2 * UCW Library -- Universal Sorter: Governing Routines
4 * (c) 2007 Martin Mares <mj@ucw.cz>
6 * This software may be freely distributed and used according to the terms
7 * of the GNU Lesser General Public License.
11 #include "lib/fastbuf.h"
12 #include "lib/mempool.h"
13 #include "lib/sorter/common.h"
18 sorter_presort(struct sort_context *ctx, struct sort_bucket *in, struct sort_bucket *out, struct sort_bucket *out_only)
20 sorter_alloc_buf(ctx);
21 if (in->flags & SBF_CUSTOM_PRESORT)
23 struct fastbuf *f = sbuck_write(out);
25 return ctx->custom_presort(f, ctx->big_buf, ctx->big_buf_size); // FIXME: out_only optimization?
27 return ctx->internal_sort(ctx, in, out, out_only);
30 static inline struct sort_bucket *
31 sbuck_join_to(struct sort_bucket *b)
33 if (sorter_debug & SORT_DEBUG_NO_JOIN)
36 struct sort_bucket *out = (struct sort_bucket *) b->n.prev; // Such bucket is guaranteed to exist
37 return (out->flags & SBF_FINAL) ? out : NULL;
41 sorter_join(struct sort_bucket *b)
43 struct sort_bucket *join = (struct sort_bucket *) b->n.prev;
44 ASSERT(join->flags & SBF_FINAL);
47 if (!sbuck_has_file(join))
49 // The final bucket doesn't have any file associated yet, so replace
50 // it with the new bucket.
51 SORT_XTRACE(2, "Replaced final bucket");
52 b->flags |= SBF_FINAL;
57 SORT_TRACE("Copying to output file: %s", F_BSIZE(b));
58 struct fastbuf *src = sbuck_read(b);
59 struct fastbuf *dest = sbuck_write(join);
60 bbcopy(src, dest, ~0U);
66 sorter_twoway(struct sort_context *ctx, struct sort_bucket *b)
68 struct sort_bucket *ins[3] = { NULL }, *outs[3] = { NULL };
69 cnode *list_pos = b->n.prev;
70 struct sort_bucket *join = sbuck_join_to(b);
72 if (!(sorter_debug & SORT_DEBUG_NO_PRESORT) || (b->flags & SBF_CUSTOM_PRESORT))
74 SORT_XTRACE(2, "%s", ((b->flags & SBF_CUSTOM_PRESORT) ? "Custom presorting" : "Presorting"));
75 ins[0] = sbuck_new(ctx);
76 if (!sorter_presort(ctx, b, ins[0], join ? : ins[0]))
78 SORT_TRACE("Sorted in memory");
82 clist_insert_after(&ins[0]->n, list_pos);
87 ins[1] = sbuck_new(ctx);
89 while (sorter_presort(ctx, b, ins[i], ins[i]))
92 SORT_TRACE("Presorting pass (%d+%d runs, %s+%s)", ins[0]->runs, ins[1]->runs, F_BSIZE(ins[0]), F_BSIZE(ins[1]));
96 SORT_XTRACE(2, "Presorting disabled");
100 SORT_XTRACE(2, "Main sorting");
104 if (ins[0]->runs == 1 && ins[1]->runs == 1 && join)
106 // This is guaranteed to produce a single run, so join if possible
109 ctx->twoway_merge(ctx, ins, outs);
110 ASSERT(outs[0]->runs == 2);
112 SORT_TRACE("Mergesort pass %d (final run, %s)", pass, F_BSIZE(outs[0]));
117 outs[0] = sbuck_new(ctx);
118 outs[1] = sbuck_new(ctx);
120 ctx->twoway_merge(ctx, ins, outs);
121 SORT_TRACE("Mergesort pass %d (%d+%d runs, %s+%s)", pass, outs[0]->runs, outs[1]->runs, F_BSIZE(outs[0]), F_BSIZE(outs[1]));
124 memcpy(ins, outs, 3*sizeof(struct sort_bucket *));
125 } while (sbuck_have(ins[1]));
128 clist_insert_after(&ins[0]->n, list_pos);
132 sorter_run(struct sort_context *ctx)
134 ctx->pool = mp_new(4096);
135 clist_init(&ctx->bucket_list);
137 /* FIXME: Remember to test sorting of empty files */
139 // Create bucket containing the source
140 struct sort_bucket *bin = sbuck_new(ctx);
141 bin->flags = SBF_SOURCE | SBF_OPEN_READ;
142 if (ctx->custom_presort)
143 bin->flags |= SBF_CUSTOM_PRESORT;
145 bin->fb = ctx->in_fb;
147 bin->size = ctx->in_size;
148 bin->hash_bits = ctx->hash_bits;
149 clist_add_tail(&ctx->bucket_list, &bin->n);
150 SORT_XTRACE(2, "Input size: %s", (ctx->in_size == ~(u64)0 ? (byte*)"unknown" : F_BSIZE(bin)));
152 // Create bucket for the output
153 struct sort_bucket *bout = sbuck_new(ctx);
154 bout->flags = SBF_FINAL;
155 if (bout->fb = ctx->out_fb)
156 bout->flags |= SBF_OPEN_WRITE;
159 clist_add_head(&ctx->bucket_list, &bout->n);
161 struct sort_bucket *b;
162 while (bout = clist_head(&ctx->bucket_list), b = clist_next(&ctx->bucket_list, &bout->n))
166 else if (b->runs == 1)
169 sorter_twoway(ctx, b);
172 sorter_free_buf(ctx);
173 sbuck_write(bout); // Force empty bucket to a file
174 SORT_XTRACE(2, "Final size: %s", F_BSIZE(bout));
175 ctx->out_fb = sbuck_read(bout);