Merge with git+ssh://git.ucw.cz/projects/sherlock/GIT/sherlock.git#dev-sorter

[libucw.git] / lib / sorter / sorter.h

diff --git a/lib/sorter/sorter.h b/lib/sorter/sorter.h
index 39bdcb97a7a2d3cf7a995f2da43de4a830e318d3..675a3d5b14fd21a0dbf5bd79f678c2a4e119e24c 100644 (file)
--- a/lib/sorter/sorter.h
+++ b/lib/sorter/sorter.h
@@ -46,6 +46,7 @@
  *                     is supplied automatically and the sorting function gets an extra
  *                     parameter specifying a range of the integers. The better the range
  *                     fits, the faster we sort. Sets up SORT_HASH_xxx automatically.
+ *  SORT_INT64(key)    the same for 64-bit integers.
  *
  *  Hashing (optional, but it can speed sorting up):
  *
@@ -60,10 +61,14 @@
  *  void PREFIX_write_merged(struct fastbuf *f, SORT_KEY **keys, void **data, uns n, void *buf)
  *                     takes n records in memory with keys which compare equal and writes
  *                     a single record to the given fastbuf. `buf' points to a buffer which
- *                     is guaranteed to hold all given records.
+ *                     is guaranteed to hold the sum of workspace requirements (see below)
+ *                     over all given records. The function is allowed to modify all its inputs.
  *  void PREFIX_copy_merged(SORT_KEY **keys, struct fastbuf **data, uns n, struct fastbuf *dest)
  *                     takes n records with keys in memory and data in fastbufs and writes
- *                     a single record.
+ *                     a single record. Used only if SORT_DATA_SIZE or SORT_UNIFY_WORKSPACE is defined.
+ *  SORT_UNIFY_WORKSPACE(key)
+ *                     gets a key and returns the amount of workspace required when merging
+ *                     the given record. Defaults to 0.
  *
  *  Input (choose one of these):
  *
@@ -71,10 +76,12 @@
  *  SORT_INPUT_FB      seekable fastbuf stream
  *  SORT_INPUT_PIPE    non-seekable fastbuf stream
  *  SORT_INPUT_PRESORT custom presorter. Calls function
- *  int PREFIX_presort(struct fastbuf *dest, void *buf, size_t bufsize);
+ *  int PREFIX_presort(struct fastbuf *dest, void *buf, size_t bufsize)
  *                     to get successive batches of pre-sorted data.
  *                     The function is passed a page-aligned presorting buffer.
  *                     It returns 1 on success or 0 on EOF.
+ *  SORT_DELETE_INPUT  A C expression, if true, then the input files are deleted
+ *                     as soon as possible.
  *
  *  Output (chose one of these):
  *
@@ -88,7 +95,7 @@
  *
  *  The function generated:
  *
- *  <outfb> PREFIX_SORT(<in>, <out> [,<range>]), where:
+ *  <outfb> PREFIX_sort(<in>, <out> [,<range>]), where:
  *                     <in> = input file name/fastbuf or NULL
  *                     <out> = output file name/fastbuf or NULL
  *                     <range> = maximum integer value for the SORT_INT mode
@@ -121,6 +128,13 @@ typedef SORT_KEY P(key);
 #error Missing definition of sorting key.
 #endif
 
+#ifdef SORT_INT64
+typedef u64 P(hash_t);
+#define SORT_INT SORT_INT64
+#else
+typedef uns P(hash_t);
+#endif
+
 #ifdef SORT_INT
 static inline int P(compare) (P(key) *x, P(key) *y)
 {
@@ -132,7 +146,7 @@ static inline int P(compare) (P(key) *x, P(key) *y)
 }
 
 #ifndef SORT_HASH_BITS
-static inline int P(hash) (P(key) *x)
+static inline P(hash_t) P(hash) (P(key) *x)
 {
   return SORT_INT((*x));
 }
@@ -172,13 +186,21 @@ static inline void P(copy_data)(P(key) *key, struct fastbuf *in, struct fastbuf
 #endif
 }
 
-#if defined(SORT_VAR_KEY) || defined(SORT_VAR_DATA) || defined(SORT_UNIFY)
+#if defined(SORT_UNIFY) && !defined(SORT_VAR_DATA) && !defined(SORT_UNIFY_WORKSPACE)
+static inline void P(copy_merged)(P(key) **keys, struct fastbuf **data UNUSED, uns n, struct fastbuf *dest)
+{
+  P(write_merged)(dest, keys, NULL, n, NULL);
+}
+#endif
+
+#if defined(SORT_VAR_KEY) || defined(SORT_VAR_DATA) || defined(SORT_UNIFY_WORKSPACE)
 #include "lib/sorter/s-internal.h"
 #else
 #include "lib/sorter/s-fixint.h"
 #endif
 
 #include "lib/sorter/s-twoway.h"
+#include "lib/sorter/s-multiway.h"
 
 #if defined(SORT_HASH_BITS) || defined(SORT_INT)
 #include "lib/sorter/s-radix.h"
@@ -196,7 +218,7 @@ static struct fastbuf *P(sort)(
                               struct fastbuf *out
 #endif
 #ifdef SORT_INT
-                              , uns int_range
+                              , u64 int_range
 #endif
                               )
 {
@@ -204,7 +226,7 @@ static struct fastbuf *P(sort)(
   bzero(&ctx, sizeof(ctx));
 
 #ifdef SORT_INPUT_FILE
-  ctx.in_fb = bopen(in, O_RDONLY, sorter_stream_bufsize);
+  ctx.in_fb = bopen_file(in, O_RDONLY, &sorter_fb_params);
   ctx.in_size = bfilesize(ctx.in_fb);
 #elif defined(SORT_INPUT_FB)
   ctx.in_fb = in;
@@ -219,6 +241,10 @@ static struct fastbuf *P(sort)(
 #else
 #error No input given.
 #endif
+#ifdef SORT_DELETE_INPUT
+  if (SORT_DELETE_INPUT)
+    bconfig(ctx.in_fb, BCONFIG_IS_TEMP_FILE, 1);
+#endif
 
 #ifdef SORT_OUTPUT_FB
   ASSERT(!out);
@@ -235,21 +261,20 @@ static struct fastbuf *P(sort)(
   ctx.radix_split = P(radix_split);
 #elif defined(SORT_INT)
   ctx.hash_bits = 0;
-  while (ctx.hash_bits < 32 && (int_range >> ctx.hash_bits))
+  while (ctx.hash_bits < 64 && (int_range >> ctx.hash_bits))
     ctx.hash_bits++;
   ctx.radix_split = P(radix_split);
 #endif
 
   ctx.internal_sort = P(internal);
+  ctx.internal_estimate = P(internal_estimate);
   ctx.twoway_merge = P(twoway_merge);
+  ctx.multiway_merge = P(multiway_merge);
 
   sorter_run(&ctx);
 
 #ifdef SORT_OUTPUT_FILE
-  if (rename(ctx.out_fb->name, out) < 0)
-    die("Cannot rename %s to %s: %m", ctx.out_fb->name, out);
-  bconfig(ctx.out_fb, BCONFIG_IS_TEMP_FILE, 0);
-  bclose(ctx.out_fb);
+  bfix_tmp_file(ctx.out_fb, out);
   ctx.out_fb = NULL;
 #endif
   return ctx.out_fb;
@@ -263,8 +288,10 @@ static struct fastbuf *P(sort)(
 #undef SORT_VAR_KEY
 #undef SORT_VAR_DATA
 #undef SORT_INT
+#undef SORT_INT64
 #undef SORT_HASH_BITS
 #undef SORT_UNIFY
+#undef SORT_UNIFY_WORKSPACE
 #undef SORT_INPUT_FILE
 #undef SORT_INPUT_FB
 #undef SORT_INPUT_PRESORT
@@ -273,6 +300,7 @@ static struct fastbuf *P(sort)(
 #undef SORT_OUTPUT_THIS_FB
 #undef SORT_UNIQUE
 #undef SORT_ASSERT_UNIQUE
+#undef SORT_DELETE_INPUT
 #undef SWAP
 #undef LESS
 #undef P