2 This modules is based on the params.c module from Samba, written by Karl Auer
3 and much modifed by Christopher Hertel.
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.
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.
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.
19 * -------------------------------------------------------------------------- **
23 * -------------------------------------------------------------------------- **
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
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.
36 * A sample parameter file might look like this:
39 * parameter one = value string
40 * parameter two = another value
42 * new parameter = some value or t'other
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
50 * <file> :== { <section> } EOF
52 * <section> :== <section header> { <parameter line> }
54 * <section header> :== '[' NAME ']'
56 * <parameter line> :== NAME '=' VALUE '\n'
58 * Blank lines and comment lines are ignored. Comment lines are lines
59 * beginning with either a semicolon (';') or a pound sign ('#').
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.
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.
73 * -------------------------------------------------------------------------- **
78 /* -------------------------------------------------------------------------- **
85 /* -------------------------------------------------------------------------- **
88 * bufr - pointer to a global buffer. This is probably a kludge,
89 * but it was the nicest kludge I could think of (for now).
90 * bSize - The size of the global buffer <bufr>.
93 static char *bufr = NULL;
96 /* -------------------------------------------------------------------------- **
100 static int EatWhitespace( FILE *InFile )
101 /* ------------------------------------------------------------------------ **
102 * Scan past whitespace (see ctype(3C)) and return the first non-whitespace
103 * character, or newline, or EOF.
105 * Input: InFile - Input source.
107 * Output: The next non-whitespace character in the input stream.
109 * Notes: Because the config files use a line-oriented grammar, we
110 * explicitly exclude the newline character from the list of
111 * whitespace characters.
112 * - Note that both EOF (-1) and the nul character ('\0') are
113 * considered end-of-file markers.
115 * ------------------------------------------------------------------------ **
120 for( c = getc( InFile ); isspace( c ) && ('\n' != c); c = getc( InFile ) )
123 } /* EatWhitespace */
125 static int EatComment( FILE *InFile )
126 /* ------------------------------------------------------------------------ **
127 * Scan to the end of a comment.
129 * Input: InFile - Input source.
131 * Output: The character that marks the end of the comment. Normally,
132 * this will be a newline, but it *might* be an EOF.
134 * Notes: Because the config files use a line-oriented grammar, we
135 * explicitly exclude the newline character from the list of
136 * whitespace characters.
137 * - Note that both EOF (-1) and the nul character ('\0') are
138 * considered end-of-file markers.
140 * ------------------------------------------------------------------------ **
145 for( c = getc( InFile ); ('\n'!=c) && (EOF!=c) && (c>0); c = getc( InFile ) )
150 static int Continuation( char *line, int pos )
151 /* ------------------------------------------------------------------------ **
152 * Scan backards within a string to discover if the last non-whitespace
153 * character is a line-continuation character ('\\').
155 * Input: line - A pointer to a buffer containing the string to be
157 * pos - This is taken to be the offset of the end of the
158 * string. This position is *not* scanned.
160 * Output: The offset of the '\\' character if it was found, or -1 to
161 * indicate that it was not.
163 * ------------------------------------------------------------------------ **
167 while( (pos >= 0) && isspace(line[pos]) )
170 return( ((pos >= 0) && ('\\' == line[pos])) ? pos : -1 );
174 static BOOL Section( FILE *InFile, BOOL (*sfunc)(char *) )
175 /* ------------------------------------------------------------------------ **
176 * Scan a section name, and pass the name to function sfunc().
178 * Input: InFile - Input source.
179 * sfunc - Pointer to the function to be called if the section
180 * name is successfully read.
182 * Output: True if the section name was read and True was returned from
183 * <sfunc>. False if <sfunc> failed or if a lexical error was
186 * ------------------------------------------------------------------------ **
192 char *func = "params.c:Section() -";
194 i = 0; /* <i> is the offset of the next free byte in bufr[] and */
195 end = 0; /* <end> is the current "end of string" offset. In most */
196 /* cases these will be the same, but if the last */
197 /* character written to bufr[] is a space, then <end> */
198 /* will be one less than <i>. */
200 c = EatWhitespace( InFile ); /* We've already got the '['. Scan */
201 /* past initial white space. */
203 while( (EOF != c) && (c > 0) )
206 /* Check that the buffer is big enough for the next character. */
207 if( i > (bSize - 2) )
210 bufr = Realloc( bufr, bSize );
213 rprintf(FERROR, "%s Memory re-allocation failure.", func);
218 /* Handle a single character. */
221 case ']': /* Found the closing bracket. */
223 if( 0 == end ) /* Don't allow an empty name. */
225 rprintf(FERROR, "%s Empty section name in configuration file.\n", func );
228 if( !sfunc( bufr ) ) /* Got a valid name. Deal with it. */
230 (void)EatComment( InFile ); /* Finish off the line. */
233 case '\n': /* Got newline before closing ']'. */
234 i = Continuation( bufr, i ); /* Check for line continuation. */
238 rprintf(FERROR, "%s Badly formed line in configuration file: %s\n",
242 end = ( (i > 0) && (' ' == bufr[i - 1]) ) ? (i - 1) : (i);
243 c = getc( InFile ); /* Continue with next line. */
246 default: /* All else are a valid name chars. */
247 if( isspace( c ) ) /* One space per whitespace region. */
251 c = EatWhitespace( InFile );
253 else /* All others copy verbatim. */
262 /* We arrive here if we've met the EOF before the closing bracket. */
263 rprintf(FERROR, "%s Unexpected EOF in the configuration file: %s\n", func, bufr );
267 static BOOL Parameter( FILE *InFile, BOOL (*pfunc)(char *, char *), int c )
268 /* ------------------------------------------------------------------------ **
269 * Scan a parameter name and value, and pass these two fields to pfunc().
271 * Input: InFile - The input source.
272 * pfunc - A pointer to the function that will be called to
273 * process the parameter, once it has been scanned.
274 * c - The first character of the parameter name, which
275 * would have been read by Parse(). Unlike a comment
276 * line or a section header, there is no lead-in
277 * character that can be discarded.
279 * Output: True if the parameter name and value were scanned and processed
280 * successfully, else False.
282 * Notes: This function is in two parts. The first loop scans the
283 * parameter name. Internal whitespace is compressed, and an
284 * equal sign (=) terminates the token. Leading and trailing
285 * whitespace is discarded. The second loop scans the parameter
286 * value. When both have been successfully identified, they are
287 * passed to pfunc() for processing.
289 * ------------------------------------------------------------------------ **
292 int i = 0; /* Position within bufr. */
293 int end = 0; /* bufr[end] is current end-of-string. */
294 int vstart = 0; /* Starting position of the parameter value. */
295 char *func = "params.c:Parameter() -";
297 /* Read the parameter name. */
298 while( 0 == vstart ) /* Loop until we've found the start of the value. */
301 if( i > (bSize - 2) ) /* Ensure there's space for next char. */
304 bufr = Realloc( bufr, bSize );
307 rprintf(FERROR, "%s Memory re-allocation failure.", func) ;
314 case '=': /* Equal sign marks end of param name. */
315 if( 0 == end ) /* Don't allow an empty name. */
317 rprintf(FERROR, "%s Invalid parameter name in config. file.\n", func );
320 bufr[end++] = '\0'; /* Mark end of string & advance. */
321 i = end; /* New string starts here. */
322 vstart = end; /* New string is parameter value. */
323 bufr[i] = '\0'; /* New string is nul, for now. */
326 case '\n': /* Find continuation char, else error. */
327 i = Continuation( bufr, i );
331 rprintf(FERROR, "%s Ignoring badly formed line in configuration file: %s\n",
335 end = ( (i > 0) && (' ' == bufr[i - 1]) ) ? (i - 1) : (i);
336 c = getc( InFile ); /* Read past eoln. */
339 case '\0': /* Shouldn't have EOF within param name. */
342 rprintf(FERROR, "%s Unexpected end-of-file at: %s\n", func, bufr );
346 if( isspace( c ) ) /* One ' ' per whitespace region. */
350 c = EatWhitespace( InFile );
352 else /* All others verbatim. */
361 /* Now parse the value. */
362 c = EatWhitespace( InFile ); /* Again, trim leading whitespace. */
363 while( (EOF !=c) && (c > 0) )
366 if( i > (bSize - 2) ) /* Make sure there's enough room. */
369 bufr = Realloc( bufr, bSize );
372 rprintf(FERROR, "%s Memory re-allocation failure.", func) ;
379 case '\r': /* Explicitly remove '\r' because the older */
380 c = getc( InFile ); /* version called fgets_slash() which also */
381 break; /* removes them. */
383 case '\n': /* Marks end of value unless there's a '\'. */
384 i = Continuation( bufr, i );
389 for( end = i; (end >= 0) && isspace(bufr[end]); end-- )
395 default: /* All others verbatim. Note that spaces do */
396 bufr[i++] = c; /* not advance <end>. This allows trimming */
397 if( !isspace( c ) ) /* of whitespace at the end of the line. */
403 bufr[end] = '\0'; /* End of value. */
405 return( pfunc( bufr, &bufr[vstart] ) ); /* Pass name & value to pfunc(). */
408 static BOOL Parse( FILE *InFile,
409 BOOL (*sfunc)(char *),
410 BOOL (*pfunc)(char *, char *) )
411 /* ------------------------------------------------------------------------ **
412 * Scan & parse the input.
414 * Input: InFile - Input source.
415 * sfunc - Function to be called when a section name is scanned.
417 * pfunc - Function to be called when a parameter is scanned.
420 * Output: True if the file was successfully scanned, else False.
422 * Notes: The input can be viewed in terms of 'lines'. There are four
424 * Blank - May contain whitespace, otherwise empty.
425 * Comment - First non-whitespace character is a ';' or '#'.
426 * The remainder of the line is ignored.
427 * Section - First non-whitespace character is a '['.
428 * Parameter - The default case.
430 * ------------------------------------------------------------------------ **
435 c = EatWhitespace( InFile );
436 while( (EOF != c) && (c > 0) )
440 case '\n': /* Blank line. */
441 c = EatWhitespace( InFile );
444 case ';': /* Comment line. */
446 c = EatComment( InFile );
449 case '[': /* Section Header. */
450 if (!sfunc) return True;
451 if( !Section( InFile, sfunc ) )
453 c = EatWhitespace( InFile );
456 case '\\': /* Bogus backslash. */
457 c = EatWhitespace( InFile );
460 default: /* Parameter line. */
461 if( !Parameter( InFile, pfunc, c ) )
463 c = EatWhitespace( InFile );
470 static FILE *OpenConfFile( char *FileName )
471 /* ------------------------------------------------------------------------ **
472 * Open a configuration file.
474 * Input: FileName - The pathname of the config file to be opened.
476 * Output: A pointer of type (FILE *) to the opened file, or NULL if the
477 * file could not be opened.
479 * ------------------------------------------------------------------------ **
483 char *func = "params.c:OpenConfFile() -";
485 if( NULL == FileName || 0 == *FileName )
487 rprintf(FERROR,"%s No configuration filename specified.\n", func);
491 OpenedFile = fopen( FileName, "r" );
492 if( NULL == OpenedFile )
494 rprintf(FERROR,"%s Unable to open configuration file \"%s\":\n\t%s\n",
495 func, FileName, strerror(errno));
498 return( OpenedFile );
501 BOOL pm_process( char *FileName,
502 BOOL (*sfunc)(char *),
503 BOOL (*pfunc)(char *, char *) )
504 /* ------------------------------------------------------------------------ **
505 * Process the named parameter file.
507 * Input: FileName - The pathname of the parameter file to be opened.
508 * sfunc - A pointer to a function that will be called when
509 * a section name is discovered.
510 * pfunc - A pointer to a function that will be called when
511 * a parameter name and value are discovered.
513 * Output: TRUE if the file was successfully parsed, else FALSE.
515 * ------------------------------------------------------------------------ **
520 char *func = "params.c:pm_process() -";
522 InFile = OpenConfFile( FileName ); /* Open the config file. */
526 if( NULL != bufr ) /* If we already have a buffer */
527 result = Parse( InFile, sfunc, pfunc ); /* (recursive call), then just */
530 else /* If we don't have a buffer */
531 { /* allocate one, then parse, */
532 bSize = BUFR_INC; /* then free. */
533 bufr = (char *)malloc( bSize );
536 rprintf(FERROR,"%s memory allocation failure.\n", func);
540 result = Parse( InFile, sfunc, pfunc );
548 if( !result ) /* Generic failure. */
550 rprintf(FERROR,"%s Failed. Error returned from params.c:parse().\n", func);
554 return( True ); /* Generic success. */
557 /* -------------------------------------------------------------------------- */