sys-file-reader: Avoid assert-fail for duplicate attribute names.

[pspp] / src / data / sys-file-reader.c

diff --git a/src/data/sys-file-reader.c b/src/data/sys-file-reader.c
index 54788deb6e217bb918d8873077e2cb305ffbea77..8abfe10b689fda6ce24df8c7d916ee3618a37aa7 100644 (file)
--- a/src/data/sys-file-reader.c
+++ b/src/data/sys-file-reader.c
@@ -203,6 +203,7 @@ struct sfm_reader
     size_t sfm_var_cnt;         /* Number of variables. */
     int case_cnt;               /* Number of cases */
     const char *encoding;       /* String encoding. */
+    bool written_by_readstat; /* From https://github.com/WizardMac/ReadStat? */
 
     /* Decompression. */
     enum any_compression compression;
@@ -861,7 +862,7 @@ sfm_decode (struct any_reader *r_, const char *encoding,
      amount that the header claims.  SPSS version 13 gets this
      wrong when very long strings are involved, so don't warn in
      that case. */
-  if (r->header.nominal_case_size != -1
+  if (r->header.nominal_case_size > 0
       && r->header.nominal_case_size != r->n_vars
       && r->info.version_major != 13)
     sys_warn (r, -1, _("File header claims %d variable positions but "
@@ -967,6 +968,8 @@ read_header (struct sfm_reader *r, struct any_read_info *info,
   if (!read_string (r, header->magic, sizeof header->magic)
       || !read_string (r, header->eye_catcher, sizeof header->eye_catcher))
     return false;
+  r->written_by_readstat = strstr (header->eye_catcher,
+                                   "https://github.com/WizardMac/ReadStat");
 
   if (!strcmp (ASCII_MAGIC, header->magic)
       || !strcmp (EBCDIC_MAGIC, header->magic))
@@ -1454,11 +1457,13 @@ parse_variable_records (struct sfm_reader *r, struct dictionary *dict,
                                    "`%s' to `%s'."),
                     name, new_name);
           var = rec->var = dict_create_var_assert (dict, new_name, rec->width);
+          var_set_short_name (var, 0, new_name);
           free (new_name);
         }
 
-      /* Set the short name the same as the long name. */
-      var_set_short_name (var, 0, name);
+      /* Set the short name the same as the long name (even if we renamed
+         it). */
+      var_set_short_name (var, 0, var_get_name (var));
 
       /* Get variable label, if any. */
       if (rec->label)
@@ -1828,10 +1833,9 @@ decode_mrsets (struct sfm_reader *r, struct dictionary *dict)
       size_t i;
 
       name = recode_string ("UTF-8", r->encoding, s->name, -1);
-      if (name[0] != '$')
+      if (!mrset_is_valid_name (name, dict_get_encoding (dict), false))
         {
-          sys_warn (r, -1, _("Multiple response set name `%s' does not begin "
-                             "with `$'."),
+          sys_warn (r, -1, _("Invalid multiple response set name `%s'."),
                     name);
           free (name);
           continue;
@@ -1993,8 +1997,9 @@ parse_display_parameters (struct sfm_reader *r,
 }
 
 static void
-rename_var_and_save_short_names (struct dictionary *dict, struct variable *var,
-                                 const char *new_name)
+rename_var_and_save_short_names (struct sfm_reader *r, off_t pos,
+                                 struct dictionary *dict,
+                                 struct variable *var, const char *new_name)
 {
   size_t n_short_names;
   char **short_names;
@@ -2012,7 +2017,8 @@ rename_var_and_save_short_names (struct dictionary *dict, struct variable *var,
     }
 
   /* Set long name. */
-  dict_rename_var (dict, var, new_name);
+  if (!dict_try_rename_var (dict, var, new_name))
+    sys_warn (r, pos, _("Duplicate long variable name `%s'."), new_name);
 
   /* Restore short names. */
   for (i = 0; i < n_short_names; i++)
@@ -2046,7 +2052,7 @@ parse_long_var_name_map (struct sfm_reader *r,
           char *new_name;
 
           new_name = utf8_to_lower (var_get_name (var));
-          rename_var_and_save_short_names (dict, var, new_name);
+          rename_var_and_save_short_names (r, -1, dict, var, new_name);
           free (new_name);
        }
 
@@ -2071,16 +2077,7 @@ parse_long_var_name_map (struct sfm_reader *r,
           continue;
         }
 
-      /* Identify any duplicates. */
-      if (utf8_strcasecmp (var_get_short_name (var, 0), long_name)
-          && dict_lookup_var (dict, long_name) != NULL)
-        {
-          sys_warn (r, record->pos,
-                    _("Duplicate long variable name `%s'."), long_name);
-          continue;
-        }
-
-      rename_var_and_save_short_names (dict, var, long_name);
+      rename_var_and_save_short_names (r, record->pos, dict, var, long_name);
     }
   close_text_record (r, text);
 }
@@ -2228,7 +2225,15 @@ parse_value_labels (struct sfm_reader *r, struct dictionary *dict,
 
           if (!var_add_value_label (var, &value, utf8_labels[j]))
             {
-              if (var_is_numeric (var))
+              if (r->written_by_readstat)
+                {
+                  /* Ignore the problem.  ReadStat is buggy and emits value
+                     labels whose values are longer than string variables'
+                     widths, that are identical in the actual width of the
+                     variable, e.g. both values "ABC123" and "ABC456" for a
+                     string variable with width 3. */
+                }
+              else if (var_is_numeric (var))
                 sys_warn (r, record->pos,
                           _("Duplicate value label for %g on %s."),
                           value.f, var_get_name (var));
@@ -2330,7 +2335,14 @@ parse_attributes (struct sfm_reader *r, struct text_record *text,
             break;
         }
       if (attrs != NULL)
-        attrset_add (attrs, attr);
+        {
+          if (!attrset_try_add (attrs, attr))
+            {
+              text_warn (r, text, _("Duplicate attribute %s."),
+                         attribute_get_name (attr));
+              attribute_destroy (attr);
+            }
+        }
       else
         attribute_destroy (attr);
     }
@@ -2975,7 +2987,7 @@ open_text_record (struct sfm_reader *r,
 }
 
 /* Closes TEXT, frees its storage, and issues a final warning
-   about suppressed warnings if necesary. */
+   about suppressed warnings if necessary. */
 static void
 close_text_record (struct sfm_reader *r, struct text_record *text)
 {