1 /* -------------------------------------------------------------------------- **
2 * Microsoft Network Services for Unix, AKA., Andrew Tridgell's SAMBA.
4 * This module Copyright (C) 1990-1998 Karl Auer
6 * Rewritten almost completely by Christopher R. Hertel, 1997.
7 * This module Copyright (C) 1997-1998 by Christopher R. Hertel
9 * -------------------------------------------------------------------------- **
11 * This program is free software; you can redistribute it and/or modify
12 * it under the terms of the GNU General Public License as published by
13 * the Free Software Foundation; either version 3 of the License, or
14 * (at your option) any later version.
16 * This program is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 * GNU General Public License for more details.
21 * You should have received a copy of the GNU General Public License
22 * along with this program; if not, write to the Free Software
23 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
25 * -------------------------------------------------------------------------- **
29 * -------------------------------------------------------------------------- **
31 * This module performs lexical analysis and initial parsing of a
32 * Windows-like parameter file. It recognizes and handles four token
33 * types: section-name, parameter-name, parameter-value, and
34 * end-of-file. Comments and line continuation are handled
37 * The entry point to the module is function pm_process(). This
38 * function opens the source file, calls the Parse() function to parse
39 * the input, and then closes the file when either the EOF is reached
40 * or a fatal error is encountered.
42 * A sample parameter file might look like this:
45 * parameter one = value string
46 * parameter two = another value
48 * new parameter = some value or t'other
50 * The parameter file is divided into sections by section headers:
51 * section names enclosed in square brackets (eg. [section one]).
52 * Each section contains parameter lines, each of which consist of a
53 * parameter name and value delimited by an equal sign. Roughly, the
56 * <file> :== { <section> } EOF
58 * <section> :== <section header> { <parameter line> }
60 * <section header> :== '[' NAME ']'
62 * <parameter line> :== NAME '=' VALUE '\n'
64 * Blank lines and comment lines are ignored. Comment lines are lines
65 * beginning with either a semicolon (';') or a pound sign ('#').
67 * All whitespace in section names and parameter names is compressed
68 * to single spaces. Leading and trailing whitespace is stipped from
69 * both names and values.
71 * Only the first equals sign in a parameter line is significant.
72 * Parameter values may contain equals signs, square brackets and
73 * semicolons. Internal whitespace is retained in parameter values,
74 * with the exception of the '\r' character, which is stripped for
75 * historic reasons. Parameter names may not start with a left square
76 * bracket, an equal sign, a pound sign, or a semicolon, because these
77 * are used to identify other tokens.
79 * -------------------------------------------------------------------------- **
84 extern BOOL in_client;
86 /* -------------------------------------------------------------------------- **
93 /* -------------------------------------------------------------------------- **
96 * DEBUGLEVEL - The ubiquitous DEBUGLEVEL. This determines which DEBUG()
97 * messages will be produced.
98 * bufr - pointer to a global buffer. This is probably a kludge,
99 * but it was the nicest kludge I could think of (for now).
100 * bSize - The size of the global buffer <bufr>.
103 static char *bufr = NULL;
104 static int bSize = 0;
106 /* we can't use FILE* due to the 256 fd limit - use this cheap hack
115 static int mygetc(myFILE *f)
117 if (f->p >= f->buf+f->size)
119 /* be sure to return chars >127 as positive values */
120 return (int)( *(f->p++) & 0x00FF );
123 static void myfile_close(myFILE *f)
131 /* Find the end of the section. We must use mb functions for this. */
132 static int FindSectionEnd(myFILE *f)
134 f->end_section_p = strchr_m(f->p, ']');
135 return f->end_section_p ? 1 : 0;
138 static int AtSectionEnd(myFILE *f)
140 if (f->p == f->end_section_p + 1) {
141 f->end_section_p = NULL;
147 /* -------------------------------------------------------------------------- **
150 /* ------------------------------------------------------------------------ **
151 * Scan past whitespace (see ctype(3C)) and return the first non-whitespace
152 * character, or newline, or EOF.
154 * Input: InFile - Input source.
156 * Output: The next non-whitespace character in the input stream.
158 * Notes: Because the config files use a line-oriented grammar, we
159 * explicitly exclude the newline character from the list of
160 * whitespace characters.
161 * - Note that both EOF (-1) and the nul character ('\0') are
162 * considered end-of-file markers.
164 * ------------------------------------------------------------------------ **
167 static int EatWhitespace( myFILE *InFile )
171 for( c = mygetc( InFile ); isspace( c ) && ('\n' != c); c = mygetc( InFile ) )
176 /* ------------------------------------------------------------------------ **
177 * Scan to the end of a comment.
179 * Input: InFile - Input source.
181 * Output: The character that marks the end of the comment. Normally,
182 * this will be a newline, but it *might* be an EOF.
184 * Notes: Because the config files use a line-oriented grammar, we
185 * explicitly exclude the newline character from the list of
186 * whitespace characters.
187 * - Note that both EOF (-1) and the nul character ('\0') are
188 * considered end-of-file markers.
190 * ------------------------------------------------------------------------ **
193 static int EatComment( myFILE *InFile )
197 for( c = mygetc( InFile ); ('\n'!=c) && (EOF!=c) && (c>0); c = mygetc( InFile ) )
202 /*****************************************************************************
203 * Scan backards within a string to discover if the last non-whitespace
204 * character is a line-continuation character ('\\').
206 * Input: line - A pointer to a buffer containing the string to be
208 * pos - This is taken to be the offset of the end of the
209 * string. This position is *not* scanned.
211 * Output: The offset of the '\\' character if it was found, or -1 to
212 * indicate that it was not.
214 *****************************************************************************/
216 static int Continuation(char *line, int pos )
219 while( (pos >= 0) && isspace((int)line[pos]))
222 return (((pos >= 0) && ('\\' == line[pos])) ? pos : -1 );
225 /* ------------------------------------------------------------------------ **
226 * Scan a section name, and pass the name to function sfunc().
228 * Input: InFile - Input source.
229 * sfunc - Pointer to the function to be called if the section
230 * name is successfully read.
232 * Output: True if the section name was read and True was returned from
233 * <sfunc>. False if <sfunc> failed or if a lexical error was
236 * ------------------------------------------------------------------------ **
239 static BOOL Section( myFILE *InFile, BOOL (*sfunc)(const char *) )
244 const char *func = "params.c:Section() -";
246 i = 0; /* <i> is the offset of the next free byte in bufr[] and */
247 end = 0; /* <end> is the current "end of string" offset. In most */
248 /* cases these will be the same, but if the last */
249 /* character written to bufr[] is a space, then <end> */
250 /* will be one less than <i>. */
253 /* Find the end of the section. We must use mb functions for this. */
254 if (!FindSectionEnd(InFile)) {
255 DEBUG(0, ("%s No terminating ']' character in section.\n", func) );
259 c = EatWhitespace( InFile ); /* We've already got the '['. Scan */
260 /* past initial white space. */
262 while( (EOF != c) && (c > 0) ) {
263 /* Check that the buffer is big enough for the next character. */
264 if( i > (bSize - 2) ) {
265 char *tb = (char *)SMB_REALLOC_KEEP_OLD_ON_ERROR( bufr, bSize +BUFR_INC );
267 DEBUG(0, ("%s Memory re-allocation failure.", func) );
274 /* Handle a single character other than section end. */
276 case '\n': /* Got newline before closing ']'. */
277 i = Continuation( bufr, i ); /* Check for line continuation. */
280 DEBUG(0, ("%s Badly formed line in configuration file: %s\n", func, bufr ));
283 end = ( (i > 0) && (' ' == bufr[i - 1]) ) ? (i - 1) : (i);
284 c = mygetc( InFile ); /* Continue with next line. */
287 default: /* All else are a valid name chars. */
289 /* One space per whitespace region. */
292 c = EatWhitespace( InFile );
296 c = mygetc( InFile );
300 if (AtSectionEnd(InFile)) {
301 /* Got to the closing bracket. */
304 /* Don't allow an empty name. */
305 DEBUG(0, ("%s Empty section name in configuration file.\n", func ));
308 if( !sfunc(bufr) ) /* Got a valid name. Deal with it. */
310 EatComment( InFile ); /* Finish off the line. */
316 /* We arrive here if we've met the EOF before the closing bracket. */
317 DEBUG(0, ("%s Unexpected EOF in the configuration file: %s\n", func, bufr ));
321 /* ------------------------------------------------------------------------ **
322 * Scan a parameter name and value, and pass these two fields to pfunc().
324 * Input: InFile - The input source.
325 * pfunc - A pointer to the function that will be called to
326 * process the parameter, once it has been scanned.
327 * c - The first character of the parameter name, which
328 * would have been read by Parse(). Unlike a comment
329 * line or a section header, there is no lead-in
330 * character that can be discarded.
332 * Output: True if the parameter name and value were scanned and processed
333 * successfully, else False.
335 * Notes: This function is in two parts. The first loop scans the
336 * parameter name. Internal whitespace is compressed, and an
337 * equal sign (=) terminates the token. Leading and trailing
338 * whitespace is discarded. The second loop scans the parameter
339 * value. When both have been successfully identified, they are
340 * passed to pfunc() for processing.
342 * ------------------------------------------------------------------------ **
345 static BOOL Parameter( myFILE *InFile, BOOL (*pfunc)(const char *, const char *), int c )
347 int i = 0; /* Position within bufr. */
348 int end = 0; /* bufr[end] is current end-of-string. */
349 int vstart = 0; /* Starting position of the parameter value. */
350 const char *func = "params.c:Parameter() -";
352 /* Read the parameter name. */
353 while( 0 == vstart ) {
354 /* Loop until we've found the start of the value. */
355 if( i > (bSize - 2) ) {
356 /* Ensure there's space for next char. */
357 char *tb = (char *)SMB_REALLOC_KEEP_OLD_ON_ERROR( bufr, bSize + BUFR_INC );
359 DEBUG(0, ("%s Memory re-allocation failure.", func) );
367 case '=': /* Equal sign marks end of param name. */
369 /* Don't allow an empty name. */
370 DEBUG(0, ("%s Invalid parameter name in config. file.\n", func ));
373 bufr[end++] = '\0'; /* Mark end of string & advance. */
374 i = end; /* New string starts here. */
375 vstart = end; /* New string is parameter value. */
376 bufr[i] = '\0'; /* New string is nul, for now. */
379 case '\n': /* Find continuation char, else error. */
380 i = Continuation( bufr, i );
383 DEBUG(1,("%s Ignoring badly formed line in configuration file: %s\n", func, bufr ));
386 end = ( (i > 0) && (' ' == bufr[i - 1]) ) ? (i - 1) : (i);
387 c = mygetc( InFile ); /* Read past eoln. */
390 case '\0': /* Shouldn't have EOF within param name. */
393 DEBUG(1,("%s Unexpected end-of-file at: %s\n", func, bufr ));
398 /* One ' ' per whitespace region. */
401 c = EatWhitespace( InFile );
405 c = mygetc( InFile );
410 /* Now parse the value. */
411 c = EatWhitespace( InFile ); /* Again, trim leading whitespace. */
412 while( (EOF !=c) && (c > 0) ) {
413 if( i > (bSize - 2) ) {
414 /* Make sure there's enough room. */
415 char *tb = (char *)SMB_REALLOC_KEEP_OLD_ON_ERROR( bufr, bSize + BUFR_INC );
417 DEBUG(0, ("%s Memory re-allocation failure.", func));
425 case '\r': /* Explicitly remove '\r' because the older */
426 c = mygetc( InFile ); /* version called fgets_slash() which also */
427 break; /* removes them. */
429 case '\n': /* Marks end of value unless there's a '\'. */
430 i = Continuation( bufr, i );
434 for( end = i; (end >= 0) && isspace((int)bufr[end]); end-- )
436 c = mygetc( InFile );
440 default: /* All others verbatim. Note that spaces do not advance <end>. This allows trimming */
442 if( !isspace( c ) ) /* of whitespace at the end of the line. */
444 c = mygetc( InFile );
448 bufr[end] = '\0'; /* End of value. */
450 return( pfunc( bufr, &bufr[vstart] ) ); /* Pass name & value to pfunc(). */
453 /* ------------------------------------------------------------------------ **
454 * Scan & parse the input.
456 * Input: InFile - Input source.
457 * sfunc - Function to be called when a section name is scanned.
459 * pfunc - Function to be called when a parameter is scanned.
462 * Output: True if the file was successfully scanned, else False.
464 * Notes: The input can be viewed in terms of 'lines'. There are four
466 * Blank - May contain whitespace, otherwise empty.
467 * Comment - First non-whitespace character is a ';' or '#'.
468 * The remainder of the line is ignored.
469 * Section - First non-whitespace character is a '['.
470 * Parameter - The default case.
472 * ------------------------------------------------------------------------ **
475 static BOOL Parse( myFILE *InFile,
476 BOOL (*sfunc)(const char *),
477 BOOL (*pfunc)(const char *, const char *) )
481 c = EatWhitespace( InFile );
482 while( (EOF != c) && (c > 0) ) {
484 case '\n': /* Blank line. */
485 c = EatWhitespace( InFile );
488 case ';': /* Comment line. */
490 c = EatComment( InFile );
493 case '[': /* Section Header. */
494 if( !Section( InFile, sfunc ) )
496 c = EatWhitespace( InFile );
499 case '\\': /* Bogus backslash. */
500 c = EatWhitespace( InFile );
503 default: /* Parameter line. */
504 if( !Parameter( InFile, pfunc, c ) )
506 c = EatWhitespace( InFile );
513 /* ------------------------------------------------------------------------ **
514 * Open a configuration file.
516 * Input: FileName - The pathname of the config file to be opened.
518 * Output: A pointer of type (char **) to the lines of the file
520 * ------------------------------------------------------------------------ **
523 static myFILE *OpenConfFile( const char *FileName )
525 const char *func = "params.c:OpenConfFile() -";
526 int lvl = in_client?1:0;
529 ret = SMB_MALLOC_P(myFILE);
533 ret->buf = file_load(FileName, &ret->size, 0);
534 if( NULL == ret->buf ) {
535 DEBUG( lvl, ("%s Unable to open configuration file \"%s\":\n\t%s\n",
536 func, FileName, strerror(errno)) );
542 ret->end_section_p = NULL;
546 /* ------------------------------------------------------------------------ **
547 * Process the named parameter file.
549 * Input: FileName - The pathname of the parameter file to be opened.
550 * sfunc - A pointer to a function that will be called when
551 * a section name is discovered.
552 * pfunc - A pointer to a function that will be called when
553 * a parameter name and value are discovered.
555 * Output: TRUE if the file was successfully parsed, else FALSE.
557 * ------------------------------------------------------------------------ **
560 BOOL pm_process( const char *FileName,
561 BOOL (*sfunc)(const char *),
562 BOOL (*pfunc)(const char *, const char *) )
566 const char *func = "params.c:pm_process() -";
568 InFile = OpenConfFile( FileName ); /* Open the config file. */
572 DEBUG( 3, ("%s Processing configuration file \"%s\"\n", func, FileName) );
575 /* If we already have a buffer */
576 /* (recursive call), then just */
578 result = Parse( InFile, sfunc, pfunc );
581 bufr = (char *)SMB_MALLOC( bSize );
583 DEBUG(0,("%s memory allocation failure.\n", func));
584 myfile_close(InFile);
588 result = Parse( InFile, sfunc, pfunc );
593 myfile_close(InFile);
596 DEBUG(0,("%s Failed. Error returned from params.c:parse().\n", func));