1 | """
|
2 | format_strings.py
|
3 |
|
4 | Parse a printf format string so we can compile it to function calls.
|
5 | """
|
6 | from __future__ import print_function
|
7 |
|
8 | import re
|
9 |
|
10 |
|
11 | def DecodeMyPyString(s):
|
12 | # type: (str) -> str
|
13 | """Workaround for MyPy's weird escaping.
|
14 |
|
15 | Used below and in cppgen_pass.py.
|
16 | """
|
17 | byte_string = bytes(s, 'utf-8')
|
18 |
|
19 | # In Python 3
|
20 | # >>> b'\\t'.decode('unicode_escape')
|
21 | # '\t'
|
22 |
|
23 | raw_string = byte_string.decode('unicode_escape')
|
24 | return raw_string
|
25 |
|
26 |
|
27 | class LiteralPart:
|
28 |
|
29 | def __init__(self, s):
|
30 | self.s = s
|
31 | self.strlen = len(s)
|
32 |
|
33 | def __repr__(self):
|
34 | return '(Literal %r)' % (self.s)
|
35 |
|
36 |
|
37 | class SubstPart:
|
38 |
|
39 | def __init__(self, width, char_code, arg_num):
|
40 | self.width = width
|
41 | self.char_code = char_code
|
42 | self.arg_num = arg_num
|
43 |
|
44 | def __repr__(self):
|
45 | return '(Subst %r %s %d)' % (self.width, self.char_code, self.arg_num)
|
46 |
|
47 |
|
48 | PAT = re.compile(
|
49 | '''
|
50 | ([^%]*)
|
51 | (?:
|
52 | %([0-9]*)(.) # optional number, and then character code
|
53 | )?
|
54 | ''', re.VERBOSE)
|
55 |
|
56 |
|
57 | def Parse(fmt):
|
58 |
|
59 | arg_num = 0
|
60 | parts = []
|
61 | for m in PAT.finditer(fmt):
|
62 | lit = m.group(1)
|
63 | width = m.group(2)
|
64 | char_code = m.group(3)
|
65 |
|
66 | if lit:
|
67 | parts.append(LiteralPart(lit))
|
68 | if char_code:
|
69 | if char_code == '%':
|
70 | part = LiteralPart('%')
|
71 | else:
|
72 | part = SubstPart(width, char_code, arg_num)
|
73 | arg_num += 1
|
74 | parts.append(part)
|
75 |
|
76 | #print('end =', m.end(0))
|
77 |
|
78 | return parts
|
79 |
|
80 |
|
81 | # Note: This would be a lot easier in Oil!
|
82 | # TODO: Should there be a char type?
|
83 | """
|
84 | enum format_part {
|
85 | case Literal(s BigStr)
|
86 | case Subst(char_code BigStr, arg_num Int)
|
87 | }
|
88 |
|
89 | let PAT = ///
|
90 | < ~['%']* : lit > # anything except %
|
91 | < '%' dot : subst > # % and then any char
|
92 | ///
|
93 |
|
94 | func Parse(fmt BigStr) {
|
95 | var arg_num = 0
|
96 | let parts = []
|
97 |
|
98 | for (m in find(fmt, PAT)) {
|
99 | if (m.lit) {
|
100 | do parts.append(format_part.Literal(m.lit))
|
101 | }
|
102 | if (m.subst) {
|
103 | if (char_code == '%') {
|
104 | part = format_part.Literal('%')
|
105 | } else {
|
106 | part = format_part.Subst(char_code, arg_num)
|
107 | }
|
108 | do parts.append(part)
|
109 | set arg_num += 1
|
110 | }
|
111 | }
|
112 | return parts
|
113 | }
|
114 | """
|