char_utf8.rs - source

Files

aho_corasick

packed

teddy

compile.rs mod.rs runtime.rs

api.rs mod.rs pattern.rs rabinkarp.rs vector.rs

ahocorasick.rs automaton.rs buffer.rs byte_frequencies.rs classes.rs dfa.rs error.rs lib.rs nfa.rs prefilter.rs state_id.rs

atty

beef

generic.rs lean.rs lib.rs traits.rs wide.rs

bitflags

bstr

byteorder

cfg_if

clap

build

app

debug_asserts.rs mod.rs settings.rs

arg

mod.rs settings.rs value_hint.rs

arg_group.rs macros.rs mod.rs usage_parser.rs

output

fmt.rs help.rs mod.rs usage.rs

parse

features

mod.rs suggestions.rs

matches

arg_matches.rs matched_arg.rs mod.rs

arg_matcher.rs errors.rs mod.rs parser.rs validator.rs

util

argstr.rs fnv.rs graph.rs id.rs mod.rs

derive.rs lib.rs macros.rs mkeymap.rs

clap_derive

derives

arg_enum.rs clap.rs from_arg_matches.rs into_app.rs mod.rs subcommand.rs

utils

doc_comments.rs mod.rs spanned.rs ty.rs

attrs.rs dummies.rs lib.rs parse.rs

codespan

file.rs index.rs lib.rs location.rs span.rs

codespan_reporting

term

config.rs renderer.rs views.rs

diagnostic.rs files.rs lib.rs term.rs

crc_any

constants

crc_u16.rs crc_u32.rs crc_u64.rs crc_u8.rs mod.rs

crc_u16.rs crc_u32.rs crc_u64.rs crc_u8.rs lib.rs

crypto_hash

imp

csv

byte_record.rs cookbook.rs deserializer.rs error.rs lib.rs reader.rs serializer.rs string_record.rs tutorial.rs writer.rs

csv_core

lib.rs reader.rs writer.rs

debug_helper

filepath

fixed

fixed_macro

fixed_macro_impl

dispatch.rs lib.rs

fixed_macro_types

fnv

foreign_types

foreign_types_shared

getrandom

error.rs error_impls.rs lib.rs linux_android.rs use_file.rs util.rs util_libc.rs

glob

hashbrown

external_trait_impls

raw

bitmask.rs mod.rs sse2.rs

lib.rs macros.rs map.rs scopeguard.rs set.rs

heck

camel.rs kebab.rs lib.rs mixed.rs shouty_kebab.rs shouty_snake.rs snake.rs title.rs

hex

indexmap

map

core

equivalent.rs lib.rs macros.rs map.rs mutable_keys.rs set.rs util.rs

itoa

lazy_static

inline_lazy.rs lib.rs

libc

unix

linux_like

linux

gnu

b64

x86_64

align.rs mod.rs not_x32.rs

align.rs mod.rs

align.rs mod.rs

align.rs mod.rs

fixed_width_ints.rs lib.rs macros.rs

linked_hash_map

linked_hash_set

logos

internal.rs lexer.rs lib.rs source.rs

logos_derive

lrl_test_compiler

config_binary

mod.rs sections.rs

environment

calibration.rs config.rs drivers.rs mod.rs

mission_control

result

diagnostic.rs mod.rs stringable.rs

test_assembly

disassembly.rs mod.rs

test_binary

instructions.rs mod.rs segments.rs

test_descriptor

ast.rs concrete_test.rs mod.rs

timeline

mod.rs relays.rs sensors.rs

bits.rs lib.rs test_builder.rs

maplit

memchr

x86

avx.rs mod.rs sse2.rs

fallback.rs iter.rs lib.rs naive.rs

memoffset

lib.rs offset_of.rs raw_field.rs span_of.rs

once_cell

imp_std.rs lib.rs race.rs

openssl

openssl_sys

os_str_bytes

common

paste

attr.rs error.rs lib.rs segment.rs

pest

iterators

flat_pairs.rs mod.rs pair.rs pairs.rs queueable_token.rs tokens.rs

unicode

binary.rs category.rs mod.rs

error.rs lib.rs macros.rs parser.rs parser_state.rs position.rs prec_climber.rs span.rs stack.rs token.rs

pest_derive

pest_generator

generator.rs lib.rs macros.rs

pest_meta

optimizer

concatenator.rs factorizer.rs mod.rs restorer.rs rotater.rs skipper.rs unroller.rs

ast.rs grammar.rs lib.rs parser.rs validator.rs

phf

lib.rs map.rs ordered_map.rs ordered_set.rs set.rs

phf_generator

phf_macros

phf_shared

ppv_lite86

x86_64

lib.rs soft.rs types.rs

proc_macro2

detection.rs fallback.rs lib.rs marker.rs parse.rs wrapper.rs

proc_macro_error

imp

diagnostic.rs dummy.rs lib.rs macros.rs sealed.rs

proc_macro_error_attr

lib.rs parse.rs settings.rs

proc_macro_hack

error.rs iter.rs lib.rs parse.rs quote.rs

quote

ext.rs format.rs ident_fragment.rs lib.rs runtime.rs spanned.rs to_tokens.rs

rand

distributions

bernoulli.rs float.rs integer.rs mod.rs other.rs uniform.rs utils.rs weighted.rs weighted_index.rs

rngs

adapter

mod.rs read.rs reseeding.rs

mock.rs mod.rs small.rs std.rs thread.rs xoshiro256plusplus.rs

seq

index.rs mod.rs

lib.rs prelude.rs rng.rs

rand_chacha

chacha.rs guts.rs lib.rs

rand_core

block.rs error.rs impls.rs le.rs lib.rs os.rs

regex

regex_automata

classes.rs dense.rs dfa.rs lib.rs regex.rs sparse.rs state_id.rs

regex_syntax

remove_dir_all

ring

rowan

green

builder.rs element.rs node.rs token.rs

api.rs cow_mut.rs cursor.rs green.rs lib.rs syntax_text.rs utility_types.rs

rustc_hash

ryu

buffer

pretty

exponent.rs mantissa.rs mod.rs

common.rs d2s.rs d2s_full_table.rs d2s_intrinsics.rs digit_table.rs f2s.rs f2s_intrinsics.rs lib.rs

semver

lib.rs version.rs version_req.rs

semver_parser

generated.rs lexer.rs lib.rs parser.rs range.rs range_set.rs version.rs

serde

de

ignored_any.rs impls.rs mod.rs seed.rs utf8.rs value.rs

private

de.rs doc.rs mod.rs ser.rs size_hint.rs

ser

fmt.rs impls.rs impossible.rs mod.rs

integer128.rs lib.rs macros.rs

serde_derive

internals

ast.rs attr.rs case.rs check.rs ctxt.rs mod.rs receiver.rs respan.rs symbol.rs

bound.rs de.rs dummy.rs fragment.rs lib.rs pretend.rs ser.rs try.rs

serde_json

features_check

io

value

de.rs from.rs index.rs mod.rs partial_eq.rs ser.rs

de.rs error.rs iter.rs lib.rs macros.rs map.rs number.rs read.rs ser.rs

siphasher

lib.rs sip.rs sip128.rs

smallvec

smawk

lib.rs monge.rs

smol_str

spin

lib.rs mutex.rs once.rs rw_lock.rs

stable_deref_trait

strsim

syn

taplo

dom

macros.rs mod.rs

formatter

macros.rs mod.rs

parser

macros.rs mod.rs

util

coords.rs escape.rs mod.rs syntax.rs

analytics.rs lib.rs syntax.rs value.rs

tempfile

file

imp

dir.rs error.rs lib.rs spooled.rs util.rs

termcolor

text_size

lib.rs range.rs size.rs traits.rs

textwrap

wrap_algorithms

core.rs indentation.rs lib.rs word_separators.rs word_splitters.rs wrap_algorithms.rs

toml

datetime.rs de.rs lib.rs macros.rs map.rs ser.rs spanned.rs tokens.rs value.rs

triomphe

arc.rs arc_borrow.rs arc_union.rs header.rs lib.rs offset_arc.rs thin_arc.rs unique_arc.rs

typenum

array.rs bit.rs int.rs lib.rs marker_traits.rs operator_aliases.rs private.rs type_operators.rs uint.rs

ucd_trie

lib.rs owned.rs

unicode_linebreak

lib.rs shared.rs

unicode_segmentation

grapheme.rs lib.rs sentence.rs tables.rs word.rs

unicode_width

lib.rs tables.rs

unicode_xid

lib.rs tables.rs

untrusted

utf8_ranges

char_utf8.rs lib.rs

vec_map

// Pulled from std::char until encode_utf8 stabilizes. ---AG

// UTF-8 ranges and tags for encoding characters
const TAG_CONT: u8    = 0b1000_0000;
const TAG_TWO_B: u8   = 0b1100_0000;
const TAG_THREE_B: u8 = 0b1110_0000;
const TAG_FOUR_B: u8  = 0b1111_0000;
const MAX_ONE_B: u32   =     0x80;
const MAX_TWO_B: u32   =    0x800;
const MAX_THREE_B: u32 =  0x10000;

#[inline]
pub fn encode_utf8(character: char, dst: &mut [u8]) -> Option<usize> {
    let code = character as u32;
    if code < MAX_ONE_B && !dst.is_empty() {
        dst[0] = code as u8;
        Some(1)
    } else if code < MAX_TWO_B && dst.len() >= 2 {
        dst[0] = (code >> 6 & 0x1F) as u8 | TAG_TWO_B;
        dst[1] = (code & 0x3F) as u8 | TAG_CONT;
        Some(2)
    } else if code < MAX_THREE_B && dst.len() >= 3  {
        dst[0] = (code >> 12 & 0x0F) as u8 | TAG_THREE_B;
        dst[1] = (code >>  6 & 0x3F) as u8 | TAG_CONT;
        dst[2] = (code & 0x3F) as u8 | TAG_CONT;
        Some(3)
    } else if dst.len() >= 4 {
        dst[0] = (code >> 18 & 0x07) as u8 | TAG_FOUR_B;
        dst[1] = (code >> 12 & 0x3F) as u8 | TAG_CONT;
        dst[2] = (code >>  6 & 0x3F) as u8 | TAG_CONT;
        dst[3] = (code & 0x3F) as u8 | TAG_CONT;
        Some(4)
    } else {
        None
    }
}