XML: Implemented a merger of element's contents.

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

diff --git a/lib/sorter/common.h b/lib/sorter/common.h
index db414dd08c6c033e99124ce00b373b68997a3881..ddd7ba6fadb407ba46803961dca92e1d61e07683 100644 (file)
--- a/lib/sorter/common.h
+++ b/lib/sorter/common.h
@@ -12,11 +12,14 @@
 
 #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, sorter_min_radix_bits, sorter_max_radix_bits;
-extern u64 sorter_bufsize;
-extern struct fb_params sorter_fb_params;
+/* Configuration variables */
+extern uns sorter_trace, sorter_trace_array, sorter_stream_bufsize;
+extern uns sorter_debug, sorter_min_radix_bits, sorter_max_radix_bits, sorter_add_radix_bits;
+extern uns sorter_min_multiway_bits, sorter_max_multiway_bits;
+extern uns sorter_threads;
+extern u64 sorter_bufsize, sorter_small_input;
+extern u64 sorter_thread_threshold, sorter_thread_chunk, sorter_radix_threshold;
+extern struct fb_params sorter_fb_params, sorter_small_fb_params;
 
 #define SORT_TRACE(x...) do { if (sorter_trace) msg(L_DEBUG, x); } while(0)
 #define SORT_XTRACE(level, x...) do { if (sorter_trace >= level) msg(L_DEBUG, x); } while(0)
@@ -26,6 +29,9 @@ enum sort_debug {
   SORT_DEBUG_NO_JOIN = 2,
   SORT_DEBUG_KEEP_BUCKETS = 4,
   SORT_DEBUG_NO_RADIX = 8,
+  SORT_DEBUG_NO_MULTIWAY = 16,
+  SORT_DEBUG_ASORT_NO_RADIX = 32,
+  SORT_DEBUG_ASORT_NO_THREADS = 64
 };
 
 struct sort_bucket;
@@ -35,6 +41,7 @@ struct sort_context {
   struct fastbuf *out_fb;
   uns hash_bits;
   u64 in_size;
+  struct fb_params *fb_params;
 
   struct mempool *pool;
   clist bucket_list;
@@ -54,6 +61,9 @@ struct sort_context {
   // Bucket arrays are NULL-terminated.
   void (*twoway_merge)(struct sort_context *ctx, struct sort_bucket **ins, struct sort_bucket **outs);
 
+  // Multi-way merge: merge an arbitrary number of source buckets to a single destination bucket.
+  void (*multiway_merge)(struct sort_context *ctx, struct sort_bucket **ins, struct sort_bucket *out);
+
   // Radix split according to hash function
   void (*radix_split)(struct sort_context *ctx, struct sort_bucket *in, struct sort_bucket **outs, uns bitpos, uns numbits);
 
@@ -108,6 +118,35 @@ 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);
-void format_size(byte *buf, u64 x);
+
+/* Contexts and helper functions for the array sorter */
+
+struct asort_context {
+  // Interface between generic code in array.c and functions generated by array.h
+  void *array;                         // Array to sort
+  void *buffer;                                // Auxiliary buffer (required when radix-sorting)
+  uns num_elts;                                // Number of elements in the array
+  uns elt_size;                                // Bytes per element
+  uns hash_bits;                       // Remaining bits of the hash function
+  uns radix_bits;                      // How many bits to process in a single radix-sort pass
+  void (*quicksort)(void *array_ptr, uns num_elts);
+  void (*quicksplit)(void *array_ptr, uns num_elts, int *leftp, int *rightp);
+  void (*radix_count)(void *src_ptr, uns num_elts, uns *cnt, uns shift);
+  void (*radix_split)(void *src_ptr, void *dest_ptr, uns num_elts, uns *ptrs, uns shift);
+
+  // Used internally by array.c
+  struct rs_work **rs_works;
+  struct work_queue *rs_work_queue;
+  struct eltpool *eltpool;
+
+  // Configured limits translated from bytes to elements
+  uns thread_threshold;
+  uns thread_chunk;
+  uns radix_threshold;
+};
+
+void asort_run(struct asort_context *ctx);
+void asort_start_threads(uns run);
+void asort_stop_threads(void);
 
 #endif