Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 1 | //===-- RegularExpression.cpp -----------------------------------*- C++ -*-===// |
| 2 | // |
| 3 | // The LLVM Compiler Infrastructure |
| 4 | // |
| 5 | // This file is distributed under the University of Illinois Open Source |
| 6 | // License. See LICENSE.TXT for details. |
| 7 | // |
| 8 | //===----------------------------------------------------------------------===// |
| 9 | |
| 10 | #include "lldb/Core/RegularExpression.h" |
Greg Clayton | 1b3815c | 2013-01-30 00:18:29 +0000 | [diff] [blame] | 11 | #include "llvm/ADT/StringRef.h" |
Eli Friedman | 8896697 | 2010-06-09 08:50:27 +0000 | [diff] [blame] | 12 | #include <string.h> |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 13 | |
| 14 | using namespace lldb_private; |
| 15 | |
| 16 | //---------------------------------------------------------------------- |
| 17 | // Default constructor |
| 18 | //---------------------------------------------------------------------- |
| 19 | RegularExpression::RegularExpression() : |
| 20 | m_re(), |
| 21 | m_comp_err (1), |
| 22 | m_preg(), |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 23 | m_compile_flags(REG_EXTENDED) |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 24 | { |
| 25 | memset(&m_preg,0,sizeof(m_preg)); |
| 26 | } |
| 27 | |
| 28 | //---------------------------------------------------------------------- |
| 29 | // Constructor that compiles "re" using "flags" and stores the |
| 30 | // resulting compiled regular expression into this object. |
| 31 | //---------------------------------------------------------------------- |
| 32 | RegularExpression::RegularExpression(const char* re, int flags) : |
| 33 | m_re(), |
| 34 | m_comp_err (1), |
Daniel Dunbar | a08823f | 2011-10-31 22:50:49 +0000 | [diff] [blame] | 35 | m_preg(), |
| 36 | m_compile_flags(flags) |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 37 | { |
| 38 | memset(&m_preg,0,sizeof(m_preg)); |
Jim Ingham | 969795f | 2011-09-21 01:17:13 +0000 | [diff] [blame] | 39 | Compile(re); |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 40 | } |
| 41 | |
| 42 | //---------------------------------------------------------------------- |
Jim Ingham | 969795f | 2011-09-21 01:17:13 +0000 | [diff] [blame] | 43 | // Constructor that compiles "re" using "flags" and stores the |
| 44 | // resulting compiled regular expression into this object. |
| 45 | //---------------------------------------------------------------------- |
| 46 | RegularExpression::RegularExpression(const char* re) : |
| 47 | m_re(), |
| 48 | m_comp_err (1), |
Daniel Dunbar | a08823f | 2011-10-31 22:50:49 +0000 | [diff] [blame] | 49 | m_preg(), |
| 50 | m_compile_flags(REG_EXTENDED) |
Jim Ingham | 969795f | 2011-09-21 01:17:13 +0000 | [diff] [blame] | 51 | { |
| 52 | memset(&m_preg,0,sizeof(m_preg)); |
| 53 | Compile(re); |
| 54 | } |
| 55 | |
| 56 | RegularExpression::RegularExpression(const RegularExpression &rhs) |
| 57 | { |
| 58 | memset(&m_preg,0,sizeof(m_preg)); |
| 59 | Compile(rhs.GetText(), rhs.GetCompileFlags()); |
| 60 | } |
| 61 | |
| 62 | const RegularExpression & |
| 63 | RegularExpression::operator= (const RegularExpression &rhs) |
| 64 | { |
| 65 | if (&rhs != this) |
| 66 | { |
| 67 | Compile (rhs.GetText(), rhs.GetCompileFlags()); |
| 68 | } |
| 69 | return *this; |
| 70 | } |
| 71 | //---------------------------------------------------------------------- |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 72 | // Destructor |
| 73 | // |
| 74 | // Any previosuly compiled regular expression contained in this |
| 75 | // object will be freed. |
| 76 | //---------------------------------------------------------------------- |
| 77 | RegularExpression::~RegularExpression() |
| 78 | { |
| 79 | Free(); |
| 80 | } |
| 81 | |
| 82 | //---------------------------------------------------------------------- |
| 83 | // Compile a regular expression using the supplied regular |
| 84 | // expression text and flags. The compied regular expression lives |
| 85 | // in this object so that it can be readily used for regular |
| 86 | // expression matches. Execute() can be called after the regular |
| 87 | // expression is compiled. Any previosuly compiled regular |
| 88 | // expression contained in this object will be freed. |
| 89 | // |
| 90 | // RETURNS |
| 91 | // True of the refular expression compiles successfully, false |
| 92 | // otherwise. |
| 93 | //---------------------------------------------------------------------- |
| 94 | bool |
Jim Ingham | 969795f | 2011-09-21 01:17:13 +0000 | [diff] [blame] | 95 | RegularExpression::Compile(const char* re) |
| 96 | { |
| 97 | return Compile (re, m_compile_flags); |
| 98 | } |
| 99 | |
| 100 | bool |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 101 | RegularExpression::Compile(const char* re, int flags) |
| 102 | { |
| 103 | Free(); |
Jim Ingham | 969795f | 2011-09-21 01:17:13 +0000 | [diff] [blame] | 104 | m_compile_flags = flags; |
| 105 | |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 106 | if (re && re[0]) |
| 107 | { |
| 108 | m_re = re; |
| 109 | m_comp_err = ::regcomp (&m_preg, re, flags); |
| 110 | } |
| 111 | else |
| 112 | { |
| 113 | // No valid regular expression |
| 114 | m_comp_err = 1; |
| 115 | } |
| 116 | |
| 117 | return m_comp_err == 0; |
| 118 | } |
| 119 | |
| 120 | //---------------------------------------------------------------------- |
| 121 | // Execute a regular expression match using the compiled regular |
| 122 | // expression that is already in this object against the match |
| 123 | // string "s". If any parens are used for regular expression |
| 124 | // matches "match_count" should indicate the number of regmatch_t |
| 125 | // values that are present in "match_ptr". The regular expression |
| 126 | // will be executed using the "execute_flags". |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 127 | //--------------------------------------------------------------------- |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 128 | bool |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 129 | RegularExpression::Execute(const char* s, Match *match, int execute_flags) const |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 130 | { |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 131 | int err = 1; |
Greg Clayton | cc24775 | 2013-05-23 20:27:15 +0000 | [diff] [blame^] | 132 | if (s != NULL && m_comp_err == 0) |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 133 | { |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 134 | if (match) |
Greg Clayton | 1b3815c | 2013-01-30 00:18:29 +0000 | [diff] [blame] | 135 | { |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 136 | err = ::regexec (&m_preg, |
| 137 | s, |
| 138 | match->GetSize(), |
| 139 | match->GetData(), |
| 140 | execute_flags); |
| 141 | } |
| 142 | else |
| 143 | { |
| 144 | err = ::regexec (&m_preg, |
| 145 | s, |
| 146 | 0, |
| 147 | NULL, |
| 148 | execute_flags); |
Greg Clayton | 1b3815c | 2013-01-30 00:18:29 +0000 | [diff] [blame] | 149 | } |
| 150 | } |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 151 | |
| 152 | if (err != 0) |
| 153 | { |
| 154 | // The regular expression didn't compile, so clear the matches |
| 155 | if (match) |
| 156 | match->Clear(); |
| 157 | return false; |
| 158 | } |
| 159 | return true; |
Greg Clayton | 1b3815c | 2013-01-30 00:18:29 +0000 | [diff] [blame] | 160 | } |
| 161 | |
| 162 | bool |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 163 | RegularExpression::Match::GetMatchAtIndex (const char* s, uint32_t idx, std::string& match_str) const |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 164 | { |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 165 | if (idx < m_matches.size()) |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 166 | { |
Greg Clayton | 6ad07dd | 2010-12-19 03:41:24 +0000 | [diff] [blame] | 167 | if (m_matches[idx].rm_eo == m_matches[idx].rm_so) |
| 168 | { |
| 169 | // Matched the empty string... |
| 170 | match_str.clear(); |
| 171 | return true; |
| 172 | } |
| 173 | else if (m_matches[idx].rm_eo > m_matches[idx].rm_so) |
| 174 | { |
| 175 | match_str.assign (s + m_matches[idx].rm_so, |
| 176 | m_matches[idx].rm_eo - m_matches[idx].rm_so); |
| 177 | return true; |
| 178 | } |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 179 | } |
| 180 | return false; |
| 181 | } |
| 182 | |
Greg Clayton | 1b3815c | 2013-01-30 00:18:29 +0000 | [diff] [blame] | 183 | bool |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 184 | RegularExpression::Match::GetMatchAtIndex (const char* s, uint32_t idx, llvm::StringRef& match_str) const |
Greg Clayton | 1b3815c | 2013-01-30 00:18:29 +0000 | [diff] [blame] | 185 | { |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 186 | if (idx < m_matches.size()) |
Greg Clayton | 1b3815c | 2013-01-30 00:18:29 +0000 | [diff] [blame] | 187 | { |
| 188 | if (m_matches[idx].rm_eo == m_matches[idx].rm_so) |
| 189 | { |
| 190 | // Matched the empty string... |
| 191 | match_str = llvm::StringRef(); |
| 192 | return true; |
| 193 | } |
| 194 | else if (m_matches[idx].rm_eo > m_matches[idx].rm_so) |
| 195 | { |
| 196 | match_str = llvm::StringRef (s + m_matches[idx].rm_so, m_matches[idx].rm_eo - m_matches[idx].rm_so); |
| 197 | return true; |
| 198 | } |
| 199 | } |
| 200 | return false; |
| 201 | } |
| 202 | |
| 203 | bool |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 204 | RegularExpression::Match::GetMatchSpanningIndices (const char* s, uint32_t idx1, uint32_t idx2, llvm::StringRef& match_str) const |
Greg Clayton | 1b3815c | 2013-01-30 00:18:29 +0000 | [diff] [blame] | 205 | { |
Greg Clayton | bc43cab | 2013-04-03 21:37:16 +0000 | [diff] [blame] | 206 | if (idx1 < m_matches.size() && idx2 < m_matches.size()) |
Greg Clayton | 1b3815c | 2013-01-30 00:18:29 +0000 | [diff] [blame] | 207 | { |
| 208 | if (m_matches[idx1].rm_so == m_matches[idx2].rm_eo) |
| 209 | { |
| 210 | // Matched the empty string... |
| 211 | match_str = llvm::StringRef(); |
| 212 | return true; |
| 213 | } |
| 214 | else if (m_matches[idx1].rm_so < m_matches[idx2].rm_eo) |
| 215 | { |
| 216 | match_str = llvm::StringRef (s + m_matches[idx1].rm_so, m_matches[idx2].rm_eo - m_matches[idx1].rm_so); |
| 217 | return true; |
| 218 | } |
| 219 | } |
| 220 | return false; |
| 221 | } |
| 222 | |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 223 | |
| 224 | //---------------------------------------------------------------------- |
| 225 | // Returns true if the regular expression compiled and is ready |
| 226 | // for execution. |
| 227 | //---------------------------------------------------------------------- |
| 228 | bool |
| 229 | RegularExpression::IsValid () const |
| 230 | { |
| 231 | return m_comp_err == 0; |
| 232 | } |
| 233 | |
| 234 | //---------------------------------------------------------------------- |
| 235 | // Returns the text that was used to compile the current regular |
| 236 | // expression. |
| 237 | //---------------------------------------------------------------------- |
| 238 | const char* |
| 239 | RegularExpression::GetText () const |
| 240 | { |
Greg Clayton | 67cc063 | 2012-08-22 17:17:09 +0000 | [diff] [blame] | 241 | if (m_re.empty()) |
| 242 | return NULL; |
Chris Lattner | 30fdc8d | 2010-06-08 16:52:24 +0000 | [diff] [blame] | 243 | return m_re.c_str(); |
| 244 | } |
| 245 | |
| 246 | //---------------------------------------------------------------------- |
| 247 | // Free any contained compiled regular expressions. |
| 248 | //---------------------------------------------------------------------- |
| 249 | void |
| 250 | RegularExpression::Free() |
| 251 | { |
| 252 | if (m_comp_err == 0) |
| 253 | { |
| 254 | m_re.clear(); |
| 255 | regfree(&m_preg); |
| 256 | // Set a compile error since we no longer have a valid regex |
| 257 | m_comp_err = 1; |
| 258 | } |
| 259 | } |
Greg Clayton | 4674702 | 2010-10-10 23:55:27 +0000 | [diff] [blame] | 260 | |
| 261 | size_t |
| 262 | RegularExpression::GetErrorAsCString (char *err_str, size_t err_str_max_len) const |
| 263 | { |
| 264 | if (m_comp_err == 0) |
| 265 | { |
| 266 | if (err_str && err_str_max_len) |
| 267 | *err_str = '\0'; |
| 268 | return 0; |
| 269 | } |
| 270 | |
| 271 | return ::regerror (m_comp_err, &m_preg, err_str, err_str_max_len); |
| 272 | } |
| 273 | |
Enrico Granata | 0a3958e | 2011-07-02 00:25:22 +0000 | [diff] [blame] | 274 | bool |
| 275 | RegularExpression::operator < (const RegularExpression& rhs) const |
| 276 | { |
| 277 | return (m_re < rhs.m_re); |
| 278 | } |
Greg Clayton | 4674702 | 2010-10-10 23:55:27 +0000 | [diff] [blame] | 279 | |