OILS / frontend / option_def.py View on Github | oilshell.org

410 lines, 205 significant
1#!/usr/bin/env python2
2from __future__ import print_function
3
4from typing import List, Dict, Optional, Any
5
6
7class Option(object):
8
9 def __init__(self,
10 index,
11 name,
12 short_flag=None,
13 builtin='shopt',
14 default=False,
15 implemented=True,
16 groups=None):
17 # type: (int, str, str, Optional[str], bool, bool, List[str]) -> None
18 self.index = index
19 self.name = name # e.g. 'errexit'
20 self.short_flag = short_flag # 'e' for -e
21
22 if short_flag:
23 self.builtin = 'set'
24 else:
25 # The 'interactive' option is the only one where builtin is None. It has
26 # a cell but you can't change it. Only the shell can.
27 self.builtin = builtin
28
29 self.default = default # default value is True in some cases
30 self.implemented = implemented
31 self.groups = groups or [] # list of groups
32
33 # for optview
34 self.is_parse = (name.startswith('parse_') or
35 name.startswith('strict_parse_') or
36 name == 'expand_aliases')
37 # interactive() is an accessor
38 self.is_exec = implemented and not self.is_parse
39
40
41class _OptionDef(object):
42 """Description of all shell options.
43
44 Similar to id_kind_def.IdSpec
45 """
46
47 def __init__(self):
48 # type: () -> None
49 self.opts = [] # type: List[Option]
50 self.index = 1 # start with 1
51 self.array_size = -1
52
53 def Add(self, *args, **kwargs):
54 # type: (Any, Any) -> None
55 self.opts.append(Option(self.index, *args, **kwargs))
56 self.index += 1
57
58 def DoneWithImplementedOptions(self):
59 # type: () -> None
60 self.array_size = self.index
61
62
63# Used by builtin
64_OTHER_SET_OPTIONS = [
65 # NOTE: set -i and +i is explicitly disallowed. Only osh -i or +i is valid
66 # https://unix.stackexchange.com/questions/339506/can-an-interactive-shell-become-non-interactive-or-vice-versa
67 ('n', 'noexec'),
68 ('x', 'xtrace'),
69 ('v', 'verbose'), # like xtrace, but prints unevaluated commands
70 ('f', 'noglob'),
71 ('C', 'noclobber'),
72
73 # A no-op for modernish.
74 (None, 'posix'),
75 (None, 'vi'),
76 (None, 'emacs'),
77]
78
79_STRICT_OPTS = [
80 # $a{[@]::} is not allowed, you need ${a[@]::0} or ${a[@]::n}
81 'strict_parse_slice',
82
83 # These are RUNTIME strict options.
84 'strict_argv', # empty argv not allowed
85 'strict_arith', # string to integer conversions, e.g. x=foo; echo $(( x ))
86
87 # No implicit conversions between string and array.
88 # - foo="$@" not allowed because it decays. Should be foo=( "$@" ).
89 # - ${a} not ${a[0]} (not implemented)
90 # sane-array? compare arrays like [[ "$@" == "${a[@]}" ]], which is
91 # incompatible because bash coerces
92 # default: do not allow
93 'strict_array',
94 'strict_control_flow', # break/continue at top level is fatal
95 # 'return $empty' and return "" are NOT accepted
96 'strict_errexit', # errexit can't be disabled in compound commands
97 'strict_nameref', # trap invalid variable names
98 'strict_word_eval', # negative slices, unicode
99 'strict_tilde', # ~nonexistent is an error (like zsh)
100
101 # Not implemented
102 'strict_glob', # glob_.py GlobParser has warnings
103]
104
105# These will break some programs, but the fix should be simple.
106
107# command_sub_errexit makes 'local foo=$(false)' and echo $(false) fail.
108# By default, we have mimic bash's undesirable behavior of ignoring
109# these failures, since ash copied it, and Alpine's abuild relies on it.
110#
111# Note that inherit_errexit is a strict option.
112
113_UPGRADE_RUNTIME_OPTS = [
114 ('simple_word_eval', False), # No splitting; arity isn't data-dependent
115 # Don't reparse program data as globs
116 ('dashglob', True), # do globs return files starting with - ?
117
118 # TODO: Should these be in strict mode?
119 # The logic was that strict_errexit improves your bash programs, but these
120 # would lead you to remove error handling. But the same could be said for
121 # strict_array?
122 ('command_sub_errexit', False), # check after command sub
123 ('process_sub_fail', False), # like pipefail, but for <(sort foo.txt)
124 ('xtrace_rich', False), # Hierarchical trace with PIDs
125 ('xtrace_details', True), # Legacy set -x stuff
126
127 # Whether status 141 in pipelines is turned into 0
128 ('sigpipe_status_ok', False),
129
130 # This applies to shell functions too
131 # It's also turned on in interactive mode
132 ('redefine_proc_func', True),
133]
134
135# TODO: Add strict_arg_parse? For example, 'trap 1 2 3' shouldn't be
136# valid, because it has an extra argument. Builtins are inconsistent about
137# checking this.
138
139_YSH_RUNTIME_OPTS = [
140 ('simple_echo', False), # echo takes 0 or 1 arguments
141 ('simple_eval_builtin', False), # eval takes exactly 1 argument
142
143 # only file tests (no strings), remove [, status 2
144 ('simple_test_builtin', False),
145
146 # TODO: simple_trap
147
148 # Turn aliases off so we can statically parse. bash has it off
149 # non-interactively, sothis shouldn't break much.
150 ('expand_aliases', True),
151]
152
153# Stuff that doesn't break too many programs.
154_UPGRADE_PARSE_OPTS = [
155 'parse_at', # @array, @[expr]
156 'parse_proc', # proc p { ... }
157 'parse_func', # func f(x) { ... }
158 'parse_brace', # cd /bin { ... }
159 'parse_bracket', # assert [42 === x]
160
161 # bare assignment 'x = 42' is allowed in Hay { } blocks, but disallowed
162 # everywhere else. It's not a command 'x' with arg '='.
163 'parse_equals',
164 'parse_paren', # if (x > 0) ...
165 'parse_ysh_string', # r'' u'' b'' and multi-line versions
166 'parse_triple_quote', # for ''' and """
167]
168
169# Extra stuff that breaks too many programs.
170_YSH_PARSE_OPTS = [
171 ('parse_at_all', False), # @ starting any word, e.g. @[] @{} @@ @_ @-
172
173 # Legacy syntax that is removed. These options are distinct from strict_*
174 # because they don't help you avoid bugs in bash programs. They just makes
175 # the language more consistent.
176 ('parse_backslash', True),
177 ('parse_backticks', True),
178 ('parse_dollar', True),
179 ('parse_ignored', True),
180 ('parse_sh_arith', True), # disallow all shell arithmetic, $(( )) etc.
181 ('parse_dparen', True), # disallow bash's ((
182 ('parse_dbracket', True), # disallow bash's [[
183 ('parse_bare_word', True), # 'case bare' and 'for x in bare'
184]
185
186# No-ops for bash compatibility
187_NO_OPS = [
188 'lastpipe', # this feature is always on
189
190 # Handled one by one
191 'progcomp',
192 'histappend', # stubbed out for issue #218
193 'hostcomplete', # complete words with '@' ?
194 'cmdhist', # multi-line commands in history
195
196 # Copied from https://www.gnu.org/software/bash/manual/bash.txt
197 # except 'compat*' because they were deemed too ugly
198 'assoc_expand_once',
199 'autocd',
200 'cdable_vars',
201 'cdspell',
202 'checkhash',
203 'checkjobs',
204 'checkwinsize',
205 'complete_fullquote', # Set by default
206 # If set, Bash quotes all shell metacharacters in filenames and
207 # directory names when performing completion. If not set, Bash
208 # removes metacharacters such as the dollar sign from the set of
209 # characters that will be quoted in completed filenames when
210 # these metacharacters appear in shell variable references in
211 # words to be completed. This means that dollar signs in
212 # variable names that expand to directories will not be quoted;
213 # however, any dollar signs appearing in filenames will not be
214 # quoted, either. This is active only when bash is using
215 # backslashes to quote completed filenames. This variable is
216 # set by default, which is the default Bash behavior in versions
217 # through 4.2.
218 'direxpand',
219 'dirspell',
220 'dotglob',
221 'execfail',
222 'extdebug', # for --debugger?
223 'extquote',
224 'force_fignore',
225 'globasciiranges',
226 'globstar', # TODO: implement **
227 'gnu_errfmt',
228 'histreedit',
229 'histverify',
230 'huponexit',
231 'interactive_comments',
232 'lithist',
233 'localvar_inherit',
234 'localvar_unset',
235 'login_shell',
236 'mailwarn',
237 'no_empty_cmd_completion',
238 'nocaseglob',
239 'progcomp_alias',
240 'promptvars',
241 'restricted_shell',
242 'shift_verbose',
243 'sourcepath',
244 'xpg_echo',
245]
246
247
248def _Init(opt_def):
249 # type: (_OptionDef) -> None
250
251 opt_def.Add('errexit',
252 short_flag='e',
253 builtin='set',
254 groups=['ysh:upgrade', 'ysh:all'])
255 opt_def.Add('nounset',
256 short_flag='u',
257 builtin='set',
258 groups=['ysh:upgrade', 'ysh:all'])
259 opt_def.Add('pipefail', builtin='set', groups=['ysh:upgrade', 'ysh:all'])
260
261 opt_def.Add('inherit_errexit', groups=['ysh:upgrade', 'ysh:all'])
262 # Hm is this subsumed by simple_word_eval?
263 opt_def.Add('nullglob', groups=['ysh:upgrade', 'ysh:all'])
264 opt_def.Add('verbose_errexit', groups=['ysh:upgrade', 'ysh:all'])
265
266 # set -o noclobber, etc.
267 for short_flag, name in _OTHER_SET_OPTIONS:
268 opt_def.Add(name, short_flag=short_flag, builtin='set')
269
270 # The only one where builtin=None. Only the shell can change it.
271 opt_def.Add('interactive', builtin=None)
272
273 # bash --norc -c 'set -o' shows this is on by default
274 opt_def.Add('hashall', short_flag='h', builtin='set', default=True)
275
276 #
277 # shopt
278 # (bash uses $BASHOPTS rather than $SHELLOPTS)
279 #
280
281 # shopt options that aren't in any groups.
282 opt_def.Add('failglob')
283 opt_def.Add('extglob')
284 opt_def.Add('nocasematch')
285
286 # recursive parsing and evaluation - for compatibility, ble.sh, etc.
287 opt_def.Add('eval_unsafe_arith')
288
289 opt_def.Add('ignore_flags_not_impl')
290 opt_def.Add('ignore_opts_not_impl')
291
292 # For implementing strict_errexit
293 # TODO: could be _no_command_sub / _no_process_sub, if we had to discourage
294 # "default True" options
295 opt_def.Add('_allow_command_sub', default=True)
296 opt_def.Add('_allow_process_sub', default=True)
297
298 # For implementing 'proc'
299 opt_def.Add('dynamic_scope', default=True)
300
301 # On in interactive shell
302 opt_def.Add('redefine_module', default=False)
303 # Hm these aren't the same?
304 #opt_def.Add('redefine_proc_func', default=False),
305
306 # For disabling strict_errexit while running traps. Because we run in the
307 # main loop, the value can be "off". Prefix with _ because it's undocumented
308 # and users shouldn't fiddle with it. We need a stack so this is a
309 # convenient place.
310 opt_def.Add('_running_trap')
311 opt_def.Add('_running_hay')
312
313 # For fixing lastpipe / job control / DEBUG trap interaction
314 opt_def.Add('_no_debug_trap')
315 # To implement ERR trap semantics - it's only run for the WHOLE pipeline,
316 # not each part (even the last part)
317 opt_def.Add('_no_err_trap')
318
319 # shopt -s strict_arith, etc.
320 for name in _STRICT_OPTS:
321 opt_def.Add(name, groups=['strict:all', 'ysh:all'])
322
323 #
324 # Options that enable YSH features
325 #
326
327 for name in _UPGRADE_PARSE_OPTS:
328 opt_def.Add(name, groups=['ysh:upgrade', 'ysh:all'])
329 # shopt -s simple_word_eval, etc.
330 for name, default in _UPGRADE_RUNTIME_OPTS:
331 opt_def.Add(name, default=default, groups=['ysh:upgrade', 'ysh:all'])
332
333 for name, default in _YSH_PARSE_OPTS:
334 opt_def.Add(name, default=default, groups=['ysh:all'])
335 for name, default in _YSH_RUNTIME_OPTS:
336 opt_def.Add(name, default=default, groups=['ysh:all'])
337
338 opt_def.DoneWithImplementedOptions()
339
340 # NO_OPS
341
342 # Stubs for shopt -s xpg_echo, etc.
343 for name in _NO_OPS:
344 opt_def.Add(name, implemented=False)
345
346
347def All():
348 # type: () -> List[Option]
349 """Return a list of options with metadata.
350
351 - Used by osh/builtin_pure.py to construct the arg spec.
352 - Used by frontend/lexer_gen.py to construct the lexer/matcher
353 """
354 return _OPTION_DEF.opts
355
356
357def ArraySize():
358 # type: () -> int
359 """Unused now, since we use opt_num::ARRAY_SIZE.
360
361 We could get rid of unimplemented options and shrink the array.
362 """
363 return _OPTION_DEF.array_size
364
365
366def OptionDict():
367 # type: () -> Dict[str, int]
368 """For the slow path in frontend/match.py."""
369 return dict((opt.name, opt.index) for opt in _OPTION_DEF.opts)
370
371
372def ParseOptNames():
373 # type: () -> List[str]
374 """Used by core/optview*.py."""
375 return [opt.name for opt in _OPTION_DEF.opts if opt.is_parse]
376
377
378def ExecOptNames():
379 # type: () -> List[str]
380 """Used by core/optview*.py."""
381 return [opt.name for opt in _OPTION_DEF.opts if opt.is_exec]
382
383
384_OPTION_DEF = _OptionDef()
385
386_Init(_OPTION_DEF)
387
388# Sort by name because we print options.
389# TODO: for MEMBERSHIP queries, we could sort by the most common? errexit
390# first?
391_SORTED = sorted(_OPTION_DEF.opts, key=lambda opt: opt.name)
392
393PARSE_OPTION_NUMS = [opt.index for opt in _SORTED if opt.is_parse]
394
395# Sorted because 'shopt -o -p' should be sorted, etc.
396VISIBLE_SHOPT_NUMS = [
397 opt.index for opt in _SORTED if opt.builtin == 'shopt' and opt.implemented
398]
399
400YSH_UPGRADE = [opt.index for opt in _SORTED if 'ysh:upgrade' in opt.groups]
401YSH_ALL = [opt.index for opt in _SORTED if 'ysh:all' in opt.groups]
402STRICT_ALL = [opt.index for opt in _SORTED if 'strict:all' in opt.groups]
403DEFAULT_TRUE = [opt.index for opt in _SORTED if opt.default]
404#print([opt.name for opt in _SORTED if opt.default])
405
406META_OPTIONS = ['strict:all', 'ysh:upgrade',
407 'ysh:all'] # Passed to flag parser
408
409# For printing option names to stdout. Wrapped by frontend/consts.
410OPTION_NAMES = dict((opt.index, opt.name) for opt in _SORTED)