Beispiel #1
0
/* Reads a record from a disk file into R.
   Returns true if successful, false on error or at end of file. */
static bool
read_file_record (struct dfm_reader *r)
{
  assert (r->fh != fh_inline_file ());

  ds_clear (&r->line);
  switch (fh_get_mode (r->fh))
    {
    case FH_MODE_TEXT:
      if (ds_read_line (&r->line, r->file, SIZE_MAX))
        {
          ds_chomp_byte (&r->line, '\n');
          return true;
        }
      else
        {
          if (ferror (r->file))
            read_error (r);
          return false;
        }

    case FH_MODE_FIXED:
      if (ds_read_stream (&r->line, 1, fh_get_record_width (r->fh), r->file))
        return true;
      else
        {
          if (ferror (r->file))
            read_error (r);
          else if (!ds_is_empty (&r->line))
            partial_record (r);
          return false;
        }

    case FH_MODE_VARIABLE:
      {
        size_t leading_size;
        size_t trailing_size;
        int status;

        /* Read leading record size. */
        status = read_size (r, &leading_size);
        if (status <= 0)
          return false;

        /* Read record data. */
        if (!ds_read_stream (&r->line, leading_size, 1, r->file))
          {
            if (ferror (r->file))
              read_error (r);
            else
              partial_record (r);
            return false;
          }

        /* Read trailing record size and check that it's the same
           as the leading record size. */
        status = read_size (r, &trailing_size);
        if (status <= 0)
          {
            if (status == 0)
              partial_record (r);
            return false;
          }
        if (leading_size != trailing_size)
          {
            corrupt_size (r);
            return false;
          }

        return true;
      }

    case FH_MODE_360_VARIABLE:
    case FH_MODE_360_SPANNED:
      for (;;)
        {
          size_t record_size;
          int segment;
          int status;

          /* If we've exhausted our current block, start another
             one by reading the new block descriptor word. */
          if (r->block_left == 0)
            {
              status = read_descriptor_word (r, BLOCK, &r->block_left, NULL);
              if (status < 0)
                return false;
              else if (status == 0)
                return !ds_is_empty (&r->line);
            }

          /* Read record descriptor. */
          if (r->block_left < 4)
            {
              partial_record (r);
              return false;
            }
          r->block_left -= 4;
          status = read_descriptor_word (r, RECORD, &record_size, &segment);
          if (status <= 0)
            {
              if (status == 0)
                partial_record (r);
              return false;
            }
          if (record_size > r->block_left)
            {
              msg (ME, _("Record exceeds remaining block length."));
              return false;
            }

          /* Read record data. */
          if (!ds_read_stream (&r->line, record_size, 1, r->file))
            {
              if (ferror (r->file))
                read_error (r);
              else
                partial_record (r);
              return false;
            }
          r->block_left -= record_size;

          /* In variable mode, read only a single record.
             In spanned mode, a segment value of 0 should
             designate a whole record without spanning, 1 the
             first segment in a record, 2 the last segment in a
             record, and 3 an intermediate segment in a record.
             For compatibility, though, we actually pay attention
             only to whether the segment value is even or odd. */
          if (fh_get_mode (r->fh) == FH_MODE_360_VARIABLE
              || (segment & 1) == 0)
            return true;
        }
    }

  NOT_REACHED ();
}
Beispiel #2
0
/* Parse all the aggregate functions. */
static bool
parse_aggregate_functions (struct lexer *lexer, const struct dictionary *dict,
			   struct agr_proc *agr)
{
  struct agr_var *tail; /* Tail of linked list starting at agr->vars. */

  /* Parse everything. */
  tail = NULL;
  for (;;)
    {
      char **dest;
      char **dest_label;
      size_t n_dest;
      struct string function_name;

      enum mv_class exclude;
      const struct agr_func *function;
      int func_index;

      union agr_argument arg[2];

      const struct variable **src;
      size_t n_src;

      size_t i;

      dest = NULL;
      dest_label = NULL;
      n_dest = 0;
      src = NULL;
      function = NULL;
      n_src = 0;
      arg[0].c = NULL;
      arg[1].c = NULL;
      ds_init_empty (&function_name);

      /* Parse the list of target variables. */
      while (!lex_match (lexer, T_EQUALS))
	{
	  size_t n_dest_prev = n_dest;

	  if (!parse_DATA_LIST_vars (lexer, dict, &dest, &n_dest,
                                     (PV_APPEND | PV_SINGLE | PV_NO_SCRATCH
                                      | PV_NO_DUPLICATE)))
	    goto error;

	  /* Assign empty labels. */
	  {
	    int j;

	    dest_label = xnrealloc (dest_label, n_dest, sizeof *dest_label);
	    for (j = n_dest_prev; j < n_dest; j++)
	      dest_label[j] = NULL;
	  }



	  if (lex_is_string (lexer))
	    {
	      dest_label[n_dest - 1] = xstrdup (lex_tokcstr (lexer));
	      lex_get (lexer);
	    }
	}

      /* Get the name of the aggregation function. */
      if (lex_token (lexer) != T_ID)
	{
	  lex_error (lexer, _("expecting aggregation function"));
	  goto error;
	}

      ds_assign_substring (&function_name, lex_tokss (lexer));
      exclude = ds_chomp_byte (&function_name, '.') ? MV_SYSTEM : MV_ANY;

      for (function = agr_func_tab; function->name; function++)
	if (!c_strcasecmp (function->name, ds_cstr (&function_name)))
	  break;
      if (NULL == function->name)
	{
	  msg (SE, _("Unknown aggregation function %s."),
	       ds_cstr (&function_name));
	  goto error;
	}
      ds_destroy (&function_name);
      func_index = function - agr_func_tab;
      lex_get (lexer);

      /* Check for leading lparen. */
      if (!lex_match (lexer, T_LPAREN))
	{
	  if (function->src_vars == AGR_SV_YES)
	    {
              lex_force_match (lexer, T_LPAREN);
	      goto error;
	    }
	}
      else
        {
	  /* Parse list of source variables. */
	  {
	    int pv_opts = PV_NO_SCRATCH;

	    if (func_index == SUM || func_index == MEAN || func_index == SD)
	      pv_opts |= PV_NUMERIC;
	    else if (function->n_args)
	      pv_opts |= PV_SAME_TYPE;

	    if (!parse_variables_const (lexer, dict, &src, &n_src, pv_opts))
	      goto error;
	  }

	  /* Parse function arguments, for those functions that
	     require arguments. */
	  if (function->n_args != 0)
	    for (i = 0; i < function->n_args; i++)
	      {
		int type;

		lex_match (lexer, T_COMMA);
		if (lex_is_string (lexer))
		  {
		    arg[i].c = recode_string (dict_get_encoding (agr->dict),
                                              "UTF-8", lex_tokcstr (lexer),
                                              -1);
		    type = VAL_STRING;
		  }
		else if (lex_is_number (lexer))
		  {
		    arg[i].f = lex_tokval (lexer);
		    type = VAL_NUMERIC;
		  }
                else
                  {
		    msg (SE, _("Missing argument %zu to %s."),
                         i + 1, function->name);
		    goto error;
		  }

		lex_get (lexer);

		if (type != var_get_type (src[0]))
		  {
		    msg (SE, _("Arguments to %s must be of same type as "
			       "source variables."),
			 function->name);
		    goto error;
		  }
	      }

	  /* Trailing rparen. */
	  if (!lex_force_match (lexer, T_RPAREN))
            goto error;

	  /* Now check that the number of source variables match
	     the number of target variables.  If we check earlier
	     than this, the user can get very misleading error
	     message, i.e. `AGGREGATE x=SUM(y t).' will get this
	     error message when a proper message would be more
	     like `unknown variable t'. */
	  if (n_src != n_dest)
	    {
	      msg (SE, _("Number of source variables (%zu) does not match "
			 "number of target variables (%zu)."),
		    n_src, n_dest);
	      goto error;
	    }

          if ((func_index == PIN || func_index == POUT
              || func_index == FIN || func_index == FOUT)
              && (var_is_numeric (src[0])
                  ? arg[0].f > arg[1].f
                  : str_compare_rpad (arg[0].c, arg[1].c) > 0))
            {
              union agr_argument t = arg[0];
              arg[0] = arg[1];
              arg[1] = t;

              msg (SW, _("The value arguments passed to the %s function "
                         "are out-of-order.  They will be treated as if "
                         "they had been specified in the correct order."),
                   function->name);
            }
	}

      /* Finally add these to the linked list of aggregation
         variables. */
      for (i = 0; i < n_dest; i++)
	{
	  struct agr_var *v = xzalloc (sizeof *v);

	  /* Add variable to chain. */
	  if (agr->agr_vars != NULL)
	    tail->next = v;
	  else
	    agr->agr_vars = v;
          tail = v;
	  tail->next = NULL;
          v->moments = NULL;

	  /* Create the target variable in the aggregate
             dictionary. */
	  {
	    struct variable *destvar;

	    v->function = func_index;

	    if (src)
	      {
		v->src = src[i];

		if (var_is_alpha (src[i]))
		  {
		    v->function |= FSTRING;
		    v->string = xmalloc (var_get_width (src[i]));
		  }

		if (function->alpha_type == VAL_STRING)
		  destvar = dict_clone_var_as (agr->dict, v->src, dest[i]);
		else
                  {
                    assert (var_is_numeric (v->src)
                            || function->alpha_type == VAL_NUMERIC);
                    destvar = dict_create_var (agr->dict, dest[i], 0);
                    if (destvar != NULL)
                      {
                        struct fmt_spec f;
                        if ((func_index == N || func_index == NMISS)
                            && dict_get_weight (dict) != NULL)
                          f = fmt_for_output (FMT_F, 8, 2);
                        else
                          f = function->format;
                        var_set_both_formats (destvar, &f);
                      }
                  }
	      } else {
                struct fmt_spec f;
		v->src = NULL;
		destvar = dict_create_var (agr->dict, dest[i], 0);
		if (destvar != NULL)
		  {
		    if ((func_index == N || func_index == NMISS)
			&& dict_get_weight (dict) != NULL)
		      f = fmt_for_output (FMT_F, 8, 2);
		    else
		      f = function->format;
		    var_set_both_formats (destvar, &f);
		  }
	    }

	    if (!destvar)
	      {
		msg (SE, _("Variable name %s is not unique within the "
			   "aggregate file dictionary, which contains "
			   "the aggregate variables and the break "
			   "variables."),
		     dest[i]);
		goto error;
	      }

	    free (dest[i]);
	    if (dest_label[i])
              var_set_label (destvar, dest_label[i]);

	    v->dest = destvar;
	  }

	  v->exclude = exclude;

	  if (v->src != NULL)
	    {
	      int j;

	      if (var_is_numeric (v->src))
		for (j = 0; j < function->n_args; j++)
		  v->arg[j].f = arg[j].f;
	      else
		for (j = 0; j < function->n_args; j++)
		  v->arg[j].c = xstrdup (arg[j].c);
	    }
	}

      if (src != NULL && var_is_alpha (src[0]))
	for (i = 0; i < function->n_args; i++)
	  {
	    free (arg[i].c);
	    arg[i].c = NULL;
	  }

      free (src);
      free (dest);
      free (dest_label);

      if (!lex_match (lexer, T_SLASH))
	{
	  if (lex_token (lexer) == T_ENDCMD)
	    return true;

	  lex_error (lexer, "expecting end of command");
	  return false;
	}
      continue;

    error:
      ds_destroy (&function_name);
      for (i = 0; i < n_dest; i++)
	{
	  free (dest[i]);
	  free (dest_label[i]);
	}
      free (dest);
      free (dest_label);
      free (arg[0].c);
      free (arg[1].c);
      if (src && n_src && var_is_alpha (src[0]))
	for (i = 0; i < function->n_args; i++)
	  {
	    free (arg[i].c);
	    arg[i].c = NULL;
	  }
      free (src);

      return false;
    }
}