1 /*
2 * CDDL HEADER START
3 *
4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License, Version 1.0 only
6 * (the "License"). You may not use this file except in compliance
7 * with the License.
8 *
9 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10 * or http://www.opensolaris.org/os/licensing.
11 * See the License for the specific language governing permissions
12 * and limitations under the License.
13 *
14 * When distributing Covered Code, include this CDDL HEADER in each
15 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16 * If applicable, add the following below this CDDL HEADER, with the
17 * fields enclosed by brackets "[]" replaced with your own identifying
18 * information: Portions Copyright [yyyy] [name of copyright owner]
19 *
20 * CDDL HEADER END
21 */
22 /*
23 * Copyright 1998-2003 Sun Microsystems, Inc. All rights reserved.
24 * Use is subject to license terms.
25 */
26
27 /*
28 * Overview of sort(1)
29 *
30 * sort(1) implements a robust sorting program, compliant with the POSIX
31 * specifications for sort, that is capable of handling large sorts and merges
32 * in single byte and multibyte locales. Like most sort(1) implementations,
33 * this implementation uses an internal algorithm for sorting subsets of the
34 * requested data set and an external algorithm for sorting the subsets into the
35 * final output. In the current implementation, the internal algorithm is a
36 * ternary radix quicksort, modified from the algorithm described in Bentley and
37 * Sedgewick [1], while the external algorithm is a priority-queue based
38 * heapsort, as outlined in Sedgewick [2].
39 *
40 * We use three major datatypes, defined in ./types.h: the line record,
41 * line_rec_t; the stream, stream_t; and the field definition, field_t.
42 * Because sort supports efficient code paths for each of the C, single-byte,
43 * and wide character/multibyte locales, each of these types contains unions
44 * and/or function pointers to describe appropriate properties or operations for
45 * each locale type.
46 *
47 * To utilize the radix quicksort algorithm with the potentially complex sort
48 * keys definable via the POSIX standard, we convert each line to a collatable
49 * string based on the key definition. This approach is somewhat different from
50 * historical implementations of sort(1), which have built a complex
51 * field-by-field comparison function. There are, of course, tradeoffs that
52 * accompany this decision, particularly when the duration of use of a given
53 * collated form is short. However, the maintenance costs of parallel
54 * conversion and collation functions are estimated to be high, and the
55 * performance costs of a shared set of functions were found to be excessive in
56 * prototype.
57 *
58 * [1] J. Bentley and R. Sedgewick, Fast Algorithms for Sorting and Searching
59 * Strings, in Eighth Annual ACM-SIAM Symposium on Discrete Algorithms,
60 * 1997 (SODA 1997),
61 * [2] R. Sedgewick, Algorithms in C, 3rd ed., vol. 1, Addison-Wesley, 1998.
62 */
63
64 #include "main.h"
65
66 static sort_t S;
67
68 int
main(int argc,char * argv[])69 main(int argc, char *argv[])
70 {
71 initialize_pre(&S);
72
73 if (options(&S, argc, argv))
74 return (2);
75
76 initialize_post(&S);
77
78 if (S.m_check_if_sorted_only)
79 check_if_sorted(&S);
80
81 if (!S.m_merge_only)
82 internal_sort(&S);
83
84 merge(&S);
85
86 return (0);
87 }
88