X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;f=lib%2Fsorter%2Fcommon.h;h=5e71be66e0f18cfb316660237f411ee01b26bb9a;hb=557dc31fa097cfe3b8fdbc087e27da6db7317819;hp=76d3e8c893ae243686216a65dd94486b9bcd2d64;hpb=116954d49373438c669eadd764e03ee8ac4211bb;p=libucw.git

diff --git a/lib/sorter/common.h b/lib/sorter/common.h
index 76d3e8c8..5e71be66 100644
--- a/lib/sorter/common.h
+++ b/lib/sorter/common.h
@@ -1,5 +1,5 @@
 /*
- *	UCW Library -- Universal Sorter
+ *	UCW Library -- Universal Sorter: Common Declarations
  *
  *	(c) 2007 Martin Mares <mj@ucw.cz>
  *
@@ -10,7 +10,91 @@
 #ifndef _UCW_SORTER_COMMON_H
 #define _UCW_SORTER_COMMON_H
 
+#include "lib/clists.h"
+
 /* Configuration, some of the variables are used by the old sorter, too. */
 extern uns sorter_trace, sorter_presort_bufsize, sorter_stream_bufsize;
+extern uns sorter_debug;
+extern u64 sorter_bufsize;
+
+#define SORT_TRACE(x...) do { if (sorter_trace) log(L_DEBUG, x); } while(0)
+#define SORT_XTRACE(level, x...) do { if (sorter_trace >= level) log(L_DEBUG, x); } while(0)
+
+enum sort_debug {
+  SORT_DEBUG_NO_PRESORT = 1,
+  SORT_DEBUG_NO_JOIN = 2,
+  SORT_DEBUG_KEEP_BUCKETS = 4,
+};
+
+struct sort_bucket;
+
+struct sort_context {
+  struct fastbuf *in_fb;
+  struct fastbuf *out_fb;
+  uns hash_bits;
+  u64 in_size;
+
+  struct mempool *pool;
+  clist bucket_list;
+  void *big_buf, *big_buf_half;
+  size_t big_buf_size, big_buf_half_size;
+
+  int (*custom_presort)(struct fastbuf *dest, void *buf, size_t bufsize);
+  // Take as much as possible from the source bucket, sort it in memory and dump to destination bucket.
+  // Return 1 if there is more data available in the source bucket.
+  int (*internal_sort)(struct sort_context *ctx, struct sort_bucket *in, struct sort_bucket *out, struct sort_bucket *out_only);
+  // Two-way split/merge: merge up to 2 source buckets to up to 2 destination buckets.
+  // Bucket arrays are NULL-terminated.
+  void (*twoway_merge)(struct sort_context *ctx, struct sort_bucket **ins, struct sort_bucket **outs);
+
+  // State variables of internal_sort
+  void *key_buf;
+  int more_keys;
+};
+
+void sorter_run(struct sort_context *ctx);
+
+/* Buffers */
+
+void *sorter_alloc(struct sort_context *ctx, uns size);
+void sorter_alloc_buf(struct sort_context *ctx);
+void sorter_free_buf(struct sort_context *ctx);
+
+/* Buckets */
+
+struct sort_bucket {
+  cnode n;
+  struct sort_context *ctx;
+  uns flags;
+  struct fastbuf *fb;
+  byte *filename;
+  u64 size;				// Size in bytes (not valid when writing)
+  uns runs;				// Number of runs, 0 if not sorted
+  uns hash_bits;			// Remaining bits of the hash function
+  byte *ident;				// Identifier used in debug messages
+};
+
+enum sort_bucket_flags {
+  SBF_FINAL = 1,			// This bucket corresponds to the final output file (always 1 run)
+  SBF_SOURCE = 2,			// Contains the source file (always 0 runs)
+  SBF_CUSTOM_PRESORT = 4,		// Contains source to read via custom presorter
+  SBF_OPEN_WRITE = 256,			// We are currently writing to the fastbuf
+  SBF_OPEN_READ = 512,			// We are reading from the fastbuf
+  SBF_DESTROYED = 1024,			// Already done with, no further references allowed
+  SBF_SWAPPED_OUT = 2048,		// Swapped out to a named file
+};
+
+struct sort_bucket *sbuck_new(struct sort_context *ctx);
+void sbuck_drop(struct sort_bucket *b);
+int sbuck_have(struct sort_bucket *b);
+int sbuck_has_file(struct sort_bucket *b);
+sh_off_t sbuck_size(struct sort_bucket *b);
+struct fastbuf *sbuck_read(struct sort_bucket *b);
+struct fastbuf *sbuck_write(struct sort_bucket *b);
+void sbuck_swap_out(struct sort_bucket *b);
+
+#define F_SIZE(x) ({ byte buf[16]; format_size(buf, x); buf; })
+#define F_BSIZE(b) F_SIZE(sbuck_size(b))
+void format_size(byte *buf, u64 x);
 
 #endif