1*101e15b5SRichard Lowe /*
2*101e15b5SRichard Lowe * CDDL HEADER START
3*101e15b5SRichard Lowe *
4*101e15b5SRichard Lowe * The contents of this file are subject to the terms of the
5*101e15b5SRichard Lowe * Common Development and Distribution License, Version 1.0 only
6*101e15b5SRichard Lowe * (the "License"). You may not use this file except in compliance
7*101e15b5SRichard Lowe * with the License.
8*101e15b5SRichard Lowe *
9*101e15b5SRichard Lowe * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10*101e15b5SRichard Lowe * or http://www.opensolaris.org/os/licensing.
11*101e15b5SRichard Lowe * See the License for the specific language governing permissions
12*101e15b5SRichard Lowe * and limitations under the License.
13*101e15b5SRichard Lowe *
14*101e15b5SRichard Lowe * When distributing Covered Code, include this CDDL HEADER in each
15*101e15b5SRichard Lowe * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16*101e15b5SRichard Lowe * If applicable, add the following below this CDDL HEADER, with the
17*101e15b5SRichard Lowe * fields enclosed by brackets "[]" replaced with your own identifying
18*101e15b5SRichard Lowe * information: Portions Copyright [yyyy] [name of copyright owner]
19*101e15b5SRichard Lowe *
20*101e15b5SRichard Lowe * CDDL HEADER END
21*101e15b5SRichard Lowe */
22*101e15b5SRichard Lowe /*
23*101e15b5SRichard Lowe * Copyright 1998-2003 Sun Microsystems, Inc. All rights reserved.
24*101e15b5SRichard Lowe * Use is subject to license terms.
25*101e15b5SRichard Lowe */
26*101e15b5SRichard Lowe
27*101e15b5SRichard Lowe /*
28*101e15b5SRichard Lowe * Overview of sort(1)
29*101e15b5SRichard Lowe *
30*101e15b5SRichard Lowe * sort(1) implements a robust sorting program, compliant with the POSIX
31*101e15b5SRichard Lowe * specifications for sort, that is capable of handling large sorts and merges
32*101e15b5SRichard Lowe * in single byte and multibyte locales. Like most sort(1) implementations,
33*101e15b5SRichard Lowe * this implementation uses an internal algorithm for sorting subsets of the
34*101e15b5SRichard Lowe * requested data set and an external algorithm for sorting the subsets into the
35*101e15b5SRichard Lowe * final output. In the current implementation, the internal algorithm is a
36*101e15b5SRichard Lowe * ternary radix quicksort, modified from the algorithm described in Bentley and
37*101e15b5SRichard Lowe * Sedgewick [1], while the external algorithm is a priority-queue based
38*101e15b5SRichard Lowe * heapsort, as outlined in Sedgewick [2].
39*101e15b5SRichard Lowe *
40*101e15b5SRichard Lowe * We use three major datatypes, defined in ./types.h: the line record,
41*101e15b5SRichard Lowe * line_rec_t; the stream, stream_t; and the field definition, field_t.
42*101e15b5SRichard Lowe * Because sort supports efficient code paths for each of the C, single-byte,
43*101e15b5SRichard Lowe * and wide character/multibyte locales, each of these types contains unions
44*101e15b5SRichard Lowe * and/or function pointers to describe appropriate properties or operations for
45*101e15b5SRichard Lowe * each locale type.
46*101e15b5SRichard Lowe *
47*101e15b5SRichard Lowe * To utilize the radix quicksort algorithm with the potentially complex sort
48*101e15b5SRichard Lowe * keys definable via the POSIX standard, we convert each line to a collatable
49*101e15b5SRichard Lowe * string based on the key definition. This approach is somewhat different from
50*101e15b5SRichard Lowe * historical implementations of sort(1), which have built a complex
51*101e15b5SRichard Lowe * field-by-field comparison function. There are, of course, tradeoffs that
52*101e15b5SRichard Lowe * accompany this decision, particularly when the duration of use of a given
53*101e15b5SRichard Lowe * collated form is short. However, the maintenance costs of parallel
54*101e15b5SRichard Lowe * conversion and collation functions are estimated to be high, and the
55*101e15b5SRichard Lowe * performance costs of a shared set of functions were found to be excessive in
56*101e15b5SRichard Lowe * prototype.
57*101e15b5SRichard Lowe *
58*101e15b5SRichard Lowe * [1] J. Bentley and R. Sedgewick, Fast Algorithms for Sorting and Searching
59*101e15b5SRichard Lowe * Strings, in Eighth Annual ACM-SIAM Symposium on Discrete Algorithms,
60*101e15b5SRichard Lowe * 1997 (SODA 1997),
61*101e15b5SRichard Lowe * [2] R. Sedgewick, Algorithms in C, 3rd ed., vol. 1, Addison-Wesley, 1998.
62*101e15b5SRichard Lowe */
63*101e15b5SRichard Lowe
64*101e15b5SRichard Lowe #include "main.h"
65*101e15b5SRichard Lowe
66*101e15b5SRichard Lowe static sort_t S;
67*101e15b5SRichard Lowe
68*101e15b5SRichard Lowe int
main(int argc,char * argv[])69*101e15b5SRichard Lowe main(int argc, char *argv[])
70*101e15b5SRichard Lowe {
71*101e15b5SRichard Lowe initialize_pre(&S);
72*101e15b5SRichard Lowe
73*101e15b5SRichard Lowe if (options(&S, argc, argv))
74*101e15b5SRichard Lowe return (2);
75*101e15b5SRichard Lowe
76*101e15b5SRichard Lowe initialize_post(&S);
77*101e15b5SRichard Lowe
78*101e15b5SRichard Lowe if (S.m_check_if_sorted_only)
79*101e15b5SRichard Lowe check_if_sorted(&S);
80*101e15b5SRichard Lowe
81*101e15b5SRichard Lowe if (!S.m_merge_only)
82*101e15b5SRichard Lowe internal_sort(&S);
83*101e15b5SRichard Lowe
84*101e15b5SRichard Lowe merge(&S);
85*101e15b5SRichard Lowe
86*101e15b5SRichard Lowe return (0);
87*101e15b5SRichard Lowe }
88