gvsafronov/rapidjson
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #907 from almavi/master

Browse Source 
Fixed bug on space hexadecimal encoding
This commit is contained in:
Milo Yip 2017-03-25 11:10:48 +08:00 committed by GitHub
commit b8f0414b9a
4 changed files with 14 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
include/rapidjson/reader.h
View File

@ -955,7 +955,7 @@ private:
if (c == '\0') if (c == '\0')
RAPIDJSON_PARSE_ERROR(kParseErrorStringMissQuotationMark, is.Tell()); RAPIDJSON_PARSE_ERROR(kParseErrorStringMissQuotationMark, is.Tell());
else else
RAPIDJSON_PARSE_ERROR(kParseErrorStringEscapeInvalid, is.Tell()); RAPIDJSON_PARSE_ERROR(kParseErrorStringInvalidEncoding, is.Tell());
} }
else { else {
size_t offset = is.Tell(); size_t offset = is.Tell();
@ -990,7 +990,7 @@ private:
// The rest of string using SIMD // The rest of string using SIMD
static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' };
static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' };
static const char space[16] = { 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19 }; static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F };
const __m128i dq = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&dquote[0])); const __m128i dq = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&dquote[0]));
const __m128i bs = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&bslash[0])); const __m128i bs = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&bslash[0]));
const __m128i sp = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&space[0])); const __m128i sp = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&space[0]));
@ -999,7 +999,7 @@ private:
const __m128i s = _mm_load_si128(reinterpret_cast<const __m128i *>(p)); const __m128i s = _mm_load_si128(reinterpret_cast<const __m128i *>(p));
const __m128i t1 = _mm_cmpeq_epi8(s, dq); const __m128i t1 = _mm_cmpeq_epi8(s, dq);
const __m128i t2 = _mm_cmpeq_epi8(s, bs); const __m128i t2 = _mm_cmpeq_epi8(s, bs);
const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x19) == 0x19 const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F
const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3);
unsigned short r = static_cast<unsigned short>(_mm_movemask_epi8(x)); unsigned short r = static_cast<unsigned short>(_mm_movemask_epi8(x));
if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped
@ -1053,7 +1053,7 @@ private:
// The rest of string using SIMD // The rest of string using SIMD
static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' };
static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' };
static const char space[16] = { 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19 }; static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F };
const __m128i dq = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&dquote[0])); const __m128i dq = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&dquote[0]));
const __m128i bs = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&bslash[0])); const __m128i bs = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&bslash[0]));
const __m128i sp = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&space[0])); const __m128i sp = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&space[0]));
@ -1062,7 +1062,7 @@ private:
const __m128i s = _mm_load_si128(reinterpret_cast<const __m128i *>(p)); const __m128i s = _mm_load_si128(reinterpret_cast<const __m128i *>(p));
const __m128i t1 = _mm_cmpeq_epi8(s, dq); const __m128i t1 = _mm_cmpeq_epi8(s, dq);
const __m128i t2 = _mm_cmpeq_epi8(s, bs); const __m128i t2 = _mm_cmpeq_epi8(s, bs);
const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x19) == 0x19 const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F
const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3);
unsigned short r = static_cast<unsigned short>(_mm_movemask_epi8(x)); unsigned short r = static_cast<unsigned short>(_mm_movemask_epi8(x));
if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped
@ -1101,7 +1101,7 @@ private:
// The rest of string using SIMD // The rest of string using SIMD
static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' };
static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' };
static const char space[16] = { 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19 }; static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F };
const __m128i dq = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&dquote[0])); const __m128i dq = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&dquote[0]));
const __m128i bs = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&bslash[0])); const __m128i bs = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&bslash[0]));
const __m128i sp = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&space[0])); const __m128i sp = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&space[0]));
@ -1110,7 +1110,7 @@ private:
const __m128i s = _mm_load_si128(reinterpret_cast<const __m128i *>(p)); const __m128i s = _mm_load_si128(reinterpret_cast<const __m128i *>(p));
const __m128i t1 = _mm_cmpeq_epi8(s, dq); const __m128i t1 = _mm_cmpeq_epi8(s, dq);
const __m128i t2 = _mm_cmpeq_epi8(s, bs); const __m128i t2 = _mm_cmpeq_epi8(s, bs);
const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x19) == 0x19 const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F
const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3);
unsigned short r = static_cast<unsigned short>(_mm_movemask_epi8(x)); unsigned short r = static_cast<unsigned short>(_mm_movemask_epi8(x));
if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped

4
include/rapidjson/writer.h
View File

@ -585,7 +585,7 @@ inline bool Writer<StringBuffer>::ScanWriteUnescapedString(StringStream& is, siz
// The rest of string using SIMD // The rest of string using SIMD
static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' };
static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' };
static const char space[16] = { 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19 }; static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F };
const __m128i dq = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&dquote[0])); const __m128i dq = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&dquote[0]));
const __m128i bs = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&bslash[0])); const __m128i bs = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&bslash[0]));
const __m128i sp = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&space[0])); const __m128i sp = _mm_loadu_si128(reinterpret_cast<const __m128i *>(&space[0]));
@ -594,7 +594,7 @@ inline bool Writer<StringBuffer>::ScanWriteUnescapedString(StringStream& is, siz
const __m128i s = _mm_load_si128(reinterpret_cast<const __m128i *>(p)); const __m128i s = _mm_load_si128(reinterpret_cast<const __m128i *>(p));
const __m128i t1 = _mm_cmpeq_epi8(s, dq); const __m128i t1 = _mm_cmpeq_epi8(s, dq);
const __m128i t2 = _mm_cmpeq_epi8(s, bs); const __m128i t2 = _mm_cmpeq_epi8(s, bs);
const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x19) == 0x19 const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F
const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3);
unsigned short r = static_cast<unsigned short>(_mm_movemask_epi8(x)); unsigned short r = static_cast<unsigned short>(_mm_movemask_epi8(x));
if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped

2
test/unittest/readertest.cpp
View File

@ -725,6 +725,8 @@ TEST(Reader, ParseString_Error) {
// Malform ASCII sequence // Malform ASCII sequence
TEST_STRINGENCODING_ERROR(ASCII<>, UTF8<>, char, ARRAY('[', '\"', char(0x80u), '\"', ']', '\0')); TEST_STRINGENCODING_ERROR(ASCII<>, UTF8<>, char, ARRAY('[', '\"', char(0x80u), '\"', ']', '\0'));
TEST_STRINGENCODING_ERROR(ASCII<>, UTF8<>, char, ARRAY('[', '\"', char(0x01u), '\"', ']', '\0'));
TEST_STRINGENCODING_ERROR(ASCII<>, UTF8<>, char, ARRAY('[', '\"', char(0x1Cu), '\"', ']', '\0'));
#undef ARRAY #undef ARRAY
#undef TEST_STRINGARRAY_ERROR #undef TEST_STRINGARRAY_ERROR

4
test/unittest/writertest.cpp
View File

@ -412,8 +412,10 @@ TEST(Writer, ValidateEncoding) {
EXPECT_TRUE(writer.String("\xC2\xA2")); // Cents sign U+00A2 EXPECT_TRUE(writer.String("\xC2\xA2")); // Cents sign U+00A2
EXPECT_TRUE(writer.String("\xE2\x82\xAC")); // Euro sign U+20AC EXPECT_TRUE(writer.String("\xE2\x82\xAC")); // Euro sign U+20AC
EXPECT_TRUE(writer.String("\xF0\x9D\x84\x9E")); // G clef sign U+1D11E EXPECT_TRUE(writer.String("\xF0\x9D\x84\x9E")); // G clef sign U+1D11E
EXPECT_TRUE(writer.String("\x01")); // SOH control U+0001
EXPECT_TRUE(writer.String("\x1B")); // Escape control U+001B
writer.EndArray(); writer.EndArray();
EXPECT_STREQ("[\"\x24\",\"\xC2\xA2\",\"\xE2\x82\xAC\",\"\xF0\x9D\x84\x9E\"]", buffer.GetString()); EXPECT_STREQ("[\"\x24\",\"\xC2\xA2\",\"\xE2\x82\xAC\",\"\xF0\x9D\x84\x9E\",\"\\u0001\",\"\\u001B\"]", buffer.GetString());
} }
// Fail in decoding invalid UTF-8 sequence http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt // Fail in decoding invalid UTF-8 sequence http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt