Fix bug #20910.

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

diff --git a/src/data/procedure.c b/src/data/procedure.c
index 5f3ad2065a4037d2ebda033a6d5b26affcffe76f..b237c68b05eae06e38db25d3fdf30162dac75b9c 100644 (file)
--- a/src/data/procedure.c
+++ b/src/data/procedure.c
@@ -22,6 +22,7 @@
 #include <unistd.h>
 
 #include <data/case.h>
+#include <data/case-map.h>
 #include <data/caseinit.h>
 #include <data/casereader.h>
 #include <data/casereader-provider.h>
@@ -37,6 +38,7 @@
 #include <libpspp/str.h>
 #include <libpspp/taint.h>
 
+
 struct dataset {
   /* Cases are read from source,
      their transformation variables are initialized,
@@ -61,6 +63,11 @@ struct dataset {
   /* Callback which occurs whenever the DICT is replaced by a new one */
   replace_dictionary_callback *replace_dict;
 
+  /* Callback which occurs whenever the transformation chain(s) have
+     been modified */
+  transformation_change_callback_func *xform_callback;
+  void *xform_callback_aux;
+
   /* If true, cases are discarded instead of being written to
      sink. */
   bool discard_output;
@@ -69,9 +76,9 @@ struct dataset {
      added to. */
   struct trns_chain *cur_trns_chain;
 
-  /* The compactor used to compact a case, if necessary;
+  /* The case map used to compact a case, if necessary;
      otherwise a null pointer. */
-  struct dict_compactor *compactor;
+  struct case_map *compactor;
 
   /* Time at which proc was last invoked. */
   time_t last_proc_invocation;
@@ -164,11 +171,13 @@ proc_open (struct dataset *ds)
   /* Prepare sink. */
   if (!ds->discard_output)
     {
-      ds->compactor = (dict_compacting_would_shrink (ds->permanent_dict)
-                       ? dict_make_compactor (ds->permanent_dict)
+      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 (dict_get_compacted_value_cnt (
-                                             ds->permanent_dict));
+      ds->sink = autopaging_writer_create (compacted_value_cnt);
     }
   else
     {
@@ -250,10 +259,7 @@ proc_casereader_read (struct casereader *reader UNUSED, void *ds_,
         {
           struct ccase tmp;
           if (ds->compactor != NULL)
-            {
-              case_create (&tmp, dict_get_compacted_value_cnt (ds->dict));
-              dict_compactor_compact (ds->compactor, &tmp, c);
-            }
+            case_map_execute (ds->compactor, c, &tmp);
           else
             case_clone (&tmp, c);
           casewriter_write (ds->sink, &tmp);
@@ -318,9 +324,11 @@ proc_commit (struct dataset *ds)
       /* Finish compacting. */
       if (ds->compactor != NULL)
         {
-          dict_compactor_destroy (ds->compactor);
-          dict_compact_values (ds->dict);
+          case_map_destroy (ds->compactor);
           ds->compactor = NULL;
+
+          dict_delete_scratch_vars (ds->dict);
+          dict_compact_values (ds->dict);
         }
 
       /* Old data sink becomes new data source. */
@@ -384,6 +392,10 @@ proc_capture_transformations (struct dataset *ds)
   assert (ds->temporary_trns_chain == NULL);
   chain = ds->permanent_trns_chain;
   ds->cur_trns_chain = ds->permanent_trns_chain = trns_chain_create ();
+
+  if ( ds->xform_callback)
+    ds->xform_callback (false, ds->xform_callback_aux);
+
   return chain;
 }
 
@@ -394,6 +406,8 @@ void
 add_transformation (struct dataset *ds, trns_proc_func *proc, trns_free_func *free, void *aux)
 {
   trns_chain_append (ds->cur_trns_chain, NULL, proc, free, aux);
+  if ( ds->xform_callback)
+    ds->xform_callback (true, ds->xform_callback_aux);
 }
 
 /* Adds a transformation that processes a case with PROC and
@@ -408,6 +422,9 @@ add_transformation_with_finalizer (struct dataset *ds,
                                    trns_free_func *free, void *aux)
 {
   trns_chain_append (ds->cur_trns_chain, finalize, proc, free, aux);
+
+  if ( ds->xform_callback)
+    ds->xform_callback (true, ds->xform_callback_aux);
 }
 
 /* Returns the index of the next transformation.
@@ -442,6 +459,9 @@ proc_start_temporary_transformations (struct dataset *ds)
 
       trns_chain_finalize (ds->permanent_trns_chain);
       ds->temporary_trns_chain = ds->cur_trns_chain = trns_chain_create ();
+
+      if ( ds->xform_callback)
+       ds->xform_callback (true, ds->xform_callback_aux);
     }
 }
 
@@ -483,6 +503,10 @@ proc_cancel_temporary_transformations (struct dataset *ds)
       trns_chain_destroy (ds->temporary_trns_chain);
       ds->temporary_trns_chain = NULL;
 
+      if ( ds->xform_callback)
+       ds->xform_callback (!trns_chain_is_empty (ds->permanent_trns_chain),
+                           ds->xform_callback_aux);
+
       return true;
     }
   else
@@ -500,23 +524,35 @@ proc_cancel_all_transformations (struct dataset *ds)
   ok = trns_chain_destroy (ds->temporary_trns_chain) && ok;
   ds->permanent_trns_chain = ds->cur_trns_chain = trns_chain_create ();
   ds->temporary_trns_chain = NULL;
+  if ( ds->xform_callback)
+    ds->xform_callback (false, ds->xform_callback_aux);
+
   return ok;
 }
 \f
 /* Initializes procedure handling. */
 struct dataset *
-create_dataset (replace_source_callback *rps,
-               replace_dictionary_callback *rds)
+create_dataset (transformation_change_callback_func *cb, void *aux)
 {
   struct dataset *ds = xzalloc (sizeof(*ds));
   ds->dict = dict_create ();
   ds->caseinit = caseinit_create ();
-  ds->replace_source = rps;
-  ds->replace_dict = rds;
+  ds->xform_callback = cb;
+  ds->xform_callback_aux = aux;
   proc_cancel_all_transformations (ds);
   return ds;
 }
 
+
+void
+dataset_add_transform_change_callback (struct dataset *ds,
+                                      transformation_change_callback_func *cb,
+                                      void *aux)
+{
+  ds->xform_callback = cb;
+  ds->xform_callback_aux = aux;
+}
+
 /* Finishes up procedure handling. */
 void
 destroy_dataset (struct dataset *ds)
@@ -525,6 +561,9 @@ destroy_dataset (struct dataset *ds)
   dict_destroy (ds->dict);
   caseinit_destroy (ds->caseinit);
   trns_chain_destroy (ds->permanent_trns_chain);
+
+  if ( ds->xform_callback)
+    ds->xform_callback (false, ds->xform_callback_aux);
   free (ds);
 }
 
@@ -597,6 +636,18 @@ proc_has_active_file (const struct dataset *ds)
   return ds->source != NULL;
 }
 
+/* Returns the active file data source from DS, or a null pointer
+   if DS has no data source, and removes it from DS. */
+struct casereader *
+proc_extract_active_file_data (struct dataset *ds)
+{
+  struct casereader *reader = ds->source;
+  ds->source = NULL;
+  if (ds->replace_source) ds->replace_source (reader);
+
+  return reader;
+}
+
 /* Checks whether DS has a corrupted active file.  If so,
    discards it and returns false.  If not, returns true without
    doing anything. */