StringView.cpp 8.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290
  1. /*
  2. * Copyright (c) 2018-2021, Andreas Kling <kling@serenityos.org>
  3. *
  4. * SPDX-License-Identifier: BSD-2-Clause
  5. */
  6. #include <AK/AnyOf.h>
  7. #include <AK/ByteBuffer.h>
  8. #include <AK/Find.h>
  9. #include <AK/Function.h>
  10. #include <AK/Memory.h>
  11. #include <AK/StringBuilder.h>
  12. #include <AK/StringView.h>
  13. #include <AK/Vector.h>
  14. #ifndef KERNEL
  15. # include <AK/FlyString.h>
  16. # include <AK/String.h>
  17. #endif
  18. namespace AK {
  19. #ifndef KERNEL
  20. StringView::StringView(String const& string)
  21. : m_characters(string.characters())
  22. , m_length(string.length())
  23. {
  24. }
  25. StringView::StringView(FlyString const& string)
  26. : m_characters(string.characters())
  27. , m_length(string.length())
  28. {
  29. }
  30. #endif
  31. StringView::StringView(ByteBuffer const& buffer)
  32. : m_characters((char const*)buffer.data())
  33. , m_length(buffer.size())
  34. {
  35. }
  36. Vector<StringView> StringView::split_view(char const separator, SplitBehavior split_behavior) const
  37. {
  38. StringView seperator_view { &separator, 1 };
  39. return split_view(seperator_view, split_behavior);
  40. }
  41. Vector<StringView> StringView::split_view(StringView separator, SplitBehavior split_behavior) const
  42. {
  43. Vector<StringView> parts;
  44. for_each_split_view(separator, split_behavior, [&](StringView view) {
  45. parts.append(view);
  46. });
  47. return parts;
  48. }
  49. Vector<StringView> StringView::lines(bool consider_cr) const
  50. {
  51. if (is_empty())
  52. return {};
  53. if (!consider_cr)
  54. return split_view('\n', SplitBehavior::KeepEmpty);
  55. Vector<StringView> v;
  56. size_t substart = 0;
  57. bool last_ch_was_cr = false;
  58. bool split_view = false;
  59. for (size_t i = 0; i < length(); ++i) {
  60. char ch = characters_without_null_termination()[i];
  61. if (ch == '\n') {
  62. split_view = true;
  63. if (last_ch_was_cr) {
  64. substart = i + 1;
  65. split_view = false;
  66. }
  67. }
  68. if (ch == '\r') {
  69. split_view = true;
  70. last_ch_was_cr = true;
  71. } else {
  72. last_ch_was_cr = false;
  73. }
  74. if (split_view) {
  75. size_t sublen = i - substart;
  76. v.append(substring_view(substart, sublen));
  77. substart = i + 1;
  78. }
  79. split_view = false;
  80. }
  81. size_t taillen = length() - substart;
  82. if (taillen != 0)
  83. v.append(substring_view(substart, taillen));
  84. return v;
  85. }
  86. bool StringView::starts_with(char ch) const
  87. {
  88. if (is_empty())
  89. return false;
  90. return ch == characters_without_null_termination()[0];
  91. }
  92. bool StringView::starts_with(StringView str, CaseSensitivity case_sensitivity) const
  93. {
  94. return StringUtils::starts_with(*this, str, case_sensitivity);
  95. }
  96. bool StringView::ends_with(char ch) const
  97. {
  98. if (is_empty())
  99. return false;
  100. return ch == characters_without_null_termination()[length() - 1];
  101. }
  102. bool StringView::ends_with(StringView str, CaseSensitivity case_sensitivity) const
  103. {
  104. return StringUtils::ends_with(*this, str, case_sensitivity);
  105. }
  106. bool StringView::matches(StringView mask, Vector<MaskSpan>& mask_spans, CaseSensitivity case_sensitivity) const
  107. {
  108. return StringUtils::matches(*this, mask, case_sensitivity, &mask_spans);
  109. }
  110. bool StringView::matches(StringView mask, CaseSensitivity case_sensitivity) const
  111. {
  112. return StringUtils::matches(*this, mask, case_sensitivity);
  113. }
  114. bool StringView::contains(char needle) const
  115. {
  116. for (char current : *this) {
  117. if (current == needle)
  118. return true;
  119. }
  120. return false;
  121. }
  122. bool StringView::contains(u32 needle) const
  123. {
  124. // A code point should be at most four UTF-8 bytes, which easily fits into StringBuilder's inline-buffer.
  125. // Therefore, this will not allocate.
  126. StringBuilder needle_builder;
  127. auto result = needle_builder.try_append_code_point(needle);
  128. if (result.is_error()) {
  129. // The needle is invalid, therefore the string does not contain it.
  130. return false;
  131. }
  132. return contains(needle_builder.string_view());
  133. }
  134. bool StringView::contains(StringView needle, CaseSensitivity case_sensitivity) const
  135. {
  136. return StringUtils::contains(*this, needle, case_sensitivity);
  137. }
  138. bool StringView::equals_ignoring_case(StringView other) const
  139. {
  140. return StringUtils::equals_ignoring_case(*this, other);
  141. }
  142. #ifndef KERNEL
  143. String StringView::to_lowercase_string() const
  144. {
  145. return StringImpl::create_lowercased(characters_without_null_termination(), length());
  146. }
  147. String StringView::to_uppercase_string() const
  148. {
  149. return StringImpl::create_uppercased(characters_without_null_termination(), length());
  150. }
  151. String StringView::to_titlecase_string() const
  152. {
  153. return StringUtils::to_titlecase(*this);
  154. }
  155. #endif
  156. StringView StringView::substring_view_starting_from_substring(StringView substring) const
  157. {
  158. char const* remaining_characters = substring.characters_without_null_termination();
  159. VERIFY(remaining_characters >= m_characters);
  160. VERIFY(remaining_characters <= m_characters + m_length);
  161. size_t remaining_length = m_length - (remaining_characters - m_characters);
  162. return { remaining_characters, remaining_length };
  163. }
  164. StringView StringView::substring_view_starting_after_substring(StringView substring) const
  165. {
  166. char const* remaining_characters = substring.characters_without_null_termination() + substring.length();
  167. VERIFY(remaining_characters >= m_characters);
  168. VERIFY(remaining_characters <= m_characters + m_length);
  169. size_t remaining_length = m_length - (remaining_characters - m_characters);
  170. return { remaining_characters, remaining_length };
  171. }
  172. bool StringView::copy_characters_to_buffer(char* buffer, size_t buffer_size) const
  173. {
  174. // We must fit at least the NUL-terminator.
  175. VERIFY(buffer_size > 0);
  176. size_t characters_to_copy = min(m_length, buffer_size - 1);
  177. __builtin_memcpy(buffer, m_characters, characters_to_copy);
  178. buffer[characters_to_copy] = 0;
  179. return characters_to_copy == m_length;
  180. }
  181. template<typename T>
  182. Optional<T> StringView::to_int() const
  183. {
  184. return StringUtils::convert_to_int<T>(*this);
  185. }
  186. template Optional<i8> StringView::to_int() const;
  187. template Optional<i16> StringView::to_int() const;
  188. template Optional<i32> StringView::to_int() const;
  189. template Optional<long> StringView::to_int() const;
  190. template Optional<long long> StringView::to_int() const;
  191. template<typename T>
  192. Optional<T> StringView::to_uint() const
  193. {
  194. return StringUtils::convert_to_uint<T>(*this);
  195. }
  196. template Optional<u8> StringView::to_uint() const;
  197. template Optional<u16> StringView::to_uint() const;
  198. template Optional<u32> StringView::to_uint() const;
  199. template Optional<unsigned long> StringView::to_uint() const;
  200. template Optional<unsigned long long> StringView::to_uint() const;
  201. template Optional<long> StringView::to_uint() const;
  202. template Optional<long long> StringView::to_uint() const;
  203. #ifndef KERNEL
  204. Optional<double> StringView::to_double(TrimWhitespace trim_whitespace) const
  205. {
  206. return StringUtils::convert_to_floating_point<double>(*this, trim_whitespace);
  207. }
  208. Optional<float> StringView::to_float(TrimWhitespace trim_whitespace) const
  209. {
  210. return StringUtils::convert_to_floating_point<float>(*this, trim_whitespace);
  211. }
  212. bool StringView::operator==(String const& string) const
  213. {
  214. return *this == string.view();
  215. }
  216. String StringView::to_string() const { return String { *this }; }
  217. String StringView::replace(StringView needle, StringView replacement, ReplaceMode replace_mode) const
  218. {
  219. return StringUtils::replace(*this, needle, replacement, replace_mode);
  220. }
  221. #endif
  222. Vector<size_t> StringView::find_all(StringView needle) const
  223. {
  224. return StringUtils::find_all(*this, needle);
  225. }
  226. Vector<StringView> StringView::split_view_if(Function<bool(char)> const& predicate, SplitBehavior split_behavior) const
  227. {
  228. if (is_empty())
  229. return {};
  230. Vector<StringView> v;
  231. size_t substart = 0;
  232. bool keep_empty = has_flag(split_behavior, SplitBehavior::KeepEmpty);
  233. for (size_t i = 0; i < length(); ++i) {
  234. char ch = characters_without_null_termination()[i];
  235. if (predicate(ch)) {
  236. size_t sublen = i - substart;
  237. if (sublen != 0 || keep_empty)
  238. v.append(substring_view(substart, sublen));
  239. substart = i + 1;
  240. }
  241. }
  242. size_t taillen = length() - substart;
  243. if (taillen != 0 || keep_empty)
  244. v.append(substring_view(substart, taillen));
  245. return v;
  246. }
  247. }