1 /*
2 * Copyright 2022 Google LLC
3 *
4 * Use of this source code is governed by a BSD-style license that can be
5 * found in the LICENSE file.
6 */
7 #include "modules/skunicode/src/SkUnicode_icu_bidi.h"
8
9 #include "include/core/SkTypes.h"
10 #include "include/private/base/SkDebug.h"
11 #include "include/private/base/SkTFitsIn.h"
12 #include "src/base/SkUTF.h"
13
14 #include <unicode/ubidi.h>
15 #include <unicode/umachine.h>
16 #include <unicode/utypes.h>
17
18 #include <cstdint>
19 #include <memory>
20 #include <string>
21 #include <utility>
22 #include <vector>
23
24 namespace {
25 using SkUnicodeBidi = std::unique_ptr<UBiDi, SkBidiFactory::BidiCloseCallback>;
26
27 class SkBidiIterator_icu : public SkBidiIterator {
28 public:
SkBidiIterator_icu(SkUnicodeBidi bidi,sk_sp<SkBidiFactory> fact)29 SkBidiIterator_icu(SkUnicodeBidi bidi, sk_sp<SkBidiFactory> fact)
30 : fBidi(std::move(bidi)), fBidiFact(std::move(fact)) {}
31
getLength()32 Position getLength() override { return fBidiFact->bidi_getLength(fBidi.get()); }
33
getLevelAt(Position pos)34 Level getLevelAt(Position pos) override { return fBidiFact->bidi_getLevelAt(fBidi.get(), pos); }
35
36 private:
37 SkUnicodeBidi fBidi;
38 sk_sp<SkBidiFactory> fBidiFact;
39 };
40 } // namespace
41
42 #ifdef ENABLE_DRAWING_ADAPTER
43 namespace SkiaRsText {
44 #endif
MakeIterator(const uint16_t utf16[],int utf16Units,SkBidiIterator::Direction dir) const45 std::unique_ptr<SkBidiIterator> SkBidiFactory::MakeIterator(const uint16_t utf16[],
46 int utf16Units,
47 SkBidiIterator::Direction dir) const {
48 UErrorCode status = U_ZERO_ERROR;
49 SkUnicodeBidi bidi(this->bidi_openSized(utf16Units, 0, &status), this->bidi_close_callback());
50 if (U_FAILURE(status)) {
51 SkDEBUGF("Bidi error: %s", this->errorName(status));
52 return nullptr;
53 }
54 SkASSERT(bidi);
55 uint8_t bidiLevel = (dir == SkBidiIterator::kLTR) ? UBIDI_LTR : UBIDI_RTL;
56 // The required lifetime of utf16 isn't well documented.
57 // It appears it isn't used after ubidi_setPara except through ubidi_getText.
58 this->bidi_setPara(bidi.get(), (const UChar*)utf16, utf16Units, bidiLevel, nullptr, &status);
59 if (U_FAILURE(status)) {
60 SkDEBUGF("Bidi error: %s", this->errorName(status));
61 return nullptr;
62 }
63 return std::unique_ptr<SkBidiIterator>(
64 new SkBidiIterator_icu(std::move(bidi), sk_ref_sp(this)));
65 }
66
MakeIterator(const char utf8[],int utf8Units,SkBidiIterator::Direction dir) const67 std::unique_ptr<SkBidiIterator> SkBidiFactory::MakeIterator(const char utf8[],
68 int utf8Units,
69 SkBidiIterator::Direction dir) const {
70 // Convert utf8 into utf16 since ubidi only accepts utf16
71 if (!SkTFitsIn<int32_t>(utf8Units)) {
72 SkDEBUGF("Bidi error: text too long");
73 return nullptr;
74 }
75
76 // Getting the length like this seems to always set U_BUFFER_OVERFLOW_ERROR
77 int utf16Units = SkUTF::UTF8ToUTF16(nullptr, 0, utf8, utf8Units);
78 if (utf16Units < 0) {
79 SkDEBUGF("Bidi error: Invalid utf8 input");
80 return nullptr;
81 }
82 std::unique_ptr<uint16_t[]> utf16(new uint16_t[utf16Units]);
83 SkDEBUGCODE(int dstLen =) SkUTF::UTF8ToUTF16(utf16.get(), utf16Units, utf8, utf8Units);
84 SkASSERT(dstLen == utf16Units);
85
86 return MakeIterator(utf16.get(), utf16Units, dir);
87 }
88
89 /** Replaces invalid utf-8 sequences with REPLACEMENT CHARACTER U+FFFD. */
utf8_next(const char ** ptr,const char * end)90 static inline SkUnichar utf8_next(const char** ptr, const char* end) {
91 SkUnichar val = SkUTF::NextUTF8(ptr, end);
92 return val < 0 ? 0xFFFD : val;
93 }
94
ExtractBidi(const char utf8[],int utf8Units,SkUnicode::TextDirection dir,std::vector<SkUnicode::BidiRegion> * bidiRegions) const95 bool SkBidiFactory::ExtractBidi(const char utf8[],
96 int utf8Units,
97 SkUnicode::TextDirection dir,
98 std::vector<SkUnicode::BidiRegion>* bidiRegions) const {
99 // Convert to UTF16 since for now bidi iterator only operates on utf16
100 auto utf16 = SkUnicode::convertUtf8ToUtf16(utf8, utf8Units);
101
102 // Create bidi iterator
103 UErrorCode status = U_ZERO_ERROR;
104 SkUnicodeBidi bidi(this->bidi_openSized(utf16.size(), 0, &status), this->bidi_close_callback());
105 if (U_FAILURE(status)) {
106 SkDEBUGF("Bidi error: %s", this->errorName(status));
107 return false;
108 }
109 SkASSERT(bidi);
110 uint8_t bidiLevel = (dir == SkUnicode::TextDirection::kLTR) ? UBIDI_LTR : UBIDI_RTL;
111 // The required lifetime of utf16 isn't well documented.
112 // It appears it isn't used after ubidi_setPara except through ubidi_getText.
113 this->bidi_setPara(
114 bidi.get(), (const UChar*)utf16.c_str(), utf16.size(), bidiLevel, nullptr, &status);
115 if (U_FAILURE(status)) {
116 SkDEBUGF("Bidi error: %s", this->errorName(status));
117 return false;
118 }
119
120 // Iterate through bidi regions and the result positions into utf8
121 const char* start8 = utf8;
122 const char* end8 = utf8 + utf8Units;
123 SkUnicode::BidiLevel currentLevel = 0;
124
125 SkUnicode::Position pos8 = 0;
126 SkUnicode::Position pos16 = 0;
127 SkUnicode::Position end16 = this->bidi_getLength(bidi.get());
128
129 if (end16 == 0) {
130 return true;
131 }
132 if (this->bidi_getDirection(bidi.get()) != UBIDI_MIXED) {
133 // The entire paragraph is unidirectional.
134 bidiRegions->emplace_back(0, utf8Units, this->bidi_getLevelAt(bidi.get(), 0));
135 return true;
136 }
137
138 while (pos16 < end16) {
139 auto level = this->bidi_getLevelAt(bidi.get(), pos16);
140 if (pos16 == 0) {
141 currentLevel = level;
142 } else if (level != currentLevel) {
143 SkUnicode::Position end = start8 - utf8;
144 bidiRegions->emplace_back(pos8, end, currentLevel);
145 currentLevel = level;
146 pos8 = end;
147 }
148 SkUnichar u = utf8_next(&start8, end8);
149 pos16 += SkUTF::ToUTF16(u);
150 }
151 SkUnicode::Position end = start8 - utf8;
152 if (end != pos8) {
153 bidiRegions->emplace_back(pos8, end, currentLevel);
154 }
155 return true;
156 }
157 #ifdef ENABLE_DRAWING_ADAPTER
158 }
159 #endif // ENABLE_DRAWING_ADAPTER