1/* Compare strings while treating digits characters numerically.
2   Copyright (C) 1997, 2002, 2005 Free Software Foundation, Inc.
3   This file is part of the libiberty library.
4   Contributed by Jean-François Bignolles <bignolle@ecoledoc.ibp.fr>, 1997.
5
6   Libiberty is free software; you can redistribute it and/or
7   modify it under the terms of the GNU Lesser General Public
8   License as published by the Free Software Foundation; either
9   version 2.1 of the License, or (at your option) any later version.
10
11   Libiberty is distributed in the hope that it will be useful,
12   but WITHOUT ANY WARRANTY; without even the implied warranty of
13   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
14   Lesser General Public License for more details.
15
16   You should have received a copy of the GNU Lesser General Public
17   License along with the GNU C Library; if not, write to the Free
18   Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
19   02110-1301 USA.  */
20
21#include <ctype.h>
22
23/*
24@deftypefun int strverscmp (const char *@var{s1}, const char *@var{s2})
25The @code{strverscmp} function compares the string @var{s1} against
26@var{s2}, considering them as holding indices/version numbers.  Return
27value follows the same conventions as found in the @code{strverscmp}
28function.  In fact, if @var{s1} and @var{s2} contain no digits,
29@code{strverscmp} behaves like @code{strcmp}.
30
31Basically, we compare strings normally (character by character), until
32we find a digit in each string - then we enter a special comparison
33mode, where each sequence of digits is taken as a whole.  If we reach the
34end of these two parts without noticing a difference, we return to the
35standard comparison mode.  There are two types of numeric parts:
36"integral" and "fractional" (those  begin with a '0'). The types
37of the numeric parts affect the way we sort them:
38
39@itemize @bullet
40@item
41integral/integral: we compare values as you would expect.
42
43@item
44fractional/integral: the fractional part is less than the integral one.
45Again, no surprise.
46
47@item
48fractional/fractional: the things become a bit more complex.
49If the common prefix contains only leading zeroes, the longest part is less
50than the other one; else the comparison behaves normally.
51@end itemize
52
53@smallexample
54strverscmp ("no digit", "no digit")
55    @result{} 0    // @r{same behavior as strcmp.}
56strverscmp ("item#99", "item#100")
57    @result{} <0   // @r{same prefix, but 99 < 100.}
58strverscmp ("alpha1", "alpha001")
59    @result{} >0   // @r{fractional part inferior to integral one.}
60strverscmp ("part1_f012", "part1_f01")
61    @result{} >0   // @r{two fractional parts.}
62strverscmp ("foo.009", "foo.0")
63    @result{} <0   // @r{idem, but with leading zeroes only.}
64@end smallexample
65
66This function is especially useful when dealing with filename sorting,
67because filenames frequently hold indices/version numbers.
68@end deftypefun
69
70*/
71
72/* states: S_N: normal, S_I: comparing integral part, S_F: comparing
73           fractional parts, S_Z: idem but with leading Zeroes only */
74#define  S_N    0x0
75#define  S_I    0x4
76#define  S_F    0x8
77#define  S_Z    0xC
78
79/* result_type: CMP: return diff; LEN: compare using len_diff/diff */
80#define  CMP    2
81#define  LEN    3
82
83
84/* Compare S1 and S2 as strings holding indices/version numbers,
85   returning less than, equal to or greater than zero if S1 is less than,
86   equal to or greater than S2 (for more info, see the Glibc texinfo doc).  */
87
88int
89strverscmp (const char *s1, const char *s2)
90{
91  const unsigned char *p1 = (const unsigned char *) s1;
92  const unsigned char *p2 = (const unsigned char *) s2;
93  unsigned char c1, c2;
94  int state;
95  int diff;
96
97  /* Symbol(s)    0       [1-9]   others  (padding)
98     Transition   (10) 0  (01) d  (00) x  (11) -   */
99  static const unsigned int next_state[] =
100    {
101      /* state    x    d    0    - */
102      /* S_N */  S_N, S_I, S_Z, S_N,
103      /* S_I */  S_N, S_I, S_I, S_I,
104      /* S_F */  S_N, S_F, S_F, S_F,
105      /* S_Z */  S_N, S_F, S_Z, S_Z
106    };
107
108  static const int result_type[] =
109    {
110      /* state   x/x  x/d  x/0  x/-  d/x  d/d  d/0  d/-
111                 0/x  0/d  0/0  0/-  -/x  -/d  -/0  -/- */
112
113      /* S_N */  CMP, CMP, CMP, CMP, CMP, LEN, CMP, CMP,
114                 CMP, CMP, CMP, CMP, CMP, CMP, CMP, CMP,
115      /* S_I */  CMP, -1,  -1,  CMP, +1,  LEN, LEN, CMP,
116                 +1,  LEN, LEN, CMP, CMP, CMP, CMP, CMP,
117      /* S_F */  CMP, CMP, CMP, CMP, CMP, LEN, CMP, CMP,
118                 CMP, CMP, CMP, CMP, CMP, CMP, CMP, CMP,
119      /* S_Z */  CMP, +1,  +1,  CMP, -1,  CMP, CMP, CMP,
120                 -1,  CMP, CMP, CMP
121    };
122
123  if (p1 == p2)
124    return 0;
125
126  c1 = *p1++;
127  c2 = *p2++;
128  /* Hint: '0' is a digit too.  */
129  state = S_N | ((c1 == '0') + (isdigit (c1) != 0));
130
131  while ((diff = c1 - c2) == 0 && c1 != '\0')
132    {
133      state = next_state[state];
134      c1 = *p1++;
135      c2 = *p2++;
136      state |= (c1 == '0') + (isdigit (c1) != 0);
137    }
138
139  state = result_type[state << 2 | (((c2 == '0') + (isdigit (c2) != 0)))];
140
141  switch (state)
142    {
143    case CMP:
144      return diff;
145
146    case LEN:
147      while (isdigit (*p1++))
148        if (!isdigit (*p2++))
149          return 1;
150
151      return isdigit (*p2) ? -1 : diff;
152
153    default:
154      return state;
155    }
156}
157