Commit | Line | Data |
---|---|---|
0f78b815 WD |
1 | /* This modules is based on the params.c module from Samba, written by Karl Auer |
2 | and much modifed by Christopher Hertel. */ | |
0b76cd63 | 3 | |
0f78b815 | 4 | /* |
0b76cd63 | 5 | * This program is free software; you can redistribute it and/or modify |
4fd842f9 | 6 | * it under the terms of the GNU General Public License version 3 as |
ba2133d6 | 7 | * published by the Free Software Foundation. |
0b76cd63 AT |
8 | * |
9 | * This program is distributed in the hope that it will be useful, | |
10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
12 | * GNU General Public License for more details. | |
13 | * | |
e7c67065 | 14 | * You should have received a copy of the GNU General Public License along |
4fd842f9 | 15 | * with this program; if not, visit the http://fsf.org website. |
0f78b815 WD |
16 | */ |
17 | ||
18 | /* -------------------------------------------------------------------------- ** | |
0b76cd63 AT |
19 | * |
20 | * Module name: params | |
21 | * | |
22 | * -------------------------------------------------------------------------- ** | |
23 | * | |
24 | * This module performs lexical analysis and initial parsing of a | |
25 | * Windows-like parameter file. It recognizes and handles four token | |
26 | * types: section-name, parameter-name, parameter-value, and | |
27 | * end-of-file. Comments and line continuation are handled | |
28 | * internally. | |
29 | * | |
30 | * The entry point to the module is function pm_process(). This | |
31 | * function opens the source file, calls the Parse() function to parse | |
32 | * the input, and then closes the file when either the EOF is reached | |
33 | * or a fatal error is encountered. | |
34 | * | |
35 | * A sample parameter file might look like this: | |
36 | * | |
37 | * [section one] | |
38 | * parameter one = value string | |
39 | * parameter two = another value | |
40 | * [section two] | |
41 | * new parameter = some value or t'other | |
42 | * | |
43 | * The parameter file is divided into sections by section headers: | |
44 | * section names enclosed in square brackets (eg. [section one]). | |
45 | * Each section contains parameter lines, each of which consist of a | |
46 | * parameter name and value delimited by an equal sign. Roughly, the | |
47 | * syntax is: | |
48 | * | |
49 | * <file> :== { <section> } EOF | |
50 | * | |
51 | * <section> :== <section header> { <parameter line> } | |
52 | * | |
53 | * <section header> :== '[' NAME ']' | |
54 | * | |
55 | * <parameter line> :== NAME '=' VALUE '\n' | |
56 | * | |
57 | * Blank lines and comment lines are ignored. Comment lines are lines | |
58 | * beginning with either a semicolon (';') or a pound sign ('#'). | |
59 | * | |
60 | * All whitespace in section names and parameter names is compressed | |
61 | * to single spaces. Leading and trailing whitespace is stipped from | |
62 | * both names and values. | |
63 | * | |
64 | * Only the first equals sign in a parameter line is significant. | |
65 | * Parameter values may contain equals signs, square brackets and | |
66 | * semicolons. Internal whitespace is retained in parameter values, | |
67 | * with the exception of the '\r' character, which is stripped for | |
68 | * historic reasons. Parameter names may not start with a left square | |
69 | * bracket, an equal sign, a pound sign, or a semicolon, because these | |
70 | * are used to identify other tokens. | |
71 | * | |
72 | * -------------------------------------------------------------------------- ** | |
73 | */ | |
74 | ||
75 | #include "rsync.h" | |
0b76cd63 AT |
76 | |
77 | /* -------------------------------------------------------------------------- ** | |
78 | * Constants... | |
79 | */ | |
80 | ||
81 | #define BUFR_INC 1024 | |
82 | ||
83 | ||
84 | /* -------------------------------------------------------------------------- ** | |
85 | * Variables... | |
86 | * | |
87 | * bufr - pointer to a global buffer. This is probably a kludge, | |
88 | * but it was the nicest kludge I could think of (for now). | |
89 | * bSize - The size of the global buffer <bufr>. | |
90 | */ | |
91 | ||
92 | static char *bufr = NULL; | |
93 | static int bSize = 0; | |
94 | ||
95 | /* -------------------------------------------------------------------------- ** | |
96 | * Functions... | |
97 | */ | |
98 | ||
99 | static int EatWhitespace( FILE *InFile ) | |
100 | /* ------------------------------------------------------------------------ ** | |
101 | * Scan past whitespace (see ctype(3C)) and return the first non-whitespace | |
102 | * character, or newline, or EOF. | |
103 | * | |
104 | * Input: InFile - Input source. | |
105 | * | |
106 | * Output: The next non-whitespace character in the input stream. | |
107 | * | |
108 | * Notes: Because the config files use a line-oriented grammar, we | |
109 | * explicitly exclude the newline character from the list of | |
110 | * whitespace characters. | |
111 | * - Note that both EOF (-1) and the nul character ('\0') are | |
112 | * considered end-of-file markers. | |
113 | * | |
114 | * ------------------------------------------------------------------------ ** | |
115 | */ | |
116 | { | |
117 | int c; | |
118 | ||
119 | for( c = getc( InFile ); isspace( c ) && ('\n' != c); c = getc( InFile ) ) | |
120 | ; | |
121 | return( c ); | |
122 | } /* EatWhitespace */ | |
123 | ||
124 | static int EatComment( FILE *InFile ) | |
125 | /* ------------------------------------------------------------------------ ** | |
126 | * Scan to the end of a comment. | |
127 | * | |
128 | * Input: InFile - Input source. | |
129 | * | |
130 | * Output: The character that marks the end of the comment. Normally, | |
131 | * this will be a newline, but it *might* be an EOF. | |
132 | * | |
133 | * Notes: Because the config files use a line-oriented grammar, we | |
134 | * explicitly exclude the newline character from the list of | |
135 | * whitespace characters. | |
136 | * - Note that both EOF (-1) and the nul character ('\0') are | |
137 | * considered end-of-file markers. | |
138 | * | |
139 | * ------------------------------------------------------------------------ ** | |
140 | */ | |
141 | { | |
142 | int c; | |
143 | ||
144 | for( c = getc( InFile ); ('\n'!=c) && (EOF!=c) && (c>0); c = getc( InFile ) ) | |
145 | ; | |
146 | return( c ); | |
147 | } /* EatComment */ | |
148 | ||
149 | static int Continuation( char *line, int pos ) | |
150 | /* ------------------------------------------------------------------------ ** | |
151 | * Scan backards within a string to discover if the last non-whitespace | |
152 | * character is a line-continuation character ('\\'). | |
153 | * | |
154 | * Input: line - A pointer to a buffer containing the string to be | |
155 | * scanned. | |
156 | * pos - This is taken to be the offset of the end of the | |
157 | * string. This position is *not* scanned. | |
158 | * | |
159 | * Output: The offset of the '\\' character if it was found, or -1 to | |
160 | * indicate that it was not. | |
161 | * | |
162 | * ------------------------------------------------------------------------ ** | |
163 | */ | |
164 | { | |
165 | pos--; | |
2dc7b8bd | 166 | while( pos >= 0 && isSpace(line + pos) ) |
0b76cd63 AT |
167 | pos--; |
168 | ||
169 | return( ((pos >= 0) && ('\\' == line[pos])) ? pos : -1 ); | |
170 | } /* Continuation */ | |
171 | ||
172 | ||
173 | static BOOL Section( FILE *InFile, BOOL (*sfunc)(char *) ) | |
174 | /* ------------------------------------------------------------------------ ** | |
175 | * Scan a section name, and pass the name to function sfunc(). | |
176 | * | |
177 | * Input: InFile - Input source. | |
178 | * sfunc - Pointer to the function to be called if the section | |
179 | * name is successfully read. | |
180 | * | |
181 | * Output: True if the section name was read and True was returned from | |
182 | * <sfunc>. False if <sfunc> failed or if a lexical error was | |
183 | * encountered. | |
184 | * | |
185 | * ------------------------------------------------------------------------ ** | |
186 | */ | |
187 | { | |
188 | int c; | |
189 | int i; | |
190 | int end; | |
191 | char *func = "params.c:Section() -"; | |
192 | ||
193 | i = 0; /* <i> is the offset of the next free byte in bufr[] and */ | |
194 | end = 0; /* <end> is the current "end of string" offset. In most */ | |
195 | /* cases these will be the same, but if the last */ | |
196 | /* character written to bufr[] is a space, then <end> */ | |
197 | /* will be one less than <i>. */ | |
198 | ||
199 | c = EatWhitespace( InFile ); /* We've already got the '['. Scan */ | |
200 | /* past initial white space. */ | |
201 | ||
202 | while( (EOF != c) && (c > 0) ) | |
203 | { | |
204 | ||
205 | /* Check that the buffer is big enough for the next character. */ | |
206 | if( i > (bSize - 2) ) | |
207 | { | |
208 | bSize += BUFR_INC; | |
58cadc86 | 209 | bufr = realloc_array( bufr, char, bSize ); |
0b76cd63 AT |
210 | if( NULL == bufr ) |
211 | { | |
212 | rprintf(FERROR, "%s Memory re-allocation failure.", func); | |
213 | return( False ); | |
214 | } | |
215 | } | |
216 | ||
217 | /* Handle a single character. */ | |
218 | switch( c ) | |
219 | { | |
220 | case ']': /* Found the closing bracket. */ | |
221 | bufr[end] = '\0'; | |
222 | if( 0 == end ) /* Don't allow an empty name. */ | |
223 | { | |
224 | rprintf(FERROR, "%s Empty section name in configuration file.\n", func ); | |
225 | return( False ); | |
226 | } | |
227 | if( !sfunc( bufr ) ) /* Got a valid name. Deal with it. */ | |
228 | return( False ); | |
229 | (void)EatComment( InFile ); /* Finish off the line. */ | |
230 | return( True ); | |
231 | ||
232 | case '\n': /* Got newline before closing ']'. */ | |
233 | i = Continuation( bufr, i ); /* Check for line continuation. */ | |
234 | if( i < 0 ) | |
235 | { | |
236 | bufr[end] = '\0'; | |
237 | rprintf(FERROR, "%s Badly formed line in configuration file: %s\n", | |
238 | func, bufr ); | |
239 | return( False ); | |
240 | } | |
241 | end = ( (i > 0) && (' ' == bufr[i - 1]) ) ? (i - 1) : (i); | |
242 | c = getc( InFile ); /* Continue with next line. */ | |
243 | break; | |
244 | ||
245 | default: /* All else are a valid name chars. */ | |
246 | if( isspace( c ) ) /* One space per whitespace region. */ | |
247 | { | |
248 | bufr[end] = ' '; | |
249 | i = end + 1; | |
250 | c = EatWhitespace( InFile ); | |
251 | } | |
252 | else /* All others copy verbatim. */ | |
253 | { | |
254 | bufr[i++] = c; | |
255 | end = i; | |
256 | c = getc( InFile ); | |
257 | } | |
258 | } | |
259 | } | |
260 | ||
261 | /* We arrive here if we've met the EOF before the closing bracket. */ | |
262 | rprintf(FERROR, "%s Unexpected EOF in the configuration file: %s\n", func, bufr ); | |
263 | return( False ); | |
264 | } /* Section */ | |
265 | ||
266 | static BOOL Parameter( FILE *InFile, BOOL (*pfunc)(char *, char *), int c ) | |
267 | /* ------------------------------------------------------------------------ ** | |
268 | * Scan a parameter name and value, and pass these two fields to pfunc(). | |
269 | * | |
270 | * Input: InFile - The input source. | |
271 | * pfunc - A pointer to the function that will be called to | |
272 | * process the parameter, once it has been scanned. | |
273 | * c - The first character of the parameter name, which | |
274 | * would have been read by Parse(). Unlike a comment | |
275 | * line or a section header, there is no lead-in | |
276 | * character that can be discarded. | |
277 | * | |
278 | * Output: True if the parameter name and value were scanned and processed | |
279 | * successfully, else False. | |
280 | * | |
281 | * Notes: This function is in two parts. The first loop scans the | |
282 | * parameter name. Internal whitespace is compressed, and an | |
283 | * equal sign (=) terminates the token. Leading and trailing | |
284 | * whitespace is discarded. The second loop scans the parameter | |
285 | * value. When both have been successfully identified, they are | |
286 | * passed to pfunc() for processing. | |
287 | * | |
288 | * ------------------------------------------------------------------------ ** | |
289 | */ | |
290 | { | |
291 | int i = 0; /* Position within bufr. */ | |
292 | int end = 0; /* bufr[end] is current end-of-string. */ | |
293 | int vstart = 0; /* Starting position of the parameter value. */ | |
294 | char *func = "params.c:Parameter() -"; | |
295 | ||
296 | /* Read the parameter name. */ | |
297 | while( 0 == vstart ) /* Loop until we've found the start of the value. */ | |
298 | { | |
299 | ||
300 | if( i > (bSize - 2) ) /* Ensure there's space for next char. */ | |
301 | { | |
302 | bSize += BUFR_INC; | |
58cadc86 | 303 | bufr = realloc_array( bufr, char, bSize ); |
0b76cd63 AT |
304 | if( NULL == bufr ) |
305 | { | |
306 | rprintf(FERROR, "%s Memory re-allocation failure.", func) ; | |
307 | return( False ); | |
308 | } | |
309 | } | |
310 | ||
311 | switch( c ) | |
312 | { | |
313 | case '=': /* Equal sign marks end of param name. */ | |
314 | if( 0 == end ) /* Don't allow an empty name. */ | |
315 | { | |
316 | rprintf(FERROR, "%s Invalid parameter name in config. file.\n", func ); | |
317 | return( False ); | |
318 | } | |
319 | bufr[end++] = '\0'; /* Mark end of string & advance. */ | |
320 | i = end; /* New string starts here. */ | |
321 | vstart = end; /* New string is parameter value. */ | |
322 | bufr[i] = '\0'; /* New string is nul, for now. */ | |
323 | break; | |
324 | ||
325 | case '\n': /* Find continuation char, else error. */ | |
326 | i = Continuation( bufr, i ); | |
327 | if( i < 0 ) | |
328 | { | |
329 | bufr[end] = '\0'; | |
330 | rprintf(FERROR, "%s Ignoring badly formed line in configuration file: %s\n", | |
331 | func, bufr ); | |
332 | return( True ); | |
333 | } | |
334 | end = ( (i > 0) && (' ' == bufr[i - 1]) ) ? (i - 1) : (i); | |
335 | c = getc( InFile ); /* Read past eoln. */ | |
336 | break; | |
337 | ||
338 | case '\0': /* Shouldn't have EOF within param name. */ | |
339 | case EOF: | |
340 | bufr[i] = '\0'; | |
341 | rprintf(FERROR, "%s Unexpected end-of-file at: %s\n", func, bufr ); | |
342 | return( True ); | |
343 | ||
344 | default: | |
345 | if( isspace( c ) ) /* One ' ' per whitespace region. */ | |
346 | { | |
347 | bufr[end] = ' '; | |
348 | i = end + 1; | |
349 | c = EatWhitespace( InFile ); | |
350 | } | |
351 | else /* All others verbatim. */ | |
352 | { | |
353 | bufr[i++] = c; | |
354 | end = i; | |
355 | c = getc( InFile ); | |
356 | } | |
357 | } | |
358 | } | |
359 | ||
360 | /* Now parse the value. */ | |
361 | c = EatWhitespace( InFile ); /* Again, trim leading whitespace. */ | |
362 | while( (EOF !=c) && (c > 0) ) | |
363 | { | |
364 | ||
365 | if( i > (bSize - 2) ) /* Make sure there's enough room. */ | |
366 | { | |
367 | bSize += BUFR_INC; | |
58cadc86 | 368 | bufr = realloc_array( bufr, char, bSize ); |
0b76cd63 AT |
369 | if( NULL == bufr ) |
370 | { | |
371 | rprintf(FERROR, "%s Memory re-allocation failure.", func) ; | |
372 | return( False ); | |
373 | } | |
374 | } | |
375 | ||
376 | switch( c ) | |
377 | { | |
378 | case '\r': /* Explicitly remove '\r' because the older */ | |
379 | c = getc( InFile ); /* version called fgets_slash() which also */ | |
380 | break; /* removes them. */ | |
381 | ||
382 | case '\n': /* Marks end of value unless there's a '\'. */ | |
383 | i = Continuation( bufr, i ); | |
384 | if( i < 0 ) | |
385 | c = 0; | |
386 | else | |
387 | { | |
2dc7b8bd | 388 | for( end = i; end >= 0 && isSpace(bufr + end); end-- ) |
0b76cd63 AT |
389 | ; |
390 | c = getc( InFile ); | |
391 | } | |
392 | break; | |
393 | ||
394 | default: /* All others verbatim. Note that spaces do */ | |
395 | bufr[i++] = c; /* not advance <end>. This allows trimming */ | |
396 | if( !isspace( c ) ) /* of whitespace at the end of the line. */ | |
397 | end = i; | |
398 | c = getc( InFile ); | |
399 | break; | |
400 | } | |
401 | } | |
402 | bufr[end] = '\0'; /* End of value. */ | |
403 | ||
404 | return( pfunc( bufr, &bufr[vstart] ) ); /* Pass name & value to pfunc(). */ | |
405 | } /* Parameter */ | |
406 | ||
407 | static BOOL Parse( FILE *InFile, | |
408 | BOOL (*sfunc)(char *), | |
409 | BOOL (*pfunc)(char *, char *) ) | |
410 | /* ------------------------------------------------------------------------ ** | |
411 | * Scan & parse the input. | |
412 | * | |
413 | * Input: InFile - Input source. | |
414 | * sfunc - Function to be called when a section name is scanned. | |
415 | * See Section(). | |
416 | * pfunc - Function to be called when a parameter is scanned. | |
417 | * See Parameter(). | |
418 | * | |
419 | * Output: True if the file was successfully scanned, else False. | |
420 | * | |
421 | * Notes: The input can be viewed in terms of 'lines'. There are four | |
422 | * types of lines: | |
423 | * Blank - May contain whitespace, otherwise empty. | |
424 | * Comment - First non-whitespace character is a ';' or '#'. | |
425 | * The remainder of the line is ignored. | |
426 | * Section - First non-whitespace character is a '['. | |
427 | * Parameter - The default case. | |
428 | * | |
429 | * ------------------------------------------------------------------------ ** | |
430 | */ | |
431 | { | |
432 | int c; | |
433 | ||
434 | c = EatWhitespace( InFile ); | |
435 | while( (EOF != c) && (c > 0) ) | |
436 | { | |
437 | switch( c ) | |
438 | { | |
439 | case '\n': /* Blank line. */ | |
440 | c = EatWhitespace( InFile ); | |
441 | break; | |
442 | ||
443 | case ';': /* Comment line. */ | |
444 | case '#': | |
445 | c = EatComment( InFile ); | |
446 | break; | |
447 | ||
448 | case '[': /* Section Header. */ | |
f9e940ef AT |
449 | if (!sfunc) return True; |
450 | if( !Section( InFile, sfunc ) ) | |
451 | return( False ); | |
452 | c = EatWhitespace( InFile ); | |
453 | break; | |
0b76cd63 AT |
454 | |
455 | case '\\': /* Bogus backslash. */ | |
456 | c = EatWhitespace( InFile ); | |
457 | break; | |
458 | ||
459 | default: /* Parameter line. */ | |
460 | if( !Parameter( InFile, pfunc, c ) ) | |
461 | return( False ); | |
462 | c = EatWhitespace( InFile ); | |
463 | break; | |
464 | } | |
465 | } | |
466 | return( True ); | |
467 | } /* Parse */ | |
468 | ||
469 | static FILE *OpenConfFile( char *FileName ) | |
470 | /* ------------------------------------------------------------------------ ** | |
471 | * Open a configuration file. | |
472 | * | |
473 | * Input: FileName - The pathname of the config file to be opened. | |
474 | * | |
475 | * Output: A pointer of type (FILE *) to the opened file, or NULL if the | |
476 | * file could not be opened. | |
477 | * | |
478 | * ------------------------------------------------------------------------ ** | |
479 | */ | |
480 | { | |
481 | FILE *OpenedFile; | |
482 | char *func = "params.c:OpenConfFile() -"; | |
483 | ||
484 | if( NULL == FileName || 0 == *FileName ) | |
485 | { | |
486 | rprintf(FERROR,"%s No configuration filename specified.\n", func); | |
487 | return( NULL ); | |
488 | } | |
489 | ||
0090cbdb | 490 | OpenedFile = fopen( FileName, "r" ); |
0b76cd63 AT |
491 | if( NULL == OpenedFile ) |
492 | { | |
415b5983 | 493 | rsyserr(FERROR, errno, "unable to open configuration file \"%s\"", |
45c49b52 | 494 | FileName); |
0b76cd63 AT |
495 | } |
496 | ||
497 | return( OpenedFile ); | |
498 | } /* OpenConfFile */ | |
499 | ||
500 | BOOL pm_process( char *FileName, | |
501 | BOOL (*sfunc)(char *), | |
502 | BOOL (*pfunc)(char *, char *) ) | |
503 | /* ------------------------------------------------------------------------ ** | |
504 | * Process the named parameter file. | |
505 | * | |
506 | * Input: FileName - The pathname of the parameter file to be opened. | |
507 | * sfunc - A pointer to a function that will be called when | |
508 | * a section name is discovered. | |
509 | * pfunc - A pointer to a function that will be called when | |
510 | * a parameter name and value are discovered. | |
511 | * | |
512 | * Output: TRUE if the file was successfully parsed, else FALSE. | |
513 | * | |
514 | * ------------------------------------------------------------------------ ** | |
515 | */ | |
516 | { | |
517 | int result; | |
518 | FILE *InFile; | |
519 | char *func = "params.c:pm_process() -"; | |
520 | ||
521 | InFile = OpenConfFile( FileName ); /* Open the config file. */ | |
522 | if( NULL == InFile ) | |
523 | return( False ); | |
524 | ||
525 | if( NULL != bufr ) /* If we already have a buffer */ | |
526 | result = Parse( InFile, sfunc, pfunc ); /* (recursive call), then just */ | |
527 | /* use it. */ | |
528 | ||
529 | else /* If we don't have a buffer */ | |
530 | { /* allocate one, then parse, */ | |
531 | bSize = BUFR_INC; /* then free. */ | |
58cadc86 | 532 | bufr = new_array( char, bSize ); |
0b76cd63 AT |
533 | if( NULL == bufr ) |
534 | { | |
535 | rprintf(FERROR,"%s memory allocation failure.\n", func); | |
536 | fclose(InFile); | |
537 | return( False ); | |
538 | } | |
539 | result = Parse( InFile, sfunc, pfunc ); | |
540 | free( bufr ); | |
541 | bufr = NULL; | |
542 | bSize = 0; | |
543 | } | |
544 | ||
545 | fclose(InFile); | |
546 | ||
547 | if( !result ) /* Generic failure. */ | |
548 | { | |
549 | rprintf(FERROR,"%s Failed. Error returned from params.c:parse().\n", func); | |
550 | return( False ); | |
551 | } | |
552 | ||
553 | return( True ); /* Generic success. */ | |
554 | } /* pm_process */ | |
555 | ||
556 | /* -------------------------------------------------------------------------- */ | |
f9e940ef | 557 |