1 | //===-- runtime/emit-encoded.h ----------------------------------*- C++ -*-===// |
2 | // |
3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
4 | // See https://llvm.org/LICENSE.txt for license information. |
5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
6 | // |
7 | //===----------------------------------------------------------------------===// |
8 | |
9 | // Templates for emitting CHARACTER values with conversion |
10 | |
11 | #ifndef FORTRAN_RUNTIME_EMIT_ENCODED_H_ |
12 | #define FORTRAN_RUNTIME_EMIT_ENCODED_H_ |
13 | |
14 | #include "connection.h" |
15 | #include "environment.h" |
16 | #include "tools.h" |
17 | #include "utf.h" |
18 | |
19 | namespace Fortran::runtime::io { |
20 | |
21 | template <typename CONTEXT, typename CHAR, bool NL_ADVANCES_RECORD = true> |
22 | RT_API_ATTRS bool EmitEncoded( |
23 | CONTEXT &to, const CHAR *data, std::size_t chars) { |
24 | ConnectionState &connection{to.GetConnectionState()}; |
25 | if constexpr (NL_ADVANCES_RECORD) { |
26 | if (connection.access == Access::Stream && |
27 | connection.internalIoCharKind == 0) { |
28 | // Stream output: treat newlines as record advancements so that the left |
29 | // tab limit is correctly managed |
30 | while (const CHAR * nl{FindCharacter(data, CHAR{'\n'}, chars)}) { |
31 | auto pos{static_cast<std::size_t>(nl - data)}; |
32 | // The [data, data + pos) does not contain the newline, |
33 | // so we can avoid the recursion by calling proper specialization. |
34 | if (!EmitEncoded<CONTEXT, CHAR, false>(to, data, pos)) { |
35 | return false; |
36 | } |
37 | data += pos + 1; |
38 | chars -= pos + 1; |
39 | to.AdvanceRecord(); |
40 | } |
41 | } |
42 | } |
43 | if (connection.useUTF8<CHAR>()) { |
44 | using UnsignedChar = std::make_unsigned_t<CHAR>; |
45 | const UnsignedChar *uData{reinterpret_cast<const UnsignedChar *>(data)}; |
46 | char buffer[256]; |
47 | std::size_t at{0}; |
48 | while (chars-- > 0) { |
49 | auto len{EncodeUTF8(buffer + at, *uData++)}; |
50 | at += len; |
51 | if (at + maxUTF8Bytes > sizeof buffer) { |
52 | if (!to.Emit(buffer, at)) { |
53 | return false; |
54 | } |
55 | at = 0; |
56 | } |
57 | } |
58 | return at == 0 || to.Emit(buffer, at); |
59 | } else { |
60 | std::size_t internalKind = connection.internalIoCharKind; |
61 | if (internalKind == 0 || internalKind == sizeof(CHAR)) { |
62 | const char *rawData{reinterpret_cast<const char *>(data)}; |
63 | return to.Emit(rawData, chars * sizeof(CHAR), sizeof(CHAR)); |
64 | } else { |
65 | // CHARACTER kind conversion for internal output |
66 | while (chars-- > 0) { |
67 | char32_t buffer = *data++; |
68 | char *p{reinterpret_cast<char *>(&buffer)}; |
69 | if constexpr (!isHostLittleEndian) { |
70 | p += sizeof(buffer) - internalKind; |
71 | } |
72 | if (!to.Emit(p, internalKind)) { |
73 | return false; |
74 | } |
75 | } |
76 | return true; |
77 | } |
78 | } |
79 | } |
80 | |
81 | template <typename CONTEXT> |
82 | RT_API_ATTRS bool EmitAscii(CONTEXT &to, const char *data, std::size_t chars) { |
83 | ConnectionState &connection{to.GetConnectionState()}; |
84 | if (connection.internalIoCharKind <= 1 && |
85 | connection.access != Access::Stream) { |
86 | return to.Emit(data, chars); |
87 | } else { |
88 | return EmitEncoded(to, data, chars); |
89 | } |
90 | } |
91 | |
92 | template <typename CONTEXT> |
93 | RT_API_ATTRS bool EmitRepeated(CONTEXT &to, char ch, std::size_t n) { |
94 | if (n <= 0) { |
95 | return true; |
96 | } |
97 | ConnectionState &connection{to.GetConnectionState()}; |
98 | if (connection.internalIoCharKind <= 1 && |
99 | connection.access != Access::Stream) { |
100 | // faster path, no encoding needed |
101 | while (n-- > 0) { |
102 | if (!to.Emit(&ch, 1)) { |
103 | return false; |
104 | } |
105 | } |
106 | } else { |
107 | while (n-- > 0) { |
108 | if (!EmitEncoded(to, &ch, 1)) { |
109 | return false; |
110 | } |
111 | } |
112 | } |
113 | return true; |
114 | } |
115 | |
116 | } // namespace Fortran::runtime::io |
117 | #endif // FORTRAN_RUNTIME_EMIT_ENCODED_H_ |
118 | |