Sun, 17 Feb 2013 19:07:15 +0100
Updated to Pygments 1.6.
0
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
1 | # -*- coding: utf-8 -*- |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
2 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
3 | pygments.scanner |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
4 | ~~~~~~~~~~~~~~~~ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
5 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
6 | This library implements a regex based scanner. Some languages |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
7 | like Pascal are easy to parse but have some keywords that |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
8 | depend on the context. Because of this it's impossible to lex |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
9 | that just by using a regular expression lexer like the |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
10 | `RegexLexer`. |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
11 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
12 | Have a look at the `DelphiLexer` to get an idea of how to use |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
13 | this scanner. |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
14 | |
2426
da76c71624de
Updated to Pygments 1.6.
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
1705
diff
changeset
|
15 | :copyright: Copyright 2006-2013 by the Pygments team, see AUTHORS. |
0
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
16 | :license: BSD, see LICENSE for details. |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
17 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
18 | import re |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
19 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
20 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
21 | class EndOfText(RuntimeError): |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
22 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
23 | Raise if end of text is reached and the user |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
24 | tried to call a match function. |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
25 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
26 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
27 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
28 | class Scanner(object): |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
29 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
30 | Simple scanner |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
31 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
32 | All method patterns are regular expression strings (not |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
33 | compiled expressions!) |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
34 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
35 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
36 | def __init__(self, text, flags=0): |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
37 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
38 | :param text: The text which should be scanned |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
39 | :param flags: default regular expression flags |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
40 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
41 | self.data = text |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
42 | self.data_length = len(text) |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
43 | self.start_pos = 0 |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
44 | self.pos = 0 |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
45 | self.flags = flags |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
46 | self.last = None |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
47 | self.match = None |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
48 | self._re_cache = {} |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
49 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
50 | def eos(self): |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
51 | """`True` if the scanner reached the end of text.""" |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
52 | return self.pos >= self.data_length |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
53 | eos = property(eos, eos.__doc__) |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
54 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
55 | def check(self, pattern): |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
56 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
57 | Apply `pattern` on the current position and return |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
58 | the match object. (Doesn't touch pos). Use this for |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
59 | lookahead. |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
60 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
61 | if self.eos: |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
62 | raise EndOfText() |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
63 | if pattern not in self._re_cache: |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
64 | self._re_cache[pattern] = re.compile(pattern, self.flags) |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
65 | return self._re_cache[pattern].match(self.data, self.pos) |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
66 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
67 | def test(self, pattern): |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
68 | """Apply a pattern on the current position and check |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
69 | if it patches. Doesn't touch pos.""" |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
70 | return self.check(pattern) is not None |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
71 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
72 | def scan(self, pattern): |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
73 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
74 | Scan the text for the given pattern and update pos/match |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
75 | and related fields. The return value is a boolen that |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
76 | indicates if the pattern matched. The matched value is |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
77 | stored on the instance as ``match``, the last value is |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
78 | stored as ``last``. ``start_pos`` is the position of the |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
79 | pointer before the pattern was matched, ``pos`` is the |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
80 | end position. |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
81 | """ |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
82 | if self.eos: |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
83 | raise EndOfText() |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
84 | if pattern not in self._re_cache: |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
85 | self._re_cache[pattern] = re.compile(pattern, self.flags) |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
86 | self.last = self.match |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
87 | m = self._re_cache[pattern].match(self.data, self.pos) |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
88 | if m is None: |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
89 | return False |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
90 | self.start_pos = m.start() |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
91 | self.pos = m.end() |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
92 | self.match = m.group() |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
93 | return True |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
94 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
95 | def get_char(self): |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
96 | """Scan exactly one char.""" |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
97 | self.scan('.') |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
98 | |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
99 | def __repr__(self): |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
100 | return '<%s %d/%d>' % ( |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
101 | self.__class__.__name__, |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
102 | self.pos, |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
103 | self.data_length |
de9c2efb9d02
Started porting eric4 to Python3
Detlev Offenbach <detlev@die-offenbachs.de>
parents:
diff
changeset
|
104 | ) |