1# -*- coding: utf-8 -*-
2"""
3    pygments.lexers.hexdump
4    ~~~~~~~~~~~~~~~~~~~~~~~
5
6    Lexers for hexadecimal dumps.
7
8    :copyright: Copyright 2006-2020 by the Pygments team, see AUTHORS.
9    :license: BSD, see LICENSE for details.
10"""
11
12from pygments.lexer import RegexLexer, bygroups, include
13from pygments.token import Text, Name, Number, String, Punctuation
14
15__all__ = ['HexdumpLexer']
16
17
18class HexdumpLexer(RegexLexer):
19    """
20    For typical hex dump output formats by the UNIX and GNU/Linux tools ``hexdump``,
21    ``hd``, ``hexcat``, ``od`` and ``xxd``, and the DOS tool ``DEBUG``. For example:
22
23    .. sourcecode:: hexdump
24
25        00000000  7f 45 4c 46 02 01 01 00  00 00 00 00 00 00 00 00  |.ELF............|
26        00000010  02 00 3e 00 01 00 00 00  c5 48 40 00 00 00 00 00  |..>......H@.....|
27
28    The specific supported formats are the outputs of:
29
30    * ``hexdump FILE``
31    * ``hexdump -C FILE`` -- the `canonical` format used in the example.
32    * ``hd FILE`` -- same as ``hexdump -C FILE``.
33    * ``hexcat FILE``
34    * ``od -t x1z FILE``
35    * ``xxd FILE``
36    * ``DEBUG.EXE FILE.COM`` and entering ``d`` to the prompt.
37
38    .. versionadded:: 2.1
39    """
40    name = 'Hexdump'
41    aliases = ['hexdump']
42
43    hd = r'[0-9A-Ha-h]'
44
45    tokens = {
46        'root': [
47            (r'\n', Text),
48            include('offset'),
49            (r'('+hd+r'{2})(\-)('+hd+r'{2})',
50             bygroups(Number.Hex, Punctuation, Number.Hex)),
51            (hd+r'{2}', Number.Hex),
52            (r'(\s{2,3})(\>)(.{16})(\<)$',
53             bygroups(Text, Punctuation, String, Punctuation), 'bracket-strings'),
54            (r'(\s{2,3})(\|)(.{16})(\|)$',
55             bygroups(Text, Punctuation, String, Punctuation), 'piped-strings'),
56            (r'(\s{2,3})(\>)(.{1,15})(\<)$',
57             bygroups(Text, Punctuation, String, Punctuation)),
58            (r'(\s{2,3})(\|)(.{1,15})(\|)$',
59             bygroups(Text, Punctuation, String, Punctuation)),
60            (r'(\s{2,3})(.{1,15})$', bygroups(Text, String)),
61            (r'(\s{2,3})(.{16}|.{20})$', bygroups(Text, String), 'nonpiped-strings'),
62            (r'\s', Text),
63            (r'^\*', Punctuation),
64        ],
65        'offset': [
66            (r'^('+hd+'+)(:)', bygroups(Name.Label, Punctuation), 'offset-mode'),
67            (r'^'+hd+'+', Name.Label),
68        ],
69        'offset-mode': [
70            (r'\s', Text, '#pop'),
71            (hd+'+', Name.Label),
72            (r':', Punctuation)
73        ],
74        'piped-strings': [
75            (r'\n', Text),
76            include('offset'),
77            (hd+r'{2}', Number.Hex),
78            (r'(\s{2,3})(\|)(.{1,16})(\|)$',
79             bygroups(Text, Punctuation, String, Punctuation)),
80            (r'\s', Text),
81            (r'^\*', Punctuation),
82        ],
83        'bracket-strings': [
84            (r'\n', Text),
85            include('offset'),
86            (hd+r'{2}', Number.Hex),
87            (r'(\s{2,3})(\>)(.{1,16})(\<)$',
88             bygroups(Text, Punctuation, String, Punctuation)),
89            (r'\s', Text),
90            (r'^\*', Punctuation),
91        ],
92        'nonpiped-strings': [
93            (r'\n', Text),
94            include('offset'),
95            (r'('+hd+r'{2})(\-)('+hd+r'{2})',
96             bygroups(Number.Hex, Punctuation, Number.Hex)),
97            (hd+r'{2}', Number.Hex),
98            (r'(\s{19,})(.{1,20}?)$', bygroups(Text, String)),
99            (r'(\s{2,3})(.{1,20})$', bygroups(Text, String)),
100            (r'\s', Text),
101            (r'^\*', Punctuation),
102        ],
103    }
104