X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Fdata%2Fdataset.c;h=afab39dc0c5e63a4bdfa2a6ba4d49b6b37c65137;hb=refs%2Fheads%2Fpxd;hp=482e6d9309aae3e06a91c55ecd51b14b7394b3cd;hpb=a9b46fb9e208c694e39d6f173bfa6fe631a30129;p=pspp

diff --git a/src/data/dataset.c b/src/data/dataset.c
index 482e6d9309..afab39dc0c 100644
--- a/src/data/dataset.c
+++ b/src/data/dataset.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, 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
@@ -30,8 +30,10 @@
 #include "data/casereader-provider.h"
 #include "data/casereader-shim.h"
 #include "data/casewriter.h"
+#include "data/control-stack.h"
 #include "data/dictionary.h"
 #include "data/file-handle-def.h"
+#include "data/session.h"
 #include "data/transformations.h"
 #include "data/variable.h"
 #include "libpspp/deque.h"
@@ -44,6 +46,15 @@
 #include "gl/xalloc.h"
 
 struct dataset {
+  /* A dataset is usually part of a session.  Within a session its name must
+     unique.  The name must either be a valid PSPP identifier or the empty
+     string.  (It must be unique within the session even if it is the empty
+     string; that is, there may only be a single dataset within a session with
+     the empty string as its name.) */
+  struct session *session;
+  char *name;
+  enum dataset_display display;
+
   /* Cases are read from source,
      their transformation variables are initialized,
      pass through permanent_trns_chain (which transforms them into
@@ -96,8 +107,8 @@ struct dataset {
   const struct dataset_callbacks *callbacks;
   void *cb_data;
 
-  /* Default encoding for reading syntax files. */
-  char *syntax_encoding;
+  /* Uniquely distinguishes datasets. */
+  unsigned int seqno;
 };
 
 static void dataset_changed__ (struct dataset *);
@@ -116,41 +127,90 @@ dict_callback (struct dictionary *d UNUSED, void *ds_)
   dataset_changed__ (ds);
 }
 
-/* Creates and returns a new dataset.  The dataset initially has an empty
-   dictionary and no data source. */
+static void
+dataset_create_finish__ (struct dataset *ds, struct session *session)
+{
+  static unsigned int seqno;
+
+  dict_set_change_callback (ds->dict, dict_callback, ds);
+  proc_cancel_all_transformations (ds);
+  dataset_set_session (ds, session);
+  ds->seqno = ++seqno;
+}
+
+/* Creates a new dataset named NAME, adds it to SESSION, and returns it.  If
+   SESSION already contains a dataset named NAME, it is deleted and replaced.
+   The dataset initially has an empty dictionary and no data source. */
 struct dataset *
-dataset_create (void)
+dataset_create (struct session *session, const char *name)
 {
   struct dataset *ds;
 
   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->name = xstrdup (name);
+  ds->display = DATASET_FRONT;
+  ds->dict = dict_create (get_default_encoding ());
 
   ds->caseinit = caseinit_create ();
-  proc_cancel_all_transformations (ds);
-  ds->syntax_encoding = xstrdup ("Auto");
+
+  dataset_create_finish__ (ds, session);
+
   return ds;
 }
 
+/* Creates and returns a new dataset that has the same data and dictionary as
+   OLD named NAME, adds it to the same session as OLD, and returns the new
+   dataset.  If SESSION already contains a dataset named NAME, it is deleted
+   and replaced.
+
+   OLD must not have any active transformations or temporary state and must
+   not be in the middle of a procedure.
+
+   Callbacks are not cloned. */
+struct dataset *
+dataset_clone (struct dataset *old, const char *name)
+{
+  struct dataset *new;
+
+  assert (old->proc_state == PROC_COMMITTED);
+  assert (trns_chain_is_empty (old->permanent_trns_chain));
+  assert (old->permanent_dict == NULL);
+  assert (old->sink == NULL);
+  assert (old->temporary_trns_chain == NULL);
+
+  new = xzalloc (sizeof *new);
+  new->name = xstrdup (name);
+  new->display = DATASET_FRONT;
+  new->source = casereader_clone (old->source);
+  new->dict = dict_clone (old->dict);
+  new->caseinit = caseinit_clone (old->caseinit);
+  new->last_proc_invocation = old->last_proc_invocation;
+  new->ok = old->ok;
+
+  dataset_create_finish__ (new, old->session);
+
+  return new;
+}
+
 /* Destroys DS. */
 void
 dataset_destroy (struct dataset *ds)
 {
   if (ds != NULL)
     {
+      dataset_set_session (ds, NULL);
       dataset_clear (ds);
       dict_destroy (ds->dict);
       caseinit_destroy (ds->caseinit);
+      ctl_stack_clear ();
       trns_chain_destroy (ds->permanent_trns_chain);
       dataset_transformations_changed__ (ds, false);
-      free (ds->syntax_encoding);
+      free (ds->name);
       free (ds);
     }
 }
 
-/* Discards the active file dictionary, data, and transformations. */
+/* Discards the active dataset's dictionary, data, and transformations. */
 void
 dataset_clear (struct dataset *ds)
 {
@@ -167,6 +227,55 @@ dataset_clear (struct dataset *ds)
   proc_cancel_all_transformations (ds);
 }
 
+const char *
+dataset_name (const struct dataset *ds)
+{
+  return ds->name;
+}
+
+void
+dataset_set_name (struct dataset *ds, const char *name)
+{
+  struct session *session = ds->session;
+  bool active = false;
+
+  if (session != NULL)
+    {
+      active = session_active_dataset (session) == ds;
+      if (active)
+        session_set_active_dataset (session, NULL);
+      dataset_set_session (ds, NULL);
+    }
+
+  free (ds->name);
+  ds->name = xstrdup (name);
+
+  if (session != NULL)
+    {
+      dataset_set_session (ds, session);
+      if (active)
+        session_set_active_dataset (session, ds);
+    }
+}
+
+struct session *
+dataset_session (const struct dataset *ds)
+{
+  return ds->session;
+}
+
+void
+dataset_set_session (struct dataset *ds, struct session *session)
+{
+  if (session != ds->session)
+    {
+      if (ds->session != NULL)
+        session_remove_dataset (ds->session, ds);
+      if (session != NULL)
+        session_add_dataset (session, ds);
+    }
+}
+
 /* Returns the dictionary within DS.  This is always nonnull, although it
    might not contain any variables. */
 struct dictionary *
@@ -205,7 +314,7 @@ dataset_has_source (const struct dataset *ds)
   return dataset_source (ds) != NULL;
 }
 
-/* Replaces the active file's data by READER.  READER's cases must have an
+/* Replaces the active dataset's data by READER.  READER's cases must have an
    appropriate format for DS's dictionary. */
 bool
 dataset_set_source (struct dataset *ds, struct casereader *reader)
@@ -230,6 +339,15 @@ dataset_steal_source (struct dataset *ds)
   return reader;
 }
 
+/* Returns a number unique to DS.  It can be used to distinguish one dataset
+   from any other within a given program run, even datasets that do not exist
+   at the same time. */
+unsigned int
+dataset_seqno (const struct dataset *ds)
+{
+  return ds->seqno;
+}
+
 void
 dataset_set_callbacks (struct dataset *ds,
                        const struct dataset_callbacks *callbacks,
@@ -239,17 +357,16 @@ dataset_set_callbacks (struct dataset *ds,
   ds->cb_data = cb_data;
 }
 
-void
-dataset_set_default_syntax_encoding (struct dataset *ds, const char *encoding)
+enum dataset_display
+dataset_get_display (const struct dataset *ds)
 {
-  free (ds->syntax_encoding);
-  ds->syntax_encoding = xstrdup (encoding);
+  return ds->display;
 }
 
-const char *
-dataset_get_default_syntax_encoding (const struct dataset *ds)
+void
+dataset_set_display (struct dataset *ds, enum dataset_display display)
 {
-  return ds->syntax_encoding;
+  ds->display = display;
 }
 
 /* Returns the last time the data was read. */
@@ -311,7 +428,7 @@ proc_open_filtering (struct dataset *ds, bool filter)
   add_case_limit_trns (ds);
   if (filter)
     add_filter_trns (ds);
-  trns_chain_finalize (ds->cur_trns_chain);
+  ctl_stack_clear ();
 
   /* Make permanent_dict refer to the dictionary right before
      data reaches the sink. */
@@ -426,7 +543,7 @@ proc_casereader_read (struct casereader *reader UNUSED, void *ds_)
           ds->lag_cases[deque_push_front (&ds->lag)] = case_ref (c);
         }
 
-      /* Write case to replacement active file. */
+      /* Write case to replacement dataset. */
       ds->cases_written++;
       if (ds->sink != NULL)
         casewriter_write (ds->sink,
@@ -459,7 +576,7 @@ proc_casereader_destroy (struct casereader *reader, void *ds_)
 
   /* Make sure transformations happen for every input case, in
      case they have side effects, and ensure that the replacement
-     active file gets all the cases it should. */
+     active dataset gets all the cases it should. */
   while ((c = casereader_read (reader)) != NULL)
     case_unref (c);
 
@@ -472,7 +589,7 @@ proc_casereader_destroy (struct casereader *reader, void *ds_)
 /* Must return false if the source casereader, a transformation,
    or the sink casewriter signaled an error.  (If a temporary
    transformation signals an error, then the return value is
-   false, but the replacement active file may still be
+   false, but the replacement active dataset may still be
    untainted.) */
 bool
 proc_commit (struct dataset *ds)
@@ -576,22 +693,7 @@ proc_capture_transformations (struct dataset *ds)
 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);
-  dataset_transformations_changed__ (ds, true);
-}
-
-/* Adds a transformation that processes a case with PROC and
-   frees itself with FREE to the current set of transformations.
-   When parsing of the block of transformations is complete,
-   FINALIZE will be called.
-   The functions are passed AUX as auxiliary data. */
-void
-add_transformation_with_finalizer (struct dataset *ds,
-				   trns_finalize_func *finalize,
-                                   trns_proc_func *proc,
-                                   trns_free_func *free, void *aux)
-{
-  trns_chain_append (ds->cur_trns_chain, finalize, proc, free, aux);
+  trns_chain_append (ds->cur_trns_chain, proc, free, aux);
   dataset_transformations_changed__ (ds, true);
 }
 
@@ -625,25 +727,31 @@ proc_start_temporary_transformations (struct dataset *ds)
 
       ds->permanent_dict = dict_clone (ds->dict);
 
-      trns_chain_finalize (ds->permanent_trns_chain);
+      ctl_stack_clear ();
       ds->temporary_trns_chain = ds->cur_trns_chain = trns_chain_create ();
       dataset_transformations_changed__ (ds, true);
     }
 }
 
-/* Converts all the temporary transformations, if any, to
-   permanent transformations.  Further transformations will be
-   permanent.
+/* Converts all the temporary transformations, if any, to permanent
+   transformations.  Further transformations will be permanent.
+
+   The FILTER command is implemented as a temporary transformation, so a
+   procedure that uses this function should usually use proc_open_filtering()
+   with FILTER false, instead of plain proc_open().
+
    Returns true if anything changed, false otherwise. */
 bool
 proc_make_temporary_transformations_permanent (struct dataset *ds)
 {
   if (proc_in_temporary_transformations (ds))
     {
-      trns_chain_finalize (ds->temporary_trns_chain);
+      ctl_stack_clear ();
       trns_chain_splice (ds->permanent_trns_chain, ds->temporary_trns_chain);
       ds->temporary_trns_chain = NULL;
 
+      ds->cur_trns_chain = ds->permanent_trns_chain;
+
       dict_destroy (ds->permanent_dict);
       ds->permanent_dict = NULL;
 
@@ -665,6 +773,7 @@ proc_cancel_temporary_transformations (struct dataset *ds)
       ds->dict = ds->permanent_dict;
       ds->permanent_dict = NULL;
 
+      ctl_stack_clear ();
       trns_chain_destroy (ds->temporary_trns_chain);
       ds->temporary_trns_chain = NULL;
       dataset_transformations_changed__ (
@@ -682,6 +791,7 @@ proc_cancel_all_transformations (struct dataset *ds)
 {
   bool ok;
   assert (ds->proc_state == PROC_COMMITTED);
+  ctl_stack_clear ();
   ok = trns_chain_destroy (ds->permanent_trns_chain);
   ok = trns_chain_destroy (ds->temporary_trns_chain) && ok;
   ds->permanent_trns_chain = ds->cur_trns_chain = trns_chain_create ();
@@ -690,6 +800,39 @@ proc_cancel_all_transformations (struct dataset *ds)
 
   return ok;
 }
+
+static int
+store_case_num (void *var_, struct ccase **cc, casenumber case_num)
+{
+  struct variable *var = var_;
+
+  *cc = case_unshare (*cc);
+  case_data_rw (*cc, var)->f = case_num;
+
+  return TRNS_CONTINUE;
+}
+
+/* Add a variable which we can sort by to get back the original order. */
+struct variable *
+add_permanent_ordering_transformation (struct dataset *ds)
+{
+  struct variable *temp_var;
+
+  temp_var = dict_create_var_assert (ds->dict, "$ORDER", 0);
+  if (proc_in_temporary_transformations (ds))
+    {
+      struct variable *perm_var;
+
+      perm_var = dict_clone_var_in_place_assert (ds->permanent_dict, temp_var);
+      trns_chain_append (ds->permanent_trns_chain, NULL, store_case_num,
+                         NULL, perm_var);
+      trns_chain_finalize (ds->permanent_trns_chain);
+    }
+  else
+    add_transformation (ds, store_case_num, NULL, temp_var);
+
+  return temp_var;
+}
 
 /* Causes output from the next procedure to be discarded, instead
    of being preserved for use as input for the next procedure. */
@@ -700,7 +843,7 @@ proc_discard_output (struct dataset *ds)
 }
 
 
-/* Checks whether DS has a corrupted active file.  If so,
+/* Checks whether DS has a corrupted active dataset.  If so,
    discards it and returns false.  If not, returns true without
    doing anything. */
 bool
@@ -814,3 +957,11 @@ dataset_transformations_changed__ (struct dataset *ds, bool non_empty)
   if (ds->callbacks != NULL && ds->callbacks->transformations_changed != NULL)
     ds->callbacks->transformations_changed (non_empty, ds->cb_data);
 }
+
+/* Private interface for use by session code. */
+
+void
+dataset_set_session__ (struct dataset *ds, struct session *session)
+{
+  ds->session = session;
+}