params.c

   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
  78 /* -------------------------------------------------------------------------- **
  79  * Constants...
  80  */
  81
  82 #define BUFR_INC 1024
  83
  84
  85 /* -------------------------------------------------------------------------- **
  86  * Variables...
  87  *
  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>.
  91  */
  92
  93 static char *bufr  = NULL;
  94 static int   bSize = 0;
  95
  96 /* -------------------------------------------------------------------------- **
  97  * Functions...
  98  */
  99
 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.
 104    *
 105    *  Input:  InFile  - Input source.
 106    *
 107    *  Output: The next non-whitespace character in the input stream.
 108    *
 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.
 114    *
 115    * ------------------------------------------------------------------------ **
 116    */
 117   {
 118   int c;
 119
 120   for( c = getc( InFile ); isspace( c ) && ('\n' != c); c = getc( InFile ) )
 121     ;
 122   return( c );
 123   } /* EatWhitespace */
 124
 125 static int EatComment( FILE *InFile )
 126   /* ------------------------------------------------------------------------ **
 127    * Scan to the end of a comment.
 128    *
 129    *  Input:  InFile  - Input source.
 130    *
 131    *  Output: The character that marks the end of the comment.  Normally,
 132    *          this will be a newline, but it *might* be an EOF.
 133    *
 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.
 139    *
 140    * ------------------------------------------------------------------------ **
 141    */
 142   {
 143   int c;
 144
 145   for( c = getc( InFile ); ('\n'!=c) && (EOF!=c) && (c>0); c = getc( InFile ) )
 146     ;
 147   return( c );
 148   } /* EatComment */
 149
 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 ('\\').
 154    *
 155    *  Input:  line  - A pointer to a buffer containing the string to be
 156    *                  scanned.
 157    *          pos   - This is taken to be the offset of the end of the
 158    *                  string.  This position is *not* scanned.
 159    *
 160    *  Output: The offset of the '\\' character if it was found, or -1 to
 161    *          indicate that it was not.
 162    *
 163    * ------------------------------------------------------------------------ **
 164    */
 165   {
 166   pos--;
 167   while( (pos >= 0) && isspace(((unsigned char *)line)[pos]) )
 168      pos--;
 169
 170   return( ((pos >= 0) && ('\\' == line[pos])) ? pos : -1 );
 171   } /* Continuation */
 172
 173
 174 static BOOL Section( FILE *InFile, BOOL (*sfunc)(char *) )
 175   /* ------------------------------------------------------------------------ **
 176    * Scan a section name, and pass the name to function sfunc().
 177    *
 178    *  Input:  InFile  - Input source.
 179    *          sfunc   - Pointer to the function to be called if the section
 180    *                    name is successfully read.
 181    *
 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
 184    *          encountered.
 185    *
 186    * ------------------------------------------------------------------------ **
 187    */
 188   {
 189   int   c;
 190   int   i;
 191   int   end;
 192   char *func  = "params.c:Section() -";
 193
 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>.                             */
 199
 200   c = EatWhitespace( InFile );    /* We've already got the '['.  Scan */
 201                                   /* past initial white space.        */
 202
 203   while( (EOF != c) && (c > 0) )
 204     {
 205
 206     /* Check that the buffer is big enough for the next character. */
 207     if( i > (bSize - 2) )
 208       {
 209       bSize += BUFR_INC;
 210       bufr   = Realloc( bufr, bSize );
 211       if( NULL == bufr )
 212         {
 213         rprintf(FERROR, "%s Memory re-allocation failure.", func);
 214         return( False );
 215         }
 216       }
 217
 218     /* Handle a single character. */
 219     switch( c )
 220       {
 221       case ']':                       /* Found the closing bracket.         */
 222         bufr[end] = '\0';
 223         if( 0 == end )                  /* Don't allow an empty name.       */
 224           {
 225           rprintf(FERROR, "%s Empty section name in configuration file.\n", func );
 226           return( False );
 227           }
 228         if( !sfunc( bufr ) )            /* Got a valid name.  Deal with it. */
 229           return( False );
 230         (void)EatComment( InFile );     /* Finish off the line.             */
 231         return( True );
 232
 233       case '\n':                      /* Got newline before closing ']'.    */
 234         i = Continuation( bufr, i );    /* Check for line continuation.     */
 235         if( i < 0 )
 236           {
 237           bufr[end] = '\0';
 238           rprintf(FERROR, "%s Badly formed line in configuration file: %s\n",
 239                    func, bufr );
 240           return( False );
 241           }
 242         end = ( (i > 0) && (' ' == bufr[i - 1]) ) ? (i - 1) : (i);
 243         c = getc( InFile );             /* Continue with next line.         */
 244         break;
 245
 246       default:                        /* All else are a valid name chars.   */
 247         if( isspace( c ) )              /* One space per whitespace region. */
 248           {
 249           bufr[end] = ' ';
 250           i = end + 1;
 251           c = EatWhitespace( InFile );
 252           }
 253         else                            /* All others copy verbatim.        */
 254           {
 255           bufr[i++] = c;
 256           end = i;
 257           c = getc( InFile );
 258           }
 259       }
 260     }
 261
 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 );
 264   return( False );
 265   } /* Section */
 266
 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().
 270    *
 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.
 278    *
 279    *  Output: True if the parameter name and value were scanned and processed
 280    *          successfully, else False.
 281    *
 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.
 288    *
 289    * ------------------------------------------------------------------------ **
 290    */
 291   {
 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() -";
 296
 297   /* Read the parameter name. */
 298   while( 0 == vstart )  /* Loop until we've found the start of the value. */
 299     {
 300
 301     if( i > (bSize - 2) )       /* Ensure there's space for next char.    */
 302       {
 303       bSize += BUFR_INC;
 304       bufr   = Realloc( bufr, bSize );
 305       if( NULL == bufr )
 306         {
 307         rprintf(FERROR, "%s Memory re-allocation failure.", func) ;
 308         return( False );
 309         }
 310       }
 311
 312     switch( c )
 313       {
 314       case '=':                 /* Equal sign marks end of param name. */
 315         if( 0 == end )              /* Don't allow an empty name.      */
 316           {
 317           rprintf(FERROR, "%s Invalid parameter name in config. file.\n", func );
 318           return( False );
 319           }
 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.     */
 324         break;
 325
 326       case '\n':                /* Find continuation char, else error. */
 327         i = Continuation( bufr, i );
 328         if( i < 0 )
 329           {
 330           bufr[end] = '\0';
 331           rprintf(FERROR, "%s Ignoring badly formed line in configuration file: %s\n",
 332                    func, bufr );
 333           return( True );
 334           }
 335         end = ( (i > 0) && (' ' == bufr[i - 1]) ) ? (i - 1) : (i);
 336         c = getc( InFile );       /* Read past eoln.                   */
 337         break;
 338
 339       case '\0':                /* Shouldn't have EOF within param name. */
 340       case EOF:
 341         bufr[i] = '\0';
 342         rprintf(FERROR, "%s Unexpected end-of-file at: %s\n", func, bufr );
 343         return( True );
 344
 345       default:
 346         if( isspace( c ) )     /* One ' ' per whitespace region.       */
 347           {
 348           bufr[end] = ' ';
 349           i = end + 1;
 350           c = EatWhitespace( InFile );
 351           }
 352         else                   /* All others verbatim.                 */
 353           {
 354           bufr[i++] = c;
 355           end = i;
 356           c = getc( InFile );
 357           }
 358       }
 359     }
 360
 361   /* Now parse the value. */
 362   c = EatWhitespace( InFile );  /* Again, trim leading whitespace. */
 363   while( (EOF !=c) && (c > 0) )
 364     {
 365
 366     if( i > (bSize - 2) )       /* Make sure there's enough room. */
 367       {
 368       bSize += BUFR_INC;
 369       bufr   = Realloc( bufr, bSize );
 370       if( NULL == bufr )
 371         {
 372         rprintf(FERROR, "%s Memory re-allocation failure.", func) ;
 373         return( False );
 374         }
 375       }
 376
 377     switch( c )
 378       {
 379       case '\r':              /* Explicitly remove '\r' because the older */
 380         c = getc( InFile );   /* version called fgets_slash() which also  */
 381         break;                /* removes them.                            */
 382
 383       case '\n':              /* Marks end of value unless there's a '\'. */
 384         i = Continuation( bufr, i );
 385         if( i < 0 )
 386           c = 0;
 387         else
 388           {
 389           for( end = i; (end >= 0) && isspace(((unsigned char *) bufr)[end]); end-- )
 390             ;
 391           c = getc( InFile );
 392           }
 393         break;
 394
 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.     */
 398           end = i;
 399         c = getc( InFile );
 400         break;
 401       }
 402     }
 403   bufr[end] = '\0';          /* End of value. */
 404
 405   return( pfunc( bufr, &bufr[vstart] ) );   /* Pass name & value to pfunc().  */
 406   } /* Parameter */
 407
 408 static BOOL Parse( FILE *InFile,
 409                    BOOL (*sfunc)(char *),
 410                    BOOL (*pfunc)(char *, char *) )
 411   /* ------------------------------------------------------------------------ **
 412    * Scan & parse the input.
 413    *
 414    *  Input:  InFile  - Input source.
 415    *          sfunc   - Function to be called when a section name is scanned.
 416    *                    See Section().
 417    *          pfunc   - Function to be called when a parameter is scanned.
 418    *                    See Parameter().
 419    *
 420    *  Output: True if the file was successfully scanned, else False.
 421    *
 422    *  Notes:  The input can be viewed in terms of 'lines'.  There are four
 423    *          types of lines:
 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.
 429    *
 430    * ------------------------------------------------------------------------ **
 431    */
 432   {
 433   int    c;
 434
 435   c = EatWhitespace( InFile );
 436   while( (EOF != c) && (c > 0) )
 437     {
 438     switch( c )
 439       {
 440       case '\n':                        /* Blank line. */
 441         c = EatWhitespace( InFile );
 442         break;
 443
 444       case ';':                         /* Comment line. */
 445       case '#':
 446         c = EatComment( InFile );
 447         break;
 448
 449       case '[':                         /* Section Header. */
 450               if (!sfunc) return True;
 451               if( !Section( InFile, sfunc ) )
 452                       return( False );
 453               c = EatWhitespace( InFile );
 454               break;
 455
 456       case '\\':                        /* Bogus backslash. */
 457         c = EatWhitespace( InFile );
 458         break;
 459
 460       default:                          /* Parameter line. */
 461         if( !Parameter( InFile, pfunc, c ) )
 462           return( False );
 463         c = EatWhitespace( InFile );
 464         break;
 465       }
 466     }
 467   return( True );
 468   } /* Parse */
 469
 470 static FILE *OpenConfFile( char *FileName )
 471   /* ------------------------------------------------------------------------ **
 472    * Open a configuration file.
 473    *
 474    *  Input:  FileName  - The pathname of the config file to be opened.
 475    *
 476    *  Output: A pointer of type (FILE *) to the opened file, or NULL if the
 477    *          file could not be opened.
 478    *
 479    * ------------------------------------------------------------------------ **
 480    */
 481   {
 482   FILE *OpenedFile;
 483   char *func = "params.c:OpenConfFile() -";
 484
 485   if( NULL == FileName || 0 == *FileName )
 486     {
 487     rprintf(FERROR,"%s No configuration filename specified.\n", func);
 488     return( NULL );
 489     }
 490
 491   OpenedFile = fopen( FileName, "r" );
 492   if( NULL == OpenedFile )
 493     {
 494     rprintf(FERROR,"rsync: unable to open configuration file \"%s\": %s\n",
 495             FileName, strerror(errno));
 496     }
 497
 498   return( OpenedFile );
 499   } /* OpenConfFile */
 500
 501 BOOL pm_process( char *FileName,
 502                  BOOL (*sfunc)(char *),
 503                  BOOL (*pfunc)(char *, char *) )
 504   /* ------------------------------------------------------------------------ **
 505    * Process the named parameter file.
 506    *
 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.
 512    *
 513    *  Output: TRUE if the file was successfully parsed, else FALSE.
 514    *
 515    * ------------------------------------------------------------------------ **
 516    */
 517   {
 518   int   result;
 519   FILE *InFile;
 520   char *func = "params.c:pm_process() -";
 521
 522   InFile = OpenConfFile( FileName );          /* Open the config file. */
 523   if( NULL == InFile )
 524     return( False );
 525
 526   if( NULL != bufr )                          /* If we already have a buffer */
 527     result = Parse( InFile, sfunc, pfunc );   /* (recursive call), then just */
 528                                               /* use it.                     */
 529
 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 );
 534     if( NULL == bufr )
 535       {
 536       rprintf(FERROR,"%s memory allocation failure.\n", func);
 537       fclose(InFile);
 538       return( False );
 539       }
 540     result = Parse( InFile, sfunc, pfunc );
 541     free( bufr );
 542     bufr  = NULL;
 543     bSize = 0;
 544     }
 545
 546   fclose(InFile);
 547
 548   if( !result )                               /* Generic failure. */
 549     {
 550     rprintf(FERROR,"%s Failed.  Error returned from params.c:parse().\n", func);
 551     return( False );
 552     }
 553
 554   return( True );                             /* Generic success. */
 555   } /* pm_process */
 556
 557 /* -------------------------------------------------------------------------- */
 558