Change "union value" to dynamically allocate long strings.

[pspp-builds.git] / src / data / procedure.c

diff --git a/src/data/procedure.c b/src/data/procedure.c
index 91e185a45a43f3d78802338d6b26536f0a9dc6b0..b762214d10e20a33d6bbdf89ac52253a67589453 100644 (file)
--- a/src/data/procedure.c
+++ b/src/data/procedure.c
@@ -36,7 +36,9 @@
 #include <libpspp/misc.h>
 #include <libpspp/str.h>
 #include <libpspp/taint.h>
+#include <libpspp/i18n.h>
 
+#include "minmax.h"
 #include "xalloc.h"
 
 struct dataset {
@@ -92,6 +94,10 @@ struct dataset {
   proc_state;
   casenumber cases_written;       /* Cases output so far. */
   bool ok;                    /* Error status. */
+
+  void (*callback) (void *); /* Callback for when the dataset changes */
+  void *cb_data;
+
 }; /* struct dataset */
 
 
@@ -99,9 +105,24 @@ static void add_case_limit_trns (struct dataset *ds);
 static void add_filter_trns (struct dataset *ds);
 
 static void update_last_proc_invocation (struct dataset *ds);
+
+static void
+dataset_set_unsaved (const struct dataset *ds)
+{
+  if (ds->callback) ds->callback (ds->cb_data);
+}
+
 \f
 /* Public functions. */
 
+void
+dataset_set_callback (struct dataset *ds, void (*cb) (void *), void *cb_data)
+{
+  ds->callback = cb;
+  ds->cb_data = cb_data;
+}
+
+
 /* Returns the last time the data was read. */
 time_t
 time_of_last_procedure (struct dataset *ds)
@@ -166,11 +187,19 @@ proc_open (struct dataset *ds)
     {
       struct dictionary *pd = ds->permanent_dict;
       size_t compacted_value_cnt = dict_count_values (pd, 1u << DC_SCRATCH);
-      bool should_compact = compacted_value_cnt < dict_get_next_value_idx (pd);
-      ds->compactor = (should_compact
-                       ? case_map_to_compact_dict (pd, 1u << DC_SCRATCH)
-                       : NULL);
-      ds->sink = autopaging_writer_create (compacted_value_cnt);
+      if (compacted_value_cnt < dict_get_next_value_idx (pd))
+        {
+          struct caseproto *compacted_proto;
+          compacted_proto = dict_get_compacted_proto (pd, 1u << DC_SCRATCH);
+          ds->compactor = case_map_to_compact_dict (pd, 1u << DC_SCRATCH);
+          ds->sink = autopaging_writer_create (compacted_proto);
+          caseproto_unref (compacted_proto);
+        }
+      else
+        {
+          ds->compactor = NULL;
+          ds->sink = autopaging_writer_create (dict_get_proto (pd));
+        }
     }
   else
     {
@@ -188,8 +217,7 @@ proc_open (struct dataset *ds)
   /* FIXME: use taint in dataset in place of `ok'? */
   /* FIXME: for trivial cases we can just return a clone of
      ds->source? */
-  return casereader_create_sequential (NULL,
-                                       dict_get_next_value_idx (ds->dict),
+  return casereader_create_sequential (NULL, dict_get_proto (ds->dict),
                                        CASENUMBER_MAX,
                                        &proc_casereader_class, ds);
 }
@@ -225,7 +253,7 @@ proc_casereader_read (struct casereader *reader UNUSED, void *ds_)
       c = casereader_read (ds->source);
       if (c == NULL)
         return NULL;
-      c = case_unshare_and_resize (c, dict_get_next_value_idx (ds->dict));
+      c = case_unshare_and_resize (c, dict_get_proto (ds->dict));
       caseinit_init_vars (ds->caseinit, c);
 
       /* Execute permanent transformations.  */
@@ -293,6 +321,8 @@ proc_commit (struct dataset *ds)
   assert (ds->proc_state == PROC_CLOSED);
   ds->proc_state = PROC_COMMITTED;
 
+  dataset_set_unsaved (ds);
+
   /* Free memory for lagged cases. */
   while (!deque_is_empty (&ds->lag))
     case_unref (ds->lag_cases[deque_pop_back (&ds->lag)]);
@@ -510,12 +540,25 @@ proc_cancel_all_transformations (struct dataset *ds)
   return ok;
 }
 \f
+
+static void
+dict_callback (struct dictionary *d UNUSED, void *ds_)
+{
+  struct dataset *ds = ds_;
+  dataset_set_unsaved (ds);
+}
+
 /* Initializes procedure handling. */
 struct dataset *
 create_dataset (void)
 {
   struct dataset *ds = xzalloc (sizeof(*ds));
   ds->dict = dict_create ();
+
+  dict_set_change_callback (ds->dict, dict_callback, ds);
+
+  dict_set_encoding (ds->dict, get_default_encoding ());
+
   ds->caseinit = caseinit_create ();
   proc_cancel_all_transformations (ds);
   return ds;
@@ -585,6 +628,7 @@ proc_set_active_file (struct dataset *ds,
 
   dict_destroy (ds->dict);
   ds->dict = dict;
+  dict_set_change_callback (ds->dict, dict_callback, ds);
 
   proc_set_active_file_data (ds, source);
 }