gecko-dev/xpcom/string/nsReadableUtils.h

821 строка
27 KiB
C
Исходник Обычный вид История

/* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* vim: set ts=8 sts=2 et sw=2 tw=80: */
2012-05-21 15:12:37 +04:00
/* This Source Code Form is subject to the terms of the Mozilla Public
* License, v. 2.0. If a copy of the MPL was not distributed with this
* file, You can obtain one at http://mozilla.org/MPL/2.0/. */
// IWYU pragma: private, include "nsString.h"
#ifndef nsReadableUtils_h___
#define nsReadableUtils_h___
/**
* I guess all the routines in this file are all mis-named.
* According to our conventions, they should be |NS_xxx|.
*/
#include "mozilla/Assertions.h"
2001-04-02 23:40:52 +04:00
#include "nsAString.h"
#include "nsTArrayForwardDeclare.h"
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
// Can't include mozilla/Encoding.h here. The implementations are in
// the encoding_rs and encoding_glue crates.
extern "C" {
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
size_t
encoding_utf8_valid_up_to(uint8_t const* buffer, size_t buffer_len);
bool
encoding_mem_is_ascii(uint8_t const* buffer, size_t buffer_len);
bool
encoding_mem_is_basic_latin(char16_t const* buffer, size_t buffer_len);
bool
encoding_mem_is_utf8_latin1(uint8_t const* buffer, size_t buffer_len);
bool
encoding_mem_is_str_latin1(uint8_t const* buffer, size_t buffer_len);
bool
encoding_mem_is_utf16_latin1(char16_t const* buffer, size_t buffer_len);
size_t
encoding_mem_utf16_valid_up_to(char16_t const* buffer, size_t buffer_len);
void
encoding_mem_ensure_utf16_validity(char16_t* buffer, size_t buffer_len);
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
void
encoding_mem_convert_utf16_to_latin1_lossy(const char16_t* src,
size_t src_len,
char* dst,
size_t dst_len);
size_t
encoding_mem_convert_utf8_to_latin1_lossy(const char* src,
size_t src_len,
char* dst,
size_t dst_len);
void
encoding_mem_convert_latin1_to_utf16(const char* src,
size_t src_len,
char16_t* dst,
size_t dst_len);
size_t
encoding_mem_convert_utf16_to_utf8(const char16_t* src,
size_t src_len,
char* dst,
size_t dst_len);
size_t
encoding_mem_convert_utf8_to_utf16(const char* src,
size_t src_len,
char16_t* dst,
size_t dst_len);
}
// From the nsstring crate
extern "C" {
bool
nsstring_fallible_append_utf8_impl(nsAString* aThis,
const char* aOther,
size_t aOtherLen,
size_t aOldLen);
bool
nsstring_fallible_append_latin1_impl(nsAString* aThis,
const char* aOther,
size_t aOtherLen,
size_t aOldLen,
bool aAllowShrinking);
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
bool
nscstring_fallible_append_utf16_to_utf8_impl(nsACString* aThis,
const char16_t*,
size_t aOtherLen,
size_t aOldLen);
bool
nscstring_fallible_append_utf16_to_latin1_lossy_impl(nsACString* aThis,
const char16_t*,
size_t aOtherLen,
size_t aOldLen,
bool aAllowShrinking);
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
bool
nscstring_fallible_append_utf8_to_latin1_lossy_check(nsACString* aThis,
const nsACString* aOther,
size_t aOldLen);
bool
nscstring_fallible_append_latin1_to_utf8_check(nsACString* aThis,
const nsACString* aOther,
size_t aOldLen);
}
/**
* If all the code points in the input are below U+0100, converts to Latin1,
* i.e. unsigned byte value is Unicode scalar value; not windows-1252. If
* there are code points above U+00FF, produces garbage in a memory-safe way
* and will likely start asserting in future debug builds. The nature of the
* garbage depends on the CPU architecture and must not be relied upon.
*
* The length of aDest must be not be less than the length of aSource.
*/
inline void
LossyConvertUTF16toLatin1(mozilla::Span<const char16_t> aSource,
mozilla::Span<char> aDest)
{
encoding_mem_convert_utf16_to_latin1_lossy(
aSource.Elements(), aSource.Length(), aDest.Elements(), aDest.Length());
}
/**
* If all the code points in the input are below U+0100, converts to Latin1,
* i.e. unsigned byte value is Unicode scalar value; not windows-1252. If
* there are code points above U+00FF, asserts in debug builds and produces
* garbage in memory-safe way in release builds. The nature of the garbage
* may depend on the CPU architecture and must not be relied upon.
*
* The length of aDest must be not be less than the length of aSource.
*/
inline size_t
LossyConvertUTF8toLatin1(mozilla::Span<const char> aSource,
mozilla::Span<char> aDest)
{
return encoding_mem_convert_utf8_to_latin1_lossy(
aSource.Elements(), aSource.Length(), aDest.Elements(), aDest.Length());
}
/**
* Interprets unsigned byte value as Unicode scalar value (i.e. not
* windows-1252!).
*
* The length of aDest must be not be less than the length of aSource.
*/
inline void
ConvertLatin1toUTF16(mozilla::Span<const char> aSource,
mozilla::Span<char16_t> aDest)
{
encoding_mem_convert_latin1_to_utf16(
aSource.Elements(), aSource.Length(), aDest.Elements(), aDest.Length());
}
/**
* Lone surrogates are replaced with the REPLACEMENT CHARACTER.
*
* The length of aDest must be at least the length of aSource times three
* _plus one_.
*
* Returns the number of code units written.
*/
inline size_t
ConvertUTF16toUTF8(mozilla::Span<const char16_t> aSource,
mozilla::Span<char> aDest)
{
return encoding_mem_convert_utf16_to_utf8(
aSource.Elements(), aSource.Length(), aDest.Elements(), aDest.Length());
}
/**
* Malformed byte sequences are replaced with the REPLACEMENT CHARACTER.
*
* The length of aDest must at least one greater than the length of aSource.
*
* Returns the number of code units written.
*/
inline size_t
ConvertUTF8toUTF16(mozilla::Span<const char> aSource,
mozilla::Span<char16_t> aDest)
{
return encoding_mem_convert_utf8_to_utf16(
aSource.Elements(), aSource.Length(), aDest.Elements(), aDest.Length());
}
inline size_t
Distance(const nsReadingIterator<char16_t>& aStart,
const nsReadingIterator<char16_t>& aEnd)
{
MOZ_ASSERT(aStart.get() <= aEnd.get());
return static_cast<size_t>(aEnd.get() - aStart.get());
}
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
inline size_t
Distance(const nsReadingIterator<char>& aStart,
const nsReadingIterator<char>& aEnd)
{
MOZ_ASSERT(aStart.get() <= aEnd.get());
return static_cast<size_t>(aEnd.get() - aStart.get());
}
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
// UTF-8 to UTF-16
// Invalid UTF-8 byte sequences are replaced with the REPLACEMENT CHARACTER.
inline MOZ_MUST_USE bool
CopyUTF8toUTF16(mozilla::Span<const char> aSource,
nsAString& aDest,
const mozilla::fallible_t&)
{
return nsstring_fallible_append_utf8_impl(
&aDest, aSource.Elements(), aSource.Length(), 0);
}
inline void
CopyUTF8toUTF16(mozilla::Span<const char> aSource, nsAString& aDest)
{
if (MOZ_UNLIKELY(!CopyUTF8toUTF16(aSource, aDest, mozilla::fallible))) {
aDest.AllocFailed(aSource.Length());
}
}
inline MOZ_MUST_USE bool
AppendUTF8toUTF16(mozilla::Span<const char> aSource,
nsAString& aDest,
const mozilla::fallible_t&)
{
return nsstring_fallible_append_utf8_impl(
&aDest, aSource.Elements(), aSource.Length(), aDest.Length());
}
inline void
AppendUTF8toUTF16(mozilla::Span<const char> aSource, nsAString& aDest)
{
if (MOZ_UNLIKELY(!AppendUTF8toUTF16(aSource, aDest, mozilla::fallible))) {
aDest.AllocFailed(aDest.Length() + aSource.Length());
}
}
// Latin1 to UTF-16
// Interpret each incoming unsigned byte value as a Unicode scalar value (not
// windows-1252!). The function names say "ASCII" instead of "Latin1" for
// legacy reasons.
inline MOZ_MUST_USE bool
CopyASCIItoUTF16(mozilla::Span<const char> aSource,
nsAString& aDest,
const mozilla::fallible_t&)
{
return nsstring_fallible_append_latin1_impl(
&aDest, aSource.Elements(), aSource.Length(), 0, true);
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
}
inline void
CopyASCIItoUTF16(mozilla::Span<const char> aSource, nsAString& aDest)
{
if (MOZ_UNLIKELY(!CopyASCIItoUTF16(aSource, aDest, mozilla::fallible))) {
aDest.AllocFailed(aSource.Length());
}
}
inline MOZ_MUST_USE bool
AppendASCIItoUTF16(mozilla::Span<const char> aSource,
nsAString& aDest,
const mozilla::fallible_t&)
{
return nsstring_fallible_append_latin1_impl(
&aDest, aSource.Elements(), aSource.Length(), aDest.Length(), false);
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
}
inline void
AppendASCIItoUTF16(mozilla::Span<const char> aSource, nsAString& aDest)
{
if (MOZ_UNLIKELY(!AppendASCIItoUTF16(aSource, aDest, mozilla::fallible))) {
aDest.AllocFailed(aDest.Length() + aSource.Length());
}
}
// UTF-16 to UTF-8
// Unpaired surrogates are replaced with the REPLACEMENT CHARACTER.
inline MOZ_MUST_USE bool
CopyUTF16toUTF8(mozilla::Span<const char16_t> aSource,
nsACString& aDest,
const mozilla::fallible_t&)
{
return nscstring_fallible_append_utf16_to_utf8_impl(
&aDest, aSource.Elements(), aSource.Length(), 0);
}
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
inline void
CopyUTF16toUTF8(mozilla::Span<const char16_t> aSource, nsACString& aDest)
{
if (MOZ_UNLIKELY(!CopyUTF16toUTF8(aSource, aDest, mozilla::fallible))) {
aDest.AllocFailed(aSource.Length());
}
}
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
inline MOZ_MUST_USE bool
AppendUTF16toUTF8(mozilla::Span<const char16_t> aSource,
nsACString& aDest,
const mozilla::fallible_t&)
{
return nscstring_fallible_append_utf16_to_utf8_impl(
&aDest, aSource.Elements(), aSource.Length(), aDest.Length());
}
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
inline void
AppendUTF16toUTF8(mozilla::Span<const char16_t> aSource, nsACString& aDest)
{
if (MOZ_UNLIKELY(!AppendUTF16toUTF8(aSource, aDest, mozilla::fallible))) {
aDest.AllocFailed(aDest.Length() + aSource.Length());
}
}
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
// UTF-16 to Latin1
// If all code points in the input are below U+0100, represents each scalar
// value as an unsigned byte. (This is not windows-1252!) If there are code
// points above U+00FF, memory-safely produces garbage and will likely start
// asserting in future debug builds. The nature of the garbage may differ
// based on CPU architecture and must not be relied upon. The names say
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
// "ASCII" instead of "Latin1" for legacy reasons.
inline MOZ_MUST_USE bool
LossyCopyUTF16toASCII(mozilla::Span<const char16_t> aSource,
nsACString& aDest,
const mozilla::fallible_t&)
{
return nscstring_fallible_append_utf16_to_latin1_lossy_impl(
&aDest, aSource.Elements(), aSource.Length(), 0, true);
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
}
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
inline void
LossyCopyUTF16toASCII(mozilla::Span<const char16_t> aSource, nsACString& aDest)
{
if (MOZ_UNLIKELY(!LossyCopyUTF16toASCII(aSource, aDest, mozilla::fallible))) {
aDest.AllocFailed(aSource.Length());
}
}
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
inline MOZ_MUST_USE bool
LossyAppendUTF16toASCII(mozilla::Span<const char16_t> aSource,
nsACString& aDest,
const mozilla::fallible_t&)
{
return nscstring_fallible_append_utf16_to_latin1_lossy_impl(
&aDest, aSource.Elements(), aSource.Length(), aDest.Length(), false);
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
}
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
inline void
LossyAppendUTF16toASCII(mozilla::Span<const char16_t> aSource,
nsACString& aDest)
{
if (MOZ_UNLIKELY(
!LossyAppendUTF16toASCII(aSource, aDest, mozilla::fallible))) {
aDest.AllocFailed(aDest.Length() + aSource.Length());
}
}
/**
* Returns a new |char| buffer containing a zero-terminated copy of |aSource|.
*
* Allocates and returns a new |char| buffer which you must free with |free|.
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Performs a conversion with LossyConvertUTF16toLatin1() writing into the
* newly-allocated buffer.
*
* The new buffer is zero-terminated, but that may not help you if |aSource|
* contains embedded nulls.
*
* @param aSource a 16-bit wide string
* @return a new |char| buffer you must free with |free|.
*/
char* ToNewCString(const nsAString& aSource);
/**
* Returns a new |char| buffer containing a zero-terminated copy of |aSource|.
*
* Allocates and returns a new |char| buffer which you must free with |free|.
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
*
* The new buffer is zero-terminated, but that may not help you if |aSource|
* contains embedded nulls.
*
* @param aSource an 8-bit wide string
* @return a new |char| buffer you must free with |free|.
*/
char* ToNewCString(const nsACString& aSource);
/**
* Returns a new |char| buffer containing a zero-terminated copy of |aSource|.
*
* Allocates and returns a new |char| buffer which you must free with
* |free|.
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Performs an encoding conversion from a UTF-16 string to a UTF-8 string with
* unpaired surrogates replaced with the REPLACEMENT CHARACTER copying
* |aSource| to your new buffer.
*
* The new buffer is zero-terminated, but that may not help you if |aSource|
* contains embedded nulls.
*
* @param aSource a UTF-16 string (made of char16_t's)
* @param aUTF8Count the number of 8-bit units that was returned
* @return a new |char| buffer you must free with |free|.
*/
char* ToNewUTF8String(const nsAString& aSource, uint32_t* aUTF8Count = nullptr);
/**
* Returns a new |char16_t| buffer containing a zero-terminated copy of
* |aSource|.
*
* Allocates and returns a new |char16_t| buffer which you must free with
* |free|.
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
*
* The new buffer is zero-terminated, but that may not help you if |aSource|
* contains embedded nulls.
*
* @param aSource a UTF-16 string
* @return a new |char16_t| buffer you must free with |free|.
*/
char16_t* ToNewUnicode(const nsAString& aSource);
/**
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Returns a new |char16_t| buffer containing a zero-terminated copy of
* |aSource|.
*
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Allocates and returns a new |char16_t| buffer which you must free with
* |free|.
*
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Performs an encoding conversion by 0-padding 8-bit wide characters up to
* 16-bits wide (i.e. Latin1 to UTF-16 conversion) while copying |aSource|
* to your new buffer.
*
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* The new buffer is zero-terminated, but that may not help you if |aSource|
* contains embedded nulls.
*
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* @param aSource a Latin1 string
* @return a new |char16_t| buffer you must free with |free|.
*/
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
char16_t* ToNewUnicode(const nsACString& aSource);
/**
* Returns a new |char16_t| buffer containing a zero-terminated copy
* of |aSource|.
*
* Allocates and returns a new |char| buffer which you must free with
* |free|. Performs an encoding conversion from UTF-8 to UTF-16
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* while copying |aSource| to your new buffer. Malformed byte sequences
* are replaced with the REPLACEMENT CHARACTER.
*
* The new buffer is zero-terminated, but that may not help you if |aSource|
* contains embedded nulls.
*
* @param aSource an 8-bit wide string, UTF-8 encoded
* @param aUTF16Count the number of 16-bit units that was returned
* @return a new |char16_t| buffer you must free with |free|.
* (UTF-16 encoded)
*/
char16_t* UTF8ToNewUnicode(const nsACString& aSource,
uint32_t* aUTF16Count = nullptr);
/**
* Copies |aLength| 16-bit code units from the start of |aSource| to the
* |char16_t| buffer |aDest|.
*
* After this operation |aDest| is not null terminated.
*
* @param aSource a UTF-16 string
* @param aSrcOffset start offset in the source string
* @param aDest a |char16_t| buffer
* @param aLength the number of 16-bit code units to copy
* @return pointer to destination buffer - identical to |aDest|
*/
char16_t* CopyUnicodeTo(const nsAString& aSource,
uint32_t aSrcOffset,
char16_t* aDest,
uint32_t aLength);
/**
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Returns |true| if |aString| contains only ASCII characters, that is,
* characters in the range (0x00, 0x7F).
*
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* @param aString a 16-bit wide string to scan
*/
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
inline bool
IsASCII(mozilla::Span<const char16_t> aString)
{
size_t length = aString.Length();
const char16_t* ptr = aString.Elements();
// For short strings, calling into Rust is a pessimization, and the SIMD
// code won't have a chance to kick in anyway.
if (length < 16) {
char16_t accu = 0;
for (size_t i = 0; i < length; i++) {
accu |= ptr[i];
}
return accu < 0x80U;
}
return encoding_mem_is_basic_latin(ptr, length);
}
/**
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Returns |true| if |aString| contains only ASCII characters, that is,
* characters in the range (0x00, 0x7F).
*
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* @param aString a 8-bit wide string to scan
*/
inline bool
IsASCII(mozilla::Span<const char> aString)
{
size_t length = aString.Length();
const uint8_t* ptr = reinterpret_cast<const uint8_t*>(aString.Elements());
// For short strings, calling into Rust is a pessimization, and the SIMD
// code won't have a chance to kick in anyway.
if (length < 16) {
uint8_t accu = 0;
for (size_t i = 0; i < length; i++) {
accu |= ptr[i];
}
return accu < 0x80U;
}
return encoding_mem_is_ascii(ptr, length);
}
/**
* Returns |true| if |aString| contains only Latin1 characters, that is,
* characters in the range (U+0000, U+00FF).
*
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* @param aString a potentially-invalid UTF-16 string to scan
*/
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
inline bool
IsUTF16Latin1(mozilla::Span<const char16_t> aString)
{
size_t length = aString.Length();
const char16_t* ptr = aString.Elements();
// For short strings, calling into Rust is a pessimization, and the SIMD
// code won't have a chance to kick in anyway.
if (length < 16) {
char16_t accu = 0;
for (size_t i = 0; i < length; i++) {
accu |= ptr[i];
}
return accu < 0x100U;
}
return encoding_mem_is_utf16_latin1(ptr, length);
}
/**
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Returns |true| if |aString| contains only Latin1 characters, that is,
* characters in the range (U+0000, U+00FF).
*
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* If you know that the argument is always absolutely guaranteed to be valid
* UTF-8, use the faster UnsafeIsValidUTF8Latin1() instead.
*
* @param aString potentially-invalid UTF-8 string to scan
*/
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
inline bool
IsUTF8Latin1(mozilla::Span<const char> aString)
{
size_t length = aString.Length();
const uint8_t* ptr = reinterpret_cast<const uint8_t*>(aString.Elements());
// For short strings, calling into Rust is a pessimization, and the SIMD
// code won't have a chance to kick in anyway.
if (length < 16) {
for (size_t i = 0; i < length; i++) {
if (ptr[i] >= 0x80U) {
ptr += i;
length -= i;
// This loop can't handle non-ASCII, but the Rust code can, so
// upon seeing non-ASCII, break the loop and let the Rust code
// handle the rest of the buffer (including the non-ASCII byte).
goto end;
}
}
return true;
}
end:
return encoding_mem_is_utf8_latin1(ptr, length);
}
/**
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Returns |true| if |aString| contains only Latin1 characters, that is,
* characters in the range (U+0000, U+00FF).
*
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* The argument MUST be valid UTF-8. If you are at all unsure, use IsUTF8Latin1
* instead!
*
* @param aString known-valid UTF-8 string to scan
*/
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
inline bool
UnsafeIsValidUTF8Latin1(mozilla::Span<const char> aString)
{
size_t length = aString.Length();
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
const uint8_t* ptr = reinterpret_cast<const uint8_t*>(aString.Elements());
// For short strings, calling into Rust is a pessimization, and the SIMD
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
// code won't have a chance to kick in anyway.
if (length < 16) {
for (size_t i = 0; i < length; i++) {
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
if (ptr[i] >= 0x80U) {
ptr += i;
length -= i;
goto end;
}
}
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
return true;
}
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
end:
return encoding_mem_is_str_latin1(ptr, length);
}
/**
* Returns |true| if |aString| is a valid UTF-8 string.
*
* Note that this doesn't check whether the string might look like a valid
* string in another encoding, too, e.g. ISO-2022-JP.
*
* @param aString an 8-bit wide string to scan
*/
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
inline bool
IsUTF8(mozilla::Span<const char> aString)
{
size_t length = aString.Length();
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
const uint8_t* ptr = reinterpret_cast<const uint8_t*>(aString.Elements());
// For short strings, calling into Rust is a pessimization, and the SIMD
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
// code won't have a chance to kick in anyway.
if (length < 16) {
for (size_t i = 0; i < length; i++) {
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
if (ptr[i] >= 0x80U) {
ptr += i;
length -= i;
goto end;
}
}
return true;
}
end:
return length == encoding_utf8_valid_up_to(ptr, length);
}
/**
* Returns the index of the first unpaired surrogate or
* the length of the string if there are none.
*/
inline uint32_t
UTF16ValidUpTo(mozilla::Span<const char16_t> aString)
{
return encoding_mem_utf16_valid_up_to(aString.Elements(), aString.Length());
}
/**
* Replaces unpaired surrogates with U+FFFD in the argument.
*/
inline void
EnsureUTF16ValiditySpan(mozilla::Span<char16_t> aString)
{
encoding_mem_ensure_utf16_validity(aString.Elements(), aString.Length());
}
/**
* Replaces unpaired surrogates with U+FFFD in the argument.
*
* Copies a shared string buffer or an otherwise read-only
* buffer only if there are unpaired surrogates.
*/
inline void
EnsureUTF16Validity(nsAString& aString)
{
uint32_t upTo = UTF16ValidUpTo(aString);
uint32_t len = aString.Length();
if (upTo == len) {
return;
}
char16_t* ptr = aString.BeginWriting();
auto span = mozilla::MakeSpan(ptr, len);
span[upTo] = 0xFFFD;
EnsureUTF16ValiditySpan(span.From(upTo + 1));
}
bool ParseString(const nsACString& aAstring, char aDelimiter,
nsTArray<nsCString>& aArray);
/**
* Converts case in place in the argument string.
*/
void ToUpperCase(nsACString&);
void ToLowerCase(nsACString&);
void ToUpperCase(nsACString&);
void ToLowerCase(nsACString&);
/**
* Converts case from string aSource to aDest.
*/
void ToUpperCase(const nsACString& aSource, nsACString& aDest);
void ToLowerCase(const nsACString& aSource, nsACString& aDest);
/**
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Finds the leftmost occurrence of |aPattern|, if any in the range
* |aSearchStart|..|aSearchEnd|.
*
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Returns |true| if a match was found, and adjusts |aSearchStart| and
* |aSearchEnd| to point to the match. If no match was found, returns |false|
* and makes |aSearchStart == aSearchEnd|.
*
* Currently, this is equivalent to the O(m*n) implementation previously on
* |ns[C]String|.
*
* If we need something faster, then we can implement that later.
*/
bool FindInReadable(const nsAString& aPattern, nsAString::const_iterator&,
nsAString::const_iterator&,
const nsStringComparator& = nsDefaultStringComparator());
bool FindInReadable(const nsACString& aPattern, nsACString::const_iterator&,
nsACString::const_iterator&,
const nsCStringComparator& = nsDefaultCStringComparator());
/* sometimes we don't care about where the string was, just that we
* found it or not */
inline bool
FindInReadable(const nsAString& aPattern, const nsAString& aSource,
const nsStringComparator& aCompare = nsDefaultStringComparator())
{
nsAString::const_iterator start, end;
aSource.BeginReading(start);
aSource.EndReading(end);
return FindInReadable(aPattern, start, end, aCompare);
}
inline bool
FindInReadable(const nsACString& aPattern, const nsACString& aSource,
const nsCStringComparator& aCompare = nsDefaultCStringComparator())
{
nsACString::const_iterator start, end;
aSource.BeginReading(start);
aSource.EndReading(end);
return FindInReadable(aPattern, start, end, aCompare);
}
bool CaseInsensitiveFindInReadable(const nsACString& aPattern,
nsACString::const_iterator&,
nsACString::const_iterator&);
/**
* Finds the rightmost occurrence of |aPattern|
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Returns |true| if a match was found, and adjusts |aSearchStart| and
* |aSearchEnd| to point to the match. If no match was found, returns |false|
* and makes |aSearchStart == aSearchEnd|.
*/
bool RFindInReadable(const nsAString& aPattern, nsAString::const_iterator&,
nsAString::const_iterator&,
const nsStringComparator& = nsDefaultStringComparator());
bool RFindInReadable(const nsACString& aPattern, nsACString::const_iterator&,
nsACString::const_iterator&,
const nsCStringComparator& = nsDefaultCStringComparator());
/**
* Finds the leftmost occurrence of |aChar|, if any in the range
* |aSearchStart|..|aSearchEnd|.
*
* Returns |true| if a match was found, and adjusts |aSearchStart| to
* point to the match. If no match was found, returns |false| and
* makes |aSearchStart == aSearchEnd|.
*/
bool FindCharInReadable(char16_t aChar, nsAString::const_iterator& aSearchStart,
const nsAString::const_iterator& aSearchEnd);
bool FindCharInReadable(char aChar, nsACString::const_iterator& aSearchStart,
const nsACString::const_iterator& aSearchEnd);
bool StringBeginsWith(const nsAString& aSource, const nsAString& aSubstring);
bool StringBeginsWith(const nsAString& aSource, const nsAString& aSubstring,
const nsStringComparator& aComparator);
bool StringBeginsWith(const nsACString& aSource, const nsACString& aSubstring);
bool StringBeginsWith(const nsACString& aSource, const nsACString& aSubstring,
const nsCStringComparator& aComparator);
bool StringEndsWith(const nsAString& aSource, const nsAString& aSubstring);
bool StringEndsWith(const nsAString& aSource, const nsAString& aSubstring,
const nsStringComparator& aComparator);
bool StringEndsWith(const nsACString& aSource, const nsACString& aSubstring);
bool StringEndsWith(const nsACString& aSource, const nsACString& aSubstring,
const nsCStringComparator& aComparator);
const nsString& EmptyString();
const nsCString& EmptyCString();
const nsString& VoidString();
const nsCString& VoidCString();
/**
Bug 1402247 - Use encoding_rs for XPCOM string encoding conversions. r=Nika,erahm,froydnj. Correctness improvements: * UTF errors are handled safely per spec instead of dangerously truncating strings. * There are fewer converter implementations. Performance improvements: * The old code did exact buffer length math, which meant doing UTF math twice on each input string (once for length calculation and another time for conversion). Exact length math is more complicated when handling errors properly, which the old code didn't do. The new code does UTF math on the string content only once (when converting) but risks allocating more than once. There are heuristics in place to lower the probability of reallocation in cases where the double math avoidance isn't enough of a saving to absorb an allocation and memcpy. * Previously, in UTF-16 <-> UTF-8 conversions, an ASCII prefix was optimized but a single non-ASCII code point pessimized the rest of the string. The new code tries to get back on the fast ASCII path. * UTF-16 to Latin1 conversion guarantees less about handling of out-of-range input to eliminate an operation from the inner loop on x86/x86_64. * When assigning to a pre-existing string, the new code tries to reuse the old buffer instead of first releasing the old buffer and then allocating a new one. * When reallocating from the new code, the memcpy covers only the data that is part of the logical length of the old string instead of memcpying the whole capacity. (For old callers old excess memcpy behavior is preserved due to bogus callers. See bug 1472113.) * UTF-8 strings in XPConnect that are in the Latin1 range are passed to SpiderMonkey as Latin1. New features: * Conversion between UTF-8 and Latin1 is added in order to enable faster future interop between Rust code (or otherwise UTF-8-using code) and text node and SpiderMonkey code that uses Latin1. MozReview-Commit-ID: JaJuExfILM9
2018-07-06 10:44:43 +03:00
* Compare a UTF-8 string to an UTF-16 string.
*
* Returns 0 if the strings are equal, -1 if aUTF8String is less
* than aUTF16Count, and 1 in the reverse case. Errors are replaced
* with U+FFFD and then the U+FFFD is compared as if it had occurred
* in the input. If aErr is not nullptr, *aErr is set to true if
* either string had malformed sequences.
*/
int32_t
CompareUTF8toUTF16(const nsACString& aUTF8String,
const nsAString& aUTF16String,
bool* aErr = nullptr);
void AppendUCS4ToUTF16(const uint32_t aSource, nsAString& aDest);
#endif // !defined(nsReadableUtils_h___)