|
| 1 | +#ifndef FWCore_Utilities_interface_compactStringSerializer_h |
| 2 | +#define FWCore_Utilities_interface_compactStringSerializer_h |
| 3 | + |
| 4 | +#include <cassert> |
| 5 | +#include <iterator> |
| 6 | +#include <numeric> |
| 7 | +#include <ranges> |
| 8 | +#include <string> |
| 9 | +#include <string_view> |
| 10 | + |
| 11 | +namespace edm::compactString { |
| 12 | + namespace detail { |
| 13 | + constexpr std::string_view kDelimiters = "\x1d\x1e"; |
| 14 | + constexpr char kContainerDelimiter = kDelimiters[0]; // "group separator" in ASCII |
| 15 | + constexpr char kElementDelimiter = kDelimiters[1]; // "record separator" in ASCII |
| 16 | + |
| 17 | + void throwIfContainsDelimiters(std::string const& str); |
| 18 | + } // namespace detail |
| 19 | + |
| 20 | + /** |
| 21 | + * Following three functions serialize a sequence of strings and containers of strings |
| 22 | + * |
| 23 | + * Each top-level string or container of strings is separated with kContainerDelimeter |
| 24 | + * In case of container of strings, each element is separated with kElementDelimeter |
| 25 | + * The serialized string will end with kContainerDelimeter and a null character |
| 26 | + * |
| 27 | + * The functions throw an exception if the serialized strings |
| 28 | + * contain any of the delimeter characters. The underlying string |
| 29 | + * operations may also throw exceptions. |
| 30 | + */ |
| 31 | + inline std::string serialize(std::string arg) noexcept(false) { |
| 32 | + detail::throwIfContainsDelimiters(arg); |
| 33 | + arg += detail::kContainerDelimiter; |
| 34 | + return arg; |
| 35 | + } |
| 36 | + |
| 37 | + template <typename R> |
| 38 | + requires std::ranges::input_range<R> and std::is_same_v<std::ranges::range_value_t<R>, std::string> |
| 39 | + std::string serialize(R const& arg) noexcept(false) { |
| 40 | + std::string ret; |
| 41 | + |
| 42 | + if (not std::ranges::empty(arg)) { |
| 43 | + for (std::string const& elem : arg) { |
| 44 | + ret.reserve(ret.size() + elem.size() + 1); |
| 45 | + detail::throwIfContainsDelimiters(elem); |
| 46 | + ret += elem; |
| 47 | + ret += detail::kElementDelimiter; |
| 48 | + } |
| 49 | + } |
| 50 | + |
| 51 | + ret += detail::kContainerDelimiter; |
| 52 | + return ret; |
| 53 | + } |
| 54 | + |
| 55 | + template <typename T, typename... Args> |
| 56 | + requires(sizeof...(Args) >= 1) |
| 57 | + std::string serialize(T&& arg0, Args&&... args) noexcept(false) { |
| 58 | + return serialize(std::forward<T>(arg0)) + serialize(std::forward<Args>(args)...); |
| 59 | + } |
| 60 | + |
| 61 | + /** |
| 62 | + * Following three functions deserialize a string 'input' into a |
| 63 | + * sequence of strings and containers of strings |
| 64 | + * |
| 65 | + * The 'input' string is assumed to be serialized with the |
| 66 | + * serialize() functions above. |
| 67 | + * |
| 68 | + * The output arguments following the 'input' define the schema of |
| 69 | + * the deserialization. |
| 70 | + * - std::string& for strings |
| 71 | + * - output iterator for containers of strings (e.g. std::back_inserter(vector)) |
| 72 | + * |
| 73 | + * Upon success, the return value is the position in `input` for the |
| 74 | + * next possible element (i.e. the position after the |
| 75 | + * kContainerDelimiter), that is also the number of characters |
| 76 | + * consumed by the deserializatiom.. |
| 77 | + * |
| 78 | + * Upon failure, returns 0 to denote the beginning of `input`. The |
| 79 | + * output arguments may have been modified. |
| 80 | + * |
| 81 | + * The functions do not explicitly throw exceptions, but underlying |
| 82 | + * operations may throw exceptions. |
| 83 | + */ |
| 84 | + inline std::string_view::size_type deserialize(std::string_view input, std::string& arg) { |
| 85 | + auto const pos = input.find_first_of(detail::kDelimiters); |
| 86 | + if (pos == std::string_view::npos or input[pos] != detail::kContainerDelimiter) { |
| 87 | + return 0; |
| 88 | + } |
| 89 | + arg = input.substr(0, pos); |
| 90 | + return pos + 1; // skip delimiter |
| 91 | + } |
| 92 | + |
| 93 | + template <std::output_iterator<std::string> I> |
| 94 | + inline std::string_view::size_type deserialize(std::string_view input, I arg) { |
| 95 | + auto pos = input.find_first_of(detail::kDelimiters); |
| 96 | + // invalid input |
| 97 | + if (pos == std::string_view::npos) { |
| 98 | + return 0; |
| 99 | + } |
| 100 | + // no elements |
| 101 | + if (input[pos] == detail::kContainerDelimiter) { |
| 102 | + // invalid input for empty container |
| 103 | + if (pos != 0) { |
| 104 | + return 0; |
| 105 | + } |
| 106 | + // skip delimiter |
| 107 | + return pos + 1; |
| 108 | + } |
| 109 | + |
| 110 | + std::string_view::size_type prev = 0; |
| 111 | + while (pos != std::string_view::npos and input[pos] == detail::kElementDelimiter) { |
| 112 | + *arg = std::string(input.substr(prev, pos - prev)); |
| 113 | + ++arg; |
| 114 | + prev = pos + 1; //skip delimiter |
| 115 | + pos = input.find_first_of(detail::kDelimiters, prev); |
| 116 | + } |
| 117 | + |
| 118 | + // every container must end with kContainerDelimiter |
| 119 | + // reaching npos is an error |
| 120 | + if (pos == std::string_view::npos) { |
| 121 | + return 0; |
| 122 | + } |
| 123 | + assert(input[pos] == detail::kContainerDelimiter); |
| 124 | + |
| 125 | + return pos + 1; // skip delimiter |
| 126 | + } |
| 127 | + |
| 128 | + template <typename T, typename... Args> |
| 129 | + requires(sizeof...(Args) >= 1) |
| 130 | + std::string_view::size_type deserialize(std::string_view input, T&& arg0, Args&&... args) { |
| 131 | + auto pos = deserialize(input, std::forward<T>(arg0)); |
| 132 | + if (pos != 0) { |
| 133 | + auto const ret = deserialize(input. substr(pos), std::forward<Args>(args)...); |
| 134 | + pos = (ret == 0) ? 0 : pos + ret; |
| 135 | + } |
| 136 | + return pos; |
| 137 | + } |
| 138 | +} // namespace edm::compactString |
| 139 | + |
| 140 | +#endif |
0 commit comments