1 /*
2  * Copyright (C) 2014 The Android Open Source Project
3  *
4  * Licensed under the Apache License, Version 2.0 (the "License");
5  * you may not use this file except in compliance with the License.
6  * You may obtain a copy of the License at
7  *
8  *      http://www.apache.org/licenses/LICENSE-2.0
9  *
10  * Unless required by applicable law or agreed to in writing, software
11  * distributed under the License is distributed on an "AS IS" BASIS,
12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13  * See the License for the specific language governing permissions and
14  * limitations under the License.
15  */
16 
17 #include "utils/utf8_utils.h"
18 
19 #include "utils/char_utils.h"
20 
21 namespace latinime {
22 namespace dicttoolkit {
23 
24 const size_t Utf8Utils::MAX_SEQUENCE_SIZE_FOR_A_CODE_POINT = 4;
25 const uint8_t Utf8Utils::FIRST_BYTE_MARKER_MASKS[] = {0, 0x80, 0xE0, 0xF0, 0xF8};
26 const uint8_t Utf8Utils::FIRST_BYTE_MARKERS[] = {0, 0x00, 0xC0, 0xE0, 0xF0};
27 const uint8_t Utf8Utils::FIRST_BYTE_CODE_POINT_BITS_MASKS[] = {0, 0x7F, 0x1F, 0x0F, 0x03};
28 const int Utf8Utils::MAX_ENCODED_CODE_POINT_VALUES[] = {-1, 0x7F, 0x7FF, 0xFFFF, 0x10FFFF};
29 
30 const uint8_t Utf8Utils::TRAILING_BYTE_CODE_POINT_BITS_MASK = 0x3F;
31 const uint8_t Utf8Utils::TRAILING_BYTE_MARKER = 0x80;
32 const size_t Utf8Utils::CODE_POINT_BIT_COUNT_IN_TRAILING_BYTE = 6;
33 
getCodePoints(const std::string & utf8Str)34 /* static */ std::vector<int> Utf8Utils::getCodePoints(const std::string &utf8Str) {
35     std::vector<int> codePoints;
36     int remainingByteCountForCurrentCodePoint = 0;
37     int currentCodePointSequenceSize = 0;
38     int codePoint = 0;
39     for (const char c : utf8Str) {
40         if (remainingByteCountForCurrentCodePoint == 0) {
41             currentCodePointSequenceSize = getSequenceSizeByCheckingFirstByte(c);
42             if (currentCodePointSequenceSize <= 0) {
43                 AKLOGE("%x is an invalid utf8 first byte value.", c);
44                 return std::vector<int>();
45             }
46             remainingByteCountForCurrentCodePoint = currentCodePointSequenceSize;
47             codePoint = maskFirstByte(c, remainingByteCountForCurrentCodePoint);
48         } else {
49             codePoint <<= CODE_POINT_BIT_COUNT_IN_TRAILING_BYTE;
50             codePoint += maskTrailingByte(c);
51         }
52         remainingByteCountForCurrentCodePoint--;
53         if (remainingByteCountForCurrentCodePoint == 0) {
54             if (codePoint <= MAX_ENCODED_CODE_POINT_VALUES[currentCodePointSequenceSize - 1]) {
55                 AKLOGE("%d bytes encode for codePoint(%x) is a redundant UTF-8 sequence.",
56                         currentCodePointSequenceSize,  codePoint);
57                 return std::vector<int>();
58             }
59             codePoints.push_back(codePoint);
60         }
61     }
62     return codePoints;
63 }
64 
getSequenceSizeByCheckingFirstByte(const uint8_t firstByte)65 /* static */ int Utf8Utils::getSequenceSizeByCheckingFirstByte(const uint8_t firstByte) {
66     for (size_t i = 1; i <= MAX_SEQUENCE_SIZE_FOR_A_CODE_POINT; ++i) {
67         if ((firstByte & FIRST_BYTE_MARKER_MASKS[i]) == FIRST_BYTE_MARKERS[i]) {
68             return i;
69         }
70     }
71     // Not a valid utf8 char first byte.
72     return -1;
73 }
74 
maskFirstByte(const uint8_t firstByte,const int sequenceSize)75 /* static */ AK_FORCE_INLINE int Utf8Utils::maskFirstByte(const uint8_t firstByte,
76         const int sequenceSize) {
77     return firstByte & FIRST_BYTE_CODE_POINT_BITS_MASKS[sequenceSize];
78 }
79 
maskTrailingByte(const uint8_t secondOrLaterByte)80 /* static */ AK_FORCE_INLINE int Utf8Utils::maskTrailingByte(const uint8_t secondOrLaterByte) {
81     return secondOrLaterByte & TRAILING_BYTE_CODE_POINT_BITS_MASK;
82 }
83 
getUtf8String(const CodePointArrayView codePoints)84 /* static */ std::string Utf8Utils::getUtf8String(const CodePointArrayView codePoints) {
85     std::string utf8String;
86     for (const int codePoint : codePoints) {
87         const int sequenceSize = getSequenceSizeToEncodeCodePoint(codePoint);
88         if (sequenceSize <= 0) {
89             AKLOGE("Cannot encode code point (%d).", codePoint);
90             return std::string();
91         }
92         const int trailingByteCount = sequenceSize - 1;
93         // Output first byte.
94         const int value = codePoint >> (trailingByteCount * CODE_POINT_BIT_COUNT_IN_TRAILING_BYTE);
95         utf8String.push_back(static_cast<char>(value | FIRST_BYTE_MARKERS[sequenceSize]));
96         // Output second and later bytes.
97         for (int i = 1; i < sequenceSize; ++i) {
98             const int shiftAmount = (trailingByteCount - i) * CODE_POINT_BIT_COUNT_IN_TRAILING_BYTE;
99             const int value = (codePoint >> shiftAmount) & TRAILING_BYTE_CODE_POINT_BITS_MASK;
100             utf8String.push_back(static_cast<char>(value | TRAILING_BYTE_MARKER));
101         }
102     }
103     return utf8String;
104 }
105 
getSequenceSizeToEncodeCodePoint(const int codePoint)106 /* static */ int Utf8Utils::getSequenceSizeToEncodeCodePoint(const int codePoint) {
107     if (codePoint < 0) {
108         return -1;
109     }
110     for (size_t i = 1; i <= MAX_SEQUENCE_SIZE_FOR_A_CODE_POINT; ++i) {
111         if (codePoint <= MAX_ENCODED_CODE_POINT_VALUES[i]) {
112             return i;
113         }
114     }
115     return -1;
116 }
117 
118 } // namespace dicttoolkit
119 } // namespace latinime
120