Show More
@@ -1,161 +1,161 b'' | |||||
1 | #!/usr/bin/env python |
|
1 | #!/usr/bin/env python | |
2 | # |
|
2 | # | |
3 | # hggettext - carefully extract docstrings for Mercurial |
|
3 | # hggettext - carefully extract docstrings for Mercurial | |
4 | # |
|
4 | # | |
5 | # Copyright 2009 Matt Mackall <mpm@selenic.com> and others |
|
5 | # Copyright 2009 Matt Mackall <mpm@selenic.com> and others | |
6 | # |
|
6 | # | |
7 | # This software may be used and distributed according to the terms of the |
|
7 | # This software may be used and distributed according to the terms of the | |
8 | # GNU General Public License version 2 or any later version. |
|
8 | # GNU General Public License version 2 or any later version. | |
9 |
|
9 | |||
10 | # The normalize function is taken from pygettext which is distributed |
|
10 | # The normalize function is taken from pygettext which is distributed | |
11 | # with Python under the Python License, which is GPL compatible. |
|
11 | # with Python under the Python License, which is GPL compatible. | |
12 |
|
12 | |||
13 | """Extract docstrings from Mercurial commands. |
|
13 | """Extract docstrings from Mercurial commands. | |
14 |
|
14 | |||
15 | Compared to pygettext, this script knows about the cmdtable and table |
|
15 | Compared to pygettext, this script knows about the cmdtable and table | |
16 | dictionaries used by Mercurial, and will only extract docstrings from |
|
16 | dictionaries used by Mercurial, and will only extract docstrings from | |
17 | functions mentioned therein. |
|
17 | functions mentioned therein. | |
18 |
|
18 | |||
19 | Use xgettext like normal to extract strings marked as translatable and |
|
19 | Use xgettext like normal to extract strings marked as translatable and | |
20 | join the message cataloges to get the final catalog. |
|
20 | join the message cataloges to get the final catalog. | |
21 | """ |
|
21 | """ | |
22 |
|
22 | |||
23 | from __future__ import absolute_import, print_function |
|
23 | from __future__ import absolute_import, print_function | |
24 |
|
24 | |||
25 | import inspect |
|
25 | import inspect | |
26 | import os |
|
26 | import os | |
27 | import re |
|
27 | import re | |
28 | import sys |
|
28 | import sys | |
29 |
|
29 | |||
30 |
|
30 | |||
31 | def escape(s): |
|
31 | def escape(s): | |
32 | # The order is important, the backslash must be escaped first |
|
32 | # The order is important, the backslash must be escaped first | |
33 | # since the other replacements introduce new backslashes |
|
33 | # since the other replacements introduce new backslashes | |
34 | # themselves. |
|
34 | # themselves. | |
35 | s = s.replace('\\', '\\\\') |
|
35 | s = s.replace('\\', '\\\\') | |
36 | s = s.replace('\n', '\\n') |
|
36 | s = s.replace('\n', '\\n') | |
37 | s = s.replace('\r', '\\r') |
|
37 | s = s.replace('\r', '\\r') | |
38 | s = s.replace('\t', '\\t') |
|
38 | s = s.replace('\t', '\\t') | |
39 | s = s.replace('"', '\\"') |
|
39 | s = s.replace('"', '\\"') | |
40 | return s |
|
40 | return s | |
41 |
|
41 | |||
42 |
|
42 | |||
43 | def normalize(s): |
|
43 | def normalize(s): | |
44 | # This converts the various Python string types into a format that |
|
44 | # This converts the various Python string types into a format that | |
45 | # is appropriate for .po files, namely much closer to C style. |
|
45 | # is appropriate for .po files, namely much closer to C style. | |
46 | lines = s.split('\n') |
|
46 | lines = s.split('\n') | |
47 | if len(lines) == 1: |
|
47 | if len(lines) == 1: | |
48 | s = '"' + escape(s) + '"' |
|
48 | s = '"' + escape(s) + '"' | |
49 | else: |
|
49 | else: | |
50 | if not lines[-1]: |
|
50 | if not lines[-1]: | |
51 | del lines[-1] |
|
51 | del lines[-1] | |
52 | lines[-1] = lines[-1] + '\n' |
|
52 | lines[-1] = lines[-1] + '\n' | |
53 | lines = map(escape, lines) |
|
53 | lines = map(escape, lines) | |
54 | lineterm = '\\n"\n"' |
|
54 | lineterm = '\\n"\n"' | |
55 | s = '""\n"' + lineterm.join(lines) + '"' |
|
55 | s = '""\n"' + lineterm.join(lines) + '"' | |
56 | return s |
|
56 | return s | |
57 |
|
57 | |||
58 |
|
58 | |||
59 | def poentry(path, lineno, s): |
|
59 | def poentry(path, lineno, s): | |
60 | return ('#: %s:%d\n' % (path, lineno) + |
|
60 | return ('#: %s:%d\n' % (path, lineno) + | |
61 | 'msgid %s\n' % normalize(s) + |
|
61 | 'msgid %s\n' % normalize(s) + | |
62 | 'msgstr ""\n') |
|
62 | 'msgstr ""\n') | |
63 |
|
63 | |||
64 | doctestre = re.compile(r'^ +>>> ', re.MULTILINE) |
|
64 | doctestre = re.compile(r'^ +>>> ', re.MULTILINE) | |
65 |
|
65 | |||
66 | def offset(src, doc, name, default): |
|
66 | def offset(src, doc, name, default): | |
67 | """Compute offset or issue a warning on stdout.""" |
|
67 | """Compute offset or issue a warning on stdout.""" | |
68 | # remove doctest part, in order to avoid backslash mismatching |
|
68 | # remove doctest part, in order to avoid backslash mismatching | |
69 | m = doctestre.search(doc) |
|
69 | m = doctestre.search(doc) | |
70 | if m: |
|
70 | if m: | |
71 | doc = doc[:m.start()] |
|
71 | doc = doc[:m.start()] | |
72 |
|
72 | |||
73 | # Backslashes in doc appear doubled in src. |
|
73 | # Backslashes in doc appear doubled in src. | |
74 | end = src.find(doc.replace('\\', '\\\\')) |
|
74 | end = src.find(doc.replace('\\', '\\\\')) | |
75 | if end == -1: |
|
75 | if end == -1: | |
76 | # This can happen if the docstring contains unnecessary escape |
|
76 | # This can happen if the docstring contains unnecessary escape | |
77 | # sequences such as \" in a triple-quoted string. The problem |
|
77 | # sequences such as \" in a triple-quoted string. The problem | |
78 | # is that \" is turned into " and so doc wont appear in src. |
|
78 | # is that \" is turned into " and so doc wont appear in src. | |
79 | sys.stderr.write("warning: unknown offset in %s, assuming %d lines\n" |
|
79 | sys.stderr.write("warning: unknown offset in %s, assuming %d lines\n" | |
80 | % (name, default)) |
|
80 | % (name, default)) | |
81 | return default |
|
81 | return default | |
82 | else: |
|
82 | else: | |
83 | return src.count('\n', 0, end) |
|
83 | return src.count('\n', 0, end) | |
84 |
|
84 | |||
85 |
|
85 | |||
86 | def importpath(path): |
|
86 | def importpath(path): | |
87 | """Import a path like foo/bar/baz.py and return the baz module.""" |
|
87 | """Import a path like foo/bar/baz.py and return the baz module.""" | |
88 | if path.endswith('.py'): |
|
88 | if path.endswith('.py'): | |
89 | path = path[:-3] |
|
89 | path = path[:-3] | |
90 | if path.endswith('/__init__'): |
|
90 | if path.endswith('/__init__'): | |
91 | path = path[:-9] |
|
91 | path = path[:-9] | |
92 | path = path.replace('/', '.') |
|
92 | path = path.replace('/', '.') | |
93 | mod = __import__(path) |
|
93 | mod = __import__(path) | |
94 | for comp in path.split('.')[1:]: |
|
94 | for comp in path.split('.')[1:]: | |
95 | mod = getattr(mod, comp) |
|
95 | mod = getattr(mod, comp) | |
96 | return mod |
|
96 | return mod | |
97 |
|
97 | |||
98 |
|
98 | |||
99 | def docstrings(path): |
|
99 | def docstrings(path): | |
100 | """Extract docstrings from path. |
|
100 | """Extract docstrings from path. | |
101 |
|
101 | |||
102 | This respects the Mercurial cmdtable/table convention and will |
|
102 | This respects the Mercurial cmdtable/table convention and will | |
103 | only extract docstrings from functions mentioned in these tables. |
|
103 | only extract docstrings from functions mentioned in these tables. | |
104 | """ |
|
104 | """ | |
105 | mod = importpath(path) |
|
105 | mod = importpath(path) | |
106 | if mod.__doc__: |
|
106 | if not path.startswith('mercurial/') and mod.__doc__: | |
107 | src = open(path).read() |
|
107 | src = open(path).read() | |
108 | lineno = 1 + offset(src, mod.__doc__, path, 7) |
|
108 | lineno = 1 + offset(src, mod.__doc__, path, 7) | |
109 | print(poentry(path, lineno, mod.__doc__)) |
|
109 | print(poentry(path, lineno, mod.__doc__)) | |
110 |
|
110 | |||
111 | functions = list(getattr(mod, 'i18nfunctions', [])) |
|
111 | functions = list(getattr(mod, 'i18nfunctions', [])) | |
112 | functions = [(f, True) for f in functions] |
|
112 | functions = [(f, True) for f in functions] | |
113 |
|
113 | |||
114 | cmdtable = getattr(mod, 'cmdtable', {}) |
|
114 | cmdtable = getattr(mod, 'cmdtable', {}) | |
115 | if not cmdtable: |
|
115 | if not cmdtable: | |
116 | # Maybe we are processing mercurial.commands? |
|
116 | # Maybe we are processing mercurial.commands? | |
117 | cmdtable = getattr(mod, 'table', {}) |
|
117 | cmdtable = getattr(mod, 'table', {}) | |
118 | functions.extend((c[0], False) for c in cmdtable.itervalues()) |
|
118 | functions.extend((c[0], False) for c in cmdtable.itervalues()) | |
119 |
|
119 | |||
120 | for func, rstrip in functions: |
|
120 | for func, rstrip in functions: | |
121 | if func.__doc__: |
|
121 | if func.__doc__: | |
122 | docobj = func # this might be a proxy to provide formatted doc |
|
122 | docobj = func # this might be a proxy to provide formatted doc | |
123 | func = getattr(func, '_origfunc', func) |
|
123 | func = getattr(func, '_origfunc', func) | |
124 | funcmod = inspect.getmodule(func) |
|
124 | funcmod = inspect.getmodule(func) | |
125 | extra = '' |
|
125 | extra = '' | |
126 | if funcmod.__package__ == funcmod.__name__: |
|
126 | if funcmod.__package__ == funcmod.__name__: | |
127 | extra = '/__init__' |
|
127 | extra = '/__init__' | |
128 | actualpath = '%s%s.py' % (funcmod.__name__.replace('.', '/'), extra) |
|
128 | actualpath = '%s%s.py' % (funcmod.__name__.replace('.', '/'), extra) | |
129 |
|
129 | |||
130 | src = inspect.getsource(func) |
|
130 | src = inspect.getsource(func) | |
131 | name = "%s.%s" % (actualpath, func.__name__) |
|
131 | name = "%s.%s" % (actualpath, func.__name__) | |
132 | lineno = inspect.getsourcelines(func)[1] |
|
132 | lineno = inspect.getsourcelines(func)[1] | |
133 | doc = docobj.__doc__ |
|
133 | doc = docobj.__doc__ | |
134 | origdoc = getattr(docobj, '_origdoc', '') |
|
134 | origdoc = getattr(docobj, '_origdoc', '') | |
135 | if rstrip: |
|
135 | if rstrip: | |
136 | doc = doc.rstrip() |
|
136 | doc = doc.rstrip() | |
137 | origdoc = origdoc.rstrip() |
|
137 | origdoc = origdoc.rstrip() | |
138 | if origdoc: |
|
138 | if origdoc: | |
139 | lineno += offset(src, origdoc, name, 1) |
|
139 | lineno += offset(src, origdoc, name, 1) | |
140 | else: |
|
140 | else: | |
141 | lineno += offset(src, doc, name, 1) |
|
141 | lineno += offset(src, doc, name, 1) | |
142 | print(poentry(actualpath, lineno, doc)) |
|
142 | print(poentry(actualpath, lineno, doc)) | |
143 |
|
143 | |||
144 |
|
144 | |||
145 | def rawtext(path): |
|
145 | def rawtext(path): | |
146 | src = open(path).read() |
|
146 | src = open(path).read() | |
147 | print(poentry(path, 1, src)) |
|
147 | print(poentry(path, 1, src)) | |
148 |
|
148 | |||
149 |
|
149 | |||
150 | if __name__ == "__main__": |
|
150 | if __name__ == "__main__": | |
151 | # It is very important that we import the Mercurial modules from |
|
151 | # It is very important that we import the Mercurial modules from | |
152 | # the source tree where hggettext is executed. Otherwise we might |
|
152 | # the source tree where hggettext is executed. Otherwise we might | |
153 | # accidentally import and extract strings from a Mercurial |
|
153 | # accidentally import and extract strings from a Mercurial | |
154 | # installation mentioned in PYTHONPATH. |
|
154 | # installation mentioned in PYTHONPATH. | |
155 | sys.path.insert(0, os.getcwd()) |
|
155 | sys.path.insert(0, os.getcwd()) | |
156 | from mercurial import demandimport; demandimport.enable() |
|
156 | from mercurial import demandimport; demandimport.enable() | |
157 | for path in sys.argv[1:]: |
|
157 | for path in sys.argv[1:]: | |
158 | if path.endswith('.txt'): |
|
158 | if path.endswith('.txt'): | |
159 | rawtext(path) |
|
159 | rawtext(path) | |
160 | else: |
|
160 | else: | |
161 | docstrings(path) |
|
161 | docstrings(path) |
General Comments 0
You need to be logged in to leave comments.
Login now