MCPcopy Create free account
hub / github.com/p-ranav/tabulate / byte_offset_for_width

Function byte_offset_for_width

include/tabulate/utf8.hpp:192–267  ·  view source on GitHub ↗

Returns the number of bytes from the start of `text` whose rendered width does not exceed max_width, without ever splitting a multi-byte character (or ANSI escape sequence) in half. Always advances past at least one character when text is non-empty, even if that character alone is wider than max_width, so callers that force-split long words always make progress. Mirrors get_sequence_length()'s wid

Source from the content-addressed store, hash-verified

190// progress. Mirrors get_sequence_length()'s width accounting exactly, so a
191// prefix of this many bytes is guaranteed to measure as <= max_width.
192inline size_t byte_offset_for_width(const std::string &text, const std::string &locale,
193 bool is_multi_byte_character_support_enabled,
194 size_t max_width) {
195 if (!is_multi_byte_character_support_enabled)
196 return (std::min)(max_width, text.size());
197
198#if defined(_WIN32) || defined(_WIN64)
199 (void)locale;
200 size_t consumed_width = 0;
201 bool first_character = true;
202 const char *p = text.c_str();
203 while (*p != '\0') {
204 if (static_cast<unsigned char>(*p) == 0x1b) {
205 p = skip_ansi_escape_sequence(p);
206 continue;
207 }
208 const char *character_start = p;
209 ++p;
210 while (*p != '\0' && (static_cast<unsigned char>(*p) & 0xC0) == 0x80)
211 ++p; // UTF-8 continuation byte: part of the same character
212 if (!first_character && consumed_width + 1 > max_width)
213 return static_cast<size_t>(character_start - text.c_str());
214 consumed_width += 1;
215 first_character = false;
216 }
217 return text.size();
218#elif defined(__unix__) || defined(__unix) || defined(__APPLE__)
219 const char *previous_locale = std::setlocale(LC_CTYPE, nullptr);
220 const std::string saved_locale = previous_locale != nullptr ? previous_locale : "";
221 std::setlocale(LC_CTYPE, locale.c_str());
222
223 const char *p = text.c_str();
224 std::mbstate_t state = std::mbstate_t();
225 size_t consumed_width = 0;
226 bool first_character = true;
227 size_t result = text.size();
228
229 while (*p != '\0') {
230 if (static_cast<unsigned char>(*p) == 0x1b) {
231 p = skip_ansi_escape_sequence(p);
232 continue;
233 }
234
235 const char *character_start = p;
236 wchar_t wide_character;
237 size_t length = std::mbrtowc(&wide_character, p, MB_CUR_MAX, &state);
238
239 if (length == static_cast<size_t>(-2))
240 break; // truncated multi-byte character at the end of the string
241
242 if (length == static_cast<size_t>(-1)) {
243 state = std::mbstate_t();
244 length = 1; // invalid byte: treat as width 1 and resynchronize
245 } else if (length == 0) {
246 length = 1; // embedded null: treat as a 1-byte character
247 }
248
249 int character_width = wcwidth(wide_character);

Callers 1

word_wrapMethod · 0.70

Calls 3

sizeMethod · 0.45
emptyMethod · 0.45

Tested by

no test coverage detected