 (0x3A) => 'u' (0x75)
118, // input 59 (0x3B) => 'v' (0x76)
119, // input 60 (0x3C) => 'w' (0x77)
120, // input 61 (0x3D) => 'x' (0x78)
121, // input 62 (0x3E) => 'y' (0x79)
122, // input 63 (0x3F) => 'z' (0x7A)
];
#[cfg_attr(rustfmt, rustfmt_skip)]
pub const CRYPT_DECODE: &'static [u8; 256] = &[
INVALID_VALUE, // input 0 (0x0)
INVALID_VALUE, // input 1 (0x1)
INVALID_VALUE, // input 2 (0x2)
INVALID_VALUE, // input 3 (0x3)
INVALID_VALUE, // input 4 (0x4)
INVALID_VALUE, // input 5 (0x5)
INVALID_VALUE, // input 6 (0x6)
INVALID_VALUE, // input 7 (0x7)
INVALID_VALUE, // input 8 (0x8)
INVALID_VALUE, // input 9 (0x9)
INVALID_VALUE, // input 10 (0xA)
INVALID_VALUE, // input 11 (0xB)
INVALID_VALUE, // input 12 (0xC)
INVALID_VALUE, // input 13 (0xD)
INVALID_VALUE, // input 14 (0xE)
INVALID_VALUE, // input 15 (0xF)
INVALID_VALUE, // input 16 (0x10)
INVALID_VALUE, // input 17 (0x11)
INVALID_VALUE, // input 18 (0x12)
INVALID_VALUE, // input 19 (0x13)
INVALID_VALUE, // input 20 (0x14)
INVALID_VALUE, // input 21 (0x15)
INVALID_VALUE, // input 22 (0x16)
INVALID_VALUE, // input 23 (0x17)
INVALID_VALUE, // input 24 (0x18)
INVALID_VALUE, // input 25 (0x19)
INVALID_VALUE, // input 26 (0x1A)
INVALID_VALUE, // input 27 (0x1B)
INVALID_VALUE, // input 28 (0x1C)
INVALID_VALUE, // input 29 (0x1D)
INVALID_VALUE, // input 30 (0x1E)
INVALID_VALUE, // input 31 (0x1F)
INVALID_VALUE, // input 32 (0x20)
INVALID_VALUE, // input 33 (0x21)
INVALID_VALUE, // input 34 (0x22)
INVALID_VALUE, // input 35 (0x23)
INVALID_VALUE, // input 36 (0x24)
INVALID_VALUE, // input 37 (0x25)
INVALID_VALUE, // input 38 (0x26)
INVALID_VALUE, // input 39 (0x27)
INVALID_VALUE, // input 40 (0x28)
INVALID_VALUE, // input 41 (0x29)
INVALID_VALUE, // input 42 (0x2A)
INVALID_VALUE, // input 43 (0x2B)
INVALID_VALUE, // input 44 (0x2C)
INVALID_VALUE, // input 45 (0x2D)
0, // input 46 (0x2E char '.') => 0 (0x0)
1, // input 47 (0x2F char '/') => 1 (0x1)
2, // input 48 (0x30 char '0') => 2 (0x2)
3, // input 49 (0x31 char '1') => 3 (0x3)
4, // input 50 (0x32 char '2') => 4 (0x4)
5, // input 51 (0x33 char '3') => 5 (0x5)
6, // input 52 (0x34 char '4') => 6 (0x6)
7, // input 53 (0x35 char '5') => 7 (0x7)
8, // input 54 (0x36 char '6') => 8 (0x8)
9, // input 55 (0x37 char '7') => 9 (0x9)
10, // input 56 (0x38 char '8') => 10 (0xA)
11, // input 57 (0x39 char '9') => 11 (0xB)
INVALID_VALUE, // input 58 (0x3A)
INVALID_VALUE, // input 59 (0x3B)
INVALID_VALUE, // input 60 (0x3C)
INVALID_VALUE, // input 61 (0x3D)
INVALID_VALUE, // input 62 (0x3E)
INVALID_VALUE, // input 63 (0x3F)
INVALID_VALUE, // input 64 (0x40)
12, // input 65 (0x41 char 'A') => 12 (0xC)
13, // input 66 (0x42 char 'B') => 13 (0xD)
14, // input 67 (0x43 char 'C') => 14 (0xE)
15, // input 68 (0x44 char 'D') => 15 (0xF)
16, // input 69 (0x45 char 'E') => 16 (0x10)
17, // input 70 (0x46 char 'F') => 17 (0x11)
18, // input 71 (0x47 char 'G') => 18 (0x12)
19, // input 72 (0x48 char 'H') => 19 (0x13)
20, // input 73 (0x49 char 'I') => 20 (0x14)
21, // input 74 (0x4A char 'J') => 21 (0x15)
22, // input 75 (0x4B char 'K') => 22 (0x16)
23, // input 76 (0x4C char 'L') => 23 (0x17)
24, // input 77 (0x4D char 'M') => 24 (0x18)
25, // input 78 (0x4E char 'N') => 25 (0x19)
26, // input 79 (0x4F char 'O') => 26 (0x1A)
27, // input 80 (0x50 char 'P') => 27 (0x1B)
28, // input 81 (0x51 char 'Q') => 28 (0x1C)
29, // input 82 (0x52 char 'R') => 29 (0x1D)
30, // input 83 (0x53 char 'S') => 30 (0x1E)
31, // input 84 (0x54 char 'T') => 31 (0x1F)
32, // input 85 (0x55 char 'U') => 32 (0x20)
33, // input 86 (0x56 char 'V') => 33 (0x21)
34, // input 87 (0x57 char 'W') => 34 (0x22)
35, // input 88 (0x58 char 'X') => 35 (0x23)
36, // input 89 (0x59 char 'Y') => 36 (0x24)
37, // input 90 (0x5A char 'Z') => 37 (0x25)
INVALID_VALUE, // input 91 (0x5B)
INVALID_VALUE, // input 92 (0x5C)
INVALID_VALUE, // input 93 (0x5D)
INVALID_VALUE, // input 94 (0x5E)
INVALID_VALUE, // input 95 (0x5F)
INVALID_VALUE, // input 96 (0x60)
38, // input 97 (0x61 char 'a') => 38 (0x26)
39, // input 98 (0x62 char 'b') => 39 (0x27)
40, // input 99 (0x63 char 'c') => 40 (0x28)
41, // input 100 (0x64 char 'd') => 41 (0x29)
42, // input 101 (0x65 char 'e') => 42 (0x2A)
43, // input 102 (0x66 char 'f') => 43 (0x2B)
44, // input 103 (0x67 char 'g') => 44 (0x2C)
45, // input 104 (0x68 char 'h') => 45 (0x2D)
46, // input 105 (0x69 char 'i') => 46 (0x2E)
47, // input 106 (0x6A char 'j') => 47 (0x2F)
48, // input 107 (0x6B char 'k') => 48 (0x30)
49, // input 108 (0x6C char 'l') => 49 (0x31)
50, // input 109 (0x6D char 'm') => 50 (0x32)
51, // input 110 (0x6E char 'n') => 51 (0x33)
52, // input 111 (0x6F char 'o') => 52 (0x34)
53, // input 112 (0x70 char 'p') => 53 (0x35)
54, // input 113 (0x71 char 'q') => 54 (0x36)
55, // input 114 (0x72 char 'r') => 55 (0x37)
56, // input 115 (0x73 char 's') => 56 (0x38)
57, // input 116 (0x74 char 't') => 57 (0x39)
58, // input 117 (0x75 char 'u') => 58 (0x3A)
59, // input 118 (0x76 char 'v') => 59 (0x3B)
60, // input 119 (0x77 char 'w') => 60 (0x3C)
61, // input 120 (0x78 char 'x') => 61 (0x3D)
62, // input 121 (0x79 char 'y') => 62 (0x3E)
63, // input 122 (0x7A char 'z') => 63 (0x3F)
INVALID_VALUE, // input 123 (0x7B)
INVALID_VALUE, // input 124 (0x7C)
INVALID_VALUE, // input 125 (0x7D)
INVALID_VALUE, // input 126 (0x7E)
INVALID_VALUE, // input 127 (0x7F)
INVALID_VALUE, // input 128 (0x80)
INVALID_VALUE, // input 129 (0x81)
INVALID_VALUE, // input 130 (0x82)
INVALID_VALUE, // input 131 (0x83)
INVALID_VALUE, // input 132 (0x84)
INVALID_VALUE, // input 133 (0x85)
INVALID_VALUE, // input 134 (0x86)
INVALID_VALUE, // input 135 (0x87)
INVALID_VALUE, // input 136 (0x88)
INVALID_VALUE, // input 137 (0x89)
INVALID_VALUE, // input 138 (0x8A)
INVALID_VALUE, // input 139 (0x8B)
INVALID_VALUE, // input 140 (0x8C)
INVALID_VALUE, // input 141 (0x8D)
INVALID_VALUE, // input 142 (0x8E)
INVALID_VALUE, // input 143 (0x8F)
INVALID_VALUE, // input 144 (0x90)
INVALID_VALUE, // input 145 (0x91)
INVALID_VALUE, // input 146 (0x92)
INVALID_VALUE, // input 147 (0x93)
INVALID_VALUE, // input 148 (0x94)
INVALID_VALUE, // input 149 (0x95)
INVALID_VALUE, // input 150 (0x96)
INVALID_VALUE, // input 151 (0x97)
INVALID_VALUE, // input 152 (0x98)
INVALID_VALUE, // input 153 (0x99)
INVALID_VALUE, // input 154 (0x9A)
INVALID_VALUE, // input 155 (0x9B)
INVALID_VALUE, // input 156 (0x9C)
INVALID_VALUE, // input 157 (0x9D)
INVALID_VALUE, // input 158 (0x9E)
INVALID_VALUE, // input 159 (0x9F)
INVALID_VALUE, // input 160 (0xA0)
INVALID_VALUE, // input 161 (0xA1)
INVALID_VALUE, // input 162 (0xA2)
INVALID_VALUE, // input 163 (0xA3)
INVALID_VALUE, // input 164 (0xA4)
INVALID_VALUE, // input 165 (0xA5)
INVALID_VALUE, // input 166 (0xA6)
INVALID_VALUE, // input 167 (0xA7)
INVALID_VALUE, // input 168 (0xA8)
INVALID_VALUE, // input 169 (0xA9)
INVALID_VALUE, // input 170 (0xAA)
INVALID_VALUE, // input 171 (0xAB)
INVALID_VALUE, // input 172 (0xAC)
INVALID_VALUE, // input 173 (0xAD)
INVALID_VALUE, // input 174 (0xAE)
INVALID_VALUE, // input 175 (0xAF)
INVALID_VALUE, // input 176 (0xB0)
INVALID_VALUE, // input 177 (0xB1)
INVALID_VALUE, // input 178 (0xB2)
INVALID_VALUE, // input 179 (0xB3)
INVALID_VALUE, // input 180 (0xB4)
INVALID_VALUE, // input 181 (0xB5)
INVALID_VALUE, // input 182 (0xB6)
INVALID_VALUE, // input 183 (0xB7)
INVALID_VALUE, // input 184 (0xB8)
INVALID_VALUE, // input 185 (0xB9)
INVALID_VALUE, // input 186 (0xBA)
INVALID_VALUE, // input 187 (0xBB)
INVALID_VALUE, // input 188 (0xBC)
INVALID_VALUE, // input 189 (0xBD)
INVALID_VALUE, // input 190 (0xBE)
INVALID_VALUE, // input 191 (0xBF)
INVALID_VALUE, // input 192 (0xC0)
INVALID_VALUE, // input 193 (0xC1)
INVALID_VALUE, // input 194 (0xC2)
INVALID_VALUE, // input 195 (0xC3)
INVALID_VALUE, // input 196 (0xC4)
INVALID_VALUE, // input 197 (0xC5)
INVALID_VALUE, // input 198 (0xC6)
INVALID_VALUE, // input 199 (0xC7)
INVALID_VALUE, // input 200 (0xC8)
INVALID_VALUE, // input 201 (0xC9)
INVALID_VALUE, // input 202 (0xCA)
INVALID_VALUE, // input 203 (0xCB)
INVALID_VALUE, // input 204 (0xCC)
INVALID_VALUE, // input 205 (0xCD)
INVALID_VALUE, // input 206 (0xCE)
INVALID_VALUE, // input 207 (0xCF)
INVALID_VALUE, // input 208 (0xD0)
INVALID_VALUE, // input 209 (0xD1)
INVALID_VALUE, // input 210 (0xD2)
INVALID_VALUE, // input 211 (0xD3)
INVALID_VALUE, // input 212 (0xD4)
INVALID_VALUE, // input 213 (0xD5)
INVALID_VALUE, // input 214 (0xD6)
INVALID_VALUE, // input 215 (0xD7)
INVALID_VALUE, // input 216 (0xD8)
INVALID_VALUE, // input 217 (0xD9)
INVALID_VALUE, // input 218 (0xDA)
INVALID_VALUE, // input 219 (0xDB)
INVALID_VALUE, // input 220 (0xDC)
INVALID_VALUE, // input 221 (0xDD)
INVALID_VALUE, // input 222 (0xDE)
INVALID_VALUE, // input 223 (0xDF)
INVALID_VALUE, // input 224 (0xE0)
INVALID_VALUE, // input 225 (0xE1)
INVALID_VALUE, // input 226 (0xE2)
INVALID_VALUE, // input 227 (0xE3)
INVALID_VALUE, // input 228 (0xE4)
INVALID_VALUE, // input 229 (0xE5)
INVALID_VALUE, // input 230 (0xE6)
INVALID_VALUE, // input 231 (0xE7)
INVALID_VALUE, // input 232 (0xE8)
INVALID_VALUE, // input 233 (0xE9)
INVALID_VALUE, // input 234 (0xEA)
INVALID_VALUE, // input 235 (0xEB)
INVALID_VALUE, // input 236 (0xEC)
INVALID_VALUE, // input 237 (0xED)
INVALID_VALUE, // input 238 (0xEE)
INVALID_VALUE, // input 239 (0xEF)
INVALID_VALUE, // input 240 (0xF0)
INVALID_VALUE, // input 241 (0xF1)
INVALID_VALUE, // input 242 (0xF2)
INVALID_VALUE, // input 243 (0xF3)
INVALID_VALUE, // input 244 (0xF4)
INVALID_VALUE, // input 245 (0xF5)
INVALID_VALUE, // input 246 (0xF6)
INVALID_VALUE, // input 247 (0xF7)
INVALID_VALUE, // input 248 (0xF8)
INVALID_VALUE, // input 249 (0xF9)
INVALID_VALUE, // input 250 (0xFA)
INVALID_VALUE, // input 251 (0xFB)
INVALID_VALUE, // input 252 (0xFC)
INVALID_VALUE, // input 253 (0xFD)
INVALID_VALUE, // input 254 (0xFE)
INVALID_VALUE, // input 255 (0xFF)
];

View File

@ -1,126 +1,13 @@
extern crate rand;
use super::*;
use super::line_wrap::{line_wrap_parameters};
use self::rand::Rng;
use encode::encoded_size;
use line_wrap::line_wrap_parameters;
use *;
use std::str;
use self::rand::distributions::{IndependentSample, Range};
#[test]
fn encoded_size_correct_standard() {
assert_encoded_length(0, 0, STANDARD);
assert_encoded_length(1, 4, STANDARD);
assert_encoded_length(2, 4, STANDARD);
assert_encoded_length(3, 4, STANDARD);
assert_encoded_length(4, 8, STANDARD);
assert_encoded_length(5, 8, STANDARD);
assert_encoded_length(6, 8, STANDARD);
assert_encoded_length(7, 12, STANDARD);
assert_encoded_length(8, 12, STANDARD);
assert_encoded_length(9, 12, STANDARD);
assert_encoded_length(54, 72, STANDARD);
assert_encoded_length(55, 76, STANDARD);
assert_encoded_length(56, 76, STANDARD);
assert_encoded_length(57, 76, STANDARD);
assert_encoded_length(58, 80, STANDARD);
}
#[test]
fn encoded_size_correct_no_pad_no_wrap() {
assert_encoded_length(0, 0, URL_SAFE_NO_PAD);
assert_encoded_length(1, 2, URL_SAFE_NO_PAD);
assert_encoded_length(2, 3, URL_SAFE_NO_PAD);
assert_encoded_length(3, 4, URL_SAFE_NO_PAD);
assert_encoded_length(4, 6, URL_SAFE_NO_PAD);
assert_encoded_length(5, 7, URL_SAFE_NO_PAD);
assert_encoded_length(6, 8, URL_SAFE_NO_PAD);
assert_encoded_length(7, 10, URL_SAFE_NO_PAD);
assert_encoded_length(8, 11, URL_SAFE_NO_PAD);
assert_encoded_length(9, 12, URL_SAFE_NO_PAD);
assert_encoded_length(54, 72, URL_SAFE_NO_PAD);
assert_encoded_length(55, 74, URL_SAFE_NO_PAD);
assert_encoded_length(56, 75, URL_SAFE_NO_PAD);
assert_encoded_length(57, 76, URL_SAFE_NO_PAD);
assert_encoded_length(58, 78, URL_SAFE_NO_PAD);
}
#[test]
fn encoded_size_correct_mime() {
assert_encoded_length(0, 0, MIME);
assert_encoded_length(1, 4, MIME);
assert_encoded_length(2, 4, MIME);
assert_encoded_length(3, 4, MIME);
assert_encoded_length(4, 8, MIME);
assert_encoded_length(5, 8, MIME);
assert_encoded_length(6, 8, MIME);
assert_encoded_length(7, 12, MIME);
assert_encoded_length(8, 12, MIME);
assert_encoded_length(9, 12, MIME);
assert_encoded_length(54, 72, MIME);
assert_encoded_length(55, 76, MIME);
assert_encoded_length(56, 76, MIME);
assert_encoded_length(57, 76, MIME);
assert_encoded_length(58, 82, MIME);
assert_encoded_length(59, 82, MIME);
assert_encoded_length(60, 82, MIME);
}
#[test]
fn encoded_size_correct_lf_pad() {
let config = Config::new(
CharacterSet::Standard,
true,
false,
LineWrap::Wrap(76, LineEnding::LF)
);
assert_encoded_length(0, 0, config);
assert_encoded_length(1, 4, config);
assert_encoded_length(2, 4, config);
assert_encoded_length(3, 4, config);
assert_encoded_length(4, 8, config);
assert_encoded_length(5, 8, config);
assert_encoded_length(6, 8, config);
assert_encoded_length(7, 12, config);
assert_encoded_length(8, 12, config);
assert_encoded_length(9, 12, config);
assert_encoded_length(54, 72, config);
assert_encoded_length(55, 76, config);
assert_encoded_length(56, 76, config);
assert_encoded_length(57, 76, config);
// one fewer than MIME
assert_encoded_length(58, 81, config);
assert_encoded_length(59, 81, config);
assert_encoded_length(60, 81, config);
}
#[test]
fn encoded_size_overflow() {
assert_eq!(None, encoded_size(std::usize::MAX, &STANDARD));
}
use self::rand::Rng;
#[test]
fn roundtrip_random_config_short() {
@ -133,263 +20,7 @@ fn roundtrip_random_config_long() {
roundtrip_random_config(Range::new(0, 1000), Range::new(0, 1000), 10_000);
}
#[test]
fn encode_into_nonempty_buffer_doesnt_clobber_existing_contents() {
let mut orig_data = Vec::new();
let mut prefix = String::new();
let mut encoded_data_no_prefix = String::new();
let mut encoded_data_with_prefix = String::new();
let mut decoded = Vec::new();
let prefix_len_range = Range::new(0, 1000);
let input_len_range = Range::new(0, 1000);
let line_len_range = Range::new(1, 1000);
let mut rng = rand::weak_rng();
for _ in 0..10_000 {
orig_data.clear();
prefix.clear();
encoded_data_no_prefix.clear();
encoded_data_with_prefix.clear();
decoded.clear();
let input_len = input_len_range.ind_sample(&mut rng);
for _ in 0..input_len {
orig_data.push(rng.gen());
}
let prefix_len = prefix_len_range.ind_sample(&mut rng);
for _ in 0..prefix_len {
// getting convenient random single-byte printable chars that aren't base64 is annoying
prefix.push('#');
}
encoded_data_with_prefix.push_str(&prefix);
let config = random_config(&mut rng, &line_len_range);
encode_config_buf(&orig_data, config, &mut encoded_data_no_prefix);
encode_config_buf(&orig_data, config, &mut encoded_data_with_prefix);
assert_eq!(encoded_data_no_prefix.len() + prefix_len, encoded_data_with_prefix.len());
assert_encode_sanity(&encoded_data_no_prefix, &config, input_len);
assert_encode_sanity(&encoded_data_with_prefix[prefix_len..], &config, input_len);
// append plain encode onto prefix
prefix.push_str(&mut encoded_data_no_prefix);
assert_eq!(prefix, encoded_data_with_prefix);
// since we know we have the correct count of line endings, it's reasonable to simply remove
// them without worrying about where they are
let encoded_no_line_endings: String = encoded_data_no_prefix.chars()
.filter(|&c| c != '\r' && c != '\n')
.collect();
decode_config_buf(&encoded_no_line_endings, config, &mut decoded).unwrap();
assert_eq!(orig_data, decoded);
}
}
#[test]
fn decode_into_nonempty_buffer_doesnt_clobber_existing_contents() {
let mut orig_data = Vec::new();
let mut encoded_data = String::new();
let mut decoded_with_prefix = Vec::new();
let mut decoded_without_prefix = Vec::new();
let mut prefix = Vec::new();
let prefix_len_range = Range::new(0, 1000);
let input_len_range = Range::new(0, 1000);
let line_len_range = Range::new(1, 1000);
let mut rng = rand::weak_rng();
for _ in 0..10_000 {
orig_data.clear();
encoded_data.clear();
decoded_with_prefix.clear();
decoded_without_prefix.clear();
prefix.clear();
let input_len = input_len_range.ind_sample(&mut rng);
for _ in 0..input_len {
orig_data.push(rng.gen());
}
let config = random_config(&mut rng, &line_len_range);
encode_config_buf(&orig_data, config, &mut encoded_data);
assert_encode_sanity(&encoded_data, &config, input_len);
let prefix_len = prefix_len_range.ind_sample(&mut rng);
// fill the buf with a prefix
for _ in 0..prefix_len {
prefix.push(rng.gen());
}
decoded_with_prefix.resize(prefix_len, 0);
decoded_with_prefix.copy_from_slice(&prefix);
// remove line wrapping
let encoded_no_line_endings: String = encoded_data.chars()
.filter(|&c| c != '\r' && c != '\n')
.collect();
// decode into the non-empty buf
decode_config_buf(&encoded_no_line_endings, config, &mut decoded_with_prefix).unwrap();
// also decode into the empty buf
decode_config_buf(&encoded_no_line_endings, config, &mut decoded_without_prefix).unwrap();
assert_eq!(prefix_len + decoded_without_prefix.len(), decoded_with_prefix.len());
assert_eq!(orig_data, decoded_without_prefix);
// append plain decode onto prefix
prefix.append(&mut decoded_without_prefix);
assert_eq!(prefix, decoded_with_prefix);
}
}
#[test]
fn encode_with_padding_random_valid_utf8() {
let mut input = Vec::new();
let mut output = Vec::new();
let input_len_range = Range::new(0, 1000);
let line_len_range = Range::new(1, 1000);
let mut rng = rand::weak_rng();
for _ in 0..10_000 {
input.clear();
output.clear();
let input_len = input_len_range.ind_sample(&mut rng);
for _ in 0..input_len {
input.push(rng.gen());
}
let config = random_config(&mut rng, &line_len_range);
// fill up the output buffer with garbage
let encoded_size = encoded_size(input_len, &config).unwrap();
for _ in 0..encoded_size {
output.push(rng.gen());
}
let orig_output_buf = output.to_vec();
let bytes_written =
encode_with_padding(&input, &mut output, config.char_set.encode_table(), config.pad);
let line_ending_bytes = total_line_ending_bytes(bytes_written, &config);
assert_eq!(encoded_size, bytes_written + line_ending_bytes);
// make sure the part beyond bytes_written is the same garbage it was before
assert_eq!(orig_output_buf[bytes_written..], output[bytes_written..]);
// make sure the encoded bytes are UTF-8
let _ = str::from_utf8(&output[0..bytes_written]).unwrap();
}
}
#[test]
fn encode_to_slice_random_valid_utf8() {
let mut input = Vec::new();
let mut output = Vec::new();
let input_len_range = Range::new(0, 1000);
let line_len_range = Range::new(1, 1000);
let mut rng = rand::weak_rng();
for _ in 0..10_000 {
input.clear();
output.clear();
let input_len = input_len_range.ind_sample(&mut rng);
for _ in 0..input_len {
input.push(rng.gen());
}
let config = random_config(&mut rng, &line_len_range);
// fill up the output buffer with garbage
let encoded_size = encoded_size(input_len, &config).unwrap();
for _ in 0..encoded_size {
output.push(rng.gen());
}
let orig_output_buf = output.to_vec();
let bytes_written =
encode_to_slice(&input, &mut output, config.char_set.encode_table());
// make sure the part beyond bytes_written is the same garbage it was before
assert_eq!(orig_output_buf[bytes_written..], output[bytes_written..]);
// make sure the encoded bytes are UTF-8
let _ = str::from_utf8(&output[0..bytes_written]).unwrap();
}
}
#[test]
fn add_padding_random_valid_utf8(){
let mut output = Vec::new();
let mut rng = rand::weak_rng();
// cover our bases for length % 3
for input_len in 0..10 {
output.clear();
// fill output with random
for _ in 0..10 {
output.push(rng.gen());
}
let orig_output_buf = output.to_vec();
let bytes_written =
add_padding(input_len, &mut output);
// make sure the part beyond bytes_written is the same garbage it was before
assert_eq!(orig_output_buf[bytes_written..], output[bytes_written..]);
// make sure the encoded bytes are UTF-8
let _ = str::from_utf8(&output[0..bytes_written]).unwrap();
}
}
fn total_line_ending_bytes(encoded_len: usize, config: &Config) -> usize {
match config.line_wrap {
LineWrap::NoWrap => 0,
LineWrap::Wrap(line_len, line_ending) =>
line_wrap_parameters(encoded_len, line_len, line_ending).total_line_endings_len
}
}
fn assert_encoded_length(input_len: usize, encoded_len: usize, config: Config) {
assert_eq!(encoded_len, encoded_size(input_len, &config).unwrap());
let mut bytes: Vec<u8> = Vec::new();
let mut rng = rand::weak_rng();
for _ in 0..input_len {
bytes.push(rng.gen());
}
let encoded = encode_config(&bytes, config);
assert_encode_sanity(&encoded, &config, input_len);
assert_eq!(encoded_len, encoded.len());
}
fn assert_encode_sanity(encoded: &str, config: &Config, input_len: usize) {
pub fn assert_encode_sanity(encoded: &str, config: &Config, input_len: usize) {
let input_rem = input_len % 3;
let (expected_padding_len, last_encoded_chunk_len) = if input_rem > 0 {
if config.pad {
@ -405,8 +36,9 @@ fn assert_encode_sanity(encoded: &str, config: &Config, input_len: usize) {
let expected_line_ending_len = match config.line_wrap {
LineWrap::NoWrap => 0,
LineWrap::Wrap(line_len, line_ending) =>
LineWrap::Wrap(line_len, line_ending) => {
line_wrap_parameters(b64_only_len, line_len, line_ending).total_line_endings_len
}
};
let expected_encoded_len = encoded_size(input_len, &config).unwrap();
@ -422,8 +54,11 @@ fn assert_encode_sanity(encoded: &str, config: &Config, input_len: usize) {
let _ = str::from_utf8(encoded.as_bytes()).expect("Base64 should be valid utf8");
}
fn roundtrip_random_config(input_len_range: Range<usize>, line_len_range: Range<usize>,
iterations: u32) {
fn roundtrip_random_config(
input_len_range: Range<usize>,
line_len_range: Range<usize>,
iterations: u32,
) {
let mut input_buf: Vec<u8> = Vec::new();
let mut encoded_buf = String::new();
let mut rng = rand::weak_rng();
@ -444,12 +79,7 @@ fn roundtrip_random_config(input_len_range: Range<usize>, line_len_range: Range<
assert_encode_sanity(&encoded_buf, &config, input_len);
// remove line wrapping
let encoded_no_line_endings: String = encoded_buf.chars()
.filter(|&c| c != '\r' && c != '\n')
.collect();
assert_eq!(input_buf, decode_config(&encoded_no_line_endings, config).unwrap());
assert_eq!(input_buf, decode_config(&encoded_buf, config).unwrap());
}
}
@ -468,11 +98,17 @@ pub fn random_config<R: Rng>(rng: &mut R, line_len_range: &Range<usize>) -> Conf
LineWrap::Wrap(line_len, line_ending)
};
let charset = if rng.gen() {
CharacterSet::UrlSafe
} else {
CharacterSet::Standard
const CHARSETS: &[CharacterSet] = &[
CharacterSet::UrlSafe,
CharacterSet::Standard,
CharacterSet::Crypt,
];
let charset = *rng.choose(CHARSETS).unwrap();
let strip_whitespace = match line_wrap {
LineWrap::NoWrap => false,
_ => true,
};
Config::new(charset, rng.gen(), rng.gen(), line_wrap)
Config::new(charset, rng.gen(), strip_whitespace, line_wrap)
}

351
third_party/rust/base64/tests/decode.rs vendored Normal file
View File

@ -0,0 +1,351 @@
extern crate base64;
use base64::*;
mod helpers;
use helpers::*;
fn compare_decode_mime(expected: &str, target: &str) {
assert_eq!(
expected,
String::from_utf8(decode_config(target, MIME).unwrap()).unwrap()
);
}
#[test]
fn decode_rfc4648_0() {
compare_decode("", "");
}
#[test]
fn decode_rfc4648_1() {
compare_decode("f", "Zg==");
}
#[test]
fn decode_rfc4648_1_just_a_bit_of_padding() {
// allows less padding than required
compare_decode("f", "Zg=");
}
#[test]
fn decode_rfc4648_1_no_padding() {
compare_decode("f", "Zg");
}
#[test]
fn decode_rfc4648_2() {
compare_decode("fo", "Zm8=");
}
#[test]
fn decode_rfc4648_2_no_padding() {
compare_decode("fo", "Zm8");
}
#[test]
fn decode_rfc4648_3() {
compare_decode("foo", "Zm9v");
}
#[test]
fn decode_rfc4648_4() {
compare_decode("foob", "Zm9vYg==");
}
#[test]
fn decode_rfc4648_4_no_padding() {
compare_decode("foob", "Zm9vYg");
}
#[test]
fn decode_rfc4648_5() {
compare_decode("fooba", "Zm9vYmE=");
}
#[test]
fn decode_rfc4648_5_no_padding() {
compare_decode("fooba", "Zm9vYmE");
}
#[test]
fn decode_rfc4648_6() {
compare_decode("foobar", "Zm9vYmFy");
}
#[test]
fn decode_mime_allow_space() {
assert!(decode_config("YWx pY2U=", MIME).is_ok());
}
#[test]
fn decode_mime_allow_tab() {
assert!(decode_config("YWx\tpY2U=", MIME).is_ok());
}
#[test]
fn decode_mime_allow_ff() {
assert!(decode_config("YWx\x0cpY2U=", MIME).is_ok());
}
#[test]
fn decode_mime_allow_vtab() {
assert!(decode_config("YWx\x0bpY2U=", MIME).is_ok());
}
#[test]
fn decode_mime_allow_nl() {
assert!(decode_config("YWx\npY2U=", MIME).is_ok());
}
#[test]
fn decode_mime_allow_crnl() {
assert!(decode_config("YWx\r\npY2U=", MIME).is_ok());
}
#[test]
fn decode_mime_reject_null() {
assert_eq!(
DecodeError::InvalidByte(3, 0x0),
decode_config("YWx\0pY2U==", MIME).unwrap_err()
);
}
#[test]
fn decode_mime_absurd_whitespace() {
compare_decode_mime(
"how could you let this happen",
"\n aG93I\n\nG\x0bNvd\r\nWxkI HlvdSB \tsZXQgdGh\rpcyBo\x0cYXBwZW4 = ",
);
}
#[test]
fn decode_single_pad_byte_after_2_chars_in_trailing_quad_ok() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("Zg=");
let input_len = num_quads * 3 + 1;
// Since there are 3 bytes in the trailing quad, want to be sure this allows for the fact
// that it could be bad padding rather than assuming that it will decode to 2 bytes and
// therefore allow 1 extra round of fast decode logic (stage 1 / 2).
let mut decoded = Vec::new();
decoded.resize(input_len, 0);
assert_eq!(
input_len,
decode_config_slice(&s, STANDARD, &mut decoded).unwrap()
);
}
}
//this is a MAY in the rfc: https://tools.ietf.org/html/rfc4648#section-3.3
#[test]
fn decode_1_pad_byte_in_fast_loop_then_extra_padding_chunk_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("YWxpY2U=====");
// since the first 8 bytes are handled in stage 1 or 2, the padding is detected as a
// generic invalid byte, not specifcally a padding issue.
// Could argue that the *next* padding byte (in the next quad) is technically the first
// erroneous one, but reporting that accurately is more complex and probably nobody cares
assert_eq!(
DecodeError::InvalidByte(num_quads * 4 + 7, b'='),
decode(&s).unwrap_err()
);
}
}
#[test]
fn decode_2_pad_bytes_in_leftovers_then_extra_padding_chunk_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("YWxpY2UABB====");
// 6 bytes (4 padding) after last 8-byte chunk, so it's decoded by stage 4.
// First padding byte is invalid.
assert_eq!(
DecodeError::InvalidByte(num_quads * 4 + 10, b'='),
decode(&s).unwrap_err()
);
}
}
#[test]
fn decode_valid_bytes_after_padding_in_leftovers_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("YWxpY2UABB=B");
// 4 bytes after last 8-byte chunk, so it's decoded by stage 4.
// First (and only) padding byte is invalid.
assert_eq!(
DecodeError::InvalidByte(num_quads * 4 + 10, b'='),
decode(&s).unwrap_err()
);
}
}
#[test]
fn decode_absurd_pad_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("==Y=Wx===pY=2U=====");
// Plenty of remaining bytes, so handled by stage 1 or 2.
// first padding byte
assert_eq!(
DecodeError::InvalidByte(num_quads * 4, b'='),
decode(&s).unwrap_err()
);
}
}
#[test]
fn decode_extra_padding_after_1_pad_bytes_in_trailing_quad_returns_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("EEE===");
// handled by stage 1, 2, or 4 depending on length
// first padding byte -- which would be legal if it was the only padding
assert_eq!(
DecodeError::InvalidByte(num_quads * 4 + 3, b'='),
decode(&s).unwrap_err()
);
}
}
#[test]
fn decode_extra_padding_after_2_pad_bytes_in_trailing_quad_2_returns_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("EE====");
// handled by stage 1, 2, or 4 depending on length
// first padding byte -- which would be legal if it was by itself
assert_eq!(
DecodeError::InvalidByte(num_quads * 4 + 2, b'='),
decode(&s).unwrap_err()
);
}
}
#[test]
fn decode_start_quad_with_padding_returns_error() {
for num_quads in 0..25 {
// add enough padding to ensure that we'll hit all 4 stages at the different lengths
for pad_bytes in 1..32 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
let padding: String = std::iter::repeat("=").take(pad_bytes).collect();
s.push_str(&padding);
if pad_bytes % 4 == 1 {
// detected in early length check
assert_eq!(DecodeError::InvalidLength, decode(&s).unwrap_err());
} else {
// padding lengths 2 - 8 are handled by stage 4
// padding length >= 8 will hit at least one chunk at stages 1, 2, 3 at different
// prefix lengths
assert_eq!(
DecodeError::InvalidByte(num_quads * 4, b'='),
decode(&s).unwrap_err()
);
}
}
}
}
#[test]
fn decode_padding_followed_by_non_padding_returns_error() {
for num_quads in 0..25 {
for pad_bytes in 0..31 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
let padding: String = std::iter::repeat("=").take(pad_bytes).collect();
s.push_str(&padding);
s.push_str("E");
if pad_bytes % 4 == 0 {
assert_eq!(DecodeError::InvalidLength, decode(&s).unwrap_err());
} else {
// pad len 1 - 8 will be handled by stage 4
// pad len 9 (suffix len 10) will have 8 bytes of padding handled by stage 3
// first padding byte
assert_eq!(
DecodeError::InvalidByte(num_quads * 4, b'='),
decode(&s).unwrap_err()
);
}
}
}
}
#[test]
fn decode_one_char_in_quad_with_padding_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("E=");
assert_eq!(
DecodeError::InvalidByte(num_quads * 4 + 1, b'='),
decode(&s).unwrap_err()
);
// more padding doesn't change the error
s.push_str("=");
assert_eq!(
DecodeError::InvalidByte(num_quads * 4 + 1, b'='),
decode(&s).unwrap_err()
);
s.push_str("=");
assert_eq!(
DecodeError::InvalidByte(num_quads * 4 + 1, b'='),
decode(&s).unwrap_err()
);
}
}
#[test]
fn decode_one_char_in_quad_without_padding_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push('E');
assert_eq!(DecodeError::InvalidLength, decode(&s).unwrap_err());
}
}
#[test]
fn decode_reject_invalid_bytes_with_correct_error() {
for length in 1..100 {
for index in 0_usize..length {
for invalid_byte in " \t\n\r\x0C\x0B\x00%*.".bytes() {
let prefix: String = std::iter::repeat("A").take(index).collect();
let suffix: String = std::iter::repeat("B").take(length - index - 1).collect();
let input = prefix + &String::from_utf8(vec![invalid_byte]).unwrap() + &suffix;
assert_eq!(
length,
input.len(),
"length {} error position {}",
length,
index
);
if length % 4 == 1 {
assert_eq!(DecodeError::InvalidLength, decode(&input).unwrap_err());
} else {
assert_eq!(
DecodeError::InvalidByte(index, invalid_byte),
decode(&input).unwrap_err()
);
}
}
}
}
}

149
third_party/rust/base64/tests/encode.rs vendored Normal file
View File

@ -0,0 +1,149 @@
extern crate base64;
use base64::*;
fn compare_encode(expected: &str, target: &[u8]) {
assert_eq!(expected, encode(target));
}
#[test]
fn encode_rfc4648_0() {
compare_encode("", b"");
}
#[test]
fn encode_rfc4648_1() {
compare_encode("Zg==", b"f");
}
#[test]
fn encode_rfc4648_2() {
compare_encode("Zm8=", b"fo");
}
#[test]
fn encode_rfc4648_3() {
compare_encode("Zm9v", b"foo");
}
#[test]
fn encode_rfc4648_4() {
compare_encode("Zm9vYg==", b"foob");
}
#[test]
fn encode_rfc4648_5() {
compare_encode("Zm9vYmE=", b"fooba");
}
#[test]
fn encode_rfc4648_6() {
compare_encode("Zm9vYmFy", b"foobar");
}
#[test]
fn encode_all_ascii() {
let mut ascii = Vec::<u8>::with_capacity(128);
for i in 0..128 {
ascii.push(i);
}
compare_encode(
"AAECAwQFBgcICQoLDA0ODxAREhMUFRYXGBkaGxwdHh8gISIjJCUmJygpKissLS4vMDEyMzQ1Njc4OTo7P\
D0+P0BBQkNERUZHSElKS0xNTk9QUVJTVFVWV1hZWltcXV5fYGFiY2RlZmdoaWprbG1ub3BxcnN0dXZ3eHl6e3x9fn8\
=",
&ascii,
);
}
#[test]
fn encode_all_bytes() {
let mut bytes = Vec::<u8>::with_capacity(256);
for i in 0..255 {
bytes.push(i);
}
bytes.push(255); //bug with "overflowing" ranges?
compare_encode(
"AAECAwQFBgcICQoLDA0ODxAREhMUFRYXGBkaGxwdHh8gISIjJCUmJygpKissLS4vMDEyMzQ1Njc4OTo7P\
D0+P0BBQkNERUZHSElKS0xNTk9QUVJTVFVWV1hZWltcXV5fYGFiY2RlZmdoaWprbG1ub3BxcnN0dXZ3eHl6e3x9fn\
+AgYKDhIWGh4iJiouMjY6PkJGSk5SVlpeYmZqbnJ2en6ChoqOkpaanqKmqq6ytrq+wsbKztLW2t7i5uru8vb6\
/wMHCw8TFxsfIycrLzM3Oz9DR0tPU1dbX2Nna29zd3t/g4eLj5OXm5+jp6uvs7e7v8PHy8/T19vf4+fr7/P3+/w==",
&bytes,
);
}
#[test]
fn encode_all_bytes_url() {
let mut bytes = Vec::<u8>::with_capacity(256);
for i in 0..255 {
bytes.push(i);
}
bytes.push(255); //bug with "overflowing" ranges?
assert_eq!(
"AAECAwQFBgcICQoLDA0ODxAREhMUFRYXGBkaGxwdHh8gISIjJCUmJygpKissLS4vMDEyMzQ1Njc4OTo7PD0\
-P0BBQkNERUZHSElKS0xNTk9QUVJTVFVWV1hZWltcXV5fYGFiY2RlZmdoaWprbG1ub3BxcnN0dXZ3eHl6e3x9fn\
-AgYKDhIWGh4iJiouMjY6PkJGSk5SVlpeYmZqbnJ2en6ChoqOkpaanqKmqq6ytrq\
-wsbKztLW2t7i5uru8vb6_wMHCw8TFxsfIycrLzM3Oz9DR0tPU1dbX2Nna29zd3t_g4eLj5OXm5-jp6uvs7e7v8PHy\
8_T19vf4-fr7_P3-_w==",
encode_config(&bytes, URL_SAFE)
);
}
#[test]
fn encode_line_ending_lf_partial_last_line() {
let config = Config::new(
CharacterSet::Standard,
true,
false,
LineWrap::Wrap(3, LineEnding::LF),
);
assert_eq!("Zm9\nvYm\nFy", encode_config(b"foobar", config));
}
#[test]
fn encode_line_ending_crlf_partial_last_line() {
let config = Config::new(
CharacterSet::Standard,
true,
false,
LineWrap::Wrap(3, LineEnding::CRLF),
);
assert_eq!("Zm9\r\nvYm\r\nFy", encode_config(b"foobar", config));
}
#[test]
fn encode_line_ending_lf_full_last_line() {
let config = Config::new(
CharacterSet::Standard,
true,
false,
LineWrap::Wrap(4, LineEnding::LF),
);
assert_eq!("Zm9v\nYmFy", encode_config(b"foobar", config));
}
#[test]
fn encode_line_ending_crlf_full_last_line() {
let config = Config::new(
CharacterSet::Standard,
true,
false,
LineWrap::Wrap(4, LineEnding::CRLF),
);
assert_eq!("Zm9v\r\nYmFy", encode_config(b"foobar", config));
}
#[test]
fn encode_url_safe_without_padding() {
let encoded = encode_config(b"alice", URL_SAFE_NO_PAD);
assert_eq!(&encoded, "YWxpY2U");
assert_eq!(
String::from_utf8(decode(&encoded).unwrap()).unwrap(),
"alice"
);
}

View File

@ -0,0 +1,14 @@
extern crate base64;
use base64::*;
pub fn compare_decode(expected: &str, target: &str) {
assert_eq!(
expected,
String::from_utf8(decode(target).unwrap()).unwrap()
);
assert_eq!(
expected,
String::from_utf8(decode(target.as_bytes()).unwrap()).unwrap()
);
}

View File

@ -5,22 +5,18 @@ use rand::Rng;
use base64::*;
fn compare_encode(expected: &str, target: &[u8]) {
assert_eq!(expected, encode(target));
}
fn compare_decode(expected: &str, target: &str) {
assert_eq!(expected, String::from_utf8(decode(target).unwrap()).unwrap());
assert_eq!(expected, String::from_utf8(decode(target.as_bytes()).unwrap()).unwrap());
}
fn compare_decode_mime(expected: &str, target: &str) {
assert_eq!(expected, String::from_utf8(decode_config(target, MIME).unwrap()).unwrap());
}
mod helpers;
use helpers::*;
// generate random contents of the specified length and test encode/decode roundtrip
fn roundtrip_random(byte_buf: &mut Vec<u8>, str_buf: &mut String, config: Config,
byte_len: usize, approx_values_per_byte: u8, max_rounds: u64) {
fn roundtrip_random(
byte_buf: &mut Vec<u8>,
str_buf: &mut String,
config: Config,
byte_len: usize,
approx_values_per_byte: u8,
max_rounds: u64,
) {
// let the short ones be short but don't let it get too crazy large
let num_rounds = calculate_number_of_rounds(byte_len, approx_values_per_byte, max_rounds);
let mut r = rand::weak_rng();
@ -60,206 +56,6 @@ fn no_pad_config() -> Config {
Config::new(CharacterSet::Standard, false, false, LineWrap::NoWrap)
}
//-------
//decode
#[test]
fn decode_rfc4648_0() {
compare_decode("", "");
}
#[test]
fn decode_rfc4648_1() {
compare_decode("f", "Zg==");
}
#[test]
fn decode_rfc4648_1_just_a_bit_of_padding() {
// allows less padding than required
compare_decode("f", "Zg=");
}
#[test]
fn decode_rfc4648_1_no_padding() {
compare_decode("f", "Zg");
}
#[test]
fn decode_rfc4648_2() {
compare_decode("fo", "Zm8=");
}
#[test]
fn decode_rfc4648_2_no_padding() {
compare_decode("fo", "Zm8");
}
#[test]
fn decode_rfc4648_3() {
compare_decode("foo", "Zm9v");
}
#[test]
fn decode_rfc4648_4() {
compare_decode("foob", "Zm9vYg==");
}
#[test]
fn decode_rfc4648_4_no_padding() {
compare_decode("foob", "Zm9vYg");
}
#[test]
fn decode_rfc4648_5() {
compare_decode("fooba", "Zm9vYmE=");
}
#[test]
fn decode_rfc4648_5_no_padding() {
compare_decode("fooba", "Zm9vYmE");
}
#[test]
fn decode_rfc4648_6() {
compare_decode("foobar", "Zm9vYmFy");
}
//this is a MAY in the rfc: https://tools.ietf.org/html/rfc4648#section-3.3
#[test]
fn decode_pad_inside_fast_loop_chunk_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("YWxpY2U=====");
// since the first 8 bytes are handled in the fast loop, the
// padding is an error. Could argue that the *next* padding
// byte is technically the first erroneous one, but reporting
// that accurately is more complex and probably nobody cares
assert_eq!(DecodeError::InvalidByte(num_quads * 4 + 7, b'='), decode(&s).unwrap_err());
}
}
#[test]
fn decode_extra_pad_after_fast_loop_chunk_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("YWxpY2UABB===");
// first padding byte
assert_eq!(DecodeError::InvalidByte(num_quads * 4 + 10, b'='), decode(&s).unwrap_err());
}
}
#[test]
fn decode_absurd_pad_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("==Y=Wx===pY=2U=====");
// first padding byte
assert_eq!(DecodeError::InvalidByte(num_quads * 4, b'='), decode(&s).unwrap_err());
}
}
#[test]
fn decode_extra_padding_in_trailing_quad_returns_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("EEE==");
// first padding byte -- which would be legal if it was by itself
assert_eq!(DecodeError::InvalidByte(num_quads * 4 + 3, b'='), decode(&s).unwrap_err());
}
}
#[test]
fn decode_extra_padding_in_trailing_quad_2_returns_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("EE===");
// first padding byte -- which would be legal if it was by itself
assert_eq!(DecodeError::InvalidByte(num_quads * 4 + 2, b'='), decode(&s).unwrap_err());
}
}
#[test]
fn decode_start_second_quad_with_padding_returns_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("=");
// first padding byte -- must have two non-padding bytes in a quad
assert_eq!(DecodeError::InvalidByte(num_quads * 4, b'='), decode(&s).unwrap_err());
// two padding bytes -- same
s.push_str("=");
assert_eq!(DecodeError::InvalidByte(num_quads * 4, b'='), decode(&s).unwrap_err());
s.push_str("=");
assert_eq!(DecodeError::InvalidByte(num_quads * 4, b'='), decode(&s).unwrap_err());
s.push_str("=");
assert_eq!(DecodeError::InvalidByte(num_quads * 4, b'='), decode(&s).unwrap_err());
}
}
#[test]
fn decode_padding_in_last_quad_followed_by_non_padding_returns_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("==E");
// first padding byte -- must have two non-padding bytes in a quad
assert_eq!(DecodeError::InvalidByte(num_quads * 4, b'='), decode(&s).unwrap_err());
}
}
#[test]
fn decode_one_char_in_quad_with_padding_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("E=");
assert_eq!(DecodeError::InvalidByte(num_quads * 4 + 1, b'='), decode(&s).unwrap_err());
// more padding doesn't change the error
s.push_str("=");
assert_eq!(DecodeError::InvalidByte(num_quads * 4 + 1, b'='), decode(&s).unwrap_err());
s.push_str("=");
assert_eq!(DecodeError::InvalidByte(num_quads * 4 + 1, b'='), decode(&s).unwrap_err());
}
}
#[test]
fn decode_one_char_in_quad_without_padding_error() {
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push('E');
assert_eq!(DecodeError::InvalidLength, decode(&s).unwrap_err());
}
}
#[test]
fn decode_reject_invalid_bytes_with_correct_error() {
for length in 1..100 {
for index in 0_usize..length {
for invalid_byte in " \t\n\r\x0C\x0B\x00%*.".bytes() {
let prefix: String = std::iter::repeat("A").take(index).collect();
let suffix: String = std::iter::repeat("B").take(length - index - 1).collect();
let input = prefix + &String::from_utf8(vec![invalid_byte]).unwrap() + &suffix;
assert_eq!(length, input.len(), "length {} error position {}", length, index);
assert_eq!(DecodeError::InvalidByte(index, invalid_byte),
decode(&input).unwrap_err());
}
}
}
}
#[test]
fn roundtrip_random_short_standard() {
let mut byte_buf: Vec<u8> = Vec::new();
@ -286,7 +82,14 @@ fn roundtrip_random_short_no_padding() {
let mut str_buf = String::new();
for input_len in 0..40 {
roundtrip_random(&mut byte_buf, &mut str_buf, no_pad_config(), input_len, 4, 10000);
roundtrip_random(
&mut byte_buf,
&mut str_buf,
no_pad_config(),
input_len,
4,
10000,
);
}
}
@ -296,150 +99,50 @@ fn roundtrip_random_no_padding() {
let mut str_buf = String::new();
for input_len in 40..100 {
roundtrip_random(&mut byte_buf, &mut str_buf, no_pad_config(), input_len, 4, 1000);
roundtrip_random(
&mut byte_buf,
&mut str_buf,
no_pad_config(),
input_len,
4,
1000,
);
}
}
#[test]
fn decode_mime_allow_space() {
assert!(decode_config("YWx pY2U=", MIME).is_ok());
}
fn roundtrip_decode_trailing_10_bytes() {
// This is a special case because we decode 8 byte blocks of input at a time as much as we can,
// ideally unrolled to 32 bytes at a time, in stages 1 and 2. Since we also write a u64's worth
// of bytes (8) to the output, we always write 2 garbage bytes that then will be overwritten by
// the NEXT block. However, if the next block only contains 2 bytes, it will decode to 1 byte,
// and therefore be too short to cover up the trailing 2 garbage bytes. Thus, we have stage 3
// to handle that case.
#[test]
fn decode_mime_allow_tab() {
assert!(decode_config("YWx\tpY2U=", MIME).is_ok());
}
for num_quads in 0..25 {
let mut s: String = std::iter::repeat("ABCD").take(num_quads).collect();
s.push_str("EFGHIJKLZg");
#[test]
fn decode_mime_allow_ff() {
assert!(decode_config("YWx\x0cpY2U=", MIME).is_ok());
}
let decoded = decode(&s).unwrap();
assert_eq!(num_quads * 3 + 7, decoded.len());
#[test]
fn decode_mime_allow_vtab() {
assert!(decode_config("YWx\x0bpY2U=", MIME).is_ok());
}
#[test]
fn decode_mime_allow_nl() {
assert!(decode_config("YWx\npY2U=", MIME).is_ok());
}
#[test]
fn decode_mime_allow_crnl() {
assert!(decode_config("YWx\r\npY2U=", MIME).is_ok());
}
#[test]
fn decode_mime_reject_null() {
assert_eq!(DecodeError::InvalidByte(3, 0x0),decode_config("YWx\0pY2U=", MIME).unwrap_err());
}
#[test]
fn decode_mime_absurd_whitespace() {
compare_decode_mime("how could you let this happen",
"\n aG93I\n\nG\x0bNvd\r\nWxkI HlvdSB \tsZXQgdGh\rpcyBo\x0cYXBwZW4 = ");
}
//-------
//encode
#[test]
fn encode_rfc4648_0() {
compare_encode("", b"");
}
#[test]
fn encode_rfc4648_1() {
compare_encode("Zg==", b"f");
}
#[test]
fn encode_rfc4648_2() {
compare_encode("Zm8=", b"fo");
}
#[test]
fn encode_rfc4648_3() {
compare_encode("Zm9v", b"foo");
}
#[test]
fn encode_rfc4648_4() {
compare_encode("Zm9vYg==", b"foob");
}
#[test]
fn encode_rfc4648_5() {
compare_encode("Zm9vYmE=", b"fooba");
}
#[test]
fn encode_rfc4648_6() {
compare_encode("Zm9vYmFy", b"foobar");
}
#[test]
fn encode_all_ascii() {
let mut ascii = Vec::<u8>::with_capacity(128);
for i in 0..128 {
ascii.push(i);
assert_eq!(s, encode_config(&decoded, STANDARD_NO_PAD));
}
compare_encode("AAECAwQFBgcICQoLDA0ODxAREhMUFRYXGBkaGxwdHh8gISIjJCUmJygpKissLS4vMDEyMzQ1Njc4OTo7PD0+P0BBQkNERUZHSElKS0xNTk9QUVJTVFVWV1hZWltcXV5fYGFiY2RlZmdoaWprbG1ub3BxcnN0dXZ3eHl6e3x9fn8=", &ascii);
}
#[test]
fn encode_all_bytes() {
fn display_wrapper_matches_normal_encode() {
let mut bytes = Vec::<u8>::with_capacity(256);
for i in 0..255 {
bytes.push(i);
}
bytes.push(255); //bug with "overflowing" ranges?
bytes.push(255);
compare_encode("AAECAwQFBgcICQoLDA0ODxAREhMUFRYXGBkaGxwdHh8gISIjJCUmJygpKissLS4vMDEyMzQ1Njc4OTo7PD0+P0BBQkNERUZHSElKS0xNTk9QUVJTVFVWV1hZWltcXV5fYGFiY2RlZmdoaWprbG1ub3BxcnN0dXZ3eHl6e3x9fn+AgYKDhIWGh4iJiouMjY6PkJGSk5SVlpeYmZqbnJ2en6ChoqOkpaanqKmqq6ytrq+wsbKztLW2t7i5uru8vb6/wMHCw8TFxsfIycrLzM3Oz9DR0tPU1dbX2Nna29zd3t/g4eLj5OXm5+jp6uvs7e7v8PHy8/T19vf4+fr7/P3+/w==", &bytes);
}
#[test]
fn encode_all_bytes_url() {
let mut bytes = Vec::<u8>::with_capacity(256);
for i in 0..255 {
bytes.push(i);
}
bytes.push(255); //bug with "overflowing" ranges?
assert_eq!("AAECAwQFBgcICQoLDA0ODxAREhMUFRYXGBkaGxwdHh8gISIjJCUmJygpKissLS4vMDEyMzQ1Njc4OTo7PD0-P0BBQkNERUZHSElKS0xNTk9QUVJTVFVWV1hZWltcXV5fYGFiY2RlZmdoaWprbG1ub3BxcnN0dXZ3eHl6e3x9fn-AgYKDhIWGh4iJiouMjY6PkJGSk5SVlpeYmZqbnJ2en6ChoqOkpaanqKmqq6ytrq-wsbKztLW2t7i5uru8vb6_wMHCw8TFxsfIycrLzM3Oz9DR0tPU1dbX2Nna29zd3t_g4eLj5OXm5-jp6uvs7e7v8PHy8_T19vf4-fr7_P3-_w==", encode_config(&bytes, URL_SAFE));
}
#[test]
fn encode_line_ending_lf_partial_last_line() {
let config = Config::new(CharacterSet::Standard, true, false,
LineWrap::Wrap(3, LineEnding::LF));
assert_eq!("Zm9\nvYm\nFy", encode_config(b"foobar", config));
}
#[test]
fn encode_line_ending_crlf_partial_last_line() {
let config = Config::new(CharacterSet::Standard, true, false,
LineWrap::Wrap(3, LineEnding::CRLF));
assert_eq!("Zm9\r\nvYm\r\nFy", encode_config(b"foobar", config));
}
#[test]
fn encode_line_ending_lf_full_last_line() {
let config = Config::new(CharacterSet::Standard, true, false,
LineWrap::Wrap(4, LineEnding::LF));
assert_eq!("Zm9v\nYmFy", encode_config(b"foobar", config));
}
#[test]
fn encode_line_ending_crlf_full_last_line() {
let config = Config::new(CharacterSet::Standard, true, false,
LineWrap::Wrap(4, LineEnding::CRLF));
assert_eq!("Zm9v\r\nYmFy", encode_config(b"foobar", config));
assert_eq!(
encode(&bytes),
format!("{}", base64::display::Base64Display::standard(&bytes))
);
}
#[test]
@ -450,8 +153,39 @@ fn because_we_can() {
}
#[test]
fn encode_url_safe_without_padding() {
let encoded = encode_config(b"alice", URL_SAFE_NO_PAD);
assert_eq!(&encoded, "YWxpY2U");
assert_eq!(String::from_utf8(decode(&encoded).unwrap()).unwrap(), "alice");
fn encode_config_slice_can_use_inline_buffer() {
let mut buf: [u8; 22] = [0; 22];
let mut larger_buf: [u8; 24] = [0; 24];
let mut input: [u8; 16] = [0; 16];
let mut rng = rand::weak_rng();
for elt in &mut input {
*elt = rng.gen();
}
assert_eq!(22, encode_config_slice(&input, STANDARD_NO_PAD, &mut buf));
let decoded = decode_config(&buf, STANDARD_NO_PAD).unwrap();
assert_eq!(decoded, input);
// let's try it again with padding
assert_eq!(24, encode_config_slice(&input, STANDARD, &mut larger_buf));
let decoded = decode_config(&buf, STANDARD).unwrap();
assert_eq!(decoded, input);
}
#[test]
#[should_panic(expected = "index 24 out of range for slice of length 22")]
fn encode_config_slice_panics_when_buffer_too_small() {
let mut buf: [u8; 22] = [0; 22];
let mut input: [u8; 16] = [0; 16];
let mut rng = rand::weak_rng();
for elt in &mut input {
*elt = rng.gen();
}
encode_config_slice(&input, STANDARD, &mut buf);
}