X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fdata%2Fdictionary.c;h=79952b58140f386432851a05f41586ff10172c79;hb=491cd3744c11da45ac6e41c734e4a33c19b060c9;hp=c00bd4face382c363990b6185741bdf1d62080c5;hpb=b401615e6db40bf74394839b96600afe3a868a95;p=pspp

diff --git a/src/data/dictionary.c b/src/data/dictionary.c
index c00bd4face..79952b5814 100644
--- a/src/data/dictionary.c
+++ b/src/data/dictionary.c
@@ -1,5 +1,5 @@
 /* PSPP - a program for statistical analysis.
-   Copyright (C) 1997-9, 2000, 2006, 2007, 2009, 2010, 2011 Free Software Foundation, Inc.
+   Copyright (C) 1997-9, 2000, 2006, 2007, 2009, 2010, 2011, 2012, 2013 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
@@ -87,16 +87,8 @@ struct dictionary
 static void dict_unset_split_var (struct dictionary *, struct variable *);
 static void dict_unset_mrset_var (struct dictionary *, struct variable *);
 
-void
-dict_set_encoding (struct dictionary *d, const char *enc)
-{
-  if (enc)
-    {
-      free (d->encoding);
-      d->encoding = xstrdup (enc);
-    }
-}
-
+/* Returns the encoding for data in dictionary D.  The return value is a
+   nonnull string that contains an IANA character set name. */
 const char *
 dict_get_encoding (const struct dictionary *d)
 {
@@ -171,14 +163,16 @@ dict_copy_callbacks (struct dictionary *dest,
   dest->cb_data = src->cb_data;
 }
 
-/* Creates and returns a new dictionary. */
+/* Creates and returns a new dictionary with the specified ENCODING. */
 struct dictionary *
-dict_create (void)
+dict_create (const char *encoding)
 {
   struct dictionary *d = xzalloc (sizeof *d);
 
+  d->encoding = xstrdup (encoding);
   hmap_init (&d->name_map);
   attrset_init (&d->attributes);
+
   return d;
 }
 
@@ -189,14 +183,16 @@ dict_create (void)
    dictionary.  If the new dictionary won't be used to access
    cases produced with the old dictionary, then the new
    dictionary's case indexes should be compacted with
-   dict_compact_values to save space. */
+   dict_compact_values to save space.
+
+   Callbacks are not cloned. */
 struct dictionary *
 dict_clone (const struct dictionary *s)
 {
   struct dictionary *d;
   size_t i;
 
-  d = dict_create ();
+  d = dict_create (s->encoding);
 
   for (i = 0; i < s->var_cnt; i++)
     {
@@ -235,9 +231,6 @@ dict_clone (const struct dictionary *s)
   for (i = 0; i < s->vector_cnt; i++)
     d->vector[i] = vector_clone (s->vector[i], s, d);
 
-  if ( s->encoding)
-    d->encoding = xstrdup (s->encoding);
-
   dict_set_attributes (d, dict_get_attributes (s));
 
   for (i = 0; i < s->n_mrsets; i++)
@@ -286,17 +279,6 @@ dict_clear (struct dictionary *d)
   attrset_clear (&d->attributes);
 }
 
-/* Destroys the aux data for every variable in D, by calling
-   var_clear_aux() for each variable. */
-void
-dict_clear_aux (struct dictionary *d)
-{
-  int i;
-
-  for (i = 0; i < d->var_cnt; i++)
-    var_clear_aux (d->var[i].var);
-}
-
 /* Clears a dictionary and destroys it. */
 void
 dict_destroy (struct dictionary *d)
@@ -308,6 +290,7 @@ dict_destroy (struct dictionary *d)
       d->callbacks  = NULL ;
 
       dict_clear (d);
+      string_array_destroy (&d->documents);
       hmap_destroy (&d->name_map);
       attrset_destroy (&d->attributes);
       dict_clear_mrsets (d);
@@ -380,10 +363,13 @@ dict_get_vars_mutable (const struct dictionary *d, struct variable ***vars,
 }
 
 static struct variable *
-add_var (struct dictionary *d, struct variable *v)
+add_var_with_case_index (struct dictionary *d, struct variable *v,
+                         int case_index)
 {
   struct vardict_info *vardict;
 
+  assert (case_index >= d->next_value_idx);
+
   /* Update dictionary. */
   if (d->var_cnt >= d->var_cap)
     {
@@ -403,20 +389,26 @@ add_var (struct dictionary *d, struct variable *v)
   vardict->dict = d;
   vardict->var = v;
   hmap_insert (&d->name_map, &vardict->name_node,
-               hash_case_string (var_get_name (v), 0));
-  vardict->case_index = d->next_value_idx;
+               utf8_hash_case_string (var_get_name (v), 0));
+  vardict->case_index = case_index;
   var_set_vardict (v, vardict);
 
   if ( d->changed ) d->changed (d, d->changed_data);
   if ( d->callbacks &&  d->callbacks->var_added )
     d->callbacks->var_added (d, var_get_dict_index (v), d->cb_data);
 
-  d->next_value_idx++;
   invalidate_proto (d);
+  d->next_value_idx = case_index + 1;
 
   return v;
 }
 
+static struct variable *
+add_var (struct dictionary *d, struct variable *v)
+{
+  return add_var_with_case_index (d, v, d->next_value_idx);
+}
+
 /* Creates and returns a new variable in D with the given NAME
    and WIDTH.  Returns a null pointer if the given NAME would
    duplicate that of an existing variable in the dictionary. */
@@ -485,6 +477,15 @@ dict_clone_var_as_assert (struct dictionary *d, const struct variable *old_var,
   return add_var (d, new_var);
 }
 
+struct variable *
+dict_clone_var_in_place_assert (struct dictionary *d,
+                                const struct variable *old_var)
+{
+  assert (dict_lookup_var (d, var_get_name (old_var)) == NULL);
+  return add_var_with_case_index (d, var_clone (old_var),
+                                  var_get_case_index (old_var));
+}
+
 /* Returns the variable named NAME in D, or a null pointer if no
    variable has that name. */
 struct variable *
@@ -493,10 +494,10 @@ dict_lookup_var (const struct dictionary *d, const char *name)
   struct vardict_info *vardict;
 
   HMAP_FOR_EACH_WITH_HASH (vardict, struct vardict_info, name_node,
-                           hash_case_string (name, 0), &d->name_map)
+                           utf8_hash_case_string (name, 0), &d->name_map)
     {
       struct variable *var = vardict->var;
-      if (!strcasecmp (var_get_name (var), name))
+      if (!utf8_strcasecmp (var_get_name (var), name))
         return var;
     }
 
@@ -546,6 +547,7 @@ static void
 reindex_var (struct dictionary *d, struct vardict_info *vardict)
 {
   struct variable *var = vardict->var;
+  struct variable *old = var_clone (var);
 
   var_set_vardict (var, vardict);
   hmap_insert_fast (&d->name_map, &vardict->name_node,
@@ -553,7 +555,8 @@ reindex_var (struct dictionary *d, struct vardict_info *vardict)
 
   if ( d->changed ) d->changed (d, d->changed_data);
   if ( d->callbacks &&  d->callbacks->var_changed )
-    d->callbacks->var_changed (d, var_get_dict_index (var), d->cb_data);
+    d->callbacks->var_changed (d, var_get_dict_index (var), VAR_TRAIT_POSITION, old, d->cb_data);
+  var_destroy (old);
 }
 
 /* Sets the case_index in V's vardict to CASE_INDEX. */
@@ -603,13 +606,9 @@ dict_delete_var (struct dictionary *d, struct variable *v)
 {
   int dict_index = var_get_dict_index (v);
   const int case_index = var_get_case_index (v);
-  const int width = var_get_width (v);
 
   assert (dict_contains_var (d, v));
 
-  /* Delete aux data. */
-  var_clear_aux (v);
-
   dict_unset_split_var (d, v);
   dict_unset_mrset_var (d, v);
 
@@ -631,13 +630,14 @@ dict_delete_var (struct dictionary *d, struct variable *v)
 
   /* Free memory. */
   var_clear_vardict (v);
-  var_destroy (v);
 
   if ( d->changed ) d->changed (d, d->changed_data);
 
   invalidate_proto (d);
   if (d->callbacks &&  d->callbacks->var_deleted )
-    d->callbacks->var_deleted (d, dict_index, case_index, width, d->cb_data);
+    d->callbacks->var_deleted (d, v, dict_index, case_index, d->cb_data);
+
+  var_destroy (v);
 }
 
 /* Deletes the COUNT variables listed in VARS from D.  This is
@@ -748,7 +748,7 @@ rename_var (struct variable *v, const char *new_name)
   struct vardict_info *vardict = var_get_vardict (v);
   var_clear_vardict (v);
   var_set_name (v, new_name);
-  vardict->name_node.hash = hash_case_string (new_name, 0);
+  vardict->name_node.hash = utf8_hash_case_string (new_name, 0);
   var_set_vardict (v, vardict);
 }
 
@@ -759,7 +759,8 @@ void
 dict_rename_var (struct dictionary *d, struct variable *v,
                  const char *new_name)
 {
-  assert (!strcasecmp (var_get_name (v), new_name)
+  struct variable *old = var_clone (v);
+  assert (!utf8_strcasecmp (var_get_name (v), new_name)
           || dict_lookup_var (d, new_name) == NULL);
 
   unindex_var (d, var_get_vardict (v));
@@ -771,7 +772,9 @@ dict_rename_var (struct dictionary *d, struct variable *v,
 
   if ( d->changed ) d->changed (d, d->changed_data);
   if ( d->callbacks &&  d->callbacks->var_changed )
-    d->callbacks->var_changed (d, var_get_dict_index (v), d->cb_data);
+    d->callbacks->var_changed (d, var_get_dict_index (v), VAR_TRAIT_NAME, old, d->cb_data);
+
+  var_destroy (old);
 }
 
 /* Renames COUNT variables specified in VARS to the names given
@@ -1253,15 +1256,18 @@ dict_get_label (const struct dictionary *d)
   return d->label;
 }
 
-/* Sets D's file label to LABEL, truncating it to a maximum of 60
-   characters.
+/* Sets D's file label to LABEL, truncating it to at most 60 bytes in D's
+   encoding.
 
    Removes D's label if LABEL is null or the empty string. */
 void
 dict_set_label (struct dictionary *d, const char *label)
 {
   free (d->label);
-  d->label = label != NULL && label[0] != '\0' ? xstrndup (label, 60) : NULL;
+  if (label == NULL || label[0] == '\0')
+    d->label = NULL;
+  else
+    d->label = utf8_encoding_trunc (label, d->encoding, 60);
 }
 
 /* Returns the documents for D, as an UTF-8 encoded string_array.  The
@@ -1414,7 +1420,7 @@ dict_lookup_vector (const struct dictionary *d, const char *name)
 {
   size_t i;
   for (i = 0; i < d->vector_cnt; i++)
-    if (!strcasecmp (vector_get_name (d->vector[i]), name))
+    if (!utf8_strcasecmp (vector_get_name (d->vector[i]), name))
       return d->vector[i];
   return NULL;
 }
@@ -1459,7 +1465,7 @@ dict_lookup_mrset_idx (const struct dictionary *dict, const char *name)
   size_t i;
 
   for (i = 0; i < dict->n_mrsets; i++)
-    if (!strcasecmp (name, dict->mrsets[i]->name))
+    if (!utf8_strcasecmp (name, dict->mrsets[i]->name))
       return i;
 
   return SIZE_MAX;
@@ -1590,10 +1596,12 @@ dict_has_attributes (const struct dictionary *d)
   return attrset_count (&d->attributes) > 0;
 }
 
-/* Called from variable.c to notify the dictionary that some property of
-   the variable has changed */
+/* Called from variable.c to notify the dictionary that some property (indicated
+   by WHAT) of the variable has changed.  OLDVAR is a copy of V as it existed
+   prior to the change.  OLDVAR is destroyed by this function.
+*/
 void
-dict_var_changed (const struct variable *v)
+dict_var_changed (const struct variable *v, unsigned int what, struct variable *oldvar)
 {
   if ( var_has_vardict (v))
     {
@@ -1605,49 +1613,12 @@ dict_var_changed (const struct variable *v)
 
       if (d->changed ) d->changed (d, d->changed_data);
       if ( d->callbacks && d->callbacks->var_changed )
-	d->callbacks->var_changed (d, var_get_dict_index (v), d->cb_data);
-    }
-}
-
-
-/* Called from variable.c to notify the dictionary that the variable's width
-   has changed */
-void
-dict_var_resized (const struct variable *v, int old_width)
-{
-  if ( var_has_vardict (v))
-    {
-      const struct vardict_info *vardict = var_get_vardict (v);
-      struct dictionary *d;
-
-      d = vardict->dict;
-
-      if (d->changed) d->changed (d, d->changed_data);
-
-      invalidate_proto (d);
-      if ( d->callbacks && d->callbacks->var_resized )
-	d->callbacks->var_resized (d, var_get_dict_index (v), old_width,
-                                   d->cb_data);
+	d->callbacks->var_changed (d, var_get_dict_index (v), what, oldvar, d->cb_data);
     }
+  var_destroy (oldvar);
 }
 
-/* Called from variable.c to notify the dictionary that the variable's display width
-   has changed */
-void
-dict_var_display_width_changed (const struct variable *v)
-{
-  if ( var_has_vardict (v))
-    {
-      const struct vardict_info *vardict = var_get_vardict (v);
-      struct dictionary *d;
-
-      d = vardict->dict;
 
-      if (d->changed) d->changed (d, d->changed_data);
-      if ( d->callbacks && d->callbacks->var_display_width_changed )
-	d->callbacks->var_display_width_changed (d, var_get_dict_index (v), d->cb_data);
-    }
-}
 
 /* Dictionary used to contain "internal variables". */
 static struct dictionary *internal_dict;
@@ -1658,7 +1629,7 @@ struct variable *
 dict_create_internal_var (int case_idx, int width)
 {
   if (internal_dict == NULL)
-    internal_dict = dict_create ();
+    internal_dict = dict_create ("UTF-8");
 
   for (;;)
     {