/ src / util / strencodings.h
strencodings.h
  1  // Copyright (c) 2009-2010 Satoshi Nakamoto
  2  // Copyright (c) 2009-present The Bitcoin Core developers
  3  // Distributed under the MIT software license, see the accompanying
  4  // file COPYING or http://www.opensource.org/licenses/mit-license.php.
  5  
  6  /**
  7   * Utilities for converting data from/to strings.
  8   */
  9  #ifndef BITCOIN_UTIL_STRENCODINGS_H
 10  #define BITCOIN_UTIL_STRENCODINGS_H
 11  
 12  #include <crypto/hex_base.h>
 13  #include <span.h>
 14  #include <util/string.h>
 15  
 16  #include <algorithm>
 17  #include <array>
 18  #include <bit>
 19  #include <charconv>
 20  #include <cstddef>
 21  #include <cstdint>
 22  #include <limits>
 23  #include <optional>
 24  #include <string>
 25  #include <string_view>
 26  #include <system_error>
 27  #include <type_traits>
 28  #include <vector>
 29  
 30  /** Used by SanitizeString() */
 31  enum SafeChars
 32  {
 33      SAFE_CHARS_DEFAULT, //!< The full set of allowed chars
 34      SAFE_CHARS_UA_COMMENT, //!< BIP-0014 subset
 35      SAFE_CHARS_FILENAME, //!< Chars allowed in filenames
 36      SAFE_CHARS_URI, //!< Chars allowed in URIs (RFC 3986)
 37  };
 38  
 39  /**
 40   * Used by ParseByteUnits()
 41   * Lowercase base 1000
 42   * Uppercase base 1024
 43  */
 44  enum class ByteUnit : uint64_t {
 45      NOOP = 1ULL,
 46      k = 1000ULL,
 47      K = 1024ULL,
 48      m = 1'000'000ULL,
 49      M = 1ULL << 20,
 50      g = 1'000'000'000ULL,
 51      G = 1ULL << 30,
 52      t = 1'000'000'000'000ULL,
 53      T = 1ULL << 40,
 54  };
 55  
 56  /**
 57  * Remove unsafe chars. Safe chars chosen to allow simple messages/URLs/email
 58  * addresses, but avoid anything even possibly remotely dangerous like & or >
 59  * @param[in] str    The string to sanitize
 60  * @param[in] rule   The set of safe chars to choose (default: least restrictive)
 61  * @return           A new string without unsafe chars
 62  */
 63  std::string SanitizeString(std::string_view str, int rule = SAFE_CHARS_DEFAULT);
 64  /** Parse the hex string into bytes (uint8_t or std::byte). Ignores whitespace. Returns nullopt on invalid input. */
 65  template <typename Byte = std::byte>
 66  std::optional<std::vector<Byte>> TryParseHex(std::string_view str);
 67  /** Like TryParseHex, but returns an empty vector on invalid input. */
 68  template <typename Byte = uint8_t>
 69  std::vector<Byte> ParseHex(std::string_view hex_str)
 70  {
 71      return TryParseHex<Byte>(hex_str).value_or(std::vector<Byte>{});
 72  }
 73  /* Returns true if each character in str is a hex character, and has an even
 74   * number of hex digits.*/
 75  bool IsHex(std::string_view str);
 76  std::optional<std::vector<unsigned char>> DecodeBase64(std::string_view str);
 77  std::string EncodeBase64(std::span<const unsigned char> input);
 78  inline std::string EncodeBase64(std::span<const std::byte> input) { return EncodeBase64(MakeUCharSpan(input)); }
 79  inline std::string EncodeBase64(std::string_view str) { return EncodeBase64(MakeUCharSpan(str)); }
 80  std::optional<std::vector<unsigned char>> DecodeBase32(std::string_view str);
 81  
 82  /**
 83   * Base32 encode.
 84   * If `pad` is true, then the output will be padded with '=' so that its length
 85   * is a multiple of 8.
 86   */
 87  std::string EncodeBase32(std::span<const unsigned char> input, bool pad = true);
 88  
 89  /**
 90   * Base32 encode.
 91   * If `pad` is true, then the output will be padded with '=' so that its length
 92   * is a multiple of 8.
 93   */
 94  std::string EncodeBase32(std::string_view str, bool pad = true);
 95  
 96  /**
 97   * Splits socket address string into host string and port value.
 98   * Validates port value.
 99   *
100   * @param[in] in        The socket address string to split.
101   * @param[out] portOut  Port-portion of the input, if found and parsable.
102   * @param[out] hostOut  Host-portion of the input, if found.
103   * @return              true if port-portion is absent or within its allowed range, otherwise false
104   */
105  bool SplitHostPort(std::string_view in, uint16_t& portOut, std::string& hostOut);
106  
107  // LocaleIndependentAtoi is provided for backwards compatibility reasons.
108  //
109  // New code should use ToIntegral.
110  //
111  // The goal of LocaleIndependentAtoi is to replicate the defined behaviour of
112  // std::atoi as it behaves under the "C" locale, and remove some undefined
113  // behavior. If the parsed value is bigger than the integer type's maximum
114  // value, or smaller than the integer type's minimum value, std::atoi has
115  // undefined behavior, while this function returns the maximum or minimum
116  // values, respectively.
117  template <typename T>
118  T LocaleIndependentAtoi(std::string_view str)
119  {
120      static_assert(std::is_integral_v<T>);
121      T result;
122      // Emulate atoi(...) handling of white space and leading +/-.
123      std::string_view s = util::TrimStringView(str);
124      if (!s.empty() && s[0] == '+') {
125          if (s.length() >= 2 && s[1] == '-') {
126              return 0;
127          }
128          s = s.substr(1);
129      }
130      auto [_, error_condition] = std::from_chars(s.data(), s.data() + s.size(), result);
131      if (error_condition == std::errc::result_out_of_range) {
132          if (s.length() >= 1 && s[0] == '-') {
133              // Saturate underflow, per strtoll's behavior.
134              return std::numeric_limits<T>::min();
135          } else {
136              // Saturate overflow, per strtoll's behavior.
137              return std::numeric_limits<T>::max();
138          }
139      } else if (error_condition != std::errc{}) {
140          return 0;
141      }
142      return result;
143  }
144  
145  /**
146   * Tests if the given character is a decimal digit.
147   * @param[in] c     character to test
148   * @return          true if the argument is a decimal digit; otherwise false.
149   */
150  constexpr bool IsDigit(char c)
151  {
152      return c >= '0' && c <= '9';
153  }
154  
155  /**
156   * Tests if the given character is a whitespace character. The whitespace characters
157   * are: space, form-feed ('\f'), newline ('\n'), carriage return ('\r'), horizontal
158   * tab ('\t'), and vertical tab ('\v').
159   *
160   * This function is locale independent. Under the C locale this function gives the
161   * same result as std::isspace.
162   *
163   * @param[in] c     character to test
164   * @return          true if the argument is a whitespace character; otherwise false
165   */
166  constexpr inline bool IsSpace(char c) noexcept {
167      return c == ' ' || c == '\f' || c == '\n' || c == '\r' || c == '\t' || c == '\v';
168  }
169  
170  /**
171   * Convert string to integral type T. Leading whitespace, a leading +, or any
172   * trailing character fail the parsing. The required format expressed as regex
173   * is `-?[0-9]+` by default (or `-?[0-9a-fA-F]+` if base = 16).
174   * The minus sign is only permitted for signed integer types.
175   *
176   * @returns std::nullopt if the entire string could not be parsed, or if the
177   *   parsed value is not in the range representable by the type T.
178   */
179  template <typename T>
180  std::optional<T> ToIntegral(std::string_view str, size_t base = 10)
181  {
182      static_assert(std::is_integral_v<T>);
183      T result;
184      const auto [first_nonmatching, error_condition] = std::from_chars(str.data(), str.data() + str.size(), result, base);
185      if (first_nonmatching != str.data() + str.size() || error_condition != std::errc{}) {
186          return std::nullopt;
187      }
188      return result;
189  }
190  
191  /**
192   * Format a paragraph of text to a fixed width, adding spaces for
193   * indentation to any added line.
194   */
195  std::string FormatParagraph(std::string_view in, size_t width = 79, size_t indent = 0);
196  
197  /**
198   * Timing-attack-resistant comparison.
199   * Takes time proportional to length
200   * of first argument.
201   */
202  template <typename T>
203  bool TimingResistantEqual(const T& a, const T& b)
204  {
205      if (b.size() == 0) return a.size() == 0;
206      size_t accumulator = a.size() ^ b.size();
207      for (size_t i = 0; i < a.size(); i++)
208          accumulator |= size_t(a[i] ^ b[i%b.size()]);
209      return accumulator == 0;
210  }
211  
212  /** Parse number as fixed point according to JSON number syntax.
213   * @returns true on success, false on error.
214   * @note The result must be in the range (-10^18,10^18), otherwise an overflow error will trigger.
215   */
216  [[nodiscard]] bool ParseFixedPoint(std::string_view, int decimals, int64_t *amount_out);
217  
218  namespace {
219  /** Helper class for the default infn argument to ConvertBits (just returns the input). */
220  struct IntIdentity
221  {
222      [[maybe_unused]] int operator()(int x) const { return x; }
223  };
224  
225  } // namespace
226  
227  /** Convert from one power-of-2 number base to another. */
228  template<int frombits, int tobits, bool pad, typename O, typename It, typename I = IntIdentity>
229  bool ConvertBits(O outfn, It it, It end, I infn = {}) {
230      size_t acc = 0;
231      size_t bits = 0;
232      constexpr size_t maxv = (1 << tobits) - 1;
233      constexpr size_t max_acc = (1 << (frombits + tobits - 1)) - 1;
234      while (it != end) {
235          int v = infn(*it);
236          if (v < 0) return false;
237          acc = ((acc << frombits) | v) & max_acc;
238          bits += frombits;
239          while (bits >= tobits) {
240              bits -= tobits;
241              outfn((acc >> bits) & maxv);
242          }
243          ++it;
244      }
245      if (pad) {
246          if (bits) outfn((acc << (tobits - bits)) & maxv);
247      } else if (bits >= frombits || ((acc << (tobits - bits)) & maxv)) {
248          return false;
249      }
250      return true;
251  }
252  
253  /**
254   * Converts the given character to its lowercase equivalent.
255   * This function is locale independent. It only converts uppercase
256   * characters in the standard 7-bit ASCII range.
257   * This is a feature, not a limitation.
258   *
259   * @param[in] c     the character to convert to lowercase.
260   * @return          the lowercase equivalent of c; or the argument
261   *                  if no conversion is possible.
262   */
263  constexpr char ToLower(char c)
264  {
265      return (c >= 'A' && c <= 'Z' ? (c - 'A') + 'a' : c);
266  }
267  
268  /**
269   * Returns the lowercase equivalent of the given string.
270   * This function is locale independent. It only converts uppercase
271   * characters in the standard 7-bit ASCII range.
272   * This is a feature, not a limitation.
273   *
274   * @param[in] str   the string to convert to lowercase.
275   * @returns         lowercased equivalent of str
276   */
277  std::string ToLower(std::string_view str);
278  
279  /**
280   * Converts the given character to its uppercase equivalent.
281   * This function is locale independent. It only converts lowercase
282   * characters in the standard 7-bit ASCII range.
283   * This is a feature, not a limitation.
284   *
285   * @param[in] c     the character to convert to uppercase.
286   * @return          the uppercase equivalent of c; or the argument
287   *                  if no conversion is possible.
288   */
289  constexpr char ToUpper(char c)
290  {
291      return (c >= 'a' && c <= 'z' ? (c - 'a') + 'A' : c);
292  }
293  
294  /**
295   * Returns the uppercase equivalent of the given string.
296   * This function is locale independent. It only converts lowercase
297   * characters in the standard 7-bit ASCII range.
298   * This is a feature, not a limitation.
299   *
300   * @param[in] str   the string to convert to uppercase.
301   * @returns         UPPERCASED EQUIVALENT OF str
302   */
303  std::string ToUpper(std::string_view str);
304  
305  /**
306   * Capitalizes the first character of the given string.
307   * This function is locale independent. It only converts lowercase
308   * characters in the standard 7-bit ASCII range.
309   * This is a feature, not a limitation.
310   *
311   * @param[in] str   the string to capitalize.
312   * @returns         string with the first letter capitalized.
313   */
314  std::string Capitalize(std::string str);
315  
316  /**
317   * Parse a string with suffix unit [k|K|m|M|g|G|t|T].
318   * Must be a whole integer, fractions not allowed (0.5t), no whitespace or +-
319   * Lowercase units are 1000 base. Uppercase units are 1024 base.
320   * Examples: 2m,27M,19g,41T
321   *
322   * @param[in] str                  the string to convert into bytes
323   * @param[in] default_multiplier   if no unit is found in str use this unit
324   * @returns                        optional uint64_t bytes from str or nullopt
325   *                                 if ToIntegral is false, str is empty, trailing whitespace or overflow
326   */
327  std::optional<uint64_t> ParseByteUnits(std::string_view str, ByteUnit default_multiplier);
328  
329  namespace util {
330  /** consteval version of HexDigit() without the lookup table. */
331  consteval uint8_t ConstevalHexDigit(const char c)
332  {
333      if (c >= '0' && c <= '9') return c - '0';
334      if (c >= 'a' && c <= 'f') return c - 'a' + 0xa;
335  
336      throw "Only lowercase hex digits are allowed, for consistency";
337  }
338  
339  namespace detail {
340  template <size_t N>
341  struct Hex {
342      std::array<std::byte, N / 2> bytes{};
343      consteval Hex(const char (&hex_str)[N])
344          // 2 hex digits required per byte + implicit null terminator
345          requires(N % 2 == 1)
346      {
347          if (hex_str[N - 1]) throw "null terminator required";
348          for (std::size_t i = 0; i < bytes.size(); ++i) {
349              bytes[i] = static_cast<std::byte>(
350                  (ConstevalHexDigit(hex_str[2 * i]) << 4) |
351                   ConstevalHexDigit(hex_str[2 * i + 1]));
352          }
353      }
354  };
355  } // namespace detail
356  
357  struct AsciiCaseInsensitiveKeyEqual {
358      bool operator()(std::string_view s1, std::string_view s2) const
359      {
360          return ToLower(s1) == ToLower(s2);
361      }
362  };
363  
364  struct AsciiCaseInsensitiveHash {
365      size_t operator()(std::string_view s) const
366      {
367          return std::hash<std::string>{}(ToLower(s));
368      }
369  };
370  
371  /**
372   * ""_hex is a compile-time user-defined literal returning a
373   * `std::array<std::byte>`, equivalent to ParseHex(). Variants provided:
374   *
375   * - ""_hex_v: Returns `std::vector<std::byte>`, useful for heap allocation or
376   *   variable-length serialization.
377   *
378   * - ""_hex_u8: Returns `std::array<uint8_t>`, for cases where `std::byte` is
379   *   incompatible.
380   *
381   * - ""_hex_v_u8: Returns `std::vector<uint8_t>`, combining heap allocation with
382   *   `uint8_t`.
383   *
384   * @warning It could be necessary to use vector instead of array variants when
385   *   serializing, or vice versa, because vectors are assumed to be variable-
386   *   length and serialized with a size prefix, while arrays are considered fixed
387   *   length and serialized with no prefix.
388   *
389   * @warning It may be preferable to use vector variants to save stack space when
390   *   declaring local variables if hex strings are large. Alternatively variables
391   *   could be declared constexpr to avoid using stack space.
392   *
393   * @warning Avoid `uint8_t` variants when not necessary, as the codebase
394   *   migrates to use `std::byte` instead of `unsigned char` and `uint8_t`.
395   *
396   * @note One reason ""_hex uses `std::array` instead of `std::vector` like
397   *   ParseHex() does is because heap-based containers cannot cross the compile-
398   *   time/runtime barrier.
399   */
400  inline namespace hex_literals {
401  
402  template <util::detail::Hex str>
403  constexpr auto operator""_hex() { return str.bytes; }
404  
405  template <util::detail::Hex str>
406  constexpr auto operator""_hex_u8() { return std::bit_cast<std::array<uint8_t, str.bytes.size()>>(str.bytes); }
407  
408  template <util::detail::Hex str>
409  constexpr auto operator""_hex_v() { return std::vector<std::byte>{str.bytes.begin(), str.bytes.end()}; }
410  
411  template <util::detail::Hex str>
412  inline auto operator""_hex_v_u8() { return std::vector<uint8_t>{UCharCast(str.bytes.data()), UCharCast(str.bytes.data() + str.bytes.size())}; }
413  
414  } // inline namespace hex_literals
415  } // namespace util
416  
417  #endif // BITCOIN_UTIL_STRENCODINGS_H