| 1 | /* |
| 2 | This modules is based on the params.c module from Samba, written by Karl Auer |
| 3 | and much modifed by Christopher Hertel. |
| 4 | |
| 5 | * This program is free software; you can redistribute it and/or modify |
| 6 | * it under the terms of the GNU General Public License as published by |
| 7 | * the Free Software Foundation; either version 2 of the License, or |
| 8 | * (at your option) any later version. |
| 9 | * |
| 10 | * This program is distributed in the hope that it will be useful, |
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 13 | * GNU General Public License for more details. |
| 14 | * |
| 15 | * You should have received a copy of the GNU General Public License |
| 16 | * along with this program; if not, write to the Free Software |
| 17 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
| 18 | * |
| 19 | * -------------------------------------------------------------------------- ** |
| 20 | * |
| 21 | * Module name: params |
| 22 | * |
| 23 | * -------------------------------------------------------------------------- ** |
| 24 | * |
| 25 | * This module performs lexical analysis and initial parsing of a |
| 26 | * Windows-like parameter file. It recognizes and handles four token |
| 27 | * types: section-name, parameter-name, parameter-value, and |
| 28 | * end-of-file. Comments and line continuation are handled |
| 29 | * internally. |
| 30 | * |
| 31 | * The entry point to the module is function pm_process(). This |
| 32 | * function opens the source file, calls the Parse() function to parse |
| 33 | * the input, and then closes the file when either the EOF is reached |
| 34 | * or a fatal error is encountered. |
| 35 | * |
| 36 | * A sample parameter file might look like this: |
| 37 | * |
| 38 | * [section one] |
| 39 | * parameter one = value string |
| 40 | * parameter two = another value |
| 41 | * [section two] |
| 42 | * new parameter = some value or t'other |
| 43 | * |
| 44 | * The parameter file is divided into sections by section headers: |
| 45 | * section names enclosed in square brackets (eg. [section one]). |
| 46 | * Each section contains parameter lines, each of which consist of a |
| 47 | * parameter name and value delimited by an equal sign. Roughly, the |
| 48 | * syntax is: |
| 49 | * |
| 50 | * <file> :== { <section> } EOF |
| 51 | * |
| 52 | * <section> :== <section header> { <parameter line> } |
| 53 | * |
| 54 | * <section header> :== '[' NAME ']' |
| 55 | * |
| 56 | * <parameter line> :== NAME '=' VALUE '\n' |
| 57 | * |
| 58 | * Blank lines and comment lines are ignored. Comment lines are lines |
| 59 | * beginning with either a semicolon (';') or a pound sign ('#'). |
| 60 | * |
| 61 | * All whitespace in section names and parameter names is compressed |
| 62 | * to single spaces. Leading and trailing whitespace is stipped from |
| 63 | * both names and values. |
| 64 | * |
| 65 | * Only the first equals sign in a parameter line is significant. |
| 66 | * Parameter values may contain equals signs, square brackets and |
| 67 | * semicolons. Internal whitespace is retained in parameter values, |
| 68 | * with the exception of the '\r' character, which is stripped for |
| 69 | * historic reasons. Parameter names may not start with a left square |
| 70 | * bracket, an equal sign, a pound sign, or a semicolon, because these |
| 71 | * are used to identify other tokens. |
| 72 | * |
| 73 | * -------------------------------------------------------------------------- ** |
| 74 | */ |
| 75 | |
| 76 | #include "rsync.h" |
| 77 | #define BOOL int |
| 78 | #define False 0 |
| 79 | #define True 1 |
| 80 | |
| 81 | /* -------------------------------------------------------------------------- ** |
| 82 | * Constants... |
| 83 | */ |
| 84 | |
| 85 | #define BUFR_INC 1024 |
| 86 | |
| 87 | |
| 88 | /* -------------------------------------------------------------------------- ** |
| 89 | * Variables... |
| 90 | * |
| 91 | * bufr - pointer to a global buffer. This is probably a kludge, |
| 92 | * but it was the nicest kludge I could think of (for now). |
| 93 | * bSize - The size of the global buffer <bufr>. |
| 94 | */ |
| 95 | |
| 96 | static char *bufr = NULL; |
| 97 | static int bSize = 0; |
| 98 | |
| 99 | /* -------------------------------------------------------------------------- ** |
| 100 | * Functions... |
| 101 | */ |
| 102 | |
| 103 | static int EatWhitespace( FILE *InFile ) |
| 104 | /* ------------------------------------------------------------------------ ** |
| 105 | * Scan past whitespace (see ctype(3C)) and return the first non-whitespace |
| 106 | * character, or newline, or EOF. |
| 107 | * |
| 108 | * Input: InFile - Input source. |
| 109 | * |
| 110 | * Output: The next non-whitespace character in the input stream. |
| 111 | * |
| 112 | * Notes: Because the config files use a line-oriented grammar, we |
| 113 | * explicitly exclude the newline character from the list of |
| 114 | * whitespace characters. |
| 115 | * - Note that both EOF (-1) and the nul character ('\0') are |
| 116 | * considered end-of-file markers. |
| 117 | * |
| 118 | * ------------------------------------------------------------------------ ** |
| 119 | */ |
| 120 | { |
| 121 | int c; |
| 122 | |
| 123 | for( c = getc( InFile ); isspace( c ) && ('\n' != c); c = getc( InFile ) ) |
| 124 | ; |
| 125 | return( c ); |
| 126 | } /* EatWhitespace */ |
| 127 | |
| 128 | static int EatComment( FILE *InFile ) |
| 129 | /* ------------------------------------------------------------------------ ** |
| 130 | * Scan to the end of a comment. |
| 131 | * |
| 132 | * Input: InFile - Input source. |
| 133 | * |
| 134 | * Output: The character that marks the end of the comment. Normally, |
| 135 | * this will be a newline, but it *might* be an EOF. |
| 136 | * |
| 137 | * Notes: Because the config files use a line-oriented grammar, we |
| 138 | * explicitly exclude the newline character from the list of |
| 139 | * whitespace characters. |
| 140 | * - Note that both EOF (-1) and the nul character ('\0') are |
| 141 | * considered end-of-file markers. |
| 142 | * |
| 143 | * ------------------------------------------------------------------------ ** |
| 144 | */ |
| 145 | { |
| 146 | int c; |
| 147 | |
| 148 | for( c = getc( InFile ); ('\n'!=c) && (EOF!=c) && (c>0); c = getc( InFile ) ) |
| 149 | ; |
| 150 | return( c ); |
| 151 | } /* EatComment */ |
| 152 | |
| 153 | static int Continuation( char *line, int pos ) |
| 154 | /* ------------------------------------------------------------------------ ** |
| 155 | * Scan backards within a string to discover if the last non-whitespace |
| 156 | * character is a line-continuation character ('\\'). |
| 157 | * |
| 158 | * Input: line - A pointer to a buffer containing the string to be |
| 159 | * scanned. |
| 160 | * pos - This is taken to be the offset of the end of the |
| 161 | * string. This position is *not* scanned. |
| 162 | * |
| 163 | * Output: The offset of the '\\' character if it was found, or -1 to |
| 164 | * indicate that it was not. |
| 165 | * |
| 166 | * ------------------------------------------------------------------------ ** |
| 167 | */ |
| 168 | { |
| 169 | pos--; |
| 170 | while( (pos >= 0) && isspace(line[pos]) ) |
| 171 | pos--; |
| 172 | |
| 173 | return( ((pos >= 0) && ('\\' == line[pos])) ? pos : -1 ); |
| 174 | } /* Continuation */ |
| 175 | |
| 176 | |
| 177 | static BOOL Section( FILE *InFile, BOOL (*sfunc)(char *) ) |
| 178 | /* ------------------------------------------------------------------------ ** |
| 179 | * Scan a section name, and pass the name to function sfunc(). |
| 180 | * |
| 181 | * Input: InFile - Input source. |
| 182 | * sfunc - Pointer to the function to be called if the section |
| 183 | * name is successfully read. |
| 184 | * |
| 185 | * Output: True if the section name was read and True was returned from |
| 186 | * <sfunc>. False if <sfunc> failed or if a lexical error was |
| 187 | * encountered. |
| 188 | * |
| 189 | * ------------------------------------------------------------------------ ** |
| 190 | */ |
| 191 | { |
| 192 | int c; |
| 193 | int i; |
| 194 | int end; |
| 195 | char *func = "params.c:Section() -"; |
| 196 | |
| 197 | i = 0; /* <i> is the offset of the next free byte in bufr[] and */ |
| 198 | end = 0; /* <end> is the current "end of string" offset. In most */ |
| 199 | /* cases these will be the same, but if the last */ |
| 200 | /* character written to bufr[] is a space, then <end> */ |
| 201 | /* will be one less than <i>. */ |
| 202 | |
| 203 | c = EatWhitespace( InFile ); /* We've already got the '['. Scan */ |
| 204 | /* past initial white space. */ |
| 205 | |
| 206 | while( (EOF != c) && (c > 0) ) |
| 207 | { |
| 208 | |
| 209 | /* Check that the buffer is big enough for the next character. */ |
| 210 | if( i > (bSize - 2) ) |
| 211 | { |
| 212 | bSize += BUFR_INC; |
| 213 | bufr = Realloc( bufr, bSize ); |
| 214 | if( NULL == bufr ) |
| 215 | { |
| 216 | rprintf(FERROR, "%s Memory re-allocation failure.", func); |
| 217 | return( False ); |
| 218 | } |
| 219 | } |
| 220 | |
| 221 | /* Handle a single character. */ |
| 222 | switch( c ) |
| 223 | { |
| 224 | case ']': /* Found the closing bracket. */ |
| 225 | bufr[end] = '\0'; |
| 226 | if( 0 == end ) /* Don't allow an empty name. */ |
| 227 | { |
| 228 | rprintf(FERROR, "%s Empty section name in configuration file.\n", func ); |
| 229 | return( False ); |
| 230 | } |
| 231 | if( !sfunc( bufr ) ) /* Got a valid name. Deal with it. */ |
| 232 | return( False ); |
| 233 | (void)EatComment( InFile ); /* Finish off the line. */ |
| 234 | return( True ); |
| 235 | |
| 236 | case '\n': /* Got newline before closing ']'. */ |
| 237 | i = Continuation( bufr, i ); /* Check for line continuation. */ |
| 238 | if( i < 0 ) |
| 239 | { |
| 240 | bufr[end] = '\0'; |
| 241 | rprintf(FERROR, "%s Badly formed line in configuration file: %s\n", |
| 242 | func, bufr ); |
| 243 | return( False ); |
| 244 | } |
| 245 | end = ( (i > 0) && (' ' == bufr[i - 1]) ) ? (i - 1) : (i); |
| 246 | c = getc( InFile ); /* Continue with next line. */ |
| 247 | break; |
| 248 | |
| 249 | default: /* All else are a valid name chars. */ |
| 250 | if( isspace( c ) ) /* One space per whitespace region. */ |
| 251 | { |
| 252 | bufr[end] = ' '; |
| 253 | i = end + 1; |
| 254 | c = EatWhitespace( InFile ); |
| 255 | } |
| 256 | else /* All others copy verbatim. */ |
| 257 | { |
| 258 | bufr[i++] = c; |
| 259 | end = i; |
| 260 | c = getc( InFile ); |
| 261 | } |
| 262 | } |
| 263 | } |
| 264 | |
| 265 | /* We arrive here if we've met the EOF before the closing bracket. */ |
| 266 | rprintf(FERROR, "%s Unexpected EOF in the configuration file: %s\n", func, bufr ); |
| 267 | return( False ); |
| 268 | } /* Section */ |
| 269 | |
| 270 | static BOOL Parameter( FILE *InFile, BOOL (*pfunc)(char *, char *), int c ) |
| 271 | /* ------------------------------------------------------------------------ ** |
| 272 | * Scan a parameter name and value, and pass these two fields to pfunc(). |
| 273 | * |
| 274 | * Input: InFile - The input source. |
| 275 | * pfunc - A pointer to the function that will be called to |
| 276 | * process the parameter, once it has been scanned. |
| 277 | * c - The first character of the parameter name, which |
| 278 | * would have been read by Parse(). Unlike a comment |
| 279 | * line or a section header, there is no lead-in |
| 280 | * character that can be discarded. |
| 281 | * |
| 282 | * Output: True if the parameter name and value were scanned and processed |
| 283 | * successfully, else False. |
| 284 | * |
| 285 | * Notes: This function is in two parts. The first loop scans the |
| 286 | * parameter name. Internal whitespace is compressed, and an |
| 287 | * equal sign (=) terminates the token. Leading and trailing |
| 288 | * whitespace is discarded. The second loop scans the parameter |
| 289 | * value. When both have been successfully identified, they are |
| 290 | * passed to pfunc() for processing. |
| 291 | * |
| 292 | * ------------------------------------------------------------------------ ** |
| 293 | */ |
| 294 | { |
| 295 | int i = 0; /* Position within bufr. */ |
| 296 | int end = 0; /* bufr[end] is current end-of-string. */ |
| 297 | int vstart = 0; /* Starting position of the parameter value. */ |
| 298 | char *func = "params.c:Parameter() -"; |
| 299 | |
| 300 | /* Read the parameter name. */ |
| 301 | while( 0 == vstart ) /* Loop until we've found the start of the value. */ |
| 302 | { |
| 303 | |
| 304 | if( i > (bSize - 2) ) /* Ensure there's space for next char. */ |
| 305 | { |
| 306 | bSize += BUFR_INC; |
| 307 | bufr = Realloc( bufr, bSize ); |
| 308 | if( NULL == bufr ) |
| 309 | { |
| 310 | rprintf(FERROR, "%s Memory re-allocation failure.", func) ; |
| 311 | return( False ); |
| 312 | } |
| 313 | } |
| 314 | |
| 315 | switch( c ) |
| 316 | { |
| 317 | case '=': /* Equal sign marks end of param name. */ |
| 318 | if( 0 == end ) /* Don't allow an empty name. */ |
| 319 | { |
| 320 | rprintf(FERROR, "%s Invalid parameter name in config. file.\n", func ); |
| 321 | return( False ); |
| 322 | } |
| 323 | bufr[end++] = '\0'; /* Mark end of string & advance. */ |
| 324 | i = end; /* New string starts here. */ |
| 325 | vstart = end; /* New string is parameter value. */ |
| 326 | bufr[i] = '\0'; /* New string is nul, for now. */ |
| 327 | break; |
| 328 | |
| 329 | case '\n': /* Find continuation char, else error. */ |
| 330 | i = Continuation( bufr, i ); |
| 331 | if( i < 0 ) |
| 332 | { |
| 333 | bufr[end] = '\0'; |
| 334 | rprintf(FERROR, "%s Ignoring badly formed line in configuration file: %s\n", |
| 335 | func, bufr ); |
| 336 | return( True ); |
| 337 | } |
| 338 | end = ( (i > 0) && (' ' == bufr[i - 1]) ) ? (i - 1) : (i); |
| 339 | c = getc( InFile ); /* Read past eoln. */ |
| 340 | break; |
| 341 | |
| 342 | case '\0': /* Shouldn't have EOF within param name. */ |
| 343 | case EOF: |
| 344 | bufr[i] = '\0'; |
| 345 | rprintf(FERROR, "%s Unexpected end-of-file at: %s\n", func, bufr ); |
| 346 | return( True ); |
| 347 | |
| 348 | default: |
| 349 | if( isspace( c ) ) /* One ' ' per whitespace region. */ |
| 350 | { |
| 351 | bufr[end] = ' '; |
| 352 | i = end + 1; |
| 353 | c = EatWhitespace( InFile ); |
| 354 | } |
| 355 | else /* All others verbatim. */ |
| 356 | { |
| 357 | bufr[i++] = c; |
| 358 | end = i; |
| 359 | c = getc( InFile ); |
| 360 | } |
| 361 | } |
| 362 | } |
| 363 | |
| 364 | /* Now parse the value. */ |
| 365 | c = EatWhitespace( InFile ); /* Again, trim leading whitespace. */ |
| 366 | while( (EOF !=c) && (c > 0) ) |
| 367 | { |
| 368 | |
| 369 | if( i > (bSize - 2) ) /* Make sure there's enough room. */ |
| 370 | { |
| 371 | bSize += BUFR_INC; |
| 372 | bufr = Realloc( bufr, bSize ); |
| 373 | if( NULL == bufr ) |
| 374 | { |
| 375 | rprintf(FERROR, "%s Memory re-allocation failure.", func) ; |
| 376 | return( False ); |
| 377 | } |
| 378 | } |
| 379 | |
| 380 | switch( c ) |
| 381 | { |
| 382 | case '\r': /* Explicitly remove '\r' because the older */ |
| 383 | c = getc( InFile ); /* version called fgets_slash() which also */ |
| 384 | break; /* removes them. */ |
| 385 | |
| 386 | case '\n': /* Marks end of value unless there's a '\'. */ |
| 387 | i = Continuation( bufr, i ); |
| 388 | if( i < 0 ) |
| 389 | c = 0; |
| 390 | else |
| 391 | { |
| 392 | for( end = i; (end >= 0) && isspace(bufr[end]); end-- ) |
| 393 | ; |
| 394 | c = getc( InFile ); |
| 395 | } |
| 396 | break; |
| 397 | |
| 398 | default: /* All others verbatim. Note that spaces do */ |
| 399 | bufr[i++] = c; /* not advance <end>. This allows trimming */ |
| 400 | if( !isspace( c ) ) /* of whitespace at the end of the line. */ |
| 401 | end = i; |
| 402 | c = getc( InFile ); |
| 403 | break; |
| 404 | } |
| 405 | } |
| 406 | bufr[end] = '\0'; /* End of value. */ |
| 407 | |
| 408 | return( pfunc( bufr, &bufr[vstart] ) ); /* Pass name & value to pfunc(). */ |
| 409 | } /* Parameter */ |
| 410 | |
| 411 | static BOOL Parse( FILE *InFile, |
| 412 | BOOL (*sfunc)(char *), |
| 413 | BOOL (*pfunc)(char *, char *) ) |
| 414 | /* ------------------------------------------------------------------------ ** |
| 415 | * Scan & parse the input. |
| 416 | * |
| 417 | * Input: InFile - Input source. |
| 418 | * sfunc - Function to be called when a section name is scanned. |
| 419 | * See Section(). |
| 420 | * pfunc - Function to be called when a parameter is scanned. |
| 421 | * See Parameter(). |
| 422 | * |
| 423 | * Output: True if the file was successfully scanned, else False. |
| 424 | * |
| 425 | * Notes: The input can be viewed in terms of 'lines'. There are four |
| 426 | * types of lines: |
| 427 | * Blank - May contain whitespace, otherwise empty. |
| 428 | * Comment - First non-whitespace character is a ';' or '#'. |
| 429 | * The remainder of the line is ignored. |
| 430 | * Section - First non-whitespace character is a '['. |
| 431 | * Parameter - The default case. |
| 432 | * |
| 433 | * ------------------------------------------------------------------------ ** |
| 434 | */ |
| 435 | { |
| 436 | int c; |
| 437 | |
| 438 | c = EatWhitespace( InFile ); |
| 439 | while( (EOF != c) && (c > 0) ) |
| 440 | { |
| 441 | switch( c ) |
| 442 | { |
| 443 | case '\n': /* Blank line. */ |
| 444 | c = EatWhitespace( InFile ); |
| 445 | break; |
| 446 | |
| 447 | case ';': /* Comment line. */ |
| 448 | case '#': |
| 449 | c = EatComment( InFile ); |
| 450 | break; |
| 451 | |
| 452 | case '[': /* Section Header. */ |
| 453 | if (!sfunc) return True; |
| 454 | if( !Section( InFile, sfunc ) ) |
| 455 | return( False ); |
| 456 | c = EatWhitespace( InFile ); |
| 457 | break; |
| 458 | |
| 459 | case '\\': /* Bogus backslash. */ |
| 460 | c = EatWhitespace( InFile ); |
| 461 | break; |
| 462 | |
| 463 | default: /* Parameter line. */ |
| 464 | if( !Parameter( InFile, pfunc, c ) ) |
| 465 | return( False ); |
| 466 | c = EatWhitespace( InFile ); |
| 467 | break; |
| 468 | } |
| 469 | } |
| 470 | return( True ); |
| 471 | } /* Parse */ |
| 472 | |
| 473 | static FILE *OpenConfFile( char *FileName ) |
| 474 | /* ------------------------------------------------------------------------ ** |
| 475 | * Open a configuration file. |
| 476 | * |
| 477 | * Input: FileName - The pathname of the config file to be opened. |
| 478 | * |
| 479 | * Output: A pointer of type (FILE *) to the opened file, or NULL if the |
| 480 | * file could not be opened. |
| 481 | * |
| 482 | * ------------------------------------------------------------------------ ** |
| 483 | */ |
| 484 | { |
| 485 | FILE *OpenedFile; |
| 486 | char *func = "params.c:OpenConfFile() -"; |
| 487 | |
| 488 | if( NULL == FileName || 0 == *FileName ) |
| 489 | { |
| 490 | rprintf(FERROR,"%s No configuration filename specified.\n", func); |
| 491 | return( NULL ); |
| 492 | } |
| 493 | |
| 494 | OpenedFile = fopen( FileName, "r" ); |
| 495 | if( NULL == OpenedFile ) |
| 496 | { |
| 497 | rprintf(FERROR,"%s Unable to open configuration file \"%s\":\n\t%s\n", |
| 498 | func, FileName, strerror(errno)); |
| 499 | } |
| 500 | |
| 501 | return( OpenedFile ); |
| 502 | } /* OpenConfFile */ |
| 503 | |
| 504 | BOOL pm_process( char *FileName, |
| 505 | BOOL (*sfunc)(char *), |
| 506 | BOOL (*pfunc)(char *, char *) ) |
| 507 | /* ------------------------------------------------------------------------ ** |
| 508 | * Process the named parameter file. |
| 509 | * |
| 510 | * Input: FileName - The pathname of the parameter file to be opened. |
| 511 | * sfunc - A pointer to a function that will be called when |
| 512 | * a section name is discovered. |
| 513 | * pfunc - A pointer to a function that will be called when |
| 514 | * a parameter name and value are discovered. |
| 515 | * |
| 516 | * Output: TRUE if the file was successfully parsed, else FALSE. |
| 517 | * |
| 518 | * ------------------------------------------------------------------------ ** |
| 519 | */ |
| 520 | { |
| 521 | int result; |
| 522 | FILE *InFile; |
| 523 | char *func = "params.c:pm_process() -"; |
| 524 | |
| 525 | InFile = OpenConfFile( FileName ); /* Open the config file. */ |
| 526 | if( NULL == InFile ) |
| 527 | return( False ); |
| 528 | |
| 529 | if( NULL != bufr ) /* If we already have a buffer */ |
| 530 | result = Parse( InFile, sfunc, pfunc ); /* (recursive call), then just */ |
| 531 | /* use it. */ |
| 532 | |
| 533 | else /* If we don't have a buffer */ |
| 534 | { /* allocate one, then parse, */ |
| 535 | bSize = BUFR_INC; /* then free. */ |
| 536 | bufr = (char *)malloc( bSize ); |
| 537 | if( NULL == bufr ) |
| 538 | { |
| 539 | rprintf(FERROR,"%s memory allocation failure.\n", func); |
| 540 | fclose(InFile); |
| 541 | return( False ); |
| 542 | } |
| 543 | result = Parse( InFile, sfunc, pfunc ); |
| 544 | free( bufr ); |
| 545 | bufr = NULL; |
| 546 | bSize = 0; |
| 547 | } |
| 548 | |
| 549 | fclose(InFile); |
| 550 | |
| 551 | if( !result ) /* Generic failure. */ |
| 552 | { |
| 553 | rprintf(FERROR,"%s Failed. Error returned from params.c:parse().\n", func); |
| 554 | return( False ); |
| 555 | } |
| 556 | |
| 557 | return( True ); /* Generic success. */ |
| 558 | } /* pm_process */ |
| 559 | |
| 560 | /* -------------------------------------------------------------------------- */ |
| 561 | |