]>
Commit | Line | Data |
---|---|---|
a4057fc9 | 1 | #!/usr/bin/env python |
2 | ||
3 | ## @package thtml2doxy_clang | |
4 | # Translates THtml C++ comments to Doxygen using libclang as parser. | |
5 | # | |
6 | # This code relies on Python bindings for libclang: libclang's interface is pretty unstable, and | |
7 | # its Python bindings are unstable as well. | |
8 | # | |
9 | # AST (Abstract Source Tree) traversal is performed entirely using libclang used as a C++ parser, | |
10 | # instead of attempting to write a parser ourselves. | |
11 | # | |
12 | # This code (expecially AST traversal) was inspired by: | |
13 | # | |
14 | # - [Implementing a code generator with libclang](http://szelei.me/code-generator/) | |
15 | # (this refers to API calls used here) | |
16 | # - [Parsing C++ in Python with Clang](http://eli.thegreenplace.net/2011/07/03/parsing-c-in-python-with-clang) | |
17 | # (outdated, API calls described there do not work anymore, but useful to understand some basic | |
18 | # concepts) | |
19 | # | |
20 | # Usage: | |
21 | # | |
22 | # `thtml2doxy_clang file1 [file2 [file3...]]` | |
23 | # | |
24 | # @author Dario Berzano <dario.berzano@cern.ch> | |
25 | # @date 2014-12-05 | |
26 | ||
27 | ||
28 | import sys | |
29 | import os | |
30 | import re | |
31 | import clang.cindex | |
32 | ||
33 | ||
34 | ## Brain-dead color output for terminal. | |
35 | class Colt(str): | |
36 | ||
37 | def red(self): | |
38 | return self.color('\033[31m') | |
39 | ||
40 | def green(self): | |
41 | return self.color('\033[32m') | |
42 | ||
43 | def yellow(self): | |
44 | return self.color('\033[33m') | |
45 | ||
46 | def blue(self): | |
47 | return self.color('\033[34m') | |
48 | ||
49 | def magenta(self): | |
50 | return self.color('\033[35m') | |
51 | ||
52 | def cyan(self): | |
53 | return self.color('\033[36m') | |
54 | ||
55 | def color(self, c): | |
56 | return c + self + '\033[m' | |
57 | ||
58 | ||
59 | ## Traverse the AST recursively starting from the current cursor. | |
60 | # | |
61 | # @param cursor A Clang parser cursor | |
62 | # @param recursion Current recursion depth | |
63 | def traverse_ast(cursor, recursion=0): | |
64 | ||
65 | text = cursor.spelling or cursor.displayname | |
66 | kind = str(cursor.kind)[str(cursor.kind).index('.')+1:] | |
67 | ||
68 | indent = '' | |
69 | for i in range(0, recursion): | |
70 | indent = indent + ' ' | |
71 | ||
72 | if cursor.kind == clang.cindex.CursorKind.CXX_METHOD: | |
73 | ||
74 | # cursor ran into a C++ method | |
75 | print "%s%s(%s)" % (indent, Colt(kind).magenta(), Colt(text).blue()) | |
76 | ||
77 | # we are looking for the following structure: method -> compound statement -> comment, i.e. we | |
78 | # need to extract the first comment in the compound statement composing the method | |
79 | ||
80 | in_compound_stmt = False | |
81 | expect_comment = False | |
9a13b5a2 | 82 | last_comment_line = -1 |
a4057fc9 | 83 | |
84 | for token in cursor.get_tokens(): | |
85 | ||
86 | if token.cursor.kind == clang.cindex.CursorKind.COMPOUND_STMT: | |
87 | if not in_compound_stmt: | |
88 | in_compound_stmt = True | |
89 | expect_comment = True | |
9a13b5a2 | 90 | last_comment_line = -1 |
a4057fc9 | 91 | else: |
92 | if in_compound_stmt: | |
93 | in_compound_stmt = False | |
94 | break | |
95 | ||
96 | # tkind = str(token.kind)[str(token.kind).index('.')+1:] | |
97 | # ckind = str(token.cursor.kind)[str(token.cursor.kind).index('.')+1:] | |
98 | ||
99 | if in_compound_stmt: | |
100 | ||
101 | if expect_comment: | |
102 | ||
9a13b5a2 | 103 | extent = token.extent |
104 | line_start = extent.start.line | |
105 | line_end = extent.end.line | |
106 | ||
107 | if token.kind == clang.cindex.TokenKind.PUNCTUATION and token.spelling == '{': | |
108 | pass | |
109 | ||
110 | elif token.kind == clang.cindex.TokenKind.COMMENT and (last_comment_line == -1 or line_start == last_comment_line+1): | |
a4057fc9 | 111 | #print Colt("%s %s:%s = %s" % (indent, ckind, tkind, token.spelling)).green() |
9a13b5a2 | 112 | last_comment_line = line_end |
4e465d49 | 113 | new_comment = refactor_comment(token.spelling) |
114 | ||
115 | for comment_line in new_comment: | |
116 | print Colt("%s [%d-%d]" % (indent, line_start, line_end)).green(), | |
117 | print Colt(comment_line).cyan() | |
a4057fc9 | 118 | |
a4057fc9 | 119 | else: |
120 | expect_comment = False | |
121 | ||
122 | # else: | |
123 | # print Colt("%s %s:%s = %s" % (indent, ckind, tkind, token.spelling)).yellow() | |
124 | ||
125 | ||
126 | else: | |
127 | ||
128 | print "%s%s(%s)" % (indent, kind, text) | |
129 | ||
4e465d49 | 130 | for child_cursor in cursor.get_children(): |
131 | traverse_ast(child_cursor, recursion+1) | |
a4057fc9 | 132 | |
4e465d49 | 133 | ## Remove garbage from comments and convert special tags from THtml to Doxygen. |
134 | # | |
135 | # @param comment The original comment | |
136 | def refactor_comment(comment): | |
a4057fc9 | 137 | |
4e465d49 | 138 | resingle = r'^/{2,}\s*(.*?)\s*(/{2,})?\s*$' |
139 | remulti_first = r'^/\*\s*(.*?)\s*\*?\s*$' | |
140 | remulti_last = r'^\s*(.*?)\s*\*/$' | |
a4057fc9 | 141 | |
4e465d49 | 142 | new_comment = comment.split('\n') |
a4057fc9 | 143 | |
4e465d49 | 144 | if len(new_comment) == 1: |
145 | msingle = re.search(resingle, comment) | |
146 | if msingle: | |
147 | new_comment[0] = msingle.group(1) | |
a4057fc9 | 148 | |
4e465d49 | 149 | else: |
a4057fc9 | 150 | |
4e465d49 | 151 | for i in range(0, len(new_comment)): |
152 | if i == 0: | |
153 | mmulti = re.search(remulti_first, new_comment[i]) | |
154 | if mmulti: | |
155 | new_comment[i] = mmulti.group(1) | |
156 | elif i == len(new_comment)-1: | |
157 | mmulti = re.search(remulti_last, new_comment[i]) | |
158 | if mmulti: | |
159 | new_comment[i] = mmulti.group(1) | |
160 | else: | |
161 | new_comment[i] = new_comment[i].strip() | |
162 | ||
163 | return new_comment | |
a4057fc9 | 164 | |
165 | ||
166 | ## The main function. | |
167 | # | |
168 | # **Note:** this program only has this function. | |
169 | def main(argv): | |
170 | ||
171 | # Attempt to load libclang from a list of known locations | |
172 | libclang_locations = [ | |
173 | '/usr/lib/llvm-3.5/lib/libclang.so.1', | |
174 | '/usr/lib/libclang.so', | |
175 | '/Library/Developer/CommandLineTools/usr/lib/libclang.dylib' | |
176 | ] | |
177 | libclang_found = False | |
178 | ||
179 | for lib in libclang_locations: | |
180 | if os.path.isfile(lib): | |
181 | clang.cindex.Config.set_library_file(lib) | |
182 | libclang_found = True | |
183 | break | |
184 | ||
185 | if not libclang_found: | |
186 | print Colt('[Error] Cannot find libclang, aborting').red() | |
187 | return 1 | |
188 | ||
189 | # Loop over all files | |
190 | for fn in argv[1:]: | |
191 | ||
192 | index = clang.cindex.Index.create() | |
193 | translation_unit = index.parse(fn, args=['-x', 'c++']) | |
194 | traverse_ast( translation_unit.cursor ) | |
195 | ||
196 | return 0 | |
197 | ||
198 | ||
199 | if __name__ == '__main__': | |
200 | sys.exit( main( sys.argv ) ) |