C++ Utilities  4.12.1
Useful C++ classes and routines such as argument parser, IO and conversion utilities
stringconversion.h
Go to the documentation of this file.
1 #ifndef CONVERSION_UTILITIES_STRINGCONVERSION_H
2 #define CONVERSION_UTILITIES_STRINGCONVERSION_H
3 
4 #include "./binaryconversion.h"
6 
7 #include "../misc/traits.h"
8 
9 #include <cstring>
10 #include <initializer_list>
11 #include <iomanip>
12 #include <list>
13 #include <memory>
14 #include <sstream>
15 #include <string>
16 #include <vector>
17 
18 namespace ConversionUtilities {
19 
28  void operator()(char *stringData)
29  {
30  std::free(stringData);
31  }
32 };
33 
37 typedef std::pair<std::unique_ptr<char[], StringDataDeleter>, std::size_t> StringData;
38 //typedef std::pair<std::unique_ptr<char>, std::size_t> StringData; // might work too
39 
41  const char *fromCharset, const char *toCharset, const char *inputBuffer, std::size_t inputBufferSize, float outputBufferSizeFactor = 1.0f);
42 CPP_UTILITIES_EXPORT StringData convertUtf8ToUtf16LE(const char *inputBuffer, std::size_t inputBufferSize);
43 CPP_UTILITIES_EXPORT StringData convertUtf16LEToUtf8(const char *inputBuffer, std::size_t inputBufferSize);
44 CPP_UTILITIES_EXPORT StringData convertUtf8ToUtf16BE(const char *inputBuffer, std::size_t inputBufferSize);
45 CPP_UTILITIES_EXPORT StringData convertUtf16BEToUtf8(const char *inputBuffer, std::size_t inputBufferSize);
46 CPP_UTILITIES_EXPORT StringData convertLatin1ToUtf8(const char *inputBuffer, std::size_t inputBufferSize);
47 CPP_UTILITIES_EXPORT StringData convertUtf8ToLatin1(const char *inputBuffer, std::size_t inputBufferSize);
48 
49 CPP_UTILITIES_EXPORT void truncateString(std::string &str, char terminationChar = '\0');
50 
64 template <class Container = std::initializer_list<std::string>>
65 typename Container::value_type joinStrings(const Container &strings,
66  const typename Container::value_type &delimiter = typename Container::value_type(), bool omitEmpty = false,
67  const typename Container::value_type &leftClosure = typename Container::value_type(),
68  const typename Container::value_type &rightClosure = typename Container::value_type())
69 {
70  typename Container::value_type res;
71  if (strings.size()) {
72  size_t entries = 0, size = 0;
73  for (const auto &str : strings) {
74  if (!omitEmpty || !str.empty()) {
75  size += str.size();
76  ++entries;
77  }
78  }
79  if (entries) {
80  size += (entries * leftClosure.size()) + (entries * rightClosure.size()) + ((entries - 1) * delimiter.size());
81  res.reserve(size);
82  for (const auto &str : strings) {
83  if (!omitEmpty || !str.empty()) {
84  if (!res.empty()) {
85  res.append(delimiter);
86  }
87  res.append(leftClosure);
88  res.append(str);
89  res.append(rightClosure);
90  }
91  }
92  }
93  }
94  return res;
95 }
96 
100 template <class Container = std::initializer_list<std::string>> inline std::vector<std::string> toMultiline(const Container &arrayOfLines)
101 {
102  return joinStrings(arrayOfLines, "\n", false);
103 }
104 
108 enum class EmptyPartsTreat {
109  Keep,
110  Omit,
111  Merge
112 };
113 
123 template <class Container = std::list<std::string>>
124 Container splitString(const typename Container::value_type &string, const typename Container::value_type &delimiter,
125  EmptyPartsTreat emptyPartsRole = EmptyPartsTreat::Keep, int maxParts = -1)
126 {
127  --maxParts;
128  Container res;
129  bool merge = false;
130  for (typename Container::value_type::size_type i = 0, end = string.size(), delimPos; i < end; i = delimPos + delimiter.size()) {
131  delimPos = string.find(delimiter, i);
132  if (!merge && maxParts >= 0 && res.size() == static_cast<typename Container::value_type::size_type>(maxParts)) {
133  if (delimPos == i && emptyPartsRole == EmptyPartsTreat::Merge) {
134  if (!res.empty()) {
135  merge = true;
136  continue;
137  }
138  }
139  delimPos = Container::value_type::npos;
140  }
141  if (delimPos == Container::value_type::npos) {
142  delimPos = string.size();
143  }
144  if (emptyPartsRole == EmptyPartsTreat::Keep || i != delimPos) {
145  if (merge) {
146  res.back().append(delimiter);
147  res.back().append(string.substr(i, delimPos - i));
148  merge = false;
149  } else {
150  res.emplace_back(string.substr(i, delimPos - i));
151  }
152  } else if (emptyPartsRole == EmptyPartsTreat::Merge) {
153  if (!res.empty()) {
154  merge = true;
155  }
156  }
157  }
158  return res;
159 }
160 
164 template <class Container = std::vector<std::string>> inline std::vector<std::string> toArrayOfLines(const std::string &multilineString)
165 {
166  return splitString<Container>(multilineString, "\n", EmptyPartsTreat::Keep);
167 }
168 
172 template <typename StringType> bool startsWith(const StringType &str, const StringType &phrase)
173 {
174  if (str.size() < phrase.size()) {
175  return false;
176  }
177  for (auto stri = str.cbegin(), strend = str.cend(), phrasei = phrase.cbegin(), phraseend = phrase.cend(); stri != strend; ++stri, ++phrasei) {
178  if (phrasei == phraseend) {
179  return true;
180  } else if (*stri != *phrasei) {
181  return false;
182  }
183  }
184  return false;
185 }
186 
190 template <typename StringType> bool startsWith(const StringType &str, const typename StringType::value_type *phrase)
191 {
192  for (auto stri = str.cbegin(), strend = str.cend(); stri != strend; ++stri, ++phrase) {
193  if (!*phrase) {
194  return true;
195  } else if (*stri != *phrase) {
196  return false;
197  }
198  }
199  return false;
200 }
201 
206 template <typename StringType> bool containsSubstrings(const StringType &str, std::initializer_list<StringType> substrings)
207 {
208  typename StringType::size_type currentPos = 0;
209  for (const auto &substr : substrings) {
210  if ((currentPos = str.find(substr, currentPos)) == StringType::npos) {
211  return false;
212  }
213  currentPos += substr.size();
214  }
215  return true;
216 }
217 
222 template <typename StringType>
223 bool containsSubstrings(const StringType &str, std::initializer_list<const typename StringType::value_type *> substrings)
224 {
225  typename StringType::size_type currentPos = 0;
226  for (const auto *substr : substrings) {
227  if ((currentPos = str.find(substr, currentPos)) == StringType::npos) {
228  return false;
229  }
230  currentPos += std::strlen(substr);
231  }
232  return true;
233 }
234 
238 template <typename StringType> void findAndReplace(StringType &str, const StringType &find, const StringType &replace)
239 {
240  for (typename StringType::size_type i = 0; (i = str.find(find, i)) != StringType::npos; i += replace.size()) {
241  str.replace(i, find.size(), replace);
242  }
243 }
244 
251 template <typename CharType> constexpr CharType digitToChar(CharType digit)
252 {
253  return digit <= 9 ? (digit + '0') : (digit + 'A' - 10);
254 }
255 
262 template <typename IntegralType, class StringType = std::string,
264 StringType numberToString(IntegralType number, typename StringType::value_type base = 10)
265 {
266  std::size_t resSize = 0;
267  for (auto n = number; n; n /= base, ++resSize)
268  ;
269  StringType res;
270  res.reserve(resSize);
271  do {
272  res.insert(res.begin(), digitToChar<typename StringType::value_type>(number % base));
273  number /= base;
274  } while (number);
275  return res;
276 }
277 
284 template <typename IntegralType, class StringType = std::string, Traits::EnableIf<std::is_integral<IntegralType>, std::is_signed<IntegralType>>...>
285 StringType numberToString(IntegralType number, typename StringType::value_type base = 10)
286 {
287  const bool negative = number < 0;
288  std::size_t resSize;
289  if (negative) {
290  number = -number, resSize = 1;
291  } else {
292  resSize = 0;
293  }
294  for (auto n = number; n; n /= base, ++resSize)
295  ;
296  StringType res;
297  res.reserve(resSize);
298  do {
299  res.insert(res.begin(), digitToChar<typename StringType::value_type>(number % base));
300  number /= base;
301  } while (number);
302  if (negative) {
303  res.insert(res.begin(), '-');
304  }
305  return res;
306 }
307 
316 template <typename FloatingType, class StringType = std::string, Traits::EnableIf<std::is_floating_point<FloatingType>>...>
317 StringType numberToString(FloatingType number, typename StringType::value_type base = 10)
318 {
319  std::basic_stringstream<typename StringType::value_type> ss;
320  ss << std::setbase(base) << number;
321  return ss.str();
322 }
323 
328 template <typename CharType> CharType charToDigit(CharType character, CharType base)
329 {
330  CharType res;
331  if (character >= '0' && character <= '9') {
332  res = character - '0';
333  } else if (character >= 'a' && character <= 'z') {
334  res = character - 'a' + 10;
335  } else if (character >= 'A' && character <= 'Z') {
336  res = character - 'A' + 10;
337  } else {
338  throw ConversionException("The string is no valid number");
339  }
340  if (res >= base) {
341  throw ConversionException("The string is no valid number");
342  }
343  return res;
344 }
345 
353 template <typename IntegralType, class StringType, Traits::EnableIf<std::is_integral<IntegralType>, Traits::Not<std::is_signed<IntegralType>>>...>
354 IntegralType stringToNumber(const StringType &string, typename StringType::value_type base = 10)
355 {
356  IntegralType result = 0;
357  for (const auto &c : string) {
358  if (c == ' ') {
359  continue;
360  }
361  result *= base;
362  result += charToDigit<typename StringType::value_type>(c, base);
363  }
364  return result;
365 }
366 
374 template <typename IntegralType, class StringType, Traits::EnableIf<std::is_integral<IntegralType>, std::is_signed<IntegralType>>...>
375 IntegralType stringToNumber(const StringType &string, typename StringType::value_type base = 10)
376 {
377  auto i = string.begin();
378  auto end = string.end();
379  for (; i != end && *i == ' '; ++i)
380  ;
381  if (i == end) {
382  return 0;
383  }
384  const bool negative = (*i == '-');
385  if (negative) {
386  ++i;
387  }
388  IntegralType result = 0;
389  for (; i != end; ++i) {
390  if (*i == ' ') {
391  continue;
392  }
393  result *= base;
394  result += charToDigit<typename StringType::value_type>(*i, base);
395  }
396  return negative ? -result : result;
397 }
398 
408 template <typename FloatingType, class StringType, Traits::EnableIf<std::is_floating_point<FloatingType>>...>
409 FloatingType stringToNumber(const StringType &string, typename StringType::value_type base = 10)
410 {
411  std::basic_stringstream<typename StringType::value_type> ss;
412  ss << std::setbase(base) << string;
413  FloatingType result;
414  if ((ss >> result) && ss.eof()) {
415  return result;
416  } else {
417  throw ConversionException("The string is no valid number.");
418  }
419 }
420 
428 template <typename IntegralType, class CharType, Traits::EnableIf<std::is_integral<IntegralType>, Traits::Not<std::is_signed<IntegralType>>>...>
429 IntegralType stringToNumber(const CharType *string, unsigned char base = 10)
430 {
431  IntegralType result = 0;
432  for (; *string; ++string) {
433  if (*string == ' ') {
434  continue;
435  }
436  result *= base;
437  result += charToDigit<CharType>(*string, base);
438  }
439  return result;
440 }
441 
449 template <typename IntegralType, class CharType, Traits::EnableIf<std::is_integral<IntegralType>, Traits::Not<std::is_signed<IntegralType>>>...>
450 IntegralType bufferToNumber(const CharType *string, std::size_t size, unsigned char base = 10)
451 {
452  IntegralType result = 0;
453  for (const CharType *end = string + size; string != end; ++string) {
454  if (*string == ' ') {
455  continue;
456  }
457  result *= base;
458  result += charToDigit<CharType>(*string, base);
459  }
460  return result;
461 }
462 
470 template <typename IntegralType, class CharType, Traits::EnableIf<std::is_integral<IntegralType>, std::is_signed<IntegralType>>...>
471 IntegralType stringToNumber(const CharType *string, unsigned char base = 10)
472 {
473  if (!*string) {
474  return 0;
475  }
476  for (; *string && *string == ' '; ++string)
477  ;
478  if (!*string) {
479  return 0;
480  }
481  const bool negative = (*string == '-');
482  if (negative) {
483  ++string;
484  }
485  IntegralType result = 0;
486  for (; *string; ++string) {
487  if (*string == ' ') {
488  continue;
489  }
490  result *= base;
491  result += charToDigit<CharType>(*string, base);
492  }
493  return negative ? -result : result;
494 }
495 
503 template <typename IntegralType, class CharType, Traits::EnableIf<std::is_integral<IntegralType>, std::is_signed<IntegralType>>...>
504 IntegralType bufferToNumber(const CharType *string, std::size_t size, unsigned char base = 10)
505 {
506  if (!size) {
507  return 0;
508  }
509  const CharType *end = string + size;
510  for (; string != end && *string == ' '; ++string)
511  ;
512  if (string == end) {
513  return 0;
514  }
515  const bool negative = (*string == '-');
516  if (negative) {
517  ++string;
518  }
519  IntegralType result = 0;
520  for (; string != end; ++string) {
521  if (*string == ' ') {
522  continue;
523  }
524  result *= base;
525  result += charToDigit<CharType>(*string, base);
526  }
527  return negative ? -result : result;
528 }
529 
539 template <typename T> std::string interpretIntegerAsString(T integer, int startOffset = 0)
540 {
541  char buffer[sizeof(T)];
542  ConversionUtilities::BE::getBytes(integer, buffer);
543  return std::string(buffer + startOffset, sizeof(T) - startOffset);
544 }
545 
546 CPP_UTILITIES_EXPORT std::string dataSizeToString(uint64 sizeInByte, bool includeByte = false);
547 CPP_UTILITIES_EXPORT std::string bitrateToString(double speedInKbitsPerSecond, bool useByteInsteadOfBits = false);
548 CPP_UTILITIES_EXPORT std::string encodeBase64(const byte *data, uint32 dataSize);
549 CPP_UTILITIES_EXPORT std::pair<std::unique_ptr<byte[]>, uint32> decodeBase64(const char *encodedStr, const uint32 strSize);
550 } // namespace ConversionUtilities
551 
552 #endif // CONVERSION_UTILITIES_STRINGCONVERSION_H
CPP_UTILITIES_EXPORT StringData convertUtf8ToUtf16LE(const char *inputBuffer, std::size_t inputBufferSize)
Converts the specified UTF-8 string to UTF-16 (little-endian).
bool startsWith(const StringType &str, const StringType &phrase)
Returns whether str starts with phrase.
void operator()(char *stringData)
Deletes the specified stringData with std::free(), because the memory has been allocated using std::m...
CPP_UTILITIES_EXPORT std::string encodeBase64(const byte *data, uint32 dataSize)
Encodes the specified data to Base64.
IntegralType stringToNumber(const StringType &string, typename StringType::value_type base=10)
Converts the given string to an unsigned number assuming string uses the specified base...
bool containsSubstrings(const StringType &str, std::initializer_list< StringType > substrings)
Returns whether str contains the specified substrings.
The ConversionException class is thrown by the various conversion functions of this library when a co...
std::vector< std::string > toArrayOfLines(const std::string &multilineString)
Converts the specified multilineString to an array of lines.
CPP_UTILITIES_EXPORT StringData convertUtf16BEToUtf8(const char *inputBuffer, std::size_t inputBufferSize)
Converts the specified UTF-16 (big-endian) string to UTF-8.
CPP_UTILITIES_EXPORT void truncateString(std::string &str, char terminationChar='\0')
Truncates all characters after the first occurrence of the specified terminationChar and the terminat...
StringType numberToString(IntegralType number, typename StringType::value_type base=10)
Converts the given number to its equivalent string representation using the specified base...
std::uint64_t uint64
unsigned 64-bit integer
Definition: types.h:49
The StringDataDeleter struct deletes the data of a StringData instance.
typename std::enable_if< All< Condition... >::value, Detail::Enabler >::type EnableIf
Definition: traits.h:33
std::string interpretIntegerAsString(T integer, int startOffset=0)
Interprets the given integer at the specified position as std::string using the specified byte order...
CPP_UTILITIES_EXPORT StringData convertUtf8ToUtf16BE(const char *inputBuffer, std::size_t inputBufferSize)
Converts the specified UTF-8 string to UTF-16 (big-endian).
EmptyPartsTreat
Specifies the role of empty parts when splitting strings.
CPP_UTILITIES_EXPORT StringData convertUtf8ToLatin1(const char *inputBuffer, std::size_t inputBufferSize)
Converts the specified UTF-8 string to Latin-1.
void findAndReplace(StringType &str, const StringType &find, const StringType &replace)
Replaces all occurences of find with relpace in the specified str.
Contains several functions providing conversions between different data types.
std::uint32_t uint32
unsigned 32-bit integer
Definition: types.h:44
std::pair< std::unique_ptr< char[], StringDataDeleter >, std::size_t > StringData
Type used to return string encoding conversion result.
IntegralType bufferToNumber(const CharType *string, std::size_t size, unsigned char base=10)
Converts the given string of size characters to an unsigned numeric value using the specified base...
CPP_UTILITIES_EXPORT StringData convertUtf16LEToUtf8(const char *inputBuffer, std::size_t inputBufferSize)
Converts the specified UTF-16 (little-endian) string to UTF-8.
CharType charToDigit(CharType character, CharType base)
Returns number/digit of the specified character representation using the specified base...
Container::value_type joinStrings(const Container &strings, const typename Container::value_type &delimiter=typename Container::value_type(), bool omitEmpty=false, const typename Container::value_type &leftClosure=typename Container::value_type(), const typename Container::value_type &rightClosure=typename Container::value_type())
Joins the given strings using the specified delimiter.
CPP_UTILITIES_EXPORT StringData convertString(const char *fromCharset, const char *toCharset, const char *inputBuffer, std::size_t inputBufferSize, float outputBufferSizeFactor=1.0f)
Converts the specified string from one character set to another.
std::uint8_t byte
unsigned byte
Definition: types.h:14
CPP_UTILITIES_EXPORT StringData convertLatin1ToUtf8(const char *inputBuffer, std::size_t inputBufferSize)
Converts the specified Latin-1 string to UTF-8.
std::vector< std::string > toMultiline(const Container &arrayOfLines)
Converts the specified arrayOfLines to a multiline string.
Container splitString(const typename Container::value_type &string, const typename Container::value_type &delimiter, EmptyPartsTreat emptyPartsRole=EmptyPartsTreat::Keep, int maxParts=-1)
Splits the given string at the specified delimiter.
CPP_UTILITIES_EXPORT std::string dataSizeToString(uint64 sizeInByte, bool includeByte=false)
Converts the specified data size in byte to its equivalent std::string representation.
#define CPP_UTILITIES_EXPORT
Marks the symbol to be exported by the c++utilities library.
CPP_UTILITIES_EXPORT std::pair< std::unique_ptr< byte[]>, uint32 > decodeBase64(const char *encodedStr, const uint32 strSize)
Decodes the specified Base64 encoded string.
constexpr CharType digitToChar(CharType digit)
Returns the character representation of the specified digit.
CPP_UTILITIES_EXPORT std::string bitrateToString(double speedInKbitsPerSecond, bool useByteInsteadOfBits=false)
Converts the specified bitrate in kbit/s to its equivalent std::string representation.