params.c

   1 /* This modules is based on the params.c module from Samba, written by Karl Auer
   2    and much modified by Christopher Hertel. */
   3
   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 3 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 along
  16  * with this program; if not, visit the http://fsf.org website.
  17  */
  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 stripped 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 #include "ifuncs.h"
  78 #include "itypes.h"
  79
  80 /* -------------------------------------------------------------------------- **
  81  * Constants...
  82  */
  83
  84 #define BUFR_INC 1024
  85
  86
  87 /* -------------------------------------------------------------------------- **
  88  * Variables...
  89  *
  90  *  bufr        - pointer to a global buffer.  This is probably a kludge,
  91  *                but it was the nicest kludge I could think of (for now).
  92  *  bSize       - The size of the global buffer <bufr>.
  93  */
  94
  95 static char *bufr  = NULL;
  96 static int   bSize = 0;
  97 static BOOL  (*the_sfunc)(char *);
  98 static BOOL  (*the_pfunc)(char *, char *);
  99
 100 /* -------------------------------------------------------------------------- **
 101  * Functions...
 102  */
 103
 104 static int EatWhitespace( FILE *InFile )
 105   /* ------------------------------------------------------------------------ **
 106    * Scan past whitespace (see ctype(3C)) and return the first non-whitespace
 107    * character, or newline, or EOF.
 108    *
 109    *  Input:  InFile  - Input source.
 110    *
 111    *  Output: The next non-whitespace character in the input stream.
 112    *
 113    *  Notes:  Because the config files use a line-oriented grammar, we
 114    *          explicitly exclude the newline character from the list of
 115    *          whitespace characters.
 116    *        - Note that both EOF (-1) and the nul character ('\0') are
 117    *          considered end-of-file markers.
 118    *
 119    * ------------------------------------------------------------------------ **
 120    */
 121   {
 122   int c;
 123
 124   for( c = getc( InFile ); isspace( c ) && ('\n' != c); c = getc( InFile ) )
 125     ;
 126   return( c );
 127   } /* EatWhitespace */
 128
 129 static int EatComment( FILE *InFile )
 130   /* ------------------------------------------------------------------------ **
 131    * Scan to the end of a comment.
 132    *
 133    *  Input:  InFile  - Input source.
 134    *
 135    *  Output: The character that marks the end of the comment.  Normally,
 136    *          this will be a newline, but it *might* be an EOF.
 137    *
 138    *  Notes:  Because the config files use a line-oriented grammar, we
 139    *          explicitly exclude the newline character from the list of
 140    *          whitespace characters.
 141    *        - Note that both EOF (-1) and the nul character ('\0') are
 142    *          considered end-of-file markers.
 143    *
 144    * ------------------------------------------------------------------------ **
 145    */
 146   {
 147   int c;
 148
 149   for( c = getc( InFile ); ('\n'!=c) && (EOF!=c) && (c>0); c = getc( InFile ) )
 150     ;
 151   return( c );
 152   } /* EatComment */
 153
 154 static int Continuation( char *line, int pos )
 155   /* ------------------------------------------------------------------------ **
 156    * Scan backwards within a string to discover if the last non-whitespace
 157    * character is a line-continuation character ('\\').
 158    *
 159    *  Input:  line  - A pointer to a buffer containing the string to be
 160    *                  scanned.
 161    *          pos   - This is taken to be the offset of the end of the
 162    *                  string.  This position is *not* scanned.
 163    *
 164    *  Output: The offset of the '\\' character if it was found, or -1 to
 165    *          indicate that it was not.
 166    *
 167    * ------------------------------------------------------------------------ **
 168    */
 169   {
 170   pos--;
 171   while( pos >= 0 && isSpace(line + pos) )
 172      pos--;
 173
 174   return( ((pos >= 0) && ('\\' == line[pos])) ? pos : -1 );
 175   } /* Continuation */
 176
 177
 178 static BOOL Section( FILE *InFile, BOOL (*sfunc)(char *) )
 179   /* ------------------------------------------------------------------------ **
 180    * Scan a section name, and pass the name to function sfunc().
 181    *
 182    *  Input:  InFile  - Input source.
 183    *          sfunc   - Pointer to the function to be called if the section
 184    *                    name is successfully read.
 185    *
 186    *  Output: True if the section name was read and True was returned from
 187    *          <sfunc>.  False if <sfunc> failed or if a lexical error was
 188    *          encountered.
 189    *
 190    * ------------------------------------------------------------------------ **
 191    */
 192   {
 193   int   c;
 194   int   i;
 195   int   end;
 196   char *func  = "params.c:Section() -";
 197
 198   i = 0;      /* <i> is the offset of the next free byte in bufr[] and  */
 199   end = 0;    /* <end> is the current "end of string" offset.  In most  */
 200               /* cases these will be the same, but if the last          */
 201               /* character written to bufr[] is a space, then <end>     */
 202               /* will be one less than <i>.                             */
 203
 204   c = EatWhitespace( InFile );    /* We've already got the '['.  Scan */
 205                                   /* past initial white space.        */
 206
 207   while( (EOF != c) && (c > 0) )
 208     {
 209
 210     /* Check that the buffer is big enough for the next character. */
 211     if( i > (bSize - 2) )
 212       {
 213       bSize += BUFR_INC;
 214       bufr   = realloc_array( bufr, char, bSize );
 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(FLOG, "%s Empty section name in config 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(FLOG, "%s Badly formed line in config 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(FLOG, "%s Unexpected EOF in the config 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;
 303       bufr   = realloc_array( bufr, char, bSize );
 304       }
 305
 306     switch( c )
 307       {
 308       case '=':                 /* Equal sign marks end of param name. */
 309         if( 0 == end )              /* Don't allow an empty name.      */
 310           {
 311           rprintf(FLOG, "%s Invalid parameter name in config file.\n", func );
 312           return( False );
 313           }
 314         bufr[end++] = '\0';         /* Mark end of string & advance.   */
 315         i = vstart = end;           /* New string starts here.         */
 316         c = EatWhitespace(InFile);
 317         break;
 318
 319       case '\n':                /* Find continuation char, else error. */
 320         i = Continuation( bufr, i );
 321         if( i < 0 )
 322           {
 323           bufr[end] = '\0';
 324           rprintf(FLOG, "%s Ignoring badly formed line in config file: %s\n",
 325                    func, bufr );
 326           return( True );
 327           }
 328         end = ( (i > 0) && (' ' == bufr[i - 1]) ) ? (i - 1) : (i);
 329         c = getc( InFile );       /* Read past eoln.                   */
 330         break;
 331
 332       case '\0':                /* Shouldn't have EOF within param name. */
 333       case EOF:
 334         bufr[i] = '\0';
 335         rprintf(FLOG, "%s Unexpected end-of-file at: %s\n", func, bufr );
 336         return( True );
 337
 338       case ' ':
 339       case '\t':
 340         /* A directive divides at the first space or tab. */
 341         if (*bufr == '&') {
 342           bufr[end++] = '\0';
 343           i = vstart = end;
 344           c = EatWhitespace(InFile);
 345           if (c == '=')
 346             c = EatWhitespace(InFile);
 347           break;
 348         }
 349         /* FALL THROUGH */
 350
 351       default:
 352         if( isspace( c ) )     /* One ' ' per whitespace region.       */
 353           {
 354           bufr[end] = ' ';
 355           i = end + 1;
 356           c = EatWhitespace( InFile );
 357           }
 358         else                   /* All others verbatim.                 */
 359           {
 360           bufr[i++] = c;
 361           end = i;
 362           c = getc( InFile );
 363           }
 364       }
 365     }
 366
 367   /* Now parse the value. */
 368   while( (EOF !=c) && (c > 0) )
 369     {
 370
 371     if( i > (bSize - 2) )       /* Make sure there's enough room. */
 372       {
 373       bSize += BUFR_INC;
 374       bufr   = realloc_array( bufr, char, bSize );
 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(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 int name_cmp(const void *n1, const void *n2)
 409 {
 410     return strcmp(*(char * const *)n1, *(char * const *)n2);
 411 }
 412
 413 static int include_config(char *include, int manage_globals)
 414 {
 415     STRUCT_STAT sb;
 416     char *match = manage_globals ? "*.conf" : "*.inc";
 417     int ret;
 418
 419     if (do_stat(include, &sb) < 0) {
 420         rsyserr(FLOG, errno, "unable to stat config file \"%s\"", include);
 421         return 0;
 422     }
 423
 424     if (S_ISREG(sb.st_mode)) {
 425         if (manage_globals && the_sfunc)
 426             the_sfunc("]push");
 427         ret = pm_process(include, the_sfunc, the_pfunc);
 428         if (manage_globals && the_sfunc)
 429             the_sfunc("]pop");
 430     } else if (S_ISDIR(sb.st_mode)) {
 431         char buf[MAXPATHLEN], **bpp;
 432         item_list conf_list;
 433         struct dirent *di;
 434         size_t j;
 435         DIR *d;
 436
 437         if (!(d = opendir(include))) {
 438             rsyserr(FLOG, errno, "unable to open config dir \"%s\"", include);
 439             return 0;
 440         }
 441
 442         memset(&conf_list, 0, sizeof conf_list);
 443
 444         while ((di = readdir(d)) != NULL) {
 445             char *dname = d_name(di);
 446             if (!wildmatch(match, dname))
 447                 continue;
 448             bpp = EXPAND_ITEM_LIST(&conf_list, char *, 32);
 449             pathjoin(buf, sizeof buf, include, dname);
 450             *bpp = strdup(buf);
 451         }
 452         closedir(d);
 453
 454         if (!(bpp = conf_list.items))
 455             return 1;
 456
 457         if (conf_list.count > 1)
 458             qsort(bpp, conf_list.count, sizeof (char *), name_cmp);
 459
 460         for (j = 0, ret = 1; j < conf_list.count; j++) {
 461             if (manage_globals && the_sfunc)
 462                 the_sfunc(j == 0 ? "]push" : "]reset");
 463             if ((ret = pm_process(bpp[j], the_sfunc, the_pfunc)) != 1)
 464                 break;
 465         }
 466
 467         if (manage_globals && the_sfunc)
 468             the_sfunc("]pop");
 469
 470         for (j = 0; j < conf_list.count; j++)
 471             free(bpp[j]);
 472         free(bpp);
 473     } else
 474         ret = 0;
 475
 476     return ret;
 477 }
 478
 479 static int parse_directives(char *name, char *val)
 480 {
 481     if (strcasecmp(name, "&include") == 0)
 482         return include_config(val, 1);
 483     if (strcasecmp(name, "&merge") == 0)
 484         return include_config(val, 0);
 485     rprintf(FLOG, "Unknown directive: %s.\n", name);
 486     return 0;
 487 }
 488
 489 static int Parse( FILE *InFile,
 490                    BOOL (*sfunc)(char *),
 491                    BOOL (*pfunc)(char *, char *) )
 492   /* ------------------------------------------------------------------------ **
 493    * Scan & parse the input.
 494    *
 495    *  Input:  InFile  - Input source.
 496    *          sfunc   - Function to be called when a section name is scanned.
 497    *                    See Section().
 498    *          pfunc   - Function to be called when a parameter is scanned.
 499    *                    See Parameter().
 500    *
 501    *  Output: 1 if the file was successfully scanned, 2 if the file was
 502    *  scanned until a section header with no section function, else 0.
 503    *
 504    *  Notes:  The input can be viewed in terms of 'lines'.  There are four
 505    *          types of lines:
 506    *            Blank      - May contain whitespace, otherwise empty.
 507    *            Comment    - First non-whitespace character is a ';' or '#'.
 508    *                         The remainder of the line is ignored.
 509    *            Section    - First non-whitespace character is a '['.
 510    *            Parameter  - The default case.
 511    *
 512    * ------------------------------------------------------------------------ **
 513    */
 514   {
 515   int    c;
 516
 517   c = EatWhitespace( InFile );
 518   while( (EOF != c) && (c > 0) )
 519     {
 520     switch( c )
 521       {
 522       case '\n':                        /* Blank line. */
 523         c = EatWhitespace( InFile );
 524         break;
 525
 526       case ';':                         /* Comment line. */
 527       case '#':
 528         c = EatComment( InFile );
 529         break;
 530
 531       case '[':                         /* Section Header. */
 532         if (!sfunc)
 533           return 2;
 534         if( !Section( InFile, sfunc ) )
 535           return 0;
 536         c = EatWhitespace( InFile );
 537         break;
 538
 539       case '\\':                        /* Bogus backslash. */
 540         c = EatWhitespace( InFile );
 541         break;
 542
 543       case '&':                         /* Handle directives */
 544         the_sfunc = sfunc;
 545         the_pfunc = pfunc;
 546         c = Parameter( InFile, parse_directives, c );
 547         if (c != 1)
 548           return c;
 549         c = EatWhitespace( InFile );
 550         break;
 551
 552       default:                          /* Parameter line. */
 553         if( !Parameter( InFile, pfunc, c ) )
 554           return 0;
 555         c = EatWhitespace( InFile );
 556         break;
 557       }
 558     }
 559   return 1;
 560   } /* Parse */
 561
 562 static FILE *OpenConfFile( char *FileName )
 563   /* ------------------------------------------------------------------------ **
 564    * Open a config file.
 565    *
 566    *  Input:  FileName  - The pathname of the config file to be opened.
 567    *
 568    *  Output: A pointer of type (FILE *) to the opened file, or NULL if the
 569    *          file could not be opened.
 570    *
 571    * ------------------------------------------------------------------------ **
 572    */
 573   {
 574   FILE *OpenedFile;
 575   char *func = "params.c:OpenConfFile() -";
 576
 577   if( NULL == FileName || 0 == *FileName )
 578     {
 579     rprintf(FLOG, "%s No config filename specified.\n", func);
 580     return( NULL );
 581     }
 582
 583   OpenedFile = fopen( FileName, "r" );
 584   if( NULL == OpenedFile )
 585     {
 586     rsyserr(FLOG, errno, "unable to open config file \"%s\"",
 587             FileName);
 588     }
 589
 590   return( OpenedFile );
 591   } /* OpenConfFile */
 592
 593 int pm_process( char *FileName,
 594                  BOOL (*sfunc)(char *),
 595                  BOOL (*pfunc)(char *, char *) )
 596   /* ------------------------------------------------------------------------ **
 597    * Process the named parameter file.
 598    *
 599    *  Input:  FileName  - The pathname of the parameter file to be opened.
 600    *          sfunc     - A pointer to a function that will be called when
 601    *                      a section name is discovered.
 602    *          pfunc     - A pointer to a function that will be called when
 603    *                      a parameter name and value are discovered.
 604    *
 605    *  Output: 1 if the file was successfully parsed, 2 if parsing ended at a
 606    *  section header w/o a section function, else 0.
 607    *
 608    * ------------------------------------------------------------------------ **
 609    */
 610   {
 611   int   result;
 612   FILE *InFile;
 613   char *func = "params.c:pm_process() -";
 614
 615   InFile = OpenConfFile( FileName );          /* Open the config file. */
 616   if( NULL == InFile )
 617     return( False );
 618
 619   if( NULL != bufr )                          /* If we already have a buffer */
 620     result = Parse( InFile, sfunc, pfunc );   /* (recursive call), then just */
 621                                               /* use it.                     */
 622
 623   else                                        /* If we don't have a buffer   */
 624     {                                         /* allocate one, then parse,   */
 625     bSize = BUFR_INC;                         /* then free.                  */
 626     bufr = new_array( char, bSize );
 627     result = Parse( InFile, sfunc, pfunc );
 628     free( bufr );
 629     bufr  = NULL;
 630     bSize = 0;
 631     }
 632
 633   fclose(InFile);
 634
 635   if( !result )                               /* Generic failure. */
 636     {
 637     rprintf(FLOG, "%s Failed.  Error returned from params.c:parse().\n", func);
 638     return 0;
 639     }
 640
 641   return result;
 642   } /* pm_process */
 643
 644 /* -------------------------------------------------------------------------- */
 645