1 /* Copyright 2015 The TensorFlow Authors. All Rights Reserved. 2 3 Licensed under the Apache License, Version 2.0 (the "License"); 4 you may not use this file except in compliance with the License. 5 You may obtain a copy of the License at 6 7 http://www.apache.org/licenses/LICENSE-2.0 8 9 Unless required by applicable law or agreed to in writing, software 10 distributed under the License is distributed on an "AS IS" BASIS, 11 WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 12 See the License for the specific language governing permissions and 13 limitations under the License. 14 ==============================================================================*/ 15 16 #ifndef TENSORFLOW_LIB_STRINGS_STR_UTIL_H_ 17 #define TENSORFLOW_LIB_STRINGS_STR_UTIL_H_ 18 19 #include <functional> 20 #include <string> 21 #include <vector> 22 #include "tensorflow/core/lib/core/stringpiece.h" 23 #include "tensorflow/core/lib/gtl/array_slice.h" 24 #include "tensorflow/core/lib/strings/strcat.h" 25 #include "tensorflow/core/platform/types.h" 26 27 // Basic string utility routines 28 namespace tensorflow { 29 namespace str_util { 30 31 // Returns a version of 'src' where unprintable characters have been 32 // escaped using C-style escape sequences. 33 string CEscape(StringPiece src); 34 35 // Copies "source" to "dest", rewriting C-style escape sequences -- 36 // '\n', '\r', '\\', '\ooo', etc -- to their ASCII equivalents. 37 // 38 // Errors: Sets the description of the first encountered error in 39 // 'error'. To disable error reporting, set 'error' to NULL. 40 // 41 // NOTE: Does not support \u or \U! 42 bool CUnescape(StringPiece source, string* dest, string* error); 43 44 // Removes any trailing whitespace from "*s". 45 void StripTrailingWhitespace(string* s); 46 47 // Removes leading ascii_isspace() characters. 48 // Returns number of characters removed. 49 size_t RemoveLeadingWhitespace(StringPiece* text); 50 51 // Removes trailing ascii_isspace() characters. 52 // Returns number of characters removed. 53 size_t RemoveTrailingWhitespace(StringPiece* text); 54 55 // Removes leading and trailing ascii_isspace() chars. 56 // Returns number of chars removed. 57 size_t RemoveWhitespaceContext(StringPiece* text); 58 59 // Consume a leading positive integer value. If any digits were 60 // found, store the value of the leading unsigned number in "*val", 61 // advance "*s" past the consumed number, and return true. If 62 // overflow occurred, returns false. Otherwise, returns false. 63 bool ConsumeLeadingDigits(StringPiece* s, uint64* val); 64 65 // Consume a leading token composed of non-whitespace characters only. 66 // If *s starts with a non-zero number of non-whitespace characters, store 67 // them in *val, advance *s past them, and return true. Else return false. 68 bool ConsumeNonWhitespace(StringPiece* s, StringPiece* val); 69 70 // If "*s" starts with "expected", consume it and return true. 71 // Otherwise, return false. 72 bool ConsumePrefix(StringPiece* s, StringPiece expected); 73 74 // If "*s" ends with "expected", remove it and return true. 75 // Otherwise, return false. 76 bool ConsumeSuffix(StringPiece* s, StringPiece expected); 77 78 // Return lower-cased version of s. 79 string Lowercase(StringPiece s); 80 81 // Return upper-cased version of s. 82 string Uppercase(StringPiece s); 83 84 // Converts "^2ILoveYou!" to "i_love_you_". More specifically: 85 // - converts all non-alphanumeric characters to underscores 86 // - replaces each occurrence of a capital letter (except the very 87 // first character and if there is already an '_' before it) with '_' 88 // followed by this letter in lower case 89 // - Skips leading non-alpha characters 90 // This method is useful for producing strings matching "[a-z][a-z0-9_]*" 91 // as required by OpDef.ArgDef.name. The resulting string is either empty or 92 // matches this regex. 93 string ArgDefCase(StringPiece s); 94 95 // Capitalize first character of each word in "*s". "delimiters" is a 96 // set of characters that can be used as word boundaries. 97 void TitlecaseString(string* s, StringPiece delimiters); 98 99 // Replaces the first occurrence (if replace_all is false) or all occurrences 100 // (if replace_all is true) of oldsub in s with newsub. 101 string StringReplace(StringPiece s, StringPiece oldsub, StringPiece newsub, 102 bool replace_all); 103 104 // Join functionality 105 template <typename T> 106 string Join(const T& s, const char* sep); 107 108 // A variant of Join where for each element of "s", f(&dest_string, elem) 109 // is invoked (f is often constructed with a lambda of the form: 110 // [](string* result, ElemType elem) 111 template <typename T, typename Formatter> 112 string Join(const T& s, const char* sep, Formatter f); 113 114 struct AllowEmpty { 115 bool operator()(StringPiece sp) const { return true; } 116 }; 117 struct SkipEmpty { 118 bool operator()(StringPiece sp) const { return !sp.empty(); } 119 }; 120 struct SkipWhitespace { 121 bool operator()(StringPiece sp) const { 122 RemoveTrailingWhitespace(&sp); 123 return !sp.empty(); 124 } 125 }; 126 127 // Split strings using any of the supplied delimiters. For example: 128 // Split("a,b.c,d", ".,") would return {"a", "b", "c", "d"}. 129 std::vector<string> Split(StringPiece text, StringPiece delims); 130 131 template <typename Predicate> 132 std::vector<string> Split(StringPiece text, StringPiece delims, Predicate p); 133 134 // Split "text" at "delim" characters, and parse each component as 135 // an integer. If successful, adds the individual numbers in order 136 // to "*result" and returns true. Otherwise returns false. 137 bool SplitAndParseAsInts(StringPiece text, char delim, 138 std::vector<int32>* result); 139 bool SplitAndParseAsInts(StringPiece text, char delim, 140 std::vector<int64>* result); 141 bool SplitAndParseAsFloats(StringPiece text, char delim, 142 std::vector<float>* result); 143 144 // ------------------------------------------------------------------ 145 // Implementation details below 146 template <typename T> 147 string Join(const T& s, const char* sep) { 148 string result; 149 bool first = true; 150 for (const auto& x : s) { 151 tensorflow::strings::StrAppend(&result, (first ? "" : sep), x); 152 first = false; 153 } 154 return result; 155 } 156 157 template <typename T> 158 class Formatter { 159 public: 160 Formatter(std::function<void(string*, T)> f) : f_(f) {} 161 void operator()(string* out, const T& t) { f_(out, t); } 162 163 private: 164 std::function<void(string*, T)> f_; 165 }; 166 167 template <typename T, typename Formatter> 168 string Join(const T& s, const char* sep, Formatter f) { 169 string result; 170 bool first = true; 171 for (const auto& x : s) { 172 if (!first) { 173 result.append(sep); 174 } 175 f(&result, x); 176 first = false; 177 } 178 return result; 179 } 180 181 inline std::vector<string> Split(StringPiece text, StringPiece delims) { 182 return Split(text, delims, AllowEmpty()); 183 } 184 185 template <typename Predicate> 186 std::vector<string> Split(StringPiece text, StringPiece delims, Predicate p) { 187 std::vector<string> result; 188 size_t token_start = 0; 189 if (!text.empty()) { 190 for (size_t i = 0; i < text.size() + 1; i++) { 191 if ((i == text.size()) || (delims.find(text[i]) != StringPiece::npos)) { 192 StringPiece token(text.data() + token_start, i - token_start); 193 if (p(token)) { 194 result.push_back(token.ToString()); 195 } 196 token_start = i + 1; 197 } 198 } 199 } 200 return result; 201 } 202 203 inline std::vector<string> Split(StringPiece text, char delim) { 204 return Split(text, StringPiece(&delim, 1)); 205 } 206 207 template <typename Predicate> 208 std::vector<string> Split(StringPiece text, char delims, Predicate p) { 209 return Split(text, StringPiece(&delims, 1), p); 210 } 211 212 } // namespace str_util 213 } // namespace tensorflow 214 215 #endif // TENSORFLOW_LIB_STRINGS_STR_UTIL_H_ 216