| 1 | // Copyright (C) 2018 Intel Corporation. |
| 2 | // SPDX-License-Identifier: LicenseRef-Qt-Commercial OR LGPL-3.0-only OR GPL-2.0-only OR GPL-3.0-only |
| 3 | |
| 4 | #include "qcborvalue.h" |
| 5 | #include "qcborvalue_p.h" |
| 6 | |
| 7 | #include "qcborarray.h" |
| 8 | #include "qcbormap.h" |
| 9 | |
| 10 | #include <private/qnumeric_p.h> |
| 11 | #include <qstack.h> |
| 12 | #include <private/qtools_p.h> |
| 13 | |
| 14 | QT_BEGIN_NAMESPACE |
| 15 | |
| 16 | using namespace Qt::StringLiterals; |
| 17 | |
| 18 | namespace { |
| 19 | class DiagnosticNotation |
| 20 | { |
| 21 | public: |
| 22 | static QString create(const QCborValue &v, QCborValue::DiagnosticNotationOptions opts) |
| 23 | { |
| 24 | DiagnosticNotation dn(opts); |
| 25 | dn.appendValue(v); |
| 26 | return dn.result; |
| 27 | } |
| 28 | |
| 29 | private: |
| 30 | QStack<int> byteArrayFormatStack; |
| 31 | QString separator; |
| 32 | QString result; |
| 33 | QCborValue::DiagnosticNotationOptions opts; |
| 34 | int nestingLevel = 0; |
| 35 | |
| 36 | struct Nest { |
| 37 | enum { IndentationWidth = 4 }; |
| 38 | DiagnosticNotation *dn; |
| 39 | Nest(DiagnosticNotation *that) : dn(that) |
| 40 | { |
| 41 | ++dn->nestingLevel; |
| 42 | static const char indent[IndentationWidth + 1] = " " ; |
| 43 | if (dn->opts & QCborValue::LineWrapped) |
| 44 | dn->separator += QLatin1StringView(indent, IndentationWidth); |
| 45 | } |
| 46 | ~Nest() |
| 47 | { |
| 48 | --dn->nestingLevel; |
| 49 | if (dn->opts & QCborValue::LineWrapped) |
| 50 | dn->separator.chop(n: IndentationWidth); |
| 51 | } |
| 52 | }; |
| 53 | |
| 54 | DiagnosticNotation(QCborValue::DiagnosticNotationOptions opts_) |
| 55 | : separator(opts_ & QCborValue::LineWrapped ? "\n"_L1 : ""_L1 ), opts(opts_) |
| 56 | { |
| 57 | byteArrayFormatStack.push(t: int(QCborKnownTags::ExpectedBase16)); |
| 58 | } |
| 59 | |
| 60 | void appendString(const QString &s); |
| 61 | void appendArray(const QCborArray &a); |
| 62 | void appendMap(const QCborMap &m); |
| 63 | void appendValue(const QCborValue &v); |
| 64 | }; |
| 65 | } |
| 66 | |
| 67 | static QString makeFpString(double d) |
| 68 | { |
| 69 | QString s; |
| 70 | quint64 v; |
| 71 | if (qt_is_inf(d)) { |
| 72 | s = (d < 0) ? QStringLiteral("-inf" ) : QStringLiteral("inf" ); |
| 73 | } else if (qt_is_nan(d)) { |
| 74 | s = QStringLiteral("nan" ); |
| 75 | } else if (convertDoubleTo(v: d, value: &v)) { |
| 76 | s = QString::fromLatin1(ba: "%1.0" ).arg(a: v); |
| 77 | if (d < 0) |
| 78 | s.prepend(c: u'-'); |
| 79 | } else { |
| 80 | s = QString::number(d, format: 'g', precision: QLocale::FloatingPointShortest); |
| 81 | if (!s.contains(c: u'.') && !s.contains(c: u'e')) |
| 82 | s += u'.'; |
| 83 | } |
| 84 | return s; |
| 85 | } |
| 86 | |
| 87 | static bool isByteArrayEncodingTag(QCborTag tag) |
| 88 | { |
| 89 | switch (quint64(tag)) { |
| 90 | case quint64(QCborKnownTags::ExpectedBase16): |
| 91 | case quint64(QCborKnownTags::ExpectedBase64): |
| 92 | case quint64(QCborKnownTags::ExpectedBase64url): |
| 93 | return true; |
| 94 | } |
| 95 | return false; |
| 96 | } |
| 97 | |
| 98 | void DiagnosticNotation::appendString(const QString &s) |
| 99 | { |
| 100 | result += u'"'; |
| 101 | |
| 102 | const QChar *begin = s.begin(); |
| 103 | const QChar *end = s.end(); |
| 104 | while (begin < end) { |
| 105 | // find the longest span comprising only non-escaped characters |
| 106 | const QChar *ptr = begin; |
| 107 | for ( ; ptr < end; ++ptr) { |
| 108 | ushort uc = ptr->unicode(); |
| 109 | if (uc == '\\' || uc == '"' || uc < ' ' || uc >= 0x7f) |
| 110 | break; |
| 111 | } |
| 112 | |
| 113 | if (ptr != begin) |
| 114 | result.append(uc: begin, len: ptr - begin); |
| 115 | |
| 116 | if (ptr == end) |
| 117 | break; |
| 118 | |
| 119 | // there's an escaped character |
| 120 | static const char escapeMap[16] = { |
| 121 | // The C escape characters \a \b \t \n \v \f and \r indexed by |
| 122 | // their ASCII values |
| 123 | 0, 0, 0, 0, |
| 124 | 0, 0, 0, 'a', |
| 125 | 'b', 't', 'n', 'v', |
| 126 | 'f', 'r', 0, 0 |
| 127 | }; |
| 128 | int buflen = 2; |
| 129 | QChar buf[10]; |
| 130 | buf[0] = u'\\'; |
| 131 | buf[1] = QChar::Null; |
| 132 | char16_t uc = ptr->unicode(); |
| 133 | |
| 134 | if (uc < sizeof(escapeMap)) |
| 135 | buf[1] = QLatin1Char(escapeMap[uc]); |
| 136 | else if (uc == '"' || uc == '\\') |
| 137 | buf[1] = QChar(uc); |
| 138 | |
| 139 | if (buf[1] == QChar::Null) { |
| 140 | const auto toHexUpper = [](char32_t value) -> QChar { |
| 141 | // QtMiscUtils::toHexUpper() returns char, we need QChar, so wrap |
| 142 | return char16_t(QtMiscUtils::toHexUpper(value)); |
| 143 | }; |
| 144 | if (ptr->isHighSurrogate() && (ptr + 1) != end && ptr[1].isLowSurrogate()) { |
| 145 | // properly-paired surrogates |
| 146 | ++ptr; |
| 147 | char32_t ucs4 = QChar::surrogateToUcs4(high: uc, low: ptr->unicode()); |
| 148 | buf[1] = u'U'; |
| 149 | buf[2] = u'0'; // toHexUpper(ucs4 >> 28); |
| 150 | buf[3] = u'0'; // toHexUpper(ucs4 >> 24); |
| 151 | buf[4] = toHexUpper(ucs4 >> 20); |
| 152 | buf[5] = toHexUpper(ucs4 >> 16); |
| 153 | buf[6] = toHexUpper(ucs4 >> 12); |
| 154 | buf[7] = toHexUpper(ucs4 >> 8); |
| 155 | buf[8] = toHexUpper(ucs4 >> 4); |
| 156 | buf[9] = toHexUpper(ucs4); |
| 157 | buflen = 10; |
| 158 | } else { |
| 159 | buf[1] = u'u'; |
| 160 | buf[2] = toHexUpper(uc >> 12); |
| 161 | buf[3] = toHexUpper(uc >> 8); |
| 162 | buf[4] = toHexUpper(uc >> 4); |
| 163 | buf[5] = toHexUpper(uc); |
| 164 | buflen = 6; |
| 165 | } |
| 166 | } |
| 167 | |
| 168 | result.append(uc: buf, len: buflen); |
| 169 | begin = ptr + 1; |
| 170 | } |
| 171 | |
| 172 | result += u'"'; |
| 173 | } |
| 174 | |
| 175 | void DiagnosticNotation::appendArray(const QCborArray &a) |
| 176 | { |
| 177 | result += u'['; |
| 178 | |
| 179 | // length 2 (including the space) when not line wrapping |
| 180 | QLatin1StringView commaValue(", " , opts & QCborValue::LineWrapped ? 1 : 2); |
| 181 | { |
| 182 | Nest n(this); |
| 183 | QLatin1StringView comma; |
| 184 | for (auto v : a) { |
| 185 | result += comma + separator; |
| 186 | comma = commaValue; |
| 187 | appendValue(v); |
| 188 | } |
| 189 | } |
| 190 | |
| 191 | result += separator + u']'; |
| 192 | } |
| 193 | |
| 194 | void DiagnosticNotation::appendMap(const QCborMap &m) |
| 195 | { |
| 196 | result += u'{'; |
| 197 | |
| 198 | // length 2 (including the space) when not line wrapping |
| 199 | QLatin1StringView commaValue(", " , opts & QCborValue::LineWrapped ? 1 : 2); |
| 200 | { |
| 201 | Nest n(this); |
| 202 | QLatin1StringView comma; |
| 203 | for (auto v : m) { |
| 204 | result += comma + separator; |
| 205 | comma = commaValue; |
| 206 | appendValue(v: v.first); |
| 207 | result += ": "_L1 ; |
| 208 | appendValue(v: v.second); |
| 209 | } |
| 210 | } |
| 211 | |
| 212 | result += separator + u'}'; |
| 213 | }; |
| 214 | |
| 215 | void DiagnosticNotation::appendValue(const QCborValue &v) |
| 216 | { |
| 217 | switch (v.type()) { |
| 218 | case QCborValue::Integer: |
| 219 | result += QString::number(v.toInteger()); |
| 220 | return; |
| 221 | case QCborValue::ByteArray: |
| 222 | switch (byteArrayFormatStack.top()) { |
| 223 | case int(QCborKnownTags::ExpectedBase16): |
| 224 | result += QString::fromLatin1(ba: "h'" + |
| 225 | v.toByteArray().toHex(separator: opts & QCborValue::ExtendedFormat ? ' ' : '\0') + |
| 226 | '\''); |
| 227 | return; |
| 228 | case int(QCborKnownTags::ExpectedBase64): |
| 229 | result += QString::fromLatin1(ba: "b64'" + v.toByteArray().toBase64() + '\''); |
| 230 | return; |
| 231 | default: |
| 232 | case int(QCborKnownTags::ExpectedBase64url): |
| 233 | result += QString::fromLatin1(ba: "b64'" + |
| 234 | v.toByteArray().toBase64(options: QByteArray::Base64UrlEncoding | QByteArray::OmitTrailingEquals) + |
| 235 | '\''); |
| 236 | return; |
| 237 | } |
| 238 | case QCborValue::String: |
| 239 | return appendString(s: v.toString()); |
| 240 | case QCborValue::Array: |
| 241 | return appendArray(a: v.toArray()); |
| 242 | case QCborValue::Map: |
| 243 | return appendMap(m: v.toMap()); |
| 244 | case QCborValue::False: |
| 245 | result += "false"_L1 ; |
| 246 | return; |
| 247 | case QCborValue::True: |
| 248 | result += "true"_L1 ; |
| 249 | return; |
| 250 | case QCborValue::Null: |
| 251 | result += "null"_L1 ; |
| 252 | return; |
| 253 | case QCborValue::Undefined: |
| 254 | result += "undefined"_L1 ; |
| 255 | return; |
| 256 | case QCborValue::Double: |
| 257 | result += makeFpString(d: v.toDouble()); |
| 258 | return; |
| 259 | case QCborValue::Invalid: |
| 260 | result += QStringLiteral("<invalid>" ); |
| 261 | return; |
| 262 | |
| 263 | default: |
| 264 | // Only tags, extended types, and simple types remain; see below. |
| 265 | break; |
| 266 | } |
| 267 | |
| 268 | if (v.isTag()) { |
| 269 | // We handle all extended types as regular tags, so it won't matter |
| 270 | // whether we understand that tag or not. |
| 271 | bool byteArrayFormat = opts & QCborValue::ExtendedFormat && isByteArrayEncodingTag(tag: v.tag()); |
| 272 | if (byteArrayFormat) |
| 273 | byteArrayFormatStack.push(t: int(v.tag())); |
| 274 | result += QString::number(quint64(v.tag())) + u'('; |
| 275 | appendValue(v: v.taggedValue()); |
| 276 | result += u')'; |
| 277 | if (byteArrayFormat) |
| 278 | byteArrayFormatStack.pop(); |
| 279 | } else { |
| 280 | // must be a simple type |
| 281 | result += QString::fromLatin1(ba: "simple(%1)" ).arg(a: quint8(v.toSimpleType())); |
| 282 | } |
| 283 | } |
| 284 | |
| 285 | /*! |
| 286 | Creates the diagnostic notation equivalent of this CBOR object and returns |
| 287 | it. The \a opts parameter controls the dialect of the notation. Diagnostic |
| 288 | notation is useful in debugging, to aid the developer in understanding what |
| 289 | value is stored in the QCborValue or in a CBOR stream. For that reason, the |
| 290 | Qt API provides no support for parsing the diagnostic back into the |
| 291 | in-memory format or CBOR stream, though the representation is unique and it |
| 292 | would be possible. |
| 293 | |
| 294 | CBOR diagnostic notation is specified by |
| 295 | \l{RFC 7049, section 6}{section 6} of RFC 7049. |
| 296 | It is a text representation of the CBOR stream and it is very similar to |
| 297 | JSON, but it supports the CBOR types not found in JSON. The extended format |
| 298 | enabled by the \l{DiagnosticNotationOption}{ExtendedFormat} flag is |
| 299 | currently in some IETF drafts and its format is subject to change. |
| 300 | |
| 301 | This function produces the equivalent representation of the stream that |
| 302 | toCbor() would produce, without any transformation option provided there. |
| 303 | This also implies this function may not produce a representation of the |
| 304 | stream that was used to create the object, if it was created using |
| 305 | fromCbor(), as that function may have applied transformations. For a |
| 306 | high-fidelity notation of a stream, without transformation, see the \c |
| 307 | cbordump example. |
| 308 | |
| 309 | \sa toCbor(), QJsonDocument::toJson() |
| 310 | */ |
| 311 | QString QCborValue::toDiagnosticNotation(DiagnosticNotationOptions opts) const |
| 312 | { |
| 313 | return DiagnosticNotation::create(v: *this, opts); |
| 314 | } |
| 315 | |
| 316 | QT_END_NAMESPACE |
| 317 | |