| 1 | #ifndef ABSL_STRINGS_INTERNAL_STR_FORMAT_CHECKER_H_ |
| 2 | #define ABSL_STRINGS_INTERNAL_STR_FORMAT_CHECKER_H_ |
| 3 | |
| 4 | #include "absl/strings/internal/str_format/arg.h" |
| 5 | #include "absl/strings/internal/str_format/extension.h" |
| 6 | |
| 7 | // Compile time check support for entry points. |
| 8 | |
| 9 | #ifndef ABSL_INTERNAL_ENABLE_FORMAT_CHECKER |
| 10 | #if defined(__clang__) && !defined(__native_client__) |
| 11 | #if __has_attribute(enable_if) |
| 12 | #define ABSL_INTERNAL_ENABLE_FORMAT_CHECKER 1 |
| 13 | #endif // __has_attribute(enable_if) |
| 14 | #endif // defined(__clang__) && !defined(__native_client__) |
| 15 | #endif // ABSL_INTERNAL_ENABLE_FORMAT_CHECKER |
| 16 | |
| 17 | namespace absl { |
| 18 | namespace str_format_internal { |
| 19 | |
| 20 | constexpr bool AllOf() { return true; } |
| 21 | |
| 22 | template <typename... T> |
| 23 | constexpr bool AllOf(bool b, T... t) { |
| 24 | return b && AllOf(t...); |
| 25 | } |
| 26 | |
| 27 | template <typename Arg> |
| 28 | constexpr Conv ArgumentToConv() { |
| 29 | return decltype(str_format_internal::FormatConvertImpl( |
| 30 | std::declval<const Arg&>(), std::declval<const ConversionSpec&>(), |
| 31 | std::declval<FormatSinkImpl*>()))::kConv; |
| 32 | } |
| 33 | |
| 34 | #if ABSL_INTERNAL_ENABLE_FORMAT_CHECKER |
| 35 | |
| 36 | constexpr bool ContainsChar(const char* chars, char c) { |
| 37 | return *chars == c || (*chars && ContainsChar(chars + 1, c)); |
| 38 | } |
| 39 | |
| 40 | // A constexpr compatible list of Convs. |
| 41 | struct ConvList { |
| 42 | const Conv* array; |
| 43 | int count; |
| 44 | |
| 45 | // We do the bound check here to avoid having to do it on the callers. |
| 46 | // Returning an empty Conv has the same effect as short circuiting because it |
| 47 | // will never match any conversion. |
| 48 | constexpr Conv operator[](int i) const { |
| 49 | return i < count ? array[i] : Conv{}; |
| 50 | } |
| 51 | |
| 52 | constexpr ConvList without_front() const { |
| 53 | return count != 0 ? ConvList{array + 1, count - 1} : *this; |
| 54 | } |
| 55 | }; |
| 56 | |
| 57 | template <size_t count> |
| 58 | struct ConvListT { |
| 59 | // Make sure the array has size > 0. |
| 60 | Conv list[count ? count : 1]; |
| 61 | }; |
| 62 | |
| 63 | constexpr char GetChar(string_view str, size_t index) { |
| 64 | return index < str.size() ? str[index] : char{}; |
| 65 | } |
| 66 | |
| 67 | constexpr string_view ConsumeFront(string_view str, size_t len = 1) { |
| 68 | return len <= str.size() ? string_view(str.data() + len, str.size() - len) |
| 69 | : string_view(); |
| 70 | } |
| 71 | |
| 72 | constexpr string_view ConsumeAnyOf(string_view format, const char* chars) { |
| 73 | return ContainsChar(chars, GetChar(format, 0)) |
| 74 | ? ConsumeAnyOf(ConsumeFront(format), chars) |
| 75 | : format; |
| 76 | } |
| 77 | |
| 78 | constexpr bool IsDigit(char c) { return c >= '0' && c <= '9'; } |
| 79 | |
| 80 | // Helper class for the ParseDigits function. |
| 81 | // It encapsulates the two return values we need there. |
| 82 | struct Integer { |
| 83 | string_view format; |
| 84 | int value; |
| 85 | |
| 86 | // If the next character is a '$', consume it. |
| 87 | // Otherwise, make `this` an invalid positional argument. |
| 88 | constexpr Integer ConsumePositionalDollar() const { |
| 89 | return GetChar(format, 0) == '$' ? Integer{ConsumeFront(format), value} |
| 90 | : Integer{format, 0}; |
| 91 | } |
| 92 | }; |
| 93 | |
| 94 | constexpr Integer ParseDigits(string_view format, int value = 0) { |
| 95 | return IsDigit(GetChar(format, 0)) |
| 96 | ? ParseDigits(ConsumeFront(format), |
| 97 | 10 * value + GetChar(format, 0) - '0') |
| 98 | : Integer{format, value}; |
| 99 | } |
| 100 | |
| 101 | // Parse digits for a positional argument. |
| 102 | // The parsing also consumes the '$'. |
| 103 | constexpr Integer ParsePositional(string_view format) { |
| 104 | return ParseDigits(format).ConsumePositionalDollar(); |
| 105 | } |
| 106 | |
| 107 | // Parses a single conversion specifier. |
| 108 | // See ConvParser::Run() for post conditions. |
| 109 | class ConvParser { |
| 110 | constexpr ConvParser SetFormat(string_view format) const { |
| 111 | return ConvParser(format, args_, error_, arg_position_, is_positional_); |
| 112 | } |
| 113 | |
| 114 | constexpr ConvParser SetArgs(ConvList args) const { |
| 115 | return ConvParser(format_, args, error_, arg_position_, is_positional_); |
| 116 | } |
| 117 | |
| 118 | constexpr ConvParser SetError(bool error) const { |
| 119 | return ConvParser(format_, args_, error_ || error, arg_position_, |
| 120 | is_positional_); |
| 121 | } |
| 122 | |
| 123 | constexpr ConvParser SetArgPosition(int arg_position) const { |
| 124 | return ConvParser(format_, args_, error_, arg_position, is_positional_); |
| 125 | } |
| 126 | |
| 127 | // Consumes the next arg and verifies that it matches `conv`. |
| 128 | // `error_` is set if there is no next arg or if it doesn't match `conv`. |
| 129 | constexpr ConvParser ConsumeNextArg(char conv) const { |
| 130 | return SetArgs(args_.without_front()).SetError(!Contains(args_[0], conv)); |
| 131 | } |
| 132 | |
| 133 | // Verify that positional argument `i.value` matches `conv`. |
| 134 | // `error_` is set if `i.value` is not a valid argument or if it doesn't |
| 135 | // match. |
| 136 | constexpr ConvParser VerifyPositional(Integer i, char conv) const { |
| 137 | return SetFormat(i.format).SetError(!Contains(args_[i.value - 1], conv)); |
| 138 | } |
| 139 | |
| 140 | // Parse the position of the arg and store it in `arg_position_`. |
| 141 | constexpr ConvParser ParseArgPosition(Integer arg) const { |
| 142 | return SetFormat(arg.format).SetArgPosition(arg.value); |
| 143 | } |
| 144 | |
| 145 | // Consume the flags. |
| 146 | constexpr ConvParser ParseFlags() const { |
| 147 | return SetFormat(ConsumeAnyOf(format_, "-+ #0" )); |
| 148 | } |
| 149 | |
| 150 | // Consume the width. |
| 151 | // If it is '*', we verify that it matches `args_`. `error_` is set if it |
| 152 | // doesn't match. |
| 153 | constexpr ConvParser ParseWidth() const { |
| 154 | return IsDigit(GetChar(format_, 0)) |
| 155 | ? SetFormat(ParseDigits(format_).format) |
| 156 | : GetChar(format_, 0) == '*' |
| 157 | ? is_positional_ |
| 158 | ? VerifyPositional( |
| 159 | ParsePositional(ConsumeFront(format_)), '*') |
| 160 | : SetFormat(ConsumeFront(format_)) |
| 161 | .ConsumeNextArg('*') |
| 162 | : *this; |
| 163 | } |
| 164 | |
| 165 | // Consume the precision. |
| 166 | // If it is '*', we verify that it matches `args_`. `error_` is set if it |
| 167 | // doesn't match. |
| 168 | constexpr ConvParser ParsePrecision() const { |
| 169 | return GetChar(format_, 0) != '.' |
| 170 | ? *this |
| 171 | : GetChar(format_, 1) == '*' |
| 172 | ? is_positional_ |
| 173 | ? VerifyPositional( |
| 174 | ParsePositional(ConsumeFront(format_, 2)), '*') |
| 175 | : SetFormat(ConsumeFront(format_, 2)) |
| 176 | .ConsumeNextArg('*') |
| 177 | : SetFormat(ParseDigits(ConsumeFront(format_)).format); |
| 178 | } |
| 179 | |
| 180 | // Consume the length characters. |
| 181 | constexpr ConvParser ParseLength() const { |
| 182 | return SetFormat(ConsumeAnyOf(format_, "lLhjztq" )); |
| 183 | } |
| 184 | |
| 185 | // Consume the conversion character and verify that it matches `args_`. |
| 186 | // `error_` is set if it doesn't match. |
| 187 | constexpr ConvParser ParseConversion() const { |
| 188 | return is_positional_ |
| 189 | ? VerifyPositional({ConsumeFront(format_), arg_position_}, |
| 190 | GetChar(format_, 0)) |
| 191 | : ConsumeNextArg(GetChar(format_, 0)) |
| 192 | .SetFormat(ConsumeFront(format_)); |
| 193 | } |
| 194 | |
| 195 | constexpr ConvParser(string_view format, ConvList args, bool error, |
| 196 | int arg_position, bool is_positional) |
| 197 | : format_(format), |
| 198 | args_(args), |
| 199 | error_(error), |
| 200 | arg_position_(arg_position), |
| 201 | is_positional_(is_positional) {} |
| 202 | |
| 203 | public: |
| 204 | constexpr ConvParser(string_view format, ConvList args, bool is_positional) |
| 205 | : format_(format), |
| 206 | args_(args), |
| 207 | error_(false), |
| 208 | arg_position_(0), |
| 209 | is_positional_(is_positional) {} |
| 210 | |
| 211 | // Consume the whole conversion specifier. |
| 212 | // `format()` will be set to the character after the conversion character. |
| 213 | // `error()` will be set if any of the arguments do not match. |
| 214 | constexpr ConvParser Run() const { |
| 215 | return (is_positional_ ? ParseArgPosition(ParsePositional(format_)) : *this) |
| 216 | .ParseFlags() |
| 217 | .ParseWidth() |
| 218 | .ParsePrecision() |
| 219 | .ParseLength() |
| 220 | .ParseConversion(); |
| 221 | } |
| 222 | |
| 223 | constexpr string_view format() const { return format_; } |
| 224 | constexpr ConvList args() const { return args_; } |
| 225 | constexpr bool error() const { return error_; } |
| 226 | constexpr bool is_positional() const { return is_positional_; } |
| 227 | |
| 228 | private: |
| 229 | string_view format_; |
| 230 | // Current list of arguments. If we are not in positional mode we will consume |
| 231 | // from the front. |
| 232 | ConvList args_; |
| 233 | bool error_; |
| 234 | // Holds the argument position of the conversion character, if we are in |
| 235 | // positional mode. Otherwise, it is unspecified. |
| 236 | int arg_position_; |
| 237 | // Whether we are in positional mode. |
| 238 | // It changes the behavior of '*' and where to find the converted argument. |
| 239 | bool is_positional_; |
| 240 | }; |
| 241 | |
| 242 | // Parses a whole format expression. |
| 243 | // See FormatParser::Run(). |
| 244 | class FormatParser { |
| 245 | static constexpr bool FoundPercent(string_view format) { |
| 246 | return format.empty() || |
| 247 | (GetChar(format, 0) == '%' && GetChar(format, 1) != '%'); |
| 248 | } |
| 249 | |
| 250 | // We use an inner function to increase the recursion limit. |
| 251 | // The inner function consumes up to `limit` characters on every run. |
| 252 | // This increases the limit from 512 to ~512*limit. |
| 253 | static constexpr string_view ConsumeNonPercentInner(string_view format, |
| 254 | int limit = 20) { |
| 255 | return FoundPercent(format) || !limit |
| 256 | ? format |
| 257 | : ConsumeNonPercentInner( |
| 258 | ConsumeFront(format, GetChar(format, 0) == '%' && |
| 259 | GetChar(format, 1) == '%' |
| 260 | ? 2 |
| 261 | : 1), |
| 262 | limit - 1); |
| 263 | } |
| 264 | |
| 265 | // Consume characters until the next conversion spec %. |
| 266 | // It skips %%. |
| 267 | static constexpr string_view ConsumeNonPercent(string_view format) { |
| 268 | return FoundPercent(format) |
| 269 | ? format |
| 270 | : ConsumeNonPercent(ConsumeNonPercentInner(format)); |
| 271 | } |
| 272 | |
| 273 | static constexpr bool IsPositional(string_view format) { |
| 274 | return IsDigit(GetChar(format, 0)) ? IsPositional(ConsumeFront(format)) |
| 275 | : GetChar(format, 0) == '$'; |
| 276 | } |
| 277 | |
| 278 | constexpr bool RunImpl(bool is_positional) const { |
| 279 | // In non-positional mode we require all arguments to be consumed. |
| 280 | // In positional mode just reaching the end of the format without errors is |
| 281 | // enough. |
| 282 | return (format_.empty() && (is_positional || args_.count == 0)) || |
| 283 | (!format_.empty() && |
| 284 | ValidateArg( |
| 285 | ConvParser(ConsumeFront(format_), args_, is_positional).Run())); |
| 286 | } |
| 287 | |
| 288 | constexpr bool ValidateArg(ConvParser conv) const { |
| 289 | return !conv.error() && FormatParser(conv.format(), conv.args()) |
| 290 | .RunImpl(conv.is_positional()); |
| 291 | } |
| 292 | |
| 293 | public: |
| 294 | constexpr FormatParser(string_view format, ConvList args) |
| 295 | : format_(ConsumeNonPercent(format)), args_(args) {} |
| 296 | |
| 297 | // Runs the parser for `format` and `args`. |
| 298 | // It verifies that the format is valid and that all conversion specifiers |
| 299 | // match the arguments passed. |
| 300 | // In non-positional mode it also verfies that all arguments are consumed. |
| 301 | constexpr bool Run() const { |
| 302 | return RunImpl(!format_.empty() && IsPositional(ConsumeFront(format_))); |
| 303 | } |
| 304 | |
| 305 | private: |
| 306 | string_view format_; |
| 307 | // Current list of arguments. |
| 308 | // If we are not in positional mode we will consume from the front and will |
| 309 | // have to be empty in the end. |
| 310 | ConvList args_; |
| 311 | }; |
| 312 | |
| 313 | template <Conv... C> |
| 314 | constexpr bool ValidFormatImpl(string_view format) { |
| 315 | return FormatParser(format, |
| 316 | {ConvListT<sizeof...(C)>{{C...}}.list, sizeof...(C)}) |
| 317 | .Run(); |
| 318 | } |
| 319 | |
| 320 | #endif // ABSL_INTERNAL_ENABLE_FORMAT_CHECKER |
| 321 | |
| 322 | } // namespace str_format_internal |
| 323 | } // namespace absl |
| 324 | |
| 325 | #endif // ABSL_STRINGS_INTERNAL_STR_FORMAT_CHECKER_H_ |
| 326 | |