Show More
@@ -1,747 +1,747 b'' | |||||
1 | #!/usr/bin/env python |
|
1 | #!/usr/bin/env python | |
2 |
|
2 | |||
3 | from __future__ import absolute_import, print_function |
|
3 | from __future__ import absolute_import, print_function | |
4 |
|
4 | |||
5 | import ast |
|
5 | import ast | |
6 | import collections |
|
6 | import collections | |
7 | import os |
|
7 | import os | |
8 | import sys |
|
8 | import sys | |
9 |
|
9 | |||
10 | # Import a minimal set of stdlib modules needed for list_stdlib_modules() |
|
10 | # Import a minimal set of stdlib modules needed for list_stdlib_modules() | |
11 | # to work when run from a virtualenv. The modules were chosen empirically |
|
11 | # to work when run from a virtualenv. The modules were chosen empirically | |
12 | # so that the return value matches the return value without virtualenv. |
|
12 | # so that the return value matches the return value without virtualenv. | |
13 | if True: # disable lexical sorting checks |
|
13 | if True: # disable lexical sorting checks | |
14 | try: |
|
14 | try: | |
15 | import BaseHTTPServer as basehttpserver |
|
15 | import BaseHTTPServer as basehttpserver | |
16 | except ImportError: |
|
16 | except ImportError: | |
17 | basehttpserver = None |
|
17 | basehttpserver = None | |
18 | import zlib |
|
18 | import zlib | |
19 |
|
19 | |||
20 | import testparseutil |
|
20 | import testparseutil | |
21 |
|
21 | |||
22 | # Whitelist of modules that symbols can be directly imported from. |
|
22 | # Whitelist of modules that symbols can be directly imported from. | |
23 | allowsymbolimports = ( |
|
23 | allowsymbolimports = ( | |
24 | '__future__', |
|
24 | '__future__', | |
25 | 'bzrlib', |
|
25 | 'bzrlib', | |
26 | 'hgclient', |
|
26 | 'hgclient', | |
27 | 'mercurial', |
|
27 | 'mercurial', | |
28 | 'mercurial.hgweb.common', |
|
28 | 'mercurial.hgweb.common', | |
29 | 'mercurial.hgweb.request', |
|
29 | 'mercurial.hgweb.request', | |
30 | 'mercurial.i18n', |
|
30 | 'mercurial.i18n', | |
31 | 'mercurial.node', |
|
31 | 'mercurial.node', | |
32 | # for revlog to re-export constant to extensions |
|
32 | # for revlog to re-export constant to extensions | |
33 | 'mercurial.revlogutils.constants', |
|
33 | 'mercurial.revlogutils.constants', | |
34 | # for cffi modules to re-export pure functions |
|
34 | # for cffi modules to re-export pure functions | |
35 | 'mercurial.pure.base85', |
|
35 | 'mercurial.pure.base85', | |
36 | 'mercurial.pure.bdiff', |
|
36 | 'mercurial.pure.bdiff', | |
37 | 'mercurial.pure.mpatch', |
|
37 | 'mercurial.pure.mpatch', | |
38 | 'mercurial.pure.osutil', |
|
38 | 'mercurial.pure.osutil', | |
39 | 'mercurial.pure.parsers', |
|
39 | 'mercurial.pure.parsers', | |
40 | # third-party imports should be directly imported |
|
40 | # third-party imports should be directly imported | |
41 | 'mercurial.thirdparty', |
|
41 | 'mercurial.thirdparty', | |
42 | 'mercurial.thirdparty.attr', |
|
42 | 'mercurial.thirdparty.attr', | |
43 | 'mercurial.thirdparty.zope', |
|
43 | 'mercurial.thirdparty.zope', | |
44 | 'mercurial.thirdparty.zope.interface', |
|
44 | 'mercurial.thirdparty.zope.interface', | |
45 | ) |
|
45 | ) | |
46 |
|
46 | |||
47 | # Whitelist of symbols that can be directly imported. |
|
47 | # Whitelist of symbols that can be directly imported. | |
48 | directsymbols = ( |
|
48 | directsymbols = ( | |
49 | 'demandimport', |
|
49 | 'demandimport', | |
50 | ) |
|
50 | ) | |
51 |
|
51 | |||
52 | # Modules that must be aliased because they are commonly confused with |
|
52 | # Modules that must be aliased because they are commonly confused with | |
53 | # common variables and can create aliasing and readability issues. |
|
53 | # common variables and can create aliasing and readability issues. | |
54 | requirealias = { |
|
54 | requirealias = { | |
55 | 'ui': 'uimod', |
|
55 | 'ui': 'uimod', | |
56 | } |
|
56 | } | |
57 |
|
57 | |||
58 | def usingabsolute(root): |
|
58 | def usingabsolute(root): | |
59 | """Whether absolute imports are being used.""" |
|
59 | """Whether absolute imports are being used.""" | |
60 | if sys.version_info[0] >= 3: |
|
60 | if sys.version_info[0] >= 3: | |
61 | return True |
|
61 | return True | |
62 |
|
62 | |||
63 | for node in ast.walk(root): |
|
63 | for node in ast.walk(root): | |
64 | if isinstance(node, ast.ImportFrom): |
|
64 | if isinstance(node, ast.ImportFrom): | |
65 | if node.module == '__future__': |
|
65 | if node.module == '__future__': | |
66 | for n in node.names: |
|
66 | for n in node.names: | |
67 | if n.name == 'absolute_import': |
|
67 | if n.name == 'absolute_import': | |
68 | return True |
|
68 | return True | |
69 |
|
69 | |||
70 | return False |
|
70 | return False | |
71 |
|
71 | |||
72 | def walklocal(root): |
|
72 | def walklocal(root): | |
73 | """Recursively yield all descendant nodes but not in a different scope""" |
|
73 | """Recursively yield all descendant nodes but not in a different scope""" | |
74 | todo = collections.deque(ast.iter_child_nodes(root)) |
|
74 | todo = collections.deque(ast.iter_child_nodes(root)) | |
75 | yield root, False |
|
75 | yield root, False | |
76 | while todo: |
|
76 | while todo: | |
77 | node = todo.popleft() |
|
77 | node = todo.popleft() | |
78 | newscope = isinstance(node, ast.FunctionDef) |
|
78 | newscope = isinstance(node, ast.FunctionDef) | |
79 | if not newscope: |
|
79 | if not newscope: | |
80 | todo.extend(ast.iter_child_nodes(node)) |
|
80 | todo.extend(ast.iter_child_nodes(node)) | |
81 | yield node, newscope |
|
81 | yield node, newscope | |
82 |
|
82 | |||
83 | def dotted_name_of_path(path): |
|
83 | def dotted_name_of_path(path): | |
84 | """Given a relative path to a source file, return its dotted module name. |
|
84 | """Given a relative path to a source file, return its dotted module name. | |
85 |
|
85 | |||
86 | >>> dotted_name_of_path('mercurial/error.py') |
|
86 | >>> dotted_name_of_path('mercurial/error.py') | |
87 | 'mercurial.error' |
|
87 | 'mercurial.error' | |
88 | >>> dotted_name_of_path('zlibmodule.so') |
|
88 | >>> dotted_name_of_path('zlibmodule.so') | |
89 | 'zlib' |
|
89 | 'zlib' | |
90 | """ |
|
90 | """ | |
91 | parts = path.replace(os.sep, '/').split('/') |
|
91 | parts = path.replace(os.sep, '/').split('/') | |
92 | parts[-1] = parts[-1].split('.', 1)[0] # remove .py and .so and .ARCH.so |
|
92 | parts[-1] = parts[-1].split('.', 1)[0] # remove .py and .so and .ARCH.so | |
93 | if parts[-1].endswith('module'): |
|
93 | if parts[-1].endswith('module'): | |
94 | parts[-1] = parts[-1][:-6] |
|
94 | parts[-1] = parts[-1][:-6] | |
95 | return '.'.join(parts) |
|
95 | return '.'.join(parts) | |
96 |
|
96 | |||
97 | def fromlocalfunc(modulename, localmods): |
|
97 | def fromlocalfunc(modulename, localmods): | |
98 | """Get a function to examine which locally defined module the |
|
98 | """Get a function to examine which locally defined module the | |
99 | target source imports via a specified name. |
|
99 | target source imports via a specified name. | |
100 |
|
100 | |||
101 | `modulename` is an `dotted_name_of_path()`-ed source file path, |
|
101 | `modulename` is an `dotted_name_of_path()`-ed source file path, | |
102 | which may have `.__init__` at the end of it, of the target source. |
|
102 | which may have `.__init__` at the end of it, of the target source. | |
103 |
|
103 | |||
104 | `localmods` is a set of absolute `dotted_name_of_path()`-ed source file |
|
104 | `localmods` is a set of absolute `dotted_name_of_path()`-ed source file | |
105 | paths of locally defined (= Mercurial specific) modules. |
|
105 | paths of locally defined (= Mercurial specific) modules. | |
106 |
|
106 | |||
107 | This function assumes that module names not existing in |
|
107 | This function assumes that module names not existing in | |
108 | `localmods` are from the Python standard library. |
|
108 | `localmods` are from the Python standard library. | |
109 |
|
109 | |||
110 | This function returns the function, which takes `name` argument, |
|
110 | This function returns the function, which takes `name` argument, | |
111 | and returns `(absname, dottedpath, hassubmod)` tuple if `name` |
|
111 | and returns `(absname, dottedpath, hassubmod)` tuple if `name` | |
112 | matches against locally defined module. Otherwise, it returns |
|
112 | matches against locally defined module. Otherwise, it returns | |
113 | False. |
|
113 | False. | |
114 |
|
114 | |||
115 | It is assumed that `name` doesn't have `.__init__`. |
|
115 | It is assumed that `name` doesn't have `.__init__`. | |
116 |
|
116 | |||
117 | `absname` is an absolute module name of specified `name` |
|
117 | `absname` is an absolute module name of specified `name` | |
118 | (e.g. "hgext.convert"). This can be used to compose prefix for sub |
|
118 | (e.g. "hgext.convert"). This can be used to compose prefix for sub | |
119 | modules or so. |
|
119 | modules or so. | |
120 |
|
120 | |||
121 | `dottedpath` is a `dotted_name_of_path()`-ed source file path |
|
121 | `dottedpath` is a `dotted_name_of_path()`-ed source file path | |
122 | (e.g. "hgext.convert.__init__") of `name`. This is used to look |
|
122 | (e.g. "hgext.convert.__init__") of `name`. This is used to look | |
123 | module up in `localmods` again. |
|
123 | module up in `localmods` again. | |
124 |
|
124 | |||
125 | `hassubmod` is whether it may have sub modules under it (for |
|
125 | `hassubmod` is whether it may have sub modules under it (for | |
126 | convenient, even though this is also equivalent to "absname != |
|
126 | convenient, even though this is also equivalent to "absname != | |
127 | dottednpath") |
|
127 | dottednpath") | |
128 |
|
128 | |||
129 | >>> localmods = {'foo.__init__', 'foo.foo1', |
|
129 | >>> localmods = {'foo.__init__', 'foo.foo1', | |
130 | ... 'foo.bar.__init__', 'foo.bar.bar1', |
|
130 | ... 'foo.bar.__init__', 'foo.bar.bar1', | |
131 | ... 'baz.__init__', 'baz.baz1'} |
|
131 | ... 'baz.__init__', 'baz.baz1'} | |
132 | >>> fromlocal = fromlocalfunc('foo.xxx', localmods) |
|
132 | >>> fromlocal = fromlocalfunc('foo.xxx', localmods) | |
133 | >>> # relative |
|
133 | >>> # relative | |
134 | >>> fromlocal('foo1') |
|
134 | >>> fromlocal('foo1') | |
135 | ('foo.foo1', 'foo.foo1', False) |
|
135 | ('foo.foo1', 'foo.foo1', False) | |
136 | >>> fromlocal('bar') |
|
136 | >>> fromlocal('bar') | |
137 | ('foo.bar', 'foo.bar.__init__', True) |
|
137 | ('foo.bar', 'foo.bar.__init__', True) | |
138 | >>> fromlocal('bar.bar1') |
|
138 | >>> fromlocal('bar.bar1') | |
139 | ('foo.bar.bar1', 'foo.bar.bar1', False) |
|
139 | ('foo.bar.bar1', 'foo.bar.bar1', False) | |
140 | >>> # absolute |
|
140 | >>> # absolute | |
141 | >>> fromlocal('baz') |
|
141 | >>> fromlocal('baz') | |
142 | ('baz', 'baz.__init__', True) |
|
142 | ('baz', 'baz.__init__', True) | |
143 | >>> fromlocal('baz.baz1') |
|
143 | >>> fromlocal('baz.baz1') | |
144 | ('baz.baz1', 'baz.baz1', False) |
|
144 | ('baz.baz1', 'baz.baz1', False) | |
145 | >>> # unknown = maybe standard library |
|
145 | >>> # unknown = maybe standard library | |
146 | >>> fromlocal('os') |
|
146 | >>> fromlocal('os') | |
147 | False |
|
147 | False | |
148 | >>> fromlocal(None, 1) |
|
148 | >>> fromlocal(None, 1) | |
149 | ('foo', 'foo.__init__', True) |
|
149 | ('foo', 'foo.__init__', True) | |
150 | >>> fromlocal('foo1', 1) |
|
150 | >>> fromlocal('foo1', 1) | |
151 | ('foo.foo1', 'foo.foo1', False) |
|
151 | ('foo.foo1', 'foo.foo1', False) | |
152 | >>> fromlocal2 = fromlocalfunc('foo.xxx.yyy', localmods) |
|
152 | >>> fromlocal2 = fromlocalfunc('foo.xxx.yyy', localmods) | |
153 | >>> fromlocal2(None, 2) |
|
153 | >>> fromlocal2(None, 2) | |
154 | ('foo', 'foo.__init__', True) |
|
154 | ('foo', 'foo.__init__', True) | |
155 | >>> fromlocal2('bar2', 1) |
|
155 | >>> fromlocal2('bar2', 1) | |
156 | False |
|
156 | False | |
157 | >>> fromlocal2('bar', 2) |
|
157 | >>> fromlocal2('bar', 2) | |
158 | ('foo.bar', 'foo.bar.__init__', True) |
|
158 | ('foo.bar', 'foo.bar.__init__', True) | |
159 | """ |
|
159 | """ | |
160 | if not isinstance(modulename, str): |
|
160 | if not isinstance(modulename, str): | |
161 | modulename = modulename.decode('ascii') |
|
161 | modulename = modulename.decode('ascii') | |
162 | prefix = '.'.join(modulename.split('.')[:-1]) |
|
162 | prefix = '.'.join(modulename.split('.')[:-1]) | |
163 | if prefix: |
|
163 | if prefix: | |
164 | prefix += '.' |
|
164 | prefix += '.' | |
165 | def fromlocal(name, level=0): |
|
165 | def fromlocal(name, level=0): | |
166 | # name is false value when relative imports are used. |
|
166 | # name is false value when relative imports are used. | |
167 | if not name: |
|
167 | if not name: | |
168 | # If relative imports are used, level must not be absolute. |
|
168 | # If relative imports are used, level must not be absolute. | |
169 | assert level > 0 |
|
169 | assert level > 0 | |
170 | candidates = ['.'.join(modulename.split('.')[:-level])] |
|
170 | candidates = ['.'.join(modulename.split('.')[:-level])] | |
171 | else: |
|
171 | else: | |
172 | if not level: |
|
172 | if not level: | |
173 | # Check relative name first. |
|
173 | # Check relative name first. | |
174 | candidates = [prefix + name, name] |
|
174 | candidates = [prefix + name, name] | |
175 | else: |
|
175 | else: | |
176 | candidates = ['.'.join(modulename.split('.')[:-level]) + |
|
176 | candidates = ['.'.join(modulename.split('.')[:-level]) + | |
177 | '.' + name] |
|
177 | '.' + name] | |
178 |
|
178 | |||
179 | for n in candidates: |
|
179 | for n in candidates: | |
180 | if n in localmods: |
|
180 | if n in localmods: | |
181 | return (n, n, False) |
|
181 | return (n, n, False) | |
182 | dottedpath = n + '.__init__' |
|
182 | dottedpath = n + '.__init__' | |
183 | if dottedpath in localmods: |
|
183 | if dottedpath in localmods: | |
184 | return (n, dottedpath, True) |
|
184 | return (n, dottedpath, True) | |
185 | return False |
|
185 | return False | |
186 | return fromlocal |
|
186 | return fromlocal | |
187 |
|
187 | |||
188 | def populateextmods(localmods): |
|
188 | def populateextmods(localmods): | |
189 | """Populate C extension modules based on pure modules""" |
|
189 | """Populate C extension modules based on pure modules""" | |
190 | newlocalmods = set(localmods) |
|
190 | newlocalmods = set(localmods) | |
191 | for n in localmods: |
|
191 | for n in localmods: | |
192 | if n.startswith('mercurial.pure.'): |
|
192 | if n.startswith('mercurial.pure.'): | |
193 | m = n[len('mercurial.pure.'):] |
|
193 | m = n[len('mercurial.pure.'):] | |
194 | newlocalmods.add('mercurial.cext.' + m) |
|
194 | newlocalmods.add('mercurial.cext.' + m) | |
195 | newlocalmods.add('mercurial.cffi._' + m) |
|
195 | newlocalmods.add('mercurial.cffi._' + m) | |
196 | return newlocalmods |
|
196 | return newlocalmods | |
197 |
|
197 | |||
198 | def list_stdlib_modules(): |
|
198 | def list_stdlib_modules(): | |
199 | """List the modules present in the stdlib. |
|
199 | """List the modules present in the stdlib. | |
200 |
|
200 | |||
201 | >>> py3 = sys.version_info[0] >= 3 |
|
201 | >>> py3 = sys.version_info[0] >= 3 | |
202 | >>> mods = set(list_stdlib_modules()) |
|
202 | >>> mods = set(list_stdlib_modules()) | |
203 | >>> 'BaseHTTPServer' in mods or py3 |
|
203 | >>> 'BaseHTTPServer' in mods or py3 | |
204 | True |
|
204 | True | |
205 |
|
205 | |||
206 | os.path isn't really a module, so it's missing: |
|
206 | os.path isn't really a module, so it's missing: | |
207 |
|
207 | |||
208 | >>> 'os.path' in mods |
|
208 | >>> 'os.path' in mods | |
209 | False |
|
209 | False | |
210 |
|
210 | |||
211 | sys requires special treatment, because it's baked into the |
|
211 | sys requires special treatment, because it's baked into the | |
212 | interpreter, but it should still appear: |
|
212 | interpreter, but it should still appear: | |
213 |
|
213 | |||
214 | >>> 'sys' in mods |
|
214 | >>> 'sys' in mods | |
215 | True |
|
215 | True | |
216 |
|
216 | |||
217 | >>> 'collections' in mods |
|
217 | >>> 'collections' in mods | |
218 | True |
|
218 | True | |
219 |
|
219 | |||
220 | >>> 'cStringIO' in mods or py3 |
|
220 | >>> 'cStringIO' in mods or py3 | |
221 | True |
|
221 | True | |
222 |
|
222 | |||
223 | >>> 'cffi' in mods |
|
223 | >>> 'cffi' in mods | |
224 | True |
|
224 | True | |
225 | """ |
|
225 | """ | |
226 | for m in sys.builtin_module_names: |
|
226 | for m in sys.builtin_module_names: | |
227 | yield m |
|
227 | yield m | |
228 | # These modules only exist on windows, but we should always |
|
228 | # These modules only exist on windows, but we should always | |
229 | # consider them stdlib. |
|
229 | # consider them stdlib. | |
230 | for m in ['msvcrt', '_winreg']: |
|
230 | for m in ['msvcrt', '_winreg']: | |
231 | yield m |
|
231 | yield m | |
232 | yield '__builtin__' |
|
232 | yield '__builtin__' | |
233 | yield 'builtins' # python3 only |
|
233 | yield 'builtins' # python3 only | |
234 | yield 'importlib.abc' # python3 only |
|
234 | yield 'importlib.abc' # python3 only | |
235 | yield 'importlib.machinery' # python3 only |
|
235 | yield 'importlib.machinery' # python3 only | |
236 | yield 'importlib.util' # python3 only |
|
236 | yield 'importlib.util' # python3 only | |
237 | for m in 'fcntl', 'grp', 'pwd', 'termios': # Unix only |
|
237 | for m in 'fcntl', 'grp', 'pwd', 'termios': # Unix only | |
238 | yield m |
|
238 | yield m | |
239 | for m in 'cPickle', 'datetime': # in Python (not C) on PyPy |
|
239 | for m in 'cPickle', 'datetime': # in Python (not C) on PyPy | |
240 | yield m |
|
240 | yield m | |
241 | for m in ['cffi']: |
|
241 | for m in ['cffi']: | |
242 | yield m |
|
242 | yield m | |
243 | stdlib_prefixes = {sys.prefix, sys.exec_prefix} |
|
243 | stdlib_prefixes = {sys.prefix, sys.exec_prefix} | |
244 | # We need to supplement the list of prefixes for the search to work |
|
244 | # We need to supplement the list of prefixes for the search to work | |
245 | # when run from within a virtualenv. |
|
245 | # when run from within a virtualenv. | |
246 | for mod in (basehttpserver, zlib): |
|
246 | for mod in (basehttpserver, zlib): | |
247 | if mod is None: |
|
247 | if mod is None: | |
248 | continue |
|
248 | continue | |
249 | try: |
|
249 | try: | |
250 | # Not all module objects have a __file__ attribute. |
|
250 | # Not all module objects have a __file__ attribute. | |
251 | filename = mod.__file__ |
|
251 | filename = mod.__file__ | |
252 | except AttributeError: |
|
252 | except AttributeError: | |
253 | continue |
|
253 | continue | |
254 | dirname = os.path.dirname(filename) |
|
254 | dirname = os.path.dirname(filename) | |
255 | for prefix in stdlib_prefixes: |
|
255 | for prefix in stdlib_prefixes: | |
256 | if dirname.startswith(prefix): |
|
256 | if dirname.startswith(prefix): | |
257 | # Then this directory is redundant. |
|
257 | # Then this directory is redundant. | |
258 | break |
|
258 | break | |
259 | else: |
|
259 | else: | |
260 | stdlib_prefixes.add(dirname) |
|
260 | stdlib_prefixes.add(dirname) | |
261 | sourceroot = os.path.abspath(os.path.dirname(os.path.dirname(__file__))) |
|
261 | sourceroot = os.path.abspath(os.path.dirname(os.path.dirname(__file__))) | |
262 | for libpath in sys.path: |
|
262 | for libpath in sys.path: | |
263 | # We want to walk everything in sys.path that starts with something in |
|
263 | # We want to walk everything in sys.path that starts with something in | |
264 | # stdlib_prefixes, but not directories from the hg sources. |
|
264 | # stdlib_prefixes, but not directories from the hg sources. | |
265 | if (os.path.abspath(libpath).startswith(sourceroot) |
|
265 | if (os.path.abspath(libpath).startswith(sourceroot) | |
266 | or not any(libpath.startswith(p) for p in stdlib_prefixes)): |
|
266 | or not any(libpath.startswith(p) for p in stdlib_prefixes)): | |
267 | continue |
|
267 | continue | |
268 | for top, dirs, files in os.walk(libpath): |
|
268 | for top, dirs, files in os.walk(libpath): | |
269 | for i, d in reversed(list(enumerate(dirs))): |
|
269 | for i, d in reversed(list(enumerate(dirs))): | |
270 | if (not os.path.exists(os.path.join(top, d, '__init__.py')) |
|
270 | if (not os.path.exists(os.path.join(top, d, '__init__.py')) | |
271 | or top == libpath and d in ('hgdemandimport', 'hgext', |
|
271 | or top == libpath and d in ('hgdemandimport', 'hgext', | |
272 | 'mercurial')): |
|
272 | 'mercurial')): | |
273 | del dirs[i] |
|
273 | del dirs[i] | |
274 | for name in files: |
|
274 | for name in files: | |
275 | if not name.endswith(('.py', '.so', '.pyc', '.pyo', '.pyd')): |
|
275 | if not name.endswith(('.py', '.so', '.pyc', '.pyo', '.pyd')): | |
276 | continue |
|
276 | continue | |
277 | if name.startswith('__init__.py'): |
|
277 | if name.startswith('__init__.py'): | |
278 | full_path = top |
|
278 | full_path = top | |
279 | else: |
|
279 | else: | |
280 | full_path = os.path.join(top, name) |
|
280 | full_path = os.path.join(top, name) | |
281 | rel_path = full_path[len(libpath) + 1:] |
|
281 | rel_path = full_path[len(libpath) + 1:] | |
282 | mod = dotted_name_of_path(rel_path) |
|
282 | mod = dotted_name_of_path(rel_path) | |
283 | yield mod |
|
283 | yield mod | |
284 |
|
284 | |||
285 | stdlib_modules = set(list_stdlib_modules()) |
|
285 | stdlib_modules = set(list_stdlib_modules()) | |
286 |
|
286 | |||
287 | def imported_modules(source, modulename, f, localmods, ignore_nested=False): |
|
287 | def imported_modules(source, modulename, f, localmods, ignore_nested=False): | |
288 | """Given the source of a file as a string, yield the names |
|
288 | """Given the source of a file as a string, yield the names | |
289 | imported by that file. |
|
289 | imported by that file. | |
290 |
|
290 | |||
291 | Args: |
|
291 | Args: | |
292 | source: The python source to examine as a string. |
|
292 | source: The python source to examine as a string. | |
293 | modulename: of specified python source (may have `__init__`) |
|
293 | modulename: of specified python source (may have `__init__`) | |
294 | localmods: set of locally defined module names (may have `__init__`) |
|
294 | localmods: set of locally defined module names (may have `__init__`) | |
295 | ignore_nested: If true, import statements that do not start in |
|
295 | ignore_nested: If true, import statements that do not start in | |
296 | column zero will be ignored. |
|
296 | column zero will be ignored. | |
297 |
|
297 | |||
298 | Returns: |
|
298 | Returns: | |
299 | A list of absolute module names imported by the given source. |
|
299 | A list of absolute module names imported by the given source. | |
300 |
|
300 | |||
301 | >>> f = 'foo/xxx.py' |
|
301 | >>> f = 'foo/xxx.py' | |
302 | >>> modulename = 'foo.xxx' |
|
302 | >>> modulename = 'foo.xxx' | |
303 | >>> localmods = {'foo.__init__': True, |
|
303 | >>> localmods = {'foo.__init__': True, | |
304 | ... 'foo.foo1': True, 'foo.foo2': True, |
|
304 | ... 'foo.foo1': True, 'foo.foo2': True, | |
305 | ... 'foo.bar.__init__': True, 'foo.bar.bar1': True, |
|
305 | ... 'foo.bar.__init__': True, 'foo.bar.bar1': True, | |
306 | ... 'baz.__init__': True, 'baz.baz1': True } |
|
306 | ... 'baz.__init__': True, 'baz.baz1': True } | |
307 | >>> # standard library (= not locally defined ones) |
|
307 | >>> # standard library (= not locally defined ones) | |
308 | >>> sorted(imported_modules( |
|
308 | >>> sorted(imported_modules( | |
309 | ... 'from stdlib1 import foo, bar; import stdlib2', |
|
309 | ... 'from stdlib1 import foo, bar; import stdlib2', | |
310 | ... modulename, f, localmods)) |
|
310 | ... modulename, f, localmods)) | |
311 | [] |
|
311 | [] | |
312 | >>> # relative importing |
|
312 | >>> # relative importing | |
313 | >>> sorted(imported_modules( |
|
313 | >>> sorted(imported_modules( | |
314 | ... 'import foo1; from bar import bar1', |
|
314 | ... 'import foo1; from bar import bar1', | |
315 | ... modulename, f, localmods)) |
|
315 | ... modulename, f, localmods)) | |
316 | ['foo.bar.bar1', 'foo.foo1'] |
|
316 | ['foo.bar.bar1', 'foo.foo1'] | |
317 | >>> sorted(imported_modules( |
|
317 | >>> sorted(imported_modules( | |
318 | ... 'from bar.bar1 import name1, name2, name3', |
|
318 | ... 'from bar.bar1 import name1, name2, name3', | |
319 | ... modulename, f, localmods)) |
|
319 | ... modulename, f, localmods)) | |
320 | ['foo.bar.bar1'] |
|
320 | ['foo.bar.bar1'] | |
321 | >>> # absolute importing |
|
321 | >>> # absolute importing | |
322 | >>> sorted(imported_modules( |
|
322 | >>> sorted(imported_modules( | |
323 | ... 'from baz import baz1, name1', |
|
323 | ... 'from baz import baz1, name1', | |
324 | ... modulename, f, localmods)) |
|
324 | ... modulename, f, localmods)) | |
325 | ['baz.__init__', 'baz.baz1'] |
|
325 | ['baz.__init__', 'baz.baz1'] | |
326 | >>> # mixed importing, even though it shouldn't be recommended |
|
326 | >>> # mixed importing, even though it shouldn't be recommended | |
327 | >>> sorted(imported_modules( |
|
327 | >>> sorted(imported_modules( | |
328 | ... 'import stdlib, foo1, baz', |
|
328 | ... 'import stdlib, foo1, baz', | |
329 | ... modulename, f, localmods)) |
|
329 | ... modulename, f, localmods)) | |
330 | ['baz.__init__', 'foo.foo1'] |
|
330 | ['baz.__init__', 'foo.foo1'] | |
331 | >>> # ignore_nested |
|
331 | >>> # ignore_nested | |
332 | >>> sorted(imported_modules( |
|
332 | >>> sorted(imported_modules( | |
333 | ... '''import foo |
|
333 | ... '''import foo | |
334 | ... def wat(): |
|
334 | ... def wat(): | |
335 | ... import bar |
|
335 | ... import bar | |
336 | ... ''', modulename, f, localmods)) |
|
336 | ... ''', modulename, f, localmods)) | |
337 | ['foo.__init__', 'foo.bar.__init__'] |
|
337 | ['foo.__init__', 'foo.bar.__init__'] | |
338 | >>> sorted(imported_modules( |
|
338 | >>> sorted(imported_modules( | |
339 | ... '''import foo |
|
339 | ... '''import foo | |
340 | ... def wat(): |
|
340 | ... def wat(): | |
341 | ... import bar |
|
341 | ... import bar | |
342 | ... ''', modulename, f, localmods, ignore_nested=True)) |
|
342 | ... ''', modulename, f, localmods, ignore_nested=True)) | |
343 | ['foo.__init__'] |
|
343 | ['foo.__init__'] | |
344 | """ |
|
344 | """ | |
345 | fromlocal = fromlocalfunc(modulename, localmods) |
|
345 | fromlocal = fromlocalfunc(modulename, localmods) | |
346 | for node in ast.walk(ast.parse(source, f)): |
|
346 | for node in ast.walk(ast.parse(source, f)): | |
347 | if ignore_nested and getattr(node, 'col_offset', 0) > 0: |
|
347 | if ignore_nested and getattr(node, 'col_offset', 0) > 0: | |
348 | continue |
|
348 | continue | |
349 | if isinstance(node, ast.Import): |
|
349 | if isinstance(node, ast.Import): | |
350 | for n in node.names: |
|
350 | for n in node.names: | |
351 | found = fromlocal(n.name) |
|
351 | found = fromlocal(n.name) | |
352 | if not found: |
|
352 | if not found: | |
353 | # this should import standard library |
|
353 | # this should import standard library | |
354 | continue |
|
354 | continue | |
355 | yield found[1] |
|
355 | yield found[1] | |
356 | elif isinstance(node, ast.ImportFrom): |
|
356 | elif isinstance(node, ast.ImportFrom): | |
357 | found = fromlocal(node.module, node.level) |
|
357 | found = fromlocal(node.module, node.level) | |
358 | if not found: |
|
358 | if not found: | |
359 | # this should import standard library |
|
359 | # this should import standard library | |
360 | continue |
|
360 | continue | |
361 |
|
361 | |||
362 | absname, dottedpath, hassubmod = found |
|
362 | absname, dottedpath, hassubmod = found | |
363 | if not hassubmod: |
|
363 | if not hassubmod: | |
364 | # "dottedpath" is not a package; must be imported |
|
364 | # "dottedpath" is not a package; must be imported | |
365 | yield dottedpath |
|
365 | yield dottedpath | |
366 | # examination of "node.names" should be redundant |
|
366 | # examination of "node.names" should be redundant | |
367 | # e.g.: from mercurial.node import nullid, nullrev |
|
367 | # e.g.: from mercurial.node import nullid, nullrev | |
368 | continue |
|
368 | continue | |
369 |
|
369 | |||
370 | modnotfound = False |
|
370 | modnotfound = False | |
371 | prefix = absname + '.' |
|
371 | prefix = absname + '.' | |
372 | for n in node.names: |
|
372 | for n in node.names: | |
373 | found = fromlocal(prefix + n.name) |
|
373 | found = fromlocal(prefix + n.name) | |
374 | if not found: |
|
374 | if not found: | |
375 | # this should be a function or a property of "node.module" |
|
375 | # this should be a function or a property of "node.module" | |
376 | modnotfound = True |
|
376 | modnotfound = True | |
377 | continue |
|
377 | continue | |
378 | yield found[1] |
|
378 | yield found[1] | |
379 | if modnotfound: |
|
379 | if modnotfound: | |
380 | # "dottedpath" is a package, but imported because of non-module |
|
380 | # "dottedpath" is a package, but imported because of non-module | |
381 | # lookup |
|
381 | # lookup | |
382 | yield dottedpath |
|
382 | yield dottedpath | |
383 |
|
383 | |||
384 | def verify_import_convention(module, source, localmods): |
|
384 | def verify_import_convention(module, source, localmods): | |
385 | """Verify imports match our established coding convention. |
|
385 | """Verify imports match our established coding convention. | |
386 |
|
386 | |||
387 | We have 2 conventions: legacy and modern. The modern convention is in |
|
387 | We have 2 conventions: legacy and modern. The modern convention is in | |
388 | effect when using absolute imports. |
|
388 | effect when using absolute imports. | |
389 |
|
389 | |||
390 | The legacy convention only looks for mixed imports. The modern convention |
|
390 | The legacy convention only looks for mixed imports. The modern convention | |
391 | is much more thorough. |
|
391 | is much more thorough. | |
392 | """ |
|
392 | """ | |
393 | root = ast.parse(source) |
|
393 | root = ast.parse(source) | |
394 | absolute = usingabsolute(root) |
|
394 | absolute = usingabsolute(root) | |
395 |
|
395 | |||
396 | if absolute: |
|
396 | if absolute: | |
397 | return verify_modern_convention(module, root, localmods) |
|
397 | return verify_modern_convention(module, root, localmods) | |
398 | else: |
|
398 | else: | |
399 | return verify_stdlib_on_own_line(root) |
|
399 | return verify_stdlib_on_own_line(root) | |
400 |
|
400 | |||
401 | def verify_modern_convention(module, root, localmods, root_col_offset=0): |
|
401 | def verify_modern_convention(module, root, localmods, root_col_offset=0): | |
402 | """Verify a file conforms to the modern import convention rules. |
|
402 | """Verify a file conforms to the modern import convention rules. | |
403 |
|
403 | |||
404 | The rules of the modern convention are: |
|
404 | The rules of the modern convention are: | |
405 |
|
405 | |||
406 | * Ordering is stdlib followed by local imports. Each group is lexically |
|
406 | * Ordering is stdlib followed by local imports. Each group is lexically | |
407 | sorted. |
|
407 | sorted. | |
408 | * Importing multiple modules via "import X, Y" is not allowed: use |
|
408 | * Importing multiple modules via "import X, Y" is not allowed: use | |
409 | separate import statements. |
|
409 | separate import statements. | |
410 | * Importing multiple modules via "from X import ..." is allowed if using |
|
410 | * Importing multiple modules via "from X import ..." is allowed if using | |
411 | parenthesis and one entry per line. |
|
411 | parenthesis and one entry per line. | |
412 | * Only 1 relative import statement per import level ("from .", "from ..") |
|
412 | * Only 1 relative import statement per import level ("from .", "from ..") | |
413 | is allowed. |
|
413 | is allowed. | |
414 | * Relative imports from higher levels must occur before lower levels. e.g. |
|
414 | * Relative imports from higher levels must occur before lower levels. e.g. | |
415 | "from .." must be before "from .". |
|
415 | "from .." must be before "from .". | |
416 | * Imports from peer packages should use relative import (e.g. do not |
|
416 | * Imports from peer packages should use relative import (e.g. do not | |
417 | "import mercurial.foo" from a "mercurial.*" module). |
|
417 | "import mercurial.foo" from a "mercurial.*" module). | |
418 | * Symbols can only be imported from specific modules (see |
|
418 | * Symbols can only be imported from specific modules (see | |
419 | `allowsymbolimports`). For other modules, first import the module then |
|
419 | `allowsymbolimports`). For other modules, first import the module then | |
420 | assign the symbol to a module-level variable. In addition, these imports |
|
420 | assign the symbol to a module-level variable. In addition, these imports | |
421 | must be performed before other local imports. This rule only |
|
421 | must be performed before other local imports. This rule only | |
422 | applies to import statements outside of any blocks. |
|
422 | applies to import statements outside of any blocks. | |
423 | * Relative imports from the standard library are not allowed, unless that |
|
423 | * Relative imports from the standard library are not allowed, unless that | |
424 | library is also a local module. |
|
424 | library is also a local module. | |
425 | * Certain modules must be aliased to alternate names to avoid aliasing |
|
425 | * Certain modules must be aliased to alternate names to avoid aliasing | |
426 | and readability problems. See `requirealias`. |
|
426 | and readability problems. See `requirealias`. | |
427 | """ |
|
427 | """ | |
428 | if not isinstance(module, str): |
|
428 | if not isinstance(module, str): | |
429 | module = module.decode('ascii') |
|
429 | module = module.decode('ascii') | |
430 | topmodule = module.split('.')[0] |
|
430 | topmodule = module.split('.')[0] | |
431 | fromlocal = fromlocalfunc(module, localmods) |
|
431 | fromlocal = fromlocalfunc(module, localmods) | |
432 |
|
432 | |||
433 | # Whether a local/non-stdlib import has been performed. |
|
433 | # Whether a local/non-stdlib import has been performed. | |
434 | seenlocal = None |
|
434 | seenlocal = None | |
435 | # Whether a local/non-stdlib, non-symbol import has been seen. |
|
435 | # Whether a local/non-stdlib, non-symbol import has been seen. | |
436 | seennonsymbollocal = False |
|
436 | seennonsymbollocal = False | |
437 | # The last name to be imported (for sorting). |
|
437 | # The last name to be imported (for sorting). | |
438 | lastname = None |
|
438 | lastname = None | |
439 | laststdlib = None |
|
439 | laststdlib = None | |
440 | # Relative import levels encountered so far. |
|
440 | # Relative import levels encountered so far. | |
441 | seenlevels = set() |
|
441 | seenlevels = set() | |
442 |
|
442 | |||
443 | for node, newscope in walklocal(root): |
|
443 | for node, newscope in walklocal(root): | |
444 | def msg(fmt, *args): |
|
444 | def msg(fmt, *args): | |
445 | return (fmt % args, node.lineno) |
|
445 | return (fmt % args, node.lineno) | |
446 | if newscope: |
|
446 | if newscope: | |
447 | # Check for local imports in function |
|
447 | # Check for local imports in function | |
448 | for r in verify_modern_convention(module, node, localmods, |
|
448 | for r in verify_modern_convention(module, node, localmods, | |
449 | node.col_offset + 4): |
|
449 | node.col_offset + 4): | |
450 | yield r |
|
450 | yield r | |
451 | elif isinstance(node, ast.Import): |
|
451 | elif isinstance(node, ast.Import): | |
452 | # Disallow "import foo, bar" and require separate imports |
|
452 | # Disallow "import foo, bar" and require separate imports | |
453 | # for each module. |
|
453 | # for each module. | |
454 | if len(node.names) > 1: |
|
454 | if len(node.names) > 1: | |
455 | yield msg('multiple imported names: %s', |
|
455 | yield msg('multiple imported names: %s', | |
456 | ', '.join(n.name for n in node.names)) |
|
456 | ', '.join(n.name for n in node.names)) | |
457 |
|
457 | |||
458 | name = node.names[0].name |
|
458 | name = node.names[0].name | |
459 | asname = node.names[0].asname |
|
459 | asname = node.names[0].asname | |
460 |
|
460 | |||
461 | stdlib = name in stdlib_modules |
|
461 | stdlib = name in stdlib_modules | |
462 |
|
462 | |||
463 | # Ignore sorting rules on imports inside blocks. |
|
463 | # Ignore sorting rules on imports inside blocks. | |
464 | if node.col_offset == root_col_offset: |
|
464 | if node.col_offset == root_col_offset: | |
465 | if lastname and name < lastname and laststdlib == stdlib: |
|
465 | if lastname and name < lastname and laststdlib == stdlib: | |
466 | yield msg('imports not lexically sorted: %s < %s', |
|
466 | yield msg('imports not lexically sorted: %s < %s', | |
467 | name, lastname) |
|
467 | name, lastname) | |
468 |
|
468 | |||
469 | lastname = name |
|
469 | lastname = name | |
470 | laststdlib = stdlib |
|
470 | laststdlib = stdlib | |
471 |
|
471 | |||
472 | # stdlib imports should be before local imports. |
|
472 | # stdlib imports should be before local imports. | |
473 | if stdlib and seenlocal and node.col_offset == root_col_offset: |
|
473 | if stdlib and seenlocal and node.col_offset == root_col_offset: | |
474 | yield msg('stdlib import "%s" follows local import: %s', |
|
474 | yield msg('stdlib import "%s" follows local import: %s', | |
475 | name, seenlocal) |
|
475 | name, seenlocal) | |
476 |
|
476 | |||
477 | if not stdlib: |
|
477 | if not stdlib: | |
478 | seenlocal = name |
|
478 | seenlocal = name | |
479 |
|
479 | |||
480 | # Import of sibling modules should use relative imports. |
|
480 | # Import of sibling modules should use relative imports. | |
481 | topname = name.split('.')[0] |
|
481 | topname = name.split('.')[0] | |
482 | if topname == topmodule: |
|
482 | if topname == topmodule: | |
483 | yield msg('import should be relative: %s', name) |
|
483 | yield msg('import should be relative: %s', name) | |
484 |
|
484 | |||
485 | if name in requirealias and asname != requirealias[name]: |
|
485 | if name in requirealias and asname != requirealias[name]: | |
486 | yield msg('%s module must be "as" aliased to %s', |
|
486 | yield msg('%s module must be "as" aliased to %s', | |
487 | name, requirealias[name]) |
|
487 | name, requirealias[name]) | |
488 |
|
488 | |||
489 | elif isinstance(node, ast.ImportFrom): |
|
489 | elif isinstance(node, ast.ImportFrom): | |
490 | # Resolve the full imported module name. |
|
490 | # Resolve the full imported module name. | |
491 | if node.level > 0: |
|
491 | if node.level > 0: | |
492 | fullname = '.'.join(module.split('.')[:-node.level]) |
|
492 | fullname = '.'.join(module.split('.')[:-node.level]) | |
493 | if node.module: |
|
493 | if node.module: | |
494 | fullname += '.%s' % node.module |
|
494 | fullname += '.%s' % node.module | |
495 | else: |
|
495 | else: | |
496 | assert node.module |
|
496 | assert node.module | |
497 | fullname = node.module |
|
497 | fullname = node.module | |
498 |
|
498 | |||
499 | topname = fullname.split('.')[0] |
|
499 | topname = fullname.split('.')[0] | |
500 | if topname == topmodule: |
|
500 | if topname == topmodule: | |
501 | yield msg('import should be relative: %s', fullname) |
|
501 | yield msg('import should be relative: %s', fullname) | |
502 |
|
502 | |||
503 | # __future__ is special since it needs to come first and use |
|
503 | # __future__ is special since it needs to come first and use | |
504 | # symbol import. |
|
504 | # symbol import. | |
505 | if fullname != '__future__': |
|
505 | if fullname != '__future__': | |
506 | if not fullname or ( |
|
506 | if not fullname or ( | |
507 | fullname in stdlib_modules |
|
507 | fullname in stdlib_modules | |
508 | and fullname not in localmods |
|
508 | and fullname not in localmods | |
509 | and fullname + '.__init__' not in localmods): |
|
509 | and fullname + '.__init__' not in localmods): | |
510 | yield msg('relative import of stdlib module') |
|
510 | yield msg('relative import of stdlib module') | |
511 | else: |
|
511 | else: | |
512 | seenlocal = fullname |
|
512 | seenlocal = fullname | |
513 |
|
513 | |||
514 | # Direct symbol import is only allowed from certain modules and |
|
514 | # Direct symbol import is only allowed from certain modules and | |
515 | # must occur before non-symbol imports. |
|
515 | # must occur before non-symbol imports. | |
516 | found = fromlocal(node.module, node.level) |
|
516 | found = fromlocal(node.module, node.level) | |
517 | if found and found[2]: # node.module is a package |
|
517 | if found and found[2]: # node.module is a package | |
518 | prefix = found[0] + '.' |
|
518 | prefix = found[0] + '.' | |
519 | symbols = (n.name for n in node.names |
|
519 | symbols = (n.name for n in node.names | |
520 | if not fromlocal(prefix + n.name)) |
|
520 | if not fromlocal(prefix + n.name)) | |
521 | else: |
|
521 | else: | |
522 | symbols = (n.name for n in node.names) |
|
522 | symbols = (n.name for n in node.names) | |
523 | symbols = [sym for sym in symbols if sym not in directsymbols] |
|
523 | symbols = [sym for sym in symbols if sym not in directsymbols] | |
524 | if node.module and node.col_offset == root_col_offset: |
|
524 | if node.module and node.col_offset == root_col_offset: | |
525 | if symbols and fullname not in allowsymbolimports: |
|
525 | if symbols and fullname not in allowsymbolimports: | |
526 | yield msg('direct symbol import %s from %s', |
|
526 | yield msg('direct symbol import %s from %s', | |
527 | ', '.join(symbols), fullname) |
|
527 | ', '.join(symbols), fullname) | |
528 |
|
528 | |||
529 | if symbols and seennonsymbollocal: |
|
529 | if symbols and seennonsymbollocal: | |
530 | yield msg('symbol import follows non-symbol import: %s', |
|
530 | yield msg('symbol import follows non-symbol import: %s', | |
531 | fullname) |
|
531 | fullname) | |
532 | if not symbols and fullname not in stdlib_modules: |
|
532 | if not symbols and fullname not in stdlib_modules: | |
533 | seennonsymbollocal = True |
|
533 | seennonsymbollocal = True | |
534 |
|
534 | |||
535 | if not node.module: |
|
535 | if not node.module: | |
536 | assert node.level |
|
536 | assert node.level | |
537 |
|
537 | |||
538 | # Only allow 1 group per level. |
|
538 | # Only allow 1 group per level. | |
539 | if (node.level in seenlevels |
|
539 | if (node.level in seenlevels | |
540 | and node.col_offset == root_col_offset): |
|
540 | and node.col_offset == root_col_offset): | |
541 | yield msg('multiple "from %s import" statements', |
|
541 | yield msg('multiple "from %s import" statements', | |
542 | '.' * node.level) |
|
542 | '.' * node.level) | |
543 |
|
543 | |||
544 | # Higher-level groups come before lower-level groups. |
|
544 | # Higher-level groups come before lower-level groups. | |
545 | if any(node.level > l for l in seenlevels): |
|
545 | if any(node.level > l for l in seenlevels): | |
546 | yield msg('higher-level import should come first: %s', |
|
546 | yield msg('higher-level import should come first: %s', | |
547 | fullname) |
|
547 | fullname) | |
548 |
|
548 | |||
549 | seenlevels.add(node.level) |
|
549 | seenlevels.add(node.level) | |
550 |
|
550 | |||
551 | # Entries in "from .X import ( ... )" lists must be lexically |
|
551 | # Entries in "from .X import ( ... )" lists must be lexically | |
552 | # sorted. |
|
552 | # sorted. | |
553 | lastentryname = None |
|
553 | lastentryname = None | |
554 |
|
554 | |||
555 | for n in node.names: |
|
555 | for n in node.names: | |
556 | if lastentryname and n.name < lastentryname: |
|
556 | if lastentryname and n.name < lastentryname: | |
557 | yield msg('imports from %s not lexically sorted: %s < %s', |
|
557 | yield msg('imports from %s not lexically sorted: %s < %s', | |
558 | fullname, n.name, lastentryname) |
|
558 | fullname, n.name, lastentryname) | |
559 |
|
559 | |||
560 | lastentryname = n.name |
|
560 | lastentryname = n.name | |
561 |
|
561 | |||
562 | if n.name in requirealias and n.asname != requirealias[n.name]: |
|
562 | if n.name in requirealias and n.asname != requirealias[n.name]: | |
563 | yield msg('%s from %s must be "as" aliased to %s', |
|
563 | yield msg('%s from %s must be "as" aliased to %s', | |
564 | n.name, fullname, requirealias[n.name]) |
|
564 | n.name, fullname, requirealias[n.name]) | |
565 |
|
565 | |||
566 | def verify_stdlib_on_own_line(root): |
|
566 | def verify_stdlib_on_own_line(root): | |
567 | """Given some python source, verify that stdlib imports are done |
|
567 | """Given some python source, verify that stdlib imports are done | |
568 | in separate statements from relative local module imports. |
|
568 | in separate statements from relative local module imports. | |
569 |
|
569 | |||
570 | >>> list(verify_stdlib_on_own_line(ast.parse('import sys, foo'))) |
|
570 | >>> list(verify_stdlib_on_own_line(ast.parse('import sys, foo'))) | |
571 | [('mixed imports\\n stdlib: sys\\n relative: foo', 1)] |
|
571 | [('mixed imports\\n stdlib: sys\\n relative: foo', 1)] | |
572 | >>> list(verify_stdlib_on_own_line(ast.parse('import sys, os'))) |
|
572 | >>> list(verify_stdlib_on_own_line(ast.parse('import sys, os'))) | |
573 | [] |
|
573 | [] | |
574 | >>> list(verify_stdlib_on_own_line(ast.parse('import foo, bar'))) |
|
574 | >>> list(verify_stdlib_on_own_line(ast.parse('import foo, bar'))) | |
575 | [] |
|
575 | [] | |
576 | """ |
|
576 | """ | |
577 | for node in ast.walk(root): |
|
577 | for node in ast.walk(root): | |
578 | if isinstance(node, ast.Import): |
|
578 | if isinstance(node, ast.Import): | |
579 | from_stdlib = {False: [], True: []} |
|
579 | from_stdlib = {False: [], True: []} | |
580 | for n in node.names: |
|
580 | for n in node.names: | |
581 | from_stdlib[n.name in stdlib_modules].append(n.name) |
|
581 | from_stdlib[n.name in stdlib_modules].append(n.name) | |
582 | if from_stdlib[True] and from_stdlib[False]: |
|
582 | if from_stdlib[True] and from_stdlib[False]: | |
583 | yield ('mixed imports\n stdlib: %s\n relative: %s' % |
|
583 | yield ('mixed imports\n stdlib: %s\n relative: %s' % | |
584 | (', '.join(sorted(from_stdlib[True])), |
|
584 | (', '.join(sorted(from_stdlib[True])), | |
585 | ', '.join(sorted(from_stdlib[False]))), node.lineno) |
|
585 | ', '.join(sorted(from_stdlib[False]))), node.lineno) | |
586 |
|
586 | |||
587 | class CircularImport(Exception): |
|
587 | class CircularImport(Exception): | |
588 | pass |
|
588 | pass | |
589 |
|
589 | |||
590 | def checkmod(mod, imports): |
|
590 | def checkmod(mod, imports): | |
591 | shortest = {} |
|
591 | shortest = {} | |
592 | visit = [[mod]] |
|
592 | visit = [[mod]] | |
593 | while visit: |
|
593 | while visit: | |
594 | path = visit.pop(0) |
|
594 | path = visit.pop(0) | |
595 | for i in sorted(imports.get(path[-1], [])): |
|
595 | for i in sorted(imports.get(path[-1], [])): | |
596 | if len(path) < shortest.get(i, 1000): |
|
596 | if len(path) < shortest.get(i, 1000): | |
597 | shortest[i] = len(path) |
|
597 | shortest[i] = len(path) | |
598 | if i in path: |
|
598 | if i in path: | |
599 | if i == path[0]: |
|
599 | if i == path[0]: | |
600 | raise CircularImport(path) |
|
600 | raise CircularImport(path) | |
601 | continue |
|
601 | continue | |
602 | visit.append(path + [i]) |
|
602 | visit.append(path + [i]) | |
603 |
|
603 | |||
604 | def rotatecycle(cycle): |
|
604 | def rotatecycle(cycle): | |
605 | """arrange a cycle so that the lexicographically first module listed first |
|
605 | """arrange a cycle so that the lexicographically first module listed first | |
606 |
|
606 | |||
607 | >>> rotatecycle(['foo', 'bar']) |
|
607 | >>> rotatecycle(['foo', 'bar']) | |
608 | ['bar', 'foo', 'bar'] |
|
608 | ['bar', 'foo', 'bar'] | |
609 | """ |
|
609 | """ | |
610 | lowest = min(cycle) |
|
610 | lowest = min(cycle) | |
611 | idx = cycle.index(lowest) |
|
611 | idx = cycle.index(lowest) | |
612 | return cycle[idx:] + cycle[:idx] + [lowest] |
|
612 | return cycle[idx:] + cycle[:idx] + [lowest] | |
613 |
|
613 | |||
614 | def find_cycles(imports): |
|
614 | def find_cycles(imports): | |
615 | """Find cycles in an already-loaded import graph. |
|
615 | """Find cycles in an already-loaded import graph. | |
616 |
|
616 | |||
617 | All module names recorded in `imports` should be absolute one. |
|
617 | All module names recorded in `imports` should be absolute one. | |
618 |
|
618 | |||
619 | >>> from __future__ import print_function |
|
619 | >>> from __future__ import print_function | |
620 | >>> imports = {'top.foo': ['top.bar', 'os.path', 'top.qux'], |
|
620 | >>> imports = {'top.foo': ['top.bar', 'os.path', 'top.qux'], | |
621 | ... 'top.bar': ['top.baz', 'sys'], |
|
621 | ... 'top.bar': ['top.baz', 'sys'], | |
622 | ... 'top.baz': ['top.foo'], |
|
622 | ... 'top.baz': ['top.foo'], | |
623 | ... 'top.qux': ['top.foo']} |
|
623 | ... 'top.qux': ['top.foo']} | |
624 | >>> print('\\n'.join(sorted(find_cycles(imports)))) |
|
624 | >>> print('\\n'.join(sorted(find_cycles(imports)))) | |
625 | top.bar -> top.baz -> top.foo -> top.bar |
|
625 | top.bar -> top.baz -> top.foo -> top.bar | |
626 | top.foo -> top.qux -> top.foo |
|
626 | top.foo -> top.qux -> top.foo | |
627 | """ |
|
627 | """ | |
628 | cycles = set() |
|
628 | cycles = set() | |
629 | for mod in sorted(imports.keys()): |
|
629 | for mod in sorted(imports.keys()): | |
630 | try: |
|
630 | try: | |
631 | checkmod(mod, imports) |
|
631 | checkmod(mod, imports) | |
632 | except CircularImport as e: |
|
632 | except CircularImport as e: | |
633 | cycle = e.args[0] |
|
633 | cycle = e.args[0] | |
634 | cycles.add(" -> ".join(rotatecycle(cycle))) |
|
634 | cycles.add(" -> ".join(rotatecycle(cycle))) | |
635 | return cycles |
|
635 | return cycles | |
636 |
|
636 | |||
637 | def _cycle_sortkey(c): |
|
637 | def _cycle_sortkey(c): | |
638 | return len(c), c |
|
638 | return len(c), c | |
639 |
|
639 | |||
640 | def embedded(f, modname, src): |
|
640 | def embedded(f, modname, src): | |
641 | """Extract embedded python code |
|
641 | """Extract embedded python code | |
642 |
|
642 | |||
643 | >>> def _forcestr(thing): |
|
643 | >>> def _forcestr(thing): | |
644 | ... if not isinstance(thing, str): |
|
644 | ... if not isinstance(thing, str): | |
645 | ... return thing.decode('ascii') |
|
645 | ... return thing.decode('ascii') | |
646 | ... return thing |
|
646 | ... return thing | |
647 | >>> def test(fn, lines): |
|
647 | >>> def test(fn, lines): | |
648 | ... for s, m, f, l in embedded(fn, b"example", lines): |
|
648 | ... for s, m, f, l in embedded(fn, b"example", lines): | |
649 | ... print("%s %s %d" % (_forcestr(m), _forcestr(f), l)) |
|
649 | ... print("%s %s %d" % (_forcestr(m), _forcestr(f), l)) | |
650 | ... print(repr(_forcestr(s))) |
|
650 | ... print(repr(_forcestr(s))) | |
651 | >>> lines = [ |
|
651 | >>> lines = [ | |
652 |
... |
|
652 | ... 'comment', | |
653 |
... |
|
653 | ... ' >>> from __future__ import print_function', | |
654 |
... |
|
654 | ... " >>> ' multiline", | |
655 |
... |
|
655 | ... " ... string'", | |
656 |
... |
|
656 | ... ' ', | |
657 |
... |
|
657 | ... 'comment', | |
658 |
... |
|
658 | ... ' $ cat > foo.py <<EOF', | |
659 |
... |
|
659 | ... ' > from __future__ import print_function', | |
660 |
... |
|
660 | ... ' > EOF', | |
661 | ... ] |
|
661 | ... ] | |
662 | >>> test(b"example.t", lines) |
|
662 | >>> test(b"example.t", lines) | |
663 | example[2] doctest.py 1 |
|
663 | example[2] doctest.py 1 | |
664 | "from __future__ import print_function\\n' multiline\\nstring'\\n\\n" |
|
664 | "from __future__ import print_function\\n' multiline\\nstring'\\n\\n" | |
665 | example[8] foo.py 7 |
|
665 | example[8] foo.py 7 | |
666 | 'from __future__ import print_function\\n' |
|
666 | 'from __future__ import print_function\\n' | |
667 | """ |
|
667 | """ | |
668 | errors = [] |
|
668 | errors = [] | |
669 | for name, starts, ends, code in testparseutil.pyembedded(f, src, errors): |
|
669 | for name, starts, ends, code in testparseutil.pyembedded(f, src, errors): | |
670 | if not name: |
|
670 | if not name: | |
671 | # use 'doctest.py', in order to make already existing |
|
671 | # use 'doctest.py', in order to make already existing | |
672 | # doctest above pass instantly |
|
672 | # doctest above pass instantly | |
673 | name = 'doctest.py' |
|
673 | name = 'doctest.py' | |
674 | # "starts" is "line number" (1-origin), but embedded() is |
|
674 | # "starts" is "line number" (1-origin), but embedded() is | |
675 | # expected to return "line offset" (0-origin). Therefore, this |
|
675 | # expected to return "line offset" (0-origin). Therefore, this | |
676 | # yields "starts - 1". |
|
676 | # yields "starts - 1". | |
677 | if not isinstance(modname, str): |
|
677 | if not isinstance(modname, str): | |
678 | modname = modname.decode('utf8') |
|
678 | modname = modname.decode('utf8') | |
679 | yield code, "%s[%d]" % (modname, starts), name, starts - 1 |
|
679 | yield code, "%s[%d]" % (modname, starts), name, starts - 1 | |
680 |
|
680 | |||
681 | def sources(f, modname): |
|
681 | def sources(f, modname): | |
682 | """Yields possibly multiple sources from a filepath |
|
682 | """Yields possibly multiple sources from a filepath | |
683 |
|
683 | |||
684 | input: filepath, modulename |
|
684 | input: filepath, modulename | |
685 | yields: script(string), modulename, filepath, linenumber |
|
685 | yields: script(string), modulename, filepath, linenumber | |
686 |
|
686 | |||
687 | For embedded scripts, the modulename and filepath will be different |
|
687 | For embedded scripts, the modulename and filepath will be different | |
688 | from the function arguments. linenumber is an offset relative to |
|
688 | from the function arguments. linenumber is an offset relative to | |
689 | the input file. |
|
689 | the input file. | |
690 | """ |
|
690 | """ | |
691 | py = False |
|
691 | py = False | |
692 | if not f.endswith('.t'): |
|
692 | if not f.endswith('.t'): | |
693 | with open(f, 'rb') as src: |
|
693 | with open(f, 'rb') as src: | |
694 | yield src.read(), modname, f, 0 |
|
694 | yield src.read(), modname, f, 0 | |
695 | py = True |
|
695 | py = True | |
696 | if py or f.endswith('.t'): |
|
696 | if py or f.endswith('.t'): | |
697 |
with open(f, 'r |
|
697 | with open(f, 'r') as src: | |
698 | for script, modname, t, line in embedded(f, modname, src): |
|
698 | for script, modname, t, line in embedded(f, modname, src): | |
699 | yield script, modname.encode('utf8'), t, line |
|
699 | yield script, modname.encode('utf8'), t, line | |
700 |
|
700 | |||
701 | def main(argv): |
|
701 | def main(argv): | |
702 | if len(argv) < 2 or (argv[1] == '-' and len(argv) > 2): |
|
702 | if len(argv) < 2 or (argv[1] == '-' and len(argv) > 2): | |
703 | print('Usage: %s {-|file [file] [file] ...}') |
|
703 | print('Usage: %s {-|file [file] [file] ...}') | |
704 | return 1 |
|
704 | return 1 | |
705 | if argv[1] == '-': |
|
705 | if argv[1] == '-': | |
706 | argv = argv[:1] |
|
706 | argv = argv[:1] | |
707 | argv.extend(l.rstrip() for l in sys.stdin.readlines()) |
|
707 | argv.extend(l.rstrip() for l in sys.stdin.readlines()) | |
708 | localmodpaths = {} |
|
708 | localmodpaths = {} | |
709 | used_imports = {} |
|
709 | used_imports = {} | |
710 | any_errors = False |
|
710 | any_errors = False | |
711 | for source_path in argv[1:]: |
|
711 | for source_path in argv[1:]: | |
712 | modname = dotted_name_of_path(source_path) |
|
712 | modname = dotted_name_of_path(source_path) | |
713 | localmodpaths[modname] = source_path |
|
713 | localmodpaths[modname] = source_path | |
714 | localmods = populateextmods(localmodpaths) |
|
714 | localmods = populateextmods(localmodpaths) | |
715 | for localmodname, source_path in sorted(localmodpaths.items()): |
|
715 | for localmodname, source_path in sorted(localmodpaths.items()): | |
716 | if not isinstance(localmodname, bytes): |
|
716 | if not isinstance(localmodname, bytes): | |
717 | # This is only safe because all hg's files are ascii |
|
717 | # This is only safe because all hg's files are ascii | |
718 | localmodname = localmodname.encode('ascii') |
|
718 | localmodname = localmodname.encode('ascii') | |
719 | for src, modname, name, line in sources(source_path, localmodname): |
|
719 | for src, modname, name, line in sources(source_path, localmodname): | |
720 | try: |
|
720 | try: | |
721 | used_imports[modname] = sorted( |
|
721 | used_imports[modname] = sorted( | |
722 | imported_modules(src, modname, name, localmods, |
|
722 | imported_modules(src, modname, name, localmods, | |
723 | ignore_nested=True)) |
|
723 | ignore_nested=True)) | |
724 | for error, lineno in verify_import_convention(modname, src, |
|
724 | for error, lineno in verify_import_convention(modname, src, | |
725 | localmods): |
|
725 | localmods): | |
726 | any_errors = True |
|
726 | any_errors = True | |
727 | print('%s:%d: %s' % (source_path, lineno + line, error)) |
|
727 | print('%s:%d: %s' % (source_path, lineno + line, error)) | |
728 | except SyntaxError as e: |
|
728 | except SyntaxError as e: | |
729 | print('%s:%d: SyntaxError: %s' % |
|
729 | print('%s:%d: SyntaxError: %s' % | |
730 | (source_path, e.lineno + line, e)) |
|
730 | (source_path, e.lineno + line, e)) | |
731 | cycles = find_cycles(used_imports) |
|
731 | cycles = find_cycles(used_imports) | |
732 | if cycles: |
|
732 | if cycles: | |
733 | firstmods = set() |
|
733 | firstmods = set() | |
734 | for c in sorted(cycles, key=_cycle_sortkey): |
|
734 | for c in sorted(cycles, key=_cycle_sortkey): | |
735 | first = c.split()[0] |
|
735 | first = c.split()[0] | |
736 | # As a rough cut, ignore any cycle that starts with the |
|
736 | # As a rough cut, ignore any cycle that starts with the | |
737 | # same module as some other cycle. Otherwise we see lots |
|
737 | # same module as some other cycle. Otherwise we see lots | |
738 | # of cycles that are effectively duplicates. |
|
738 | # of cycles that are effectively duplicates. | |
739 | if first in firstmods: |
|
739 | if first in firstmods: | |
740 | continue |
|
740 | continue | |
741 | print('Import cycle:', c) |
|
741 | print('Import cycle:', c) | |
742 | firstmods.add(first) |
|
742 | firstmods.add(first) | |
743 | any_errors = True |
|
743 | any_errors = True | |
744 | return any_errors != 0 |
|
744 | return any_errors != 0 | |
745 |
|
745 | |||
746 | if __name__ == '__main__': |
|
746 | if __name__ == '__main__': | |
747 | sys.exit(int(main(sys.argv))) |
|
747 | sys.exit(int(main(sys.argv))) |
General Comments 0
You need to be logged in to leave comments.
Login now