params.c

Rsync 3.0.5 source code

/* This modules is based on the params.c module from Samba, written by Karl Auer
   and much modifed by Christopher Hertel. */

/*
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 3 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License along
 * with this program; if not, visit the http://fsf.org website.
 */

/* -------------------------------------------------------------------------- **
 *
 * Module name: params
 *
 * -------------------------------------------------------------------------- **
 *
 *  This module performs lexical analysis and initial parsing of a
 *  Windows-like parameter file.  It recognizes and handles four token
 *  types:  section-name, parameter-name, parameter-value, and
 *  end-of-file.  Comments and line continuation are handled
 *  internally.
 *
 *  The entry point to the module is function pm_process().  This
 *  function opens the source file, calls the Parse() function to parse
 *  the input, and then closes the file when either the EOF is reached
 *  or a fatal error is encountered.
 *
 *  A sample parameter file might look like this:
 *
 *  [section one]
 *  parameter one = value string
 *  parameter two = another value
 *  [section two]
 *  new parameter = some value or t'other
 *
 *  The parameter file is divided into sections by section headers:
 *  section names enclosed in square brackets (eg. [section one]).
 *  Each section contains parameter lines, each of which consist of a
 *  parameter name and value delimited by an equal sign.  Roughly, the
 *  syntax is:
 *
 *    <file>            :==  { <section> } EOF
 *
 *    <section>         :==  <section header> { <parameter line> }
 *
 *    <section header>  :==  '[' NAME ']'
 *
 *    <parameter line>  :==  NAME '=' VALUE '\n'
 *
 *  Blank lines and comment lines are ignored.  Comment lines are lines
 *  beginning with either a semicolon (';') or a pound sign ('#').
 *
 *  All whitespace in section names and parameter names is compressed
 *  to single spaces.  Leading and trailing whitespace is stipped from
 *  both names and values.
 *
 *  Only the first equals sign in a parameter line is significant.
 *  Parameter values may contain equals signs, square brackets and
 *  semicolons.  Internal whitespace is retained in parameter values,
 *  with the exception of the '\r' character, which is stripped for
 *  historic reasons.  Parameter names may not start with a left square
 *  bracket, an equal sign, a pound sign, or a semicolon, because these
 *  are used to identify other tokens.
 *
 * -------------------------------------------------------------------------- **
 */

#include "rsync.h"
#include "ifuncs.h"

/* -------------------------------------------------------------------------- **
 * Constants...
 */

#define BUFR_INC 1024


/* -------------------------------------------------------------------------- **
 * Variables...
 *
 *  bufr        - pointer to a global buffer.  This is probably a kludge,
 *                but it was the nicest kludge I could think of (for now).
 *  bSize       - The size of the global buffer <bufr>.
 */

static char *bufr  = NULL;
static int   bSize = 0;

/* -------------------------------------------------------------------------- **
 * Functions...
 */

static int EatWhitespace( FILE *InFile )
  /* ------------------------------------------------------------------------ **
   * Scan past whitespace (see ctype(3C)) and return the first non-whitespace
   * character, or newline, or EOF.
   *
   *  Input:  InFile  - Input source.
   *
   *  Output: The next non-whitespace character in the input stream.
   *
   *  Notes:  Because the config files use a line-oriented grammar, we
   *          explicitly exclude the newline character from the list of
   *          whitespace characters.
   *        - Note that both EOF (-1) and the nul character ('\0') are
   *          considered end-of-file markers.
   *
   * ------------------------------------------------------------------------ **
   */
  {
  int c;

  for( c = getc( InFile ); isspace( c ) && ('\n' != c); c = getc( InFile ) )
    ;
  return( c );
  } /* EatWhitespace */

static int EatComment( FILE *InFile )
  /* ------------------------------------------------------------------------ **
   * Scan to the end of a comment.
   *
   *  Input:  InFile  - Input source.
   *
   *  Output: The character that marks the end of the comment.  Normally,
   *          this will be a newline, but it *might* be an EOF.
   *
   *  Notes:  Because the config files use a line-oriented grammar, we
   *          explicitly exclude the newline character from the list of
   *          whitespace characters.
   *        - Note that both EOF (-1) and the nul character ('\0') are
   *          considered end-of-file markers.
   *
   * ------------------------------------------------------------------------ **
   */
  {
  int c;

  for( c = getc( InFile ); ('\n'!=c) && (EOF!=c) && (c>0); c = getc( InFile ) )
    ;
  return( c );
  } /* EatComment */

static int Continuation( char *line, int pos )
  /* ------------------------------------------------------------------------ **
   * Scan backards within a string to discover if the last non-whitespace
   * character is a line-continuation character ('\\').
   *
   *  Input:  line  - A pointer to a buffer containing the string to be
   *                  scanned.
   *          pos   - This is taken to be the offset of the end of the
   *                  string.  This position is *not* scanned.
   *
   *  Output: The offset of the '\\' character if it was found, or -1 to
   *          indicate that it was not.
   *
   * ------------------------------------------------------------------------ **
   */
  {
  pos--;
  while( pos >= 0 && isSpace(line + pos) )
     pos--;

  return( ((pos >= 0) && ('\\' == line[pos])) ? pos : -1 );
  } /* Continuation */


static BOOL Section( FILE *InFile, BOOL (*sfunc)(char *) )
  /* ------------------------------------------------------------------------ **
   * Scan a section name, and pass the name to function sfunc().
   *
   *  Input:  InFile  - Input source.
   *          sfunc   - Pointer to the function to be called if the section
   *                    name is successfully read.
   *
   *  Output: True if the section name was read and True was returned from
   *          <sfunc>.  False if <sfunc> failed or if a lexical error was
   *          encountered.
   *
   * ------------------------------------------------------------------------ **
   */
  {
  int   c;
  int   i;
  int   end;
  char *func  = "params.c:Section() -";

  i = 0;      /* <i> is the offset of the next free byte in bufr[] and  */
  end = 0;    /* <end> is the current "end of string" offset.  In most  */
              /* cases these will be the same, but if the last          */
              /* character written to bufr[] is a space, then <end>     */
              /* will be one less than <i>.                             */

  c = EatWhitespace( InFile );    /* We've already got the '['.  Scan */
                                  /* past initial white space.        */

  while( (EOF != c) && (c > 0) )
    {

    /* Check that the buffer is big enough for the next character. */
    if( i > (bSize - 2) )
      {
      bSize += BUFR_INC;
      bufr   = realloc_array( bufr, char, bSize );
      if( NULL == bufr )
        {
        rprintf(FLOG, "%s Memory re-allocation failure.", func);
        return( False );
        }
      }

    /* Handle a single character. */
    switch( c )
      {
      case ']':                       /* Found the closing bracket.         */
        bufr[end] = '\0';
        if( 0 == end )                  /* Don't allow an empty name.       */
          {
          rprintf(FLOG, "%s Empty section name in configuration file.\n", func );
          return( False );
          }
        if( !sfunc( bufr ) )            /* Got a valid name.  Deal with it. */
          return( False );
        (void)EatComment( InFile );     /* Finish off the line.             */
        return( True );

      case '\n':                      /* Got newline before closing ']'.    */
        i = Continuation( bufr, i );    /* Check for line continuation.     */
        if( i < 0 )
          {
          bufr[end] = '\0';
          rprintf(FLOG, "%s Badly formed line in configuration file: %s\n",
                   func, bufr );
          return( False );
          }
        end = ( (i > 0) && (' ' == bufr[i - 1]) ) ? (i - 1) : (i);
        c = getc( InFile );             /* Continue with next line.         */
        break;

      default:                        /* All else are a valid name chars.   */
        if( isspace( c ) )              /* One space per whitespace region. */
          {
          bufr[end] = ' ';
          i = end + 1;
          c = EatWhitespace( InFile );
          }
        else                            /* All others copy verbatim.        */
          {
          bufr[i++] = c;
          end = i;
          c = getc( InFile );
          }
      }
    }

  /* We arrive here if we've met the EOF before the closing bracket. */
  rprintf(FLOG, "%s Unexpected EOF in the configuration file: %s\n", func, bufr );
  return( False );
  } /* Section */

static BOOL Parameter( FILE *InFile, BOOL (*pfunc)(char *, char *), int c )
  /* ------------------------------------------------------------------------ **
   * Scan a parameter name and value, and pass these two fields to pfunc().
   *
   *  Input:  InFile  - The input source.
   *          pfunc   - A pointer to the function that will be called to
   *                    process the parameter, once it has been scanned.
   *          c       - The first character of the parameter name, which
   *                    would have been read by Parse().  Unlike a comment
   *                    line or a section header, there is no lead-in
   *                    character that can be discarded.
   *
   *  Output: True if the parameter name and value were scanned and processed