Change how checking for missing values works.

[pspp] / src / language / data-io / combine-files.c

diff --git a/src/language/data-io/combine-files.c b/src/language/data-io/combine-files.c
index b7ba87ab7ccd39fd445121deea9d63eb1b4b4473..8322f5c03307f7e82597bcc5975a90621c0c7186 100644 (file)
--- a/src/language/data-io/combine-files.c
+++ b/src/language/data-io/combine-files.c
@@ -1,5 +1,5 @@
 /* PSPP - a program for statistical analysis.
-   Copyright (C) 1997-9, 2000, 2006, 2007, 2008, 2009, 2010, 2011, 2012 Free Software Foundation, Inc.
+   Copyright (C) 1997-9, 2000, 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014 Free Software Foundation, Inc.
 
    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
@@ -229,7 +229,8 @@ combine_files (enum comb_command_type command,
           if (file->handle == NULL)
             goto error;
 
-          file->reader = any_reader_open (file->handle, NULL, &file->dict);
+          file->reader = any_reader_open_and_decode (file->handle, NULL,
+                                                     &file->dict, NULL);
           if (file->reader == NULL)
             goto error;
         }
@@ -237,7 +238,7 @@ combine_files (enum comb_command_type command,
       while (lex_match (lexer, T_SLASH))
         if (lex_match_id (lexer, "RENAME"))
           {
-            if (!parse_dict_rename (lexer, file->dict))
+            if (!parse_dict_rename (lexer, file->dict, false))
               goto error;
           }
         else if (lex_match_id (lexer, "IN"))
@@ -264,7 +265,8 @@ combine_files (enum comb_command_type command,
             saw_sort = true;
           }
 
-      merge_dictionary (proc.dict, file);
+      if (!merge_dictionary (proc.dict, file))
+        goto error;
     }
 
   while (lex_token (lexer) != T_ENDCMD)
@@ -412,11 +414,11 @@ combine_files (enum comb_command_type command,
   for (i = 0; i < proc.n_files; i++)
     {
       struct comb_file *file = &proc.files[i];
-      size_t src_var_cnt = dict_get_var_cnt (file->dict);
+      size_t src_n_vars = dict_get_n_vars (file->dict);
       size_t j;
 
-      file->mv = xnmalloc (src_var_cnt, sizeof *file->mv);
-      for (j = 0; j < src_var_cnt; j++)
+      file->mv = xnmalloc (src_n_vars, sizeof *file->mv);
+      for (j = 0; j < src_n_vars; j++)
         {
           struct variable *src_var = dict_get_var (file->dict, j);
           struct variable *dst_var = dict_lookup_var (proc.dict,
@@ -444,7 +446,7 @@ combine_files (enum comb_command_type command,
           if (active_file == NULL)
             {
               proc_discard_output (ds);
-              file->reader = active_file = proc_open (ds);
+              file->reader = active_file = proc_open_filtering (ds, false);
             }
           else
             file->reader = casereader_clone (active_file);
@@ -515,7 +517,7 @@ merge_dictionary (struct dictionary *const m, struct comb_file *f)
      The correct thing to do would be to convert to an encoding
      which can cope with all the input files (eg UTF-8).
    */
-  if ( 0 != strcmp (dict_get_encoding (f->dict), dict_get_encoding (m)))
+  if (0 != strcmp (dict_get_encoding (f->dict), dict_get_encoding (m)))
     msg (MW, _("Combining files with incompatible encodings. String data may "
                "not be represented correctly."));
 
@@ -541,7 +543,7 @@ merge_dictionary (struct dictionary *const m, struct comb_file *f)
         }
     }
 
-  for (i = 0; i < dict_get_var_cnt (d); i++)
+  for (i = 0; i < dict_get_n_vars (d); i++)
     {
       struct variable *dv = dict_get_var (d, i);
       struct variable *mv = dict_lookup_var (m, var_get_name (dv));
@@ -554,8 +556,10 @@ merge_dictionary (struct dictionary *const m, struct comb_file *f)
           if (var_get_width (mv) != var_get_width (dv))
             {
               const char *var_name = var_get_name (dv);
-              const char *file_name = fh_get_name (f->handle);
               struct string s = DS_EMPTY_INITIALIZER;
+              const char *file_name;
+
+              file_name = f->handle ? fh_get_name (f->handle) : "*";
               ds_put_format (&s,
                              _("Variable %s in file %s has different "
                                "type or width from the same variable in "
@@ -587,7 +591,7 @@ merge_dictionary (struct dictionary *const m, struct comb_file *f)
           if (var_has_missing_values (dv) && !var_has_missing_values (mv))
             var_set_missing_values (mv, var_get_missing_values (dv));
           if (var_get_label (dv) && !var_get_label (mv))
-            var_set_label (mv, var_get_label (dv), false);
+            var_set_label (mv, var_get_label (dv));
         }
       else
         mv = dict_clone_var_assert (m, dv);
@@ -640,7 +644,7 @@ close_all_comb_files (struct comb_proc *proc)
       subcase_destroy (&file->dst);
       free (file->mv);
       fh_unref (file->handle);
-      dict_destroy (file->dict);
+      dict_unref (file->dict);
       casereader_destroy (file->reader);
       case_unref (file->data);
       free (file->in_name);
@@ -655,7 +659,7 @@ static void
 free_comb_proc (struct comb_proc *proc)
 {
   close_all_comb_files (proc);
-  dict_destroy (proc->dict);
+  dict_unref (proc->dict);
   casewriter_destroy (proc->output);
   case_matcher_destroy (proc->matcher);
   if (proc->prev_BY)
@@ -713,7 +717,7 @@ execute_match_files (struct comb_proc *proc)
       size_t i;
 
       output = create_output_case (proc);
-      for (i = proc->n_files; i-- > 0; )
+      for (i = proc->n_files; i-- > 0;)
         {
           struct comb_file *file = &proc->files[i];
           if (file->type == COMB_FILE)
@@ -817,7 +821,7 @@ scan_table (struct comb_file *file, union value by[])
 static struct ccase *
 create_output_case (const struct comb_proc *proc)
 {
-  size_t n_vars = dict_get_var_cnt (proc->dict);
+  size_t n_vars = dict_get_n_vars (proc->dict);
   struct ccase *output;
   size_t i;
 
@@ -831,7 +835,7 @@ create_output_case (const struct comb_proc *proc)
     {
       struct comb_file *file = &proc->files[i];
       if (file->in_var != NULL)
-        case_data_rw (output, file->in_var)->f = false;
+        *case_num_rw (output, file->in_var) = false;
     }
   return output;
 }
@@ -840,7 +844,7 @@ static void
 mark_file_used (const struct comb_file *file, struct ccase *output)
 {
   if (file->in_var != NULL)
-    case_data_rw (output, file->in_var)->f = true;
+    *case_num_rw (output, file->in_var) = true;
 }
 
 /* Copies the data from FILE's case into output case OUTPUT.
@@ -869,7 +873,7 @@ apply_nonmissing_case (const struct comb_file *file, struct ccase *output)
         = case_data_idx (file->data, src_field->case_index);
       int width = src_field->width;
 
-      if (!mv_is_value_missing (file->mv[i], src_value, MV_ANY)
+      if (!mv_is_value_missing (file->mv[i], src_value)
           && !(width > 0 && value_is_spaces (src_value, width)))
         value_copy (case_data_rw_idx (output, dst_field->case_index),
                     src_value, width);
@@ -909,7 +913,7 @@ output_case (struct comb_proc *proc, struct ccase *output, union value by[])
         {
           new_BY = !subcase_equal_xx (&proc->by_vars, proc->prev_BY, by);
           if (proc->last != NULL)
-            case_data_rw (proc->buffered_case, proc->last)->f = new_BY;
+            *case_num_rw (proc->buffered_case, proc->last) = new_BY;
           casewriter_write (proc->output, proc->buffered_case);
         }
       else
@@ -917,7 +921,7 @@ output_case (struct comb_proc *proc, struct ccase *output, union value by[])
 
       proc->buffered_case = output;
       if (proc->first != NULL)
-        case_data_rw (proc->buffered_case, proc->first)->f = new_BY;
+        *case_num_rw (proc->buffered_case, proc->first) = new_BY;
 
       if (new_BY)
         {
@@ -942,7 +946,7 @@ output_buffered_case (struct comb_proc *proc)
   if (proc->prev_BY != NULL)
     {
       if (proc->last != NULL)
-        case_data_rw (proc->buffered_case, proc->last)->f = 1.0;
+        *case_num_rw (proc->buffered_case, proc->last) = 1.0;
       casewriter_write (proc->output, proc->buffered_case);
       proc->buffered_case = NULL;
     }