| 1 | // ================================================================= // |
|---|
| 2 | // // |
|---|
| 3 | // File : arb_match.cxx // |
|---|
| 4 | // Purpose : POSIX ERE // |
|---|
| 5 | // // |
|---|
| 6 | // Coded by Ralf Westram (coder@reallysoft.de) in September 2013 // |
|---|
| 7 | // Institute of Microbiology (Technical University Munich) // |
|---|
| 8 | // http://www.arb-home.de/ // |
|---|
| 9 | // // |
|---|
| 10 | // ================================================================= // |
|---|
| 11 | |
|---|
| 12 | // AISC_MKPT_PROMOTE:#ifndef ARB_CORE_H |
|---|
| 13 | // AISC_MKPT_PROMOTE:#include "arb_core.h" |
|---|
| 14 | // AISC_MKPT_PROMOTE:#endif |
|---|
| 15 | |
|---|
| 16 | #include "arb_match.h" |
|---|
| 17 | #include "arb_msg.h" |
|---|
| 18 | #include "arb_string.h" |
|---|
| 19 | #include "arb_strbuf.h" |
|---|
| 20 | |
|---|
| 21 | #include <regex.h> |
|---|
| 22 | |
|---|
| 23 | // --------------------------------------------- |
|---|
| 24 | // Regular Expressions search/replace |
|---|
| 25 | |
|---|
| 26 | struct GBS_regex { regex_t compiled; }; // definition exists twice (see ../SL/REGEXPR/RegExpr.cxx) |
|---|
| 27 | |
|---|
| 28 | inline char *give_buffer(size_t size) { |
|---|
| 29 | static char *buf = NULp; |
|---|
| 30 | static size_t bufsize = 0; |
|---|
| 31 | |
|---|
| 32 | if (size<1) size = 1; |
|---|
| 33 | if (bufsize<size) { |
|---|
| 34 | bufsize = size; |
|---|
| 35 | freeset(buf, ARB_alloc<char>(bufsize)); |
|---|
| 36 | } |
|---|
| 37 | return buf; |
|---|
| 38 | } |
|---|
| 39 | |
|---|
| 40 | GBS_regex *GBS_compile_regexpr(const char *regexpr, GB_CASE case_flag, GB_ERROR *error) { |
|---|
| 41 | GBS_regex *comreg = ARB_alloc<GBS_regex>(1); |
|---|
| 42 | int cflags = REG_EXTENDED|(case_flag == GB_IGNORE_CASE ? REG_ICASE : 0)|REG_NEWLINE; |
|---|
| 43 | int errcode = regcomp(&comreg->compiled, regexpr, cflags); |
|---|
| 44 | |
|---|
| 45 | if (errcode != 0) { // error compiling regexpr |
|---|
| 46 | size_t size = regerror(errcode, &comreg->compiled, NULp, 0); |
|---|
| 47 | char *buf = give_buffer(size); |
|---|
| 48 | |
|---|
| 49 | regerror(errcode, &comreg->compiled, buf, size); |
|---|
| 50 | *error = buf; |
|---|
| 51 | |
|---|
| 52 | free(comreg); |
|---|
| 53 | comreg = NULp; |
|---|
| 54 | } |
|---|
| 55 | else { |
|---|
| 56 | *error = NULp; |
|---|
| 57 | } |
|---|
| 58 | |
|---|
| 59 | return comreg; |
|---|
| 60 | } |
|---|
| 61 | |
|---|
| 62 | void GBS_free_regexpr(GBS_regex *toFree) { |
|---|
| 63 | if (toFree) { |
|---|
| 64 | regfree(&toFree->compiled); |
|---|
| 65 | free(toFree); |
|---|
| 66 | } |
|---|
| 67 | } |
|---|
| 68 | |
|---|
| 69 | const char *GBS_unwrap_regexpr(const char *regexpr_in_slashes, GB_CASE *case_flag, GB_ERROR *error) { |
|---|
| 70 | /* unwraps 'expr' from '/expr/[i]' |
|---|
| 71 | * if slashes are not present, 'error' is set |
|---|
| 72 | * 'case_flag' is set to GB_MIND_CASE if format is '/expr/' or |
|---|
| 73 | * to GB_IGNORE_CASE if format is '/expr/i' |
|---|
| 74 | * |
|---|
| 75 | * returns a pointer to a static buffer (containing the unwrapped expression) |
|---|
| 76 | * (Note: The content is invalidated by the next call to GBS_unwrap_regexpr) |
|---|
| 77 | */ |
|---|
| 78 | |
|---|
| 79 | arb_assert(error && !*error); |
|---|
| 80 | |
|---|
| 81 | const char *result = NULp; |
|---|
| 82 | const char *end = strchr(regexpr_in_slashes, 0); |
|---|
| 83 | |
|---|
| 84 | if (end >= (regexpr_in_slashes+3)) { |
|---|
| 85 | *case_flag = GB_MIND_CASE; |
|---|
| 86 | if (end[-1] == 'i') { |
|---|
| 87 | *case_flag = GB_IGNORE_CASE; |
|---|
| 88 | end--; |
|---|
| 89 | } |
|---|
| 90 | if (regexpr_in_slashes[0] == '/' && end[-1] == '/') { |
|---|
| 91 | arb_assert(!*error); |
|---|
| 92 | |
|---|
| 93 | static char *result_buffer = NULp; |
|---|
| 94 | static size_t max_len = 0; |
|---|
| 95 | |
|---|
| 96 | size_t len = end-regexpr_in_slashes-2; |
|---|
| 97 | arb_assert(len>0); // don't accept empty expression |
|---|
| 98 | |
|---|
| 99 | if (len>max_len) { |
|---|
| 100 | max_len = len*3/2; |
|---|
| 101 | freeset(result_buffer, ARB_alloc<char>(max_len+1)); |
|---|
| 102 | } |
|---|
| 103 | |
|---|
| 104 | memcpy(result_buffer, regexpr_in_slashes+1, len); |
|---|
| 105 | result_buffer[len] = 0; |
|---|
| 106 | |
|---|
| 107 | result = result_buffer; |
|---|
| 108 | } |
|---|
| 109 | } |
|---|
| 110 | |
|---|
| 111 | if (!result) { |
|---|
| 112 | *error = GBS_global_string("Regular expression format is '/expr/' or '/expr/i', not '%s'", |
|---|
| 113 | regexpr_in_slashes); |
|---|
| 114 | } |
|---|
| 115 | |
|---|
| 116 | arb_assert(contradicted(result, *error)); |
|---|
| 117 | return result; |
|---|
| 118 | } |
|---|
| 119 | |
|---|
| 120 | const char *GBS_regmatch_compiled(const char *str, GBS_regex *comreg, size_t *matchlen) { |
|---|
| 121 | /* like GBS_regmatch, |
|---|
| 122 | * - but uses a precompiled regular expression |
|---|
| 123 | * - no errors can occur here (beside out of memory, which is not handled) |
|---|
| 124 | */ |
|---|
| 125 | |
|---|
| 126 | regmatch_t match; |
|---|
| 127 | int res = regexec(&comreg->compiled, str, 1, &match, 0); |
|---|
| 128 | const char *matchpos = NULp; |
|---|
| 129 | |
|---|
| 130 | if (res == 0) { // matched |
|---|
| 131 | matchpos = str+match.rm_so; |
|---|
| 132 | if (matchlen) *matchlen = match.rm_eo-match.rm_so; |
|---|
| 133 | } |
|---|
| 134 | |
|---|
| 135 | return matchpos; |
|---|
| 136 | } |
|---|
| 137 | |
|---|
| 138 | const char *GBS_regmatch(const char *str, const char *regExpr, size_t *matchlen, GB_ERROR *error) { |
|---|
| 139 | /* searches 'str' for first occurrence of 'regExpr' |
|---|
| 140 | * 'regExpr' has to be in format "/expr/[i]", where 'expr' is a POSIX extended regular expression |
|---|
| 141 | * |
|---|
| 142 | * for regexpression format see http://help.arb-home.de/reg.html#Syntax_of_POSIX_extended_regular_expressions_as_used_in_ARB |
|---|
| 143 | * |
|---|
| 144 | * returns |
|---|
| 145 | * - pointer to start of first match in 'str' and |
|---|
| 146 | * length of match in 'matchlen' ('matchlen' may be NULp, then no len is reported) |
|---|
| 147 | * or |
|---|
| 148 | * - NULp if nothing matched (in this case 'matchlen' is undefined) |
|---|
| 149 | * |
|---|
| 150 | * 'error' will be set if sth is wrong |
|---|
| 151 | * |
|---|
| 152 | * Note: Only use this function if you do exactly ONE match. |
|---|
| 153 | * Use GBS_regmatch_compiled if you use the regexpr twice or more! |
|---|
| 154 | */ |
|---|
| 155 | arb_assert(error && !*error); |
|---|
| 156 | |
|---|
| 157 | const char *firstMatch = NULp; |
|---|
| 158 | GB_CASE case_flag; |
|---|
| 159 | const char *unwrapped_expr = GBS_unwrap_regexpr(regExpr, &case_flag, error); |
|---|
| 160 | |
|---|
| 161 | if (unwrapped_expr) { |
|---|
| 162 | GBS_regex *comreg = GBS_compile_regexpr(unwrapped_expr, case_flag, error); |
|---|
| 163 | if (comreg) { |
|---|
| 164 | firstMatch = GBS_regmatch_compiled(str, comreg, matchlen); |
|---|
| 165 | GBS_free_regexpr(comreg); |
|---|
| 166 | } |
|---|
| 167 | } |
|---|
| 168 | |
|---|
| 169 | arb_assert(implicated(firstMatch, !*error)); |
|---|
| 170 | arb_assert(implicated(*error, !firstMatch)); |
|---|
| 171 | |
|---|
| 172 | return firstMatch; |
|---|
| 173 | } |
|---|
| 174 | |
|---|
| 175 | char *GBS_regreplace(const char *str, const char *regReplExpr, GB_ERROR *error) { |
|---|
| 176 | /* search and replace all matches in 'str' using POSIX extended regular expression |
|---|
| 177 | * 'regReplExpr' has to be in format '/regexpr/replace/[i]' |
|---|
| 178 | * |
|---|
| 179 | * returns |
|---|
| 180 | * - a heap copy of the modified string or |
|---|
| 181 | * - NULp if something went wrong (in this case 'error' contains the reason) |
|---|
| 182 | * |
|---|
| 183 | * 'replace' may contain several special substrings: |
|---|
| 184 | * |
|---|
| 185 | * "\n" gets replaced by '\n' |
|---|
| 186 | * "\t" -------''------- '\t' |
|---|
| 187 | * "\\" -------''------- '\\' |
|---|
| 188 | * "\0" -------''------- the complete match to regexpr |
|---|
| 189 | * "\1" -------''------- the match to the first subexpression |
|---|
| 190 | * "\2" -------''------- the match to the second subexpression |
|---|
| 191 | * ... |
|---|
| 192 | * "\9" -------''------- the match to the ninth subexpression |
|---|
| 193 | */ |
|---|
| 194 | |
|---|
| 195 | // test performed via ACI in ../ARBDB/gb_aci.cxx@GBS_REGREPLACE_TESTS |
|---|
| 196 | |
|---|
| 197 | arb_assert(error && !*error); |
|---|
| 198 | |
|---|
| 199 | GB_CASE case_flag; |
|---|
| 200 | const char *unwrapped_expr = GBS_unwrap_regexpr(regReplExpr, &case_flag, error); |
|---|
| 201 | char *result = NULp; |
|---|
| 202 | |
|---|
| 203 | if (unwrapped_expr) { |
|---|
| 204 | const char *sep = unwrapped_expr; |
|---|
| 205 | while (sep) { |
|---|
| 206 | sep = strchr(sep, '/'); |
|---|
| 207 | if (!sep) break; |
|---|
| 208 | if (sep>unwrapped_expr && sep[-1] != '\\') break; |
|---|
| 209 | ++sep; |
|---|
| 210 | } |
|---|
| 211 | |
|---|
| 212 | if (!sep) { |
|---|
| 213 | // Warning: GB_command_interpreter() tests for this error message - don't change |
|---|
| 214 | *error = "Missing '/' between search and replace string"; |
|---|
| 215 | } |
|---|
| 216 | else { |
|---|
| 217 | char *regexpr = ARB_strpartdup(unwrapped_expr, sep-1); |
|---|
| 218 | char *replexpr = ARB_strpartdup(sep+1, NULp); |
|---|
| 219 | GBS_regex *comreg = GBS_compile_regexpr(regexpr, case_flag, error); |
|---|
| 220 | |
|---|
| 221 | const bool exprMatchesBorderOfLine = regexpr[0] == '^' || sep[-1] == '$'; |
|---|
| 222 | |
|---|
| 223 | if (comreg) { |
|---|
| 224 | GBS_strstruct *out = GBS_stropen(1000); |
|---|
| 225 | int eflags = 0; |
|---|
| 226 | |
|---|
| 227 | while (str) { |
|---|
| 228 | regmatch_t match[10]; |
|---|
| 229 | int res = regexec(&comreg->compiled, str, 10, match, eflags); |
|---|
| 230 | |
|---|
| 231 | if (res == REG_NOMATCH) { // did not match |
|---|
| 232 | GBS_strcat(out, str); // copy original |
|---|
| 233 | str = NULp; |
|---|
| 234 | } |
|---|
| 235 | else if (match[0].rm_so == match[0].rm_eo && !exprMatchesBorderOfLine) { // found empty match |
|---|
| 236 | *error = GBS_global_string("The regular expression '%s' matched an empty string (not allowed for replace)", regexpr); |
|---|
| 237 | str = NULp; |
|---|
| 238 | } |
|---|
| 239 | else { // found non-empty match |
|---|
| 240 | size_t p; |
|---|
| 241 | char c; |
|---|
| 242 | |
|---|
| 243 | GBS_strncat(out, str, match[0].rm_so); |
|---|
| 244 | |
|---|
| 245 | for (p = 0; (c = replexpr[p]); ++p) { |
|---|
| 246 | if (c == '\\') { |
|---|
| 247 | c = replexpr[++p]; |
|---|
| 248 | if (!c) break; |
|---|
| 249 | if (c >= '0' && c <= '9') { |
|---|
| 250 | regoff_t start = match[c-'0'].rm_so; |
|---|
| 251 | GBS_strncat(out, str+start, match[c-'0'].rm_eo-start); |
|---|
| 252 | } |
|---|
| 253 | else { |
|---|
| 254 | switch (c) { |
|---|
| 255 | case 'n': c = '\n'; break; |
|---|
| 256 | case 't': c = '\t'; break; |
|---|
| 257 | default: break; |
|---|
| 258 | } |
|---|
| 259 | GBS_chrcat(out, c); |
|---|
| 260 | } |
|---|
| 261 | } |
|---|
| 262 | else { |
|---|
| 263 | GBS_chrcat(out, c); |
|---|
| 264 | } |
|---|
| 265 | } |
|---|
| 266 | |
|---|
| 267 | str = str+match[0].rm_eo; // continue behind match |
|---|
| 268 | eflags = REG_NOTBOL|REG_NOTEOL; // for futher matches, do not regard 'str' as "beginning/end of line" |
|---|
| 269 | } |
|---|
| 270 | } |
|---|
| 271 | |
|---|
| 272 | GBS_free_regexpr(comreg); |
|---|
| 273 | if (*error) { |
|---|
| 274 | GBS_strforget(out); |
|---|
| 275 | } |
|---|
| 276 | else { |
|---|
| 277 | result = GBS_strclose(out); |
|---|
| 278 | } |
|---|
| 279 | } |
|---|
| 280 | free(replexpr); |
|---|
| 281 | free(regexpr); |
|---|
| 282 | } |
|---|
| 283 | } |
|---|
| 284 | |
|---|
| 285 | arb_assert(contradicted(result, *error)); |
|---|
| 286 | |
|---|
| 287 | return result; |
|---|
| 288 | } |
|---|
| 289 | |
|---|