X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;ds=sidebyside;f=lib%2Fsorter.h;h=c2da83113646d0d1762786e491e67c731f9e02c7;hb=eb3d130d22a4235cbfc8fd547e4e303780149752;hp=d1f059fccec58f0240748dc7dc0777cb78eaa27d;hpb=4714386ed8888c1edc4e13ed976dd2c3d67c63ee;p=libucw.git

diff --git a/lib/sorter.h b/lib/sorter.h
index d1f059fc..c2da8311 100644
--- a/lib/sorter.h
+++ b/lib/sorter.h
@@ -1,7 +1,10 @@
 /*
  *	Sherlock Library -- Universal Sorter
  *
- *	(c) 2001 Martin Mares <mj@ucw.cz>
+ *	(c) 2001--2004 Martin Mares <mj@ucw.cz>
+ *
+ *	This software may be freely distributed and used according to the terms
+ *	of the GNU Lesser General Public License.
  */
 
 /*
@@ -15,8 +18,17 @@
  *  SORT_KEY	    [*]	data type capable of storing a single key
  *  SORT_PREFIX(x)  [*] add a name prefix (used on all global names
  *			defined by the sorter)
- *  SORT_PRESORT	include an in-core presorting pass
+ *  SORT_PRESORT	include an in-core presorting pass. Beware, when in
+ *			the pre-sorting mode, it's quite possible that the
+ *			comparison function will be called with both arguments
+ *			identical.
  *  SORT_UNIFY		merge items with identical keys
+ *  SORT_UNIQUE		all items have distinct keys (checked in debug mode)
+ *  SORT_REGULAR	all items are equally long and they don't contain
+ *			anything else than the key. In this case, the sorter
+ *			automatically supplies fetch_key, copy_data, fetch_item
+ *			and store_item functions. Item size is also expected
+ *			to be small.
  *  SORT_DELETE_INPUT	a C expression, if true, the input files are
  *			deleted as soon as possible
  *  SORT_INPUT_FILE	input is a file with this name
@@ -64,6 +76,9 @@
  *			from the list of items, but not deallocated, so
  *			the remaining item can freely reference data of the
  *			other one.
+ *
+ *  After including this file, all parameter macros are automatically
+ *  undef'd.
  */
 
 /* Declarations of externals from sorter.c */
@@ -75,8 +90,7 @@ extern uns sorter_trace;
 extern uns sorter_presort_bufsize;
 extern uns sorter_stream_bufsize;
 
-extern uns sorter_pass_counter, sorter_file_counter;
-struct fastbuf *sorter_open_tmp(void);
+extern uns sorter_pass_counter;
 
 #endif		/* !SORT_DECLS_READ */
 
@@ -102,14 +116,43 @@ struct fastbuf *sorter_open_tmp(void);
 #define LESS <=
 #endif
 
+#if defined(SORT_UNIQUE) && defined(DEBUG)
+#define SORT_ASSERT_UNIQUE
+#endif
+
+#ifdef SORT_REGULAR
+
+static inline int
+P(fetch_key)(struct fastbuf *in, SORT_KEY *x)
+{
+  return breadb(in, x, sizeof(*x));
+}
+
+static inline void
+P(copy_data)(struct fastbuf *in UNUSED, struct fastbuf *out, SORT_KEY *x)
+{
+  bwrite(out, x, sizeof(*x));
+}
+
+static inline byte *
+P(fetch_item)(struct fastbuf *in UNUSED, SORT_KEY *x UNUSED, byte *limit UNUSED)
+{
+  return (byte *)(x+1);
+}
+
+static inline void
+P(store_item)(struct fastbuf *out, SORT_KEY *x)
+{
+  bwrite(out, x, sizeof(*x));
+}
+
+#endif
+
 static struct fastbuf *
 P(flush_out)(struct fastbuf *out)
 {
   if (out)
-    {
-      bflush(out);
-      bsetpos(out, 0);
-    }
+    brewind(out);
   return out;
 }
 
@@ -147,7 +190,7 @@ P(pass)(struct fastbuf **fb1, struct fastbuf **fb2)
 	  struct fastbuf *t;
 	  SWAP(out1, out2, t);
 	  if (!out1)
-	    out1 = sorter_open_tmp();
+	    out1 = bopen_tmp(sorter_stream_bufsize);
 	  run_count++;
 	}
       if (comp LESS 0)
@@ -170,6 +213,10 @@ P(pass)(struct fastbuf **fb1, struct fastbuf **fb2)
 	  run2 = next2 && (P(compare)(kprev2, kin2) LESS 0);
 	  kout = kprev2;
 	}
+#endif
+#ifdef SORT_ASSERT_UNIQUE
+      else if (unlikely(comp == 0))
+	ASSERT(0);
 #endif
       else
 	{
@@ -198,6 +245,73 @@ P(pass)(struct fastbuf **fb1, struct fastbuf **fb2)
 
 #ifdef SORT_PRESORT
 
+#if defined(SORT_REGULAR) && !defined(SORT_UNIFY)
+
+/* If we are doing a simple sort on a regular file, we can use a faster presorting strategy */
+
+static SORT_KEY *P(array);
+
+#define ASORT_PREFIX(x) SORT_PREFIX(x##_array)
+#define ASORT_KEY_TYPE SORT_KEY
+#define ASORT_ELT(i) P(array)[i]
+#define ASORT_LT(x,y) (P(compare)(&(x),&(y)) < 0)
+
+#include "lib/arraysort.h"
+
+static void
+P(presort)(struct fastbuf **fb1, struct fastbuf **fb2)
+{
+  struct fastbuf *in = *fb1;
+  struct fastbuf *out1 = NULL;
+  struct fastbuf *out2 = NULL;
+  struct fastbuf *tbuf;
+  uns buf_items = sorter_presort_bufsize / sizeof(SORT_KEY);
+  uns run_count = 0;
+  SORT_KEY last_out, *array;
+
+  ASSERT(!*fb2);
+  if (buf_items < 2)
+    die("PresortBuffer set too low");
+  P(array) = array = xmalloc(buf_items * sizeof(SORT_KEY));
+
+  for(;;)
+    {
+      uns s = bread(in, array, buf_items * sizeof(SORT_KEY));
+      uns n = s / sizeof(SORT_KEY);
+      ASSERT(!(s % sizeof(SORT_KEY)));
+      if (!n)
+	break;
+      P(sort_array)(n);
+#ifdef SORT_ASSERT_UNIQUE
+      for (uns i=0; i<n-1; i++)
+	if (unlikely(P(compare)(&array[i], &array[i+1]) >= 0))
+	  ASSERT(0);
+      ASSERT(!run_count || P(compare)(&last_out, &array[0]));
+#endif
+      if (!run_count || P(compare)(&last_out, &array[0]) > 0)
+	{
+	  run_count++;
+	  SWAP(out1, out2, tbuf);
+	  if (!out1)
+	    out1 = bopen_tmp(sorter_stream_bufsize);
+	}
+      last_out = array[n-1];
+      bwrite(out1, array, n * sizeof(SORT_KEY));
+    }
+
+  bclose(in);
+  if (sorter_trace)
+    log(L_INFO, "Pass 0: %d runs, %d+%d KB",
+	run_count,
+	(out1 ? (int)((btell(out1) + 1023) / 1024) : 0),
+	(out2 ? (int)((btell(out2) + 1023) / 1024) : 0));
+  *fb1 = P(flush_out)(out1);
+  *fb2 = P(flush_out)(out2);
+  xfree(array);
+}
+
+#else
+
 #define SORT_NODE struct P(presort_node)
 
 SORT_NODE {
@@ -273,7 +387,7 @@ P(presort)(struct fastbuf **fb1, struct fastbuf **fb2)
     {
       SWAP(out1, out2, tbuf);
       if (!out1)
-	out1 = sorter_open_tmp();
+	out1 = bopen_tmp(sorter_stream_bufsize);
       current = buffer;
       last = &first;
       if (leftover)
@@ -332,6 +446,9 @@ P(presort)(struct fastbuf **fb1, struct fastbuf **fb2)
 		first = second->next;
 	      continue;
 	    }
+#endif
+#ifdef SORT_ASSERT_UNIQUE
+	  ASSERT(!first->next || P(compare)(&first->key, &first->next->key));
 #endif
 	  P(store_item)(out1, &first->key);
 	  first = first->next;
@@ -346,8 +463,11 @@ P(presort)(struct fastbuf **fb1, struct fastbuf **fb2)
 	(out2 ? (int)((btell(out2) + 1023) / 1024) : 0));
   *fb1 = P(flush_out)(out1);
   *fb2 = P(flush_out)(out2);
+  xfree(buffer);
 }
 
+#endif		/* SORT_REGULAR && !SORT_UNIFY */
+
 #endif		/* SORT_PRESORT */
 
 static
@@ -382,7 +502,7 @@ struct fastbuf *fb1, struct fastbuf *fb2
 #endif
 
 #ifdef SORT_DELETE_INPUT
-  fb1->is_temp_file = SORT_DELETE_INPUT;
+  bconfig(fb1, BCONFIG_IS_TEMP_FILE, SORT_DELETE_INPUT);
 #endif
   sorter_pass_counter = 1;
 #ifdef SORT_PRESORT
@@ -391,12 +511,12 @@ struct fastbuf *fb1, struct fastbuf *fb2
 #endif
     do P(pass)(&fb1, &fb2); while (fb1 && fb2);
   if (!fb1)
-    fb1 = sorter_open_tmp();
+    fb1 = bopen_tmp(sorter_stream_bufsize);
 
 #ifdef SORT_OUTPUT_FB
   return fb1;
 #else
-  fb1->is_temp_file = 0;
+  bconfig(fb1, BCONFIG_IS_TEMP_FILE, 0);
   if (rename(fb1->name, outname) < 0)
     die("rename(%s,%s): %m", fb1->name, outname);
   bclose(fb1);
@@ -407,5 +527,17 @@ struct fastbuf *fb1, struct fastbuf *fb2
 #undef LESS
 #undef SWAP
 #undef SORT_NODE
+#undef SORT_KEY
+#undef SORT_PREFIX
+#undef SORT_UNIFY
+#undef SORT_UNIQUE
+#undef SORT_ASSERT_UNIQUE
+#undef SORT_REGULAR
+#undef SORT_DELETE_INPUT
+#undef SORT_INPUT_FILE
+#undef SORT_INPUT_FB
+#undef SORT_INPUT_FBPAIR
+#undef SORT_OUTPUT_FILE
+#undef SORT_OUTPUT_FB
 
 #endif		/* !SORT_DECLARE_ONLY */