Show More
@@ -1,466 +1,426 b'' | |||
|
1 | 1 | # pycompat.py - portability shim for python 3 |
|
2 | 2 | # |
|
3 | 3 | # This software may be used and distributed according to the terms of the |
|
4 | 4 | # GNU General Public License version 2 or any later version. |
|
5 | 5 | |
|
6 | 6 | """Mercurial portability shim for python 3. |
|
7 | 7 | |
|
8 | 8 | This contains aliases to hide python version-specific details from the core. |
|
9 | 9 | """ |
|
10 | 10 | |
|
11 | 11 | |
|
12 | 12 | import builtins |
|
13 | 13 | import codecs |
|
14 | 14 | import concurrent.futures as futures |
|
15 | 15 | import functools |
|
16 | 16 | import getopt |
|
17 | 17 | import http.client as httplib |
|
18 | 18 | import http.cookiejar as cookielib |
|
19 | 19 | import inspect |
|
20 | 20 | import io |
|
21 | 21 | import json |
|
22 | 22 | import os |
|
23 | 23 | import pickle |
|
24 | 24 | import queue |
|
25 | 25 | import shlex |
|
26 | 26 | import socketserver |
|
27 | 27 | import struct |
|
28 | 28 | import sys |
|
29 | 29 | import tempfile |
|
30 | 30 | import xmlrpc.client as xmlrpclib |
|
31 | 31 | |
|
32 | 32 | |
|
33 | 33 | ispy3 = sys.version_info[0] >= 3 |
|
34 | 34 | ispypy = '__pypy__' in sys.builtin_module_names |
|
35 | 35 | TYPE_CHECKING = False |
|
36 | 36 | |
|
37 | 37 | if not globals(): # hide this from non-pytype users |
|
38 | 38 | import typing |
|
39 | 39 | |
|
40 | 40 | TYPE_CHECKING = typing.TYPE_CHECKING |
|
41 | 41 | |
|
42 | 42 | |
|
43 | 43 | def future_set_exception_info(f, exc_info): |
|
44 | 44 | f.set_exception(exc_info[0]) |
|
45 | 45 | |
|
46 | 46 | |
|
47 | 47 | FileNotFoundError = builtins.FileNotFoundError |
|
48 | 48 | |
|
49 | 49 | |
|
50 | 50 | def identity(a): |
|
51 | 51 | return a |
|
52 | 52 | |
|
53 | 53 | |
|
54 | 54 | def _rapply(f, xs): |
|
55 | 55 | if xs is None: |
|
56 | 56 | # assume None means non-value of optional data |
|
57 | 57 | return xs |
|
58 | 58 | if isinstance(xs, (list, set, tuple)): |
|
59 | 59 | return type(xs)(_rapply(f, x) for x in xs) |
|
60 | 60 | if isinstance(xs, dict): |
|
61 | 61 | return type(xs)((_rapply(f, k), _rapply(f, v)) for k, v in xs.items()) |
|
62 | 62 | return f(xs) |
|
63 | 63 | |
|
64 | 64 | |
|
65 | 65 | def rapply(f, xs): |
|
66 | 66 | """Apply function recursively to every item preserving the data structure |
|
67 | 67 | |
|
68 | 68 | >>> def f(x): |
|
69 | 69 | ... return 'f(%s)' % x |
|
70 | 70 | >>> rapply(f, None) is None |
|
71 | 71 | True |
|
72 | 72 | >>> rapply(f, 'a') |
|
73 | 73 | 'f(a)' |
|
74 | 74 | >>> rapply(f, {'a'}) == {'f(a)'} |
|
75 | 75 | True |
|
76 | 76 | >>> rapply(f, ['a', 'b', None, {'c': 'd'}, []]) |
|
77 | 77 | ['f(a)', 'f(b)', None, {'f(c)': 'f(d)'}, []] |
|
78 | 78 | |
|
79 | 79 | >>> xs = [object()] |
|
80 | 80 | >>> rapply(identity, xs) is xs |
|
81 | 81 | True |
|
82 | 82 | """ |
|
83 | 83 | if f is identity: |
|
84 | 84 | # fast path mainly for py2 |
|
85 | 85 | return xs |
|
86 | 86 | return _rapply(f, xs) |
|
87 | 87 | |
|
88 | 88 | |
|
89 | 89 | if os.name == r'nt': |
|
90 | 90 | # MBCS (or ANSI) filesystem encoding must be used as before. |
|
91 | 91 | # Otherwise non-ASCII filenames in existing repositories would be |
|
92 | 92 | # corrupted. |
|
93 | 93 | # This must be set once prior to any fsencode/fsdecode calls. |
|
94 | 94 | sys._enablelegacywindowsfsencoding() # pytype: disable=module-attr |
|
95 | 95 | |
|
96 | 96 | fsencode = os.fsencode |
|
97 | 97 | fsdecode = os.fsdecode |
|
98 | 98 | oscurdir = os.curdir.encode('ascii') |
|
99 | 99 | oslinesep = os.linesep.encode('ascii') |
|
100 | 100 | osname = os.name.encode('ascii') |
|
101 | 101 | ospathsep = os.pathsep.encode('ascii') |
|
102 | 102 | ospardir = os.pardir.encode('ascii') |
|
103 | 103 | ossep = os.sep.encode('ascii') |
|
104 | 104 | osaltsep = os.altsep |
|
105 | 105 | if osaltsep: |
|
106 | 106 | osaltsep = osaltsep.encode('ascii') |
|
107 | 107 | osdevnull = os.devnull.encode('ascii') |
|
108 | 108 | |
|
109 | 109 | sysplatform = sys.platform.encode('ascii') |
|
110 | 110 | sysexecutable = sys.executable |
|
111 | 111 | if sysexecutable: |
|
112 | 112 | sysexecutable = os.fsencode(sysexecutable) |
|
113 | 113 | |
|
114 | 114 | |
|
115 | 115 | def maplist(*args): |
|
116 | 116 | return list(map(*args)) |
|
117 | 117 | |
|
118 | 118 | |
|
119 | 119 | def rangelist(*args): |
|
120 | 120 | return list(range(*args)) |
|
121 | 121 | |
|
122 | 122 | |
|
123 | 123 | def ziplist(*args): |
|
124 | 124 | return list(zip(*args)) |
|
125 | 125 | |
|
126 | 126 | |
|
127 | 127 | rawinput = input |
|
128 | 128 | getargspec = inspect.getfullargspec |
|
129 | 129 | |
|
130 | 130 | long = int |
|
131 | 131 | |
|
132 | 132 | if getattr(sys, 'argv', None) is not None: |
|
133 | 133 | # On POSIX, the char** argv array is converted to Python str using |
|
134 | 134 | # Py_DecodeLocale(). The inverse of this is Py_EncodeLocale(), which |
|
135 | 135 | # isn't directly callable from Python code. In practice, os.fsencode() |
|
136 | 136 | # can be used instead (this is recommended by Python's documentation |
|
137 | 137 | # for sys.argv). |
|
138 | 138 | # |
|
139 | 139 | # On Windows, the wchar_t **argv is passed into the interpreter as-is. |
|
140 | 140 | # Like POSIX, we need to emulate what Py_EncodeLocale() would do. But |
|
141 | 141 | # there's an additional wrinkle. What we really want to access is the |
|
142 | 142 | # ANSI codepage representation of the arguments, as this is what |
|
143 | 143 | # `int main()` would receive if Python 3 didn't define `int wmain()` |
|
144 | 144 | # (this is how Python 2 worked). To get that, we encode with the mbcs |
|
145 | 145 | # encoding, which will pass CP_ACP to the underlying Windows API to |
|
146 | 146 | # produce bytes. |
|
147 | 147 | if os.name == r'nt': |
|
148 | 148 | sysargv = [a.encode("mbcs", "ignore") for a in sys.argv] |
|
149 | 149 | else: |
|
150 | 150 | sysargv = [fsencode(a) for a in sys.argv] |
|
151 | 151 | |
|
152 | 152 | bytechr = struct.Struct('>B').pack |
|
153 | 153 | byterepr = b'%r'.__mod__ |
|
154 | 154 | |
|
155 | 155 | |
|
156 | 156 | class bytestr(bytes): |
|
157 | 157 | """A bytes which mostly acts as a Python 2 str |
|
158 | 158 | |
|
159 | 159 | >>> bytestr(), bytestr(bytearray(b'foo')), bytestr(u'ascii'), bytestr(1) |
|
160 | 160 | ('', 'foo', 'ascii', '1') |
|
161 | 161 | >>> s = bytestr(b'foo') |
|
162 | 162 | >>> assert s is bytestr(s) |
|
163 | 163 | |
|
164 | 164 | __bytes__() should be called if provided: |
|
165 | 165 | |
|
166 | 166 | >>> class bytesable: |
|
167 | 167 | ... def __bytes__(self): |
|
168 | 168 | ... return b'bytes' |
|
169 | 169 | >>> bytestr(bytesable()) |
|
170 | 170 | 'bytes' |
|
171 | 171 | |
|
172 | 172 | There's no implicit conversion from non-ascii str as its encoding is |
|
173 | 173 | unknown: |
|
174 | 174 | |
|
175 | 175 | >>> bytestr(chr(0x80)) # doctest: +ELLIPSIS |
|
176 | 176 | Traceback (most recent call last): |
|
177 | 177 | ... |
|
178 | 178 | UnicodeEncodeError: ... |
|
179 | 179 | |
|
180 | 180 | Comparison between bytestr and bytes should work: |
|
181 | 181 | |
|
182 | 182 | >>> assert bytestr(b'foo') == b'foo' |
|
183 | 183 | >>> assert b'foo' == bytestr(b'foo') |
|
184 | 184 | >>> assert b'f' in bytestr(b'foo') |
|
185 | 185 | >>> assert bytestr(b'f') in b'foo' |
|
186 | 186 | |
|
187 | 187 | Sliced elements should be bytes, not integer: |
|
188 | 188 | |
|
189 | 189 | >>> s[1], s[:2] |
|
190 | 190 | (b'o', b'fo') |
|
191 | 191 | >>> list(s), list(reversed(s)) |
|
192 | 192 | ([b'f', b'o', b'o'], [b'o', b'o', b'f']) |
|
193 | 193 | |
|
194 | 194 | As bytestr type isn't propagated across operations, you need to cast |
|
195 | 195 | bytes to bytestr explicitly: |
|
196 | 196 | |
|
197 | 197 | >>> s = bytestr(b'foo').upper() |
|
198 | 198 | >>> t = bytestr(s) |
|
199 | 199 | >>> s[0], t[0] |
|
200 | 200 | (70, b'F') |
|
201 | 201 | |
|
202 | 202 | Be careful to not pass a bytestr object to a function which expects |
|
203 | 203 | bytearray-like behavior. |
|
204 | 204 | |
|
205 | 205 | >>> t = bytes(t) # cast to bytes |
|
206 | 206 | >>> assert type(t) is bytes |
|
207 | 207 | """ |
|
208 | 208 | |
|
209 | 209 | # Trick pytype into not demanding Iterable[int] be passed to __new__(), |
|
210 | 210 | # since the appropriate bytes format is done internally. |
|
211 | 211 | # |
|
212 | 212 | # https://github.com/google/pytype/issues/500 |
|
213 | 213 | if TYPE_CHECKING: |
|
214 | 214 | |
|
215 | 215 | def __init__(self, s=b''): |
|
216 | 216 | pass |
|
217 | 217 | |
|
218 | 218 | def __new__(cls, s=b''): |
|
219 | 219 | if isinstance(s, bytestr): |
|
220 | 220 | return s |
|
221 | 221 | if not isinstance( |
|
222 | 222 | s, (bytes, bytearray) |
|
223 | 223 | ) and not hasattr( # hasattr-py3-only |
|
224 | 224 | s, u'__bytes__' |
|
225 | 225 | ): |
|
226 | 226 | s = str(s).encode('ascii') |
|
227 | 227 | return bytes.__new__(cls, s) |
|
228 | 228 | |
|
229 | 229 | def __getitem__(self, key): |
|
230 | 230 | s = bytes.__getitem__(self, key) |
|
231 | 231 | if not isinstance(s, bytes): |
|
232 | 232 | s = bytechr(s) |
|
233 | 233 | return s |
|
234 | 234 | |
|
235 | 235 | def __iter__(self): |
|
236 | 236 | return iterbytestr(bytes.__iter__(self)) |
|
237 | 237 | |
|
238 | 238 | def __repr__(self): |
|
239 | 239 | return bytes.__repr__(self)[1:] # drop b'' |
|
240 | 240 | |
|
241 | 241 | |
|
242 | 242 | def iterbytestr(s): |
|
243 | 243 | """Iterate bytes as if it were a str object of Python 2""" |
|
244 | 244 | return map(bytechr, s) |
|
245 | 245 | |
|
246 | 246 | |
|
247 | 247 | def maybebytestr(s): |
|
248 | 248 | """Promote bytes to bytestr""" |
|
249 | 249 | if isinstance(s, bytes): |
|
250 | 250 | return bytestr(s) |
|
251 | 251 | return s |
|
252 | 252 | |
|
253 | 253 | |
|
254 | 254 | def sysbytes(s): |
|
255 | 255 | """Convert an internal str (e.g. keyword, __doc__) back to bytes |
|
256 | 256 | |
|
257 | 257 | This never raises UnicodeEncodeError, but only ASCII characters |
|
258 | 258 | can be round-trip by sysstr(sysbytes(s)). |
|
259 | 259 | """ |
|
260 | 260 | if isinstance(s, bytes): |
|
261 | 261 | return s |
|
262 | 262 | return s.encode('utf-8') |
|
263 | 263 | |
|
264 | 264 | |
|
265 | 265 | def sysstr(s): |
|
266 | 266 | """Return a keyword str to be passed to Python functions such as |
|
267 | 267 | getattr() and str.encode() |
|
268 | 268 | |
|
269 | 269 | This never raises UnicodeDecodeError. Non-ascii characters are |
|
270 | 270 | considered invalid and mapped to arbitrary but unique code points |
|
271 | 271 | such that 'sysstr(a) != sysstr(b)' for all 'a != b'. |
|
272 | 272 | """ |
|
273 | 273 | if isinstance(s, builtins.str): |
|
274 | 274 | return s |
|
275 | 275 | return s.decode('latin-1') |
|
276 | 276 | |
|
277 | 277 | |
|
278 | 278 | def strurl(url): |
|
279 | 279 | """Converts a bytes url back to str""" |
|
280 | 280 | if isinstance(url, bytes): |
|
281 | 281 | return url.decode('ascii') |
|
282 | 282 | return url |
|
283 | 283 | |
|
284 | 284 | |
|
285 | 285 | def bytesurl(url): |
|
286 | 286 | """Converts a str url to bytes by encoding in ascii""" |
|
287 | 287 | if isinstance(url, str): |
|
288 | 288 | return url.encode('ascii') |
|
289 | 289 | return url |
|
290 | 290 | |
|
291 | 291 | |
|
292 | 292 | def raisewithtb(exc, tb): |
|
293 | 293 | """Raise exception with the given traceback""" |
|
294 | 294 | raise exc.with_traceback(tb) |
|
295 | 295 | |
|
296 | 296 | |
|
297 | 297 | def getdoc(obj): |
|
298 | 298 | """Get docstring as bytes; may be None so gettext() won't confuse it |
|
299 | 299 | with _('')""" |
|
300 | 300 | doc = getattr(obj, '__doc__', None) |
|
301 | 301 | if doc is None: |
|
302 | 302 | return doc |
|
303 | 303 | return sysbytes(doc) |
|
304 | 304 | |
|
305 | 305 | |
|
306 | 306 | def _wrapattrfunc(f): |
|
307 | 307 | @functools.wraps(f) |
|
308 | 308 | def w(object, name, *args): |
|
309 | 309 | return f(object, sysstr(name), *args) |
|
310 | 310 | |
|
311 | 311 | return w |
|
312 | 312 | |
|
313 | 313 | |
|
314 | 314 | # these wrappers are automagically imported by hgloader |
|
315 | 315 | delattr = _wrapattrfunc(builtins.delattr) |
|
316 | 316 | getattr = _wrapattrfunc(builtins.getattr) |
|
317 | 317 | hasattr = _wrapattrfunc(builtins.hasattr) |
|
318 | 318 | setattr = _wrapattrfunc(builtins.setattr) |
|
319 | 319 | xrange = builtins.range |
|
320 | 320 | unicode = str |
|
321 | 321 | |
|
322 | 322 | |
|
323 | 323 | def open(name, mode=b'r', buffering=-1, encoding=None): |
|
324 | 324 | return builtins.open(name, sysstr(mode), buffering, encoding) |
|
325 | 325 | |
|
326 | 326 | |
|
327 | 327 | safehasattr = _wrapattrfunc(builtins.hasattr) |
|
328 | 328 | |
|
329 | 329 | |
|
330 | 330 | def _getoptbwrapper(orig, args, shortlist, namelist): |
|
331 | 331 | """ |
|
332 | 332 | Takes bytes arguments, converts them to unicode, pass them to |
|
333 | 333 | getopt.getopt(), convert the returned values back to bytes and then |
|
334 | 334 | return them for Python 3 compatibility as getopt.getopt() don't accepts |
|
335 | 335 | bytes on Python 3. |
|
336 | 336 | """ |
|
337 | 337 | args = [a.decode('latin-1') for a in args] |
|
338 | 338 | shortlist = shortlist.decode('latin-1') |
|
339 | 339 | namelist = [a.decode('latin-1') for a in namelist] |
|
340 | 340 | opts, args = orig(args, shortlist, namelist) |
|
341 | 341 | opts = [(a[0].encode('latin-1'), a[1].encode('latin-1')) for a in opts] |
|
342 | 342 | args = [a.encode('latin-1') for a in args] |
|
343 | 343 | return opts, args |
|
344 | 344 | |
|
345 | 345 | |
|
346 | 346 | def strkwargs(dic): |
|
347 | 347 | """ |
|
348 | 348 | Converts the keys of a python dictonary to str i.e. unicodes so that |
|
349 | 349 | they can be passed as keyword arguments as dictionaries with bytes keys |
|
350 | 350 | can't be passed as keyword arguments to functions on Python 3. |
|
351 | 351 | """ |
|
352 | 352 | dic = {k.decode('latin-1'): v for k, v in dic.items()} |
|
353 | 353 | return dic |
|
354 | 354 | |
|
355 | 355 | |
|
356 | 356 | def byteskwargs(dic): |
|
357 | 357 | """ |
|
358 | 358 | Converts keys of python dictionaries to bytes as they were converted to |
|
359 | 359 | str to pass that dictonary as a keyword argument on Python 3. |
|
360 | 360 | """ |
|
361 | 361 | dic = {k.encode('latin-1'): v for k, v in dic.items()} |
|
362 | 362 | return dic |
|
363 | 363 | |
|
364 | 364 | |
|
365 | 365 | # TODO: handle shlex.shlex(). |
|
366 | 366 | def shlexsplit(s, comments=False, posix=True): |
|
367 | 367 | """ |
|
368 | 368 | Takes bytes argument, convert it to str i.e. unicodes, pass that into |
|
369 | 369 | shlex.split(), convert the returned value to bytes and return that for |
|
370 | 370 | Python 3 compatibility as shelx.split() don't accept bytes on Python 3. |
|
371 | 371 | """ |
|
372 | 372 | ret = shlex.split(s.decode('latin-1'), comments, posix) |
|
373 | 373 | return [a.encode('latin-1') for a in ret] |
|
374 | 374 | |
|
375 | 375 | |
|
376 | 376 | iteritems = lambda x: x.items() |
|
377 | 377 | itervalues = lambda x: x.values() |
|
378 | 378 | |
|
379 | # Python 3.5's json.load and json.loads require str. We polyfill its | |
|
380 | # code for detecting encoding from bytes. | |
|
381 | if sys.version_info[0:2] < (3, 6): | |
|
382 | ||
|
383 | def _detect_encoding(b): | |
|
384 | bstartswith = b.startswith | |
|
385 | if bstartswith((codecs.BOM_UTF32_BE, codecs.BOM_UTF32_LE)): | |
|
386 | return 'utf-32' | |
|
387 | if bstartswith((codecs.BOM_UTF16_BE, codecs.BOM_UTF16_LE)): | |
|
388 | return 'utf-16' | |
|
389 | if bstartswith(codecs.BOM_UTF8): | |
|
390 | return 'utf-8-sig' | |
|
391 | ||
|
392 | if len(b) >= 4: | |
|
393 | if not b[0]: | |
|
394 | # 00 00 -- -- - utf-32-be | |
|
395 | # 00 XX -- -- - utf-16-be | |
|
396 | return 'utf-16-be' if b[1] else 'utf-32-be' | |
|
397 | if not b[1]: | |
|
398 | # XX 00 00 00 - utf-32-le | |
|
399 | # XX 00 00 XX - utf-16-le | |
|
400 | # XX 00 XX -- - utf-16-le | |
|
401 | return 'utf-16-le' if b[2] or b[3] else 'utf-32-le' | |
|
402 | elif len(b) == 2: | |
|
403 | if not b[0]: | |
|
404 | # 00 XX - utf-16-be | |
|
405 | return 'utf-16-be' | |
|
406 | if not b[1]: | |
|
407 | # XX 00 - utf-16-le | |
|
408 | return 'utf-16-le' | |
|
409 | # default | |
|
410 | return 'utf-8' | |
|
411 | ||
|
412 | def json_loads(s, *args, **kwargs): | |
|
413 | if isinstance(s, (bytes, bytearray)): | |
|
414 | s = s.decode(_detect_encoding(s), 'surrogatepass') | |
|
415 | ||
|
416 | return json.loads(s, *args, **kwargs) | |
|
417 | ||
|
418 | else: | |
|
419 | json_loads = json.loads | |
|
379 | json_loads = json.loads | |
|
420 | 380 | |
|
421 | 381 | isjython = sysplatform.startswith(b'java') |
|
422 | 382 | |
|
423 | 383 | isdarwin = sysplatform.startswith(b'darwin') |
|
424 | 384 | islinux = sysplatform.startswith(b'linux') |
|
425 | 385 | isposix = osname == b'posix' |
|
426 | 386 | iswindows = osname == b'nt' |
|
427 | 387 | |
|
428 | 388 | |
|
429 | 389 | def getoptb(args, shortlist, namelist): |
|
430 | 390 | return _getoptbwrapper(getopt.getopt, args, shortlist, namelist) |
|
431 | 391 | |
|
432 | 392 | |
|
433 | 393 | def gnugetoptb(args, shortlist, namelist): |
|
434 | 394 | return _getoptbwrapper(getopt.gnu_getopt, args, shortlist, namelist) |
|
435 | 395 | |
|
436 | 396 | |
|
437 | 397 | def mkdtemp(suffix=b'', prefix=b'tmp', dir=None): |
|
438 | 398 | return tempfile.mkdtemp(suffix, prefix, dir) |
|
439 | 399 | |
|
440 | 400 | |
|
441 | 401 | # text=True is not supported; use util.from/tonativeeol() instead |
|
442 | 402 | def mkstemp(suffix=b'', prefix=b'tmp', dir=None): |
|
443 | 403 | return tempfile.mkstemp(suffix, prefix, dir) |
|
444 | 404 | |
|
445 | 405 | |
|
446 | 406 | # TemporaryFile does not support an "encoding=" argument on python2. |
|
447 | 407 | # This wrapper file are always open in byte mode. |
|
448 | 408 | def unnamedtempfile(mode=None, *args, **kwargs): |
|
449 | 409 | if mode is None: |
|
450 | 410 | mode = 'w+b' |
|
451 | 411 | else: |
|
452 | 412 | mode = sysstr(mode) |
|
453 | 413 | assert 'b' in mode |
|
454 | 414 | return tempfile.TemporaryFile(mode, *args, **kwargs) |
|
455 | 415 | |
|
456 | 416 | |
|
457 | 417 | # NamedTemporaryFile does not support an "encoding=" argument on python2. |
|
458 | 418 | # This wrapper file are always open in byte mode. |
|
459 | 419 | def namedtempfile( |
|
460 | 420 | mode=b'w+b', bufsize=-1, suffix=b'', prefix=b'tmp', dir=None, delete=True |
|
461 | 421 | ): |
|
462 | 422 | mode = sysstr(mode) |
|
463 | 423 | assert 'b' in mode |
|
464 | 424 | return tempfile.NamedTemporaryFile( |
|
465 | 425 | mode, bufsize, suffix=suffix, prefix=prefix, dir=dir, delete=delete |
|
466 | 426 | ) |
General Comments 0
You need to be logged in to leave comments.
Login now