src/utils/StringOp.hh at main · doomsdayone/openmsx

◆ main openmsx / src / utils / StringOp.hh

9.0 KiB · C++ · last changed chore(): initial commit 1 month ago source blame history raw

#ifndef STRINGOP_HH
#define STRINGOP_HH

#include "IterableBitSet.hh"
#include "stringsp.hh"

#include <algorithm>
#include <charconv>
#include <concepts>
#include <cstdint>
#include <cstdlib>
#include <cstring>
#include <limits>
#include <optional>
#include <string>
#include <string_view>
#include <type_traits>
#include <utility>

#ifdef __APPLE__
// Only forward declare 'CFStringRef' (for fromCFString() below). Including
// <CoreFoundation/CoreFoundation.h> here would drag it into the ~75% of all
// translation units that (indirectly) include this header, and it declares a
// lot of names in the global namespace. E.g. MacTypes.h declares a global
// 'FixedPoint', which clashes with openmsx::FixedPoint.
using CFStringRef = const struct __CFString*;
#endif

namespace StringOp
{
	/** Convert a string to an integral type 'T' (int, uint64_t, ...).
	  * This is similar to, but not quite the same as the family of
	  * 'strtoll()' functions:
	  * - Leading whitespace is NOT accepted (unlike strtoll()).
	  * - There may NOT be any trailing character after the value (unlike
	  *   strtoll()).
	  * - (Only) if 'T' is a signed type, the value may start with a '-'
	  *   character.
	  * - A leading '+' character is NOT accepted.
	  * - It's an error if the value cannot be represented by the type 'T'.
	  * - This function tries to detect the base of the value by looking at
	  *   the prefix. (Only) these prefixes are accepted:
	  *    - '0x' or '0X': hexadecimal
	  *    - '0b' or '0B': binary
	  *    - no prefix: decimal
	  *   Note that prefix '0' for octal is NOT supported.
	  * - The input is a 'string_view' (rather than a 'const char*'), so it
	  *   is not required to be zero-terminated.
	  */
	template<std::integral T> [[nodiscard]] std::optional<T> stringTo(std::string_view s);

	/** As above, but without dynamic base detection. Moreover leading
	  * prefixes like '0x' for hexadecimal are seen as invalid input.
	  */
	template<int BASE, std::integral T> [[nodiscard]] std::optional<T> stringToBase(std::string_view s);

	[[nodiscard]] bool stringToBool(std::string_view str);

	[[nodiscard]] std::string toLower(std::string_view str);

	void trimRight(std::string& str, const char* chars);
	void trimRight(std::string& str, char chars);
	void trimRight(std::string_view& str, std::string_view chars);
	void trimRight(std::string_view& str, char chars);
	void trimLeft (std::string& str, const char* chars);
	void trimLeft (std::string& str, char chars);
	void trimLeft (std::string_view& str, std::string_view chars);
	void trimLeft (std::string_view& str, char chars);
	void trim     (std::string_view& str, std::string_view chars);
	void trim     (std::string_view& str, char chars);

	[[nodiscard]] std::pair<std::string_view, std::string_view> splitOnFirst(
		std::string_view str, std::string_view chars);
	[[nodiscard]] std::pair<std::string_view, std::string_view> splitOnFirst(
		std::string_view str, char chars);
	[[nodiscard]] std::pair<std::string_view, std::string_view> splitOnLast(
		std::string_view str, std::string_view chars);
	[[nodiscard]] std::pair<std::string_view, std::string_view> splitOnLast(
		std::string_view str, char chars);
	[[nodiscard]] IterableBitSet<64> parseRange(std::string_view str,
	                                            unsigned min, unsigned max);

	//[[nodiscard]] std::vector<std::string_view> split(std::string_view str, char chars);

	enum class EmptyParts : uint8_t {
		KEEP,  // "a,b,,c" -> "a", "b", "", "c"
		REMOVE // "a,b,,c" -> "a", "b", "c"       BUT  ",,a,b" -> "", "a", "b"  (keeps one empty part in front)
	};
	template<EmptyParts keepOrRemove = EmptyParts::KEEP, typename Separators>
	[[nodiscard]] inline auto split_view(std::string_view str, Separators separators) {
		struct Iterator {
			using value_type      = std::string_view;
			using difference_type = ptrdiff_t;

			Iterator() = default;
			Iterator(const Iterator&) = default;
			Iterator& operator=(const Iterator&) = default;

			Iterator(std::string_view str_, Separators separators_)
				: str(str_), separators(separators_) {
				str.remove_prefix(skipSeparators(0));
				next_p();
			}

			[[nodiscard]] value_type operator*() const {
				difference_type dummy; (void)dummy; // avoid warning about unused typedef
				return {str.data(), p};
			}

			Iterator& operator++() {
				if (p < str.size()) {
					str.remove_prefix(skipSeparators(p + 1));
					next_p();
				} else {
					str = "";
				}
				return *this;
			}
			Iterator operator++(int) { auto copy = *this; ++(*this); return copy; }

			[[nodiscard]] bool operator==(const Iterator&) const = default;
			[[nodiscard]] bool operator==(std::default_sentinel_t) const { return str.empty(); }

		private:
			static bool isSeparator(char c, char separators) {
				return c == separators;
			}
			static bool isSeparator(char c, std::string_view separators) {
				return separators.contains(c);
			}

			void next_p() {
				p = 0;
				while ((p < str.size()) && !isSeparator(str[p], separators)) ++p;
			}

			[[nodiscard]] std::string_view::size_type skipSeparators(std::string_view::size_type pos) const {
				if (keepOrRemove == EmptyParts::REMOVE) {
					while ((pos < str.size()) && isSeparator(str[pos], separators)) ++pos;
				}
				return pos;
			}

		private:
			std::string_view str;
			std::string_view::size_type p;
			Separators separators;
		};
		static_assert(std::forward_iterator<Iterator>);
		static_assert(std::sentinel_for<std::default_sentinel_t, Iterator>);

		struct Splitter {
			std::string_view str;
			Separators separators;

			[[nodiscard]] auto begin() const { return Iterator{str, separators}; }
			[[nodiscard]] auto end()   const { return std::default_sentinel_t{}; }
		};

		return Splitter{str, separators};
	}

	// case insensitive less-than operator
	struct caseless {
		[[nodiscard]] bool operator()(std::string_view s1, std::string_view s2) const {
			auto m = std::min(s1.size(), s2.size());
			int r = strncasecmp(s1.data(), s2.data(), m);
			return (r != 0) ? (r < 0) : (s1.size() < s2.size());
		}
	};
	// case insensitive greater-than operator
	struct inv_caseless {
		[[nodiscard]] bool operator()(std::string_view s1, std::string_view s2) const {
			auto m = std::min(s1.size(), s2.size());
			int r = strncasecmp(s1.data(), s2.data(), m);
			return (r != 0) ? (r > 0) : (s1.size() > s2.size());
		}
	};
	struct casecmp {
		[[nodiscard]] bool operator()(std::string_view s1, std::string_view s2) const {
			if (s1.size() != s2.size()) return false;
			return strncasecmp(s1.data(), s2.data(), s1.size()) == 0;
		}
	};

	[[nodiscard]] inline bool containsCaseInsensitive(std::string_view haystack, std::string_view needle)
	{
		return !std::ranges::search(haystack, needle,
		                   [](char x, char y) { return toupper(x) == toupper(y); }
		        ).empty();
	}

#ifdef __APPLE__
	[[nodiscard]] std::string fromCFString(CFStringRef str);
#endif

	template<int BASE, std::integral T>
	[[nodiscard]] std::optional<T> stringToBase(std::string_view s)
	{
		T result = {}; // dummy init to avoid warning
		const auto* b = s.data();
		const auto* e = s.data() + s.size();
		if (auto [p, ec] = std::from_chars(b, e, result, BASE);
		    (ec == std::errc()) && (p == e)) {
			return result;
		}
		return std::nullopt;
	}

	template<std::integral T>
	[[nodiscard]] std::optional<T> stringTo(std::string_view s)
	{
		if (s.empty()) [[unlikely]] return {};
		if constexpr (std::is_signed_v<T>) {
			bool negate = false;
			if (s[0] == '-') [[unlikely]] {
				negate = true;
				s.remove_prefix(1);
			}

			using U = std::make_unsigned_t<T>;
			auto tmp = stringTo<U>(s);
			if (!tmp) [[unlikely]] return {};

			U max = U(std::numeric_limits<T>::max()) + 1; // 0x8000
			if (negate) [[unlikely]] {
				if (*tmp > max) [[unlikely]] return {}; // 0x8000
				return T(~*tmp + 1);
			} else {
				if (*tmp >= max) [[unlikely]] return {}; // 0x7fff
				return T(*tmp);
			}
		} else {
			if (s[0] != '0') [[likely]] {
				return stringToBase<10, T>(s);
			} else {
				if (s.size() == 1) return T(0);
				if ((s[1] == 'x') || (s[1] == 'X')) [[likely]] {
					s.remove_prefix(2);
					return stringToBase<16, T>(s);
				} else if ((s[1] == 'b') || (s[1] == 'B')) {
					s.remove_prefix(2);
					return stringToBase<2, T>(s);
				} else {
					return stringToBase<10, T>(s);
				}
			}
		}
	}

	#if 0
	// Doesn't work yet in MacOS :(
	template<std::floating_point T>
	[[nodiscard]] std::optional<T> stringTo(std::string_view s)
	{
		T result{}; // dummy init
		const char* b = s.data();
		const char* e = s.data() + s.size();
		if (auto [p, ec] = std::from_chars(b, e, result);
		    (ec == std::errc{}) && (p == e)) {
			return result;
		}
		return std::nullopt;
	}
	#else
	template<typename T>
		requires std::is_same_v<T, double>
	[[nodiscard]] std::optional<T> stringTo(std::string_view s)
	{
		// strtod expects a null-terminated string
		char buf[128];
		if (s.size() >= sizeof(buf)) return std::nullopt;
		memcpy(buf, s.data(), s.size());
		buf[s.size()] = '\0';

		errno = 0;
		char* end = nullptr;
		double value = strtod(buf, &end);

		if (errno != 0) return std::nullopt;
		if (end != (buf + s.size())) return std::nullopt;
		return value;
	}
	#endif
}

#endif

`/`	Search
`g` `c`	Code
`g` `h`	History
`g` `b`	Bookmarks
`g` `t`	Trust
`y`	Permalink to this state (files and trees)
`?`	This sheet

e8555c89 Andy McCall 1 month ago	1	#ifndef STRINGOP_HH
	2	#define STRINGOP_HH
	3
	4	#include "IterableBitSet.hh"
	5	#include "stringsp.hh"
	6
	7	#include <algorithm>
	8	#include <charconv>
	9	#include <concepts>
	10	#include <cstdint>
	11	#include <cstdlib>
	12	#include <cstring>
	13	#include <limits>
	14	#include <optional>
	15	#include <string>
	16	#include <string_view>
	17	#include <type_traits>
	18	#include <utility>
	19
	20	#ifdef __APPLE__
	21	// Only forward declare 'CFStringRef' (for fromCFString() below). Including
	22	// <CoreFoundation/CoreFoundation.h> here would drag it into the ~75% of all
	23	// translation units that (indirectly) include this header, and it declares a
	24	// lot of names in the global namespace. E.g. MacTypes.h declares a global
	25	// 'FixedPoint', which clashes with openmsx::FixedPoint.
	26	using CFStringRef = const struct __CFString*;
	27	#endif
	28
	29	namespace StringOp
	30	{
	31	/** Convert a string to an integral type 'T' (int, uint64_t, ...).
	32	* This is similar to, but not quite the same as the family of
	33	* 'strtoll()' functions:
	34	* - Leading whitespace is NOT accepted (unlike strtoll()).
	35	* - There may NOT be any trailing character after the value (unlike
	36	* strtoll()).
	37	* - (Only) if 'T' is a signed type, the value may start with a '-'
	38	* character.
	39	* - A leading '+' character is NOT accepted.
	40	* - It's an error if the value cannot be represented by the type 'T'.
	41	* - This function tries to detect the base of the value by looking at
	42	* the prefix. (Only) these prefixes are accepted:
	43	* - '0x' or '0X': hexadecimal
	44	* - '0b' or '0B': binary
	45	* - no prefix: decimal
	46	* Note that prefix '0' for octal is NOT supported.
	47	* - The input is a 'string_view' (rather than a 'const char*'), so it
	48	* is not required to be zero-terminated.
	49	*/
	50	template<std::integral T> [[nodiscard]] std::optional<T> stringTo(std::string_view s);
	51
	52	/** As above, but without dynamic base detection. Moreover leading
	53	* prefixes like '0x' for hexadecimal are seen as invalid input.
	54	*/
	55	template<int BASE, std::integral T> [[nodiscard]] std::optional<T> stringToBase(std::string_view s);
	56
	57	[[nodiscard]] bool stringToBool(std::string_view str);
	58
	59	[[nodiscard]] std::string toLower(std::string_view str);
	60
	61	void trimRight(std::string& str, const char* chars);
	62	void trimRight(std::string& str, char chars);
	63	void trimRight(std::string_view& str, std::string_view chars);
	64	void trimRight(std::string_view& str, char chars);
65	void trimLeft (std::string& str, const char* chars);
66	void trimLeft (std::string& str, char chars);
67	void trimLeft (std::string_view& str, std::string_view chars);
68	void trimLeft (std::string_view& str, char chars);
69	void trim (std::string_view& str, std::string_view chars);
70	void trim (std::string_view& str, char chars);
71
72	[[nodiscard]] std::pair<std::string_view, std::string_view> splitOnFirst(
73	std::string_view str, std::string_view chars);
74	[[nodiscard]] std::pair<std::string_view, std::string_view> splitOnFirst(
75	std::string_view str, char chars);
76	[[nodiscard]] std::pair<std::string_view, std::string_view> splitOnLast(
77	std::string_view str, std::string_view chars);
78	[[nodiscard]] std::pair<std::string_view, std::string_view> splitOnLast(
79	std::string_view str, char chars);
80	[[nodiscard]] IterableBitSet<64> parseRange(std::string_view str,
81	unsigned min, unsigned max);
82
83	//[[nodiscard]] std::vector<std::string_view> split(std::string_view str, char chars);
84
85	enum class EmptyParts : uint8_t {
86	KEEP, // "a,b,,c" -> "a", "b", "", "c"
87	REMOVE // "a,b,,c" -> "a", "b", "c" BUT ",,a,b" -> "", "a", "b" (keeps one empty part in front)
88	};
89	template<EmptyParts keepOrRemove = EmptyParts::KEEP, typename Separators>
90	[[nodiscard]] inline auto split_view(std::string_view str, Separators separators) {
91	struct Iterator {
92	using value_type = std::string_view;
93	using difference_type = ptrdiff_t;
94
95	Iterator() = default;
96	Iterator(const Iterator&) = default;
97	Iterator& operator=(const Iterator&) = default;
98
99	Iterator(std::string_view str_, Separators separators_)
100	: str(str_), separators(separators_) {
101	str.remove_prefix(skipSeparators(0));
102	next_p();
103	}
104
105	[[nodiscard]] value_type operator*() const {
106	difference_type dummy; (void)dummy; // avoid warning about unused typedef
107	return {str.data(), p};
108	}
109
110	Iterator& operator++() {
111	if (p < str.size()) {
112	str.remove_prefix(skipSeparators(p + 1));
113	next_p();
114	} else {
115	str = "";
116	}
117	return *this;
118	}
119	Iterator operator++(int) { auto copy = this; ++(this); return copy; }
120
121	[[nodiscard]] bool operator==(const Iterator&) const = default;
122	[[nodiscard]] bool operator==(std::default_sentinel_t) const { return str.empty(); }
123
124	private:
125	static bool isSeparator(char c, char separators) {
126	return c == separators;
127	}
128	static bool isSeparator(char c, std::string_view separators) {
129	return separators.contains(c);
130	}
131
132	void next_p() {
133	p = 0;
134	while ((p < str.size()) && !isSeparator(str[p], separators)) ++p;
135	}
136
137	[[nodiscard]] std::string_view::size_type skipSeparators(std::string_view::size_type pos) const {
138	if (keepOrRemove == EmptyParts::REMOVE) {
139	while ((pos < str.size()) && isSeparator(str[pos], separators)) ++pos;
140	}
141	return pos;
142	}
143
144	private:
145	std::string_view str;
146	std::string_view::size_type p;
147	Separators separators;
148	};
149	static_assert(std::forward_iterator<Iterator>);
150	static_assert(std::sentinel_for<std::default_sentinel_t, Iterator>);
151
152	struct Splitter {
153	std::string_view str;
154	Separators separators;
155
156	[[nodiscard]] auto begin() const { return Iterator{str, separators}; }
157	[[nodiscard]] auto end() const { return std::default_sentinel_t{}; }
158	};
159
160	return Splitter{str, separators};
161	}
162
163	// case insensitive less-than operator
164	struct caseless {
165	[[nodiscard]] bool operator()(std::string_view s1, std::string_view s2) const {
166	auto m = std::min(s1.size(), s2.size());
167	int r = strncasecmp(s1.data(), s2.data(), m);
168	return (r != 0) ? (r < 0) : (s1.size() < s2.size());
169	}
170	};
171	// case insensitive greater-than operator
172	struct inv_caseless {
173	[[nodiscard]] bool operator()(std::string_view s1, std::string_view s2) const {
174	auto m = std::min(s1.size(), s2.size());
175	int r = strncasecmp(s1.data(), s2.data(), m);
176	return (r != 0) ? (r > 0) : (s1.size() > s2.size());
177	}
178	};
179	struct casecmp {
180	[[nodiscard]] bool operator()(std::string_view s1, std::string_view s2) const {
181	if (s1.size() != s2.size()) return false;
182	return strncasecmp(s1.data(), s2.data(), s1.size()) == 0;
183	}
184	};
185
186	[[nodiscard]] inline bool containsCaseInsensitive(std::string_view haystack, std::string_view needle)
187	{
188	return !std::ranges::search(haystack, needle,
189	[](char x, char y) { return toupper(x) == toupper(y); }
190	).empty();
191	}
192
193	#ifdef __APPLE__
194	[[nodiscard]] std::string fromCFString(CFStringRef str);
195	#endif
196
197	template<int BASE, std::integral T>
198	[[nodiscard]] std::optional<T> stringToBase(std::string_view s)
199	{
200	T result = {}; // dummy init to avoid warning
201	const auto* b = s.data();
202	const auto* e = s.data() + s.size();
203	if (auto [p, ec] = std::from_chars(b, e, result, BASE);
204	(ec == std::errc()) && (p == e)) {
205	return result;
206	}
207	return std::nullopt;
208	}
209
210	template<std::integral T>
211	[[nodiscard]] std::optional<T> stringTo(std::string_view s)
212	{
213	if (s.empty()) [[unlikely]] return {};
214	if constexpr (std::is_signed_v<T>) {
215	bool negate = false;
216	if (s[0] == '-') [[unlikely]] {
217	negate = true;
218	s.remove_prefix(1);
219	}
220
221	using U = std::make_unsigned_t<T>;
222	auto tmp = stringTo<U>(s);
223	if (!tmp) [[unlikely]] return {};
224
225	U max = U(std::numeric_limits<T>::max()) + 1; // 0x8000
226	if (negate) [[unlikely]] {
227	if (*tmp > max) [[unlikely]] return {}; // 0x8000
228	return T(~*tmp + 1);
229	} else {
230	if (*tmp >= max) [[unlikely]] return {}; // 0x7fff
231	return T(*tmp);
232	}
233	} else {
234	if (s[0] != '0') [[likely]] {
235	return stringToBase<10, T>(s);
236	} else {
237	if (s.size() == 1) return T(0);
238	if ((s[1] == 'x') \|\| (s[1] == 'X')) [[likely]] {
239	s.remove_prefix(2);
240	return stringToBase<16, T>(s);
241	} else if ((s[1] == 'b') \|\| (s[1] == 'B')) {
242	s.remove_prefix(2);
243	return stringToBase<2, T>(s);
244	} else {
245	return stringToBase<10, T>(s);
246	}
247	}
248	}
249	}
250
251	#if 0
252	// Doesn't work yet in MacOS :(
253	template<std::floating_point T>
254	[[nodiscard]] std::optional<T> stringTo(std::string_view s)
255	{
256	T result{}; // dummy init
257	const char* b = s.data();
258	const char* e = s.data() + s.size();
259	if (auto [p, ec] = std::from_chars(b, e, result);
260	(ec == std::errc{}) && (p == e)) {
261	return result;
262	}
263	return std::nullopt;
264	}
265	#else
266	template<typename T>
267	requires std::is_same_v<T, double>
268	[[nodiscard]] std::optional<T> stringTo(std::string_view s)
269	{
270	// strtod expects a null-terminated string
271	char buf[128];
272	if (s.size() >= sizeof(buf)) return std::nullopt;
273	memcpy(buf, s.data(), s.size());
274	buf[s.size()] = '\0';
275
276	errno = 0;
277	char* end = nullptr;
278	double value = strtod(buf, &end);
279
280	if (errno != 0) return std::nullopt;
281	if (end != (buf + s.size())) return std::nullopt;
282	return value;
283	}
284	#endif
285	}
286
287	#endif

Keyboard