OILS
/
frontend
/
location.py
1 |
#!/usr/bin/env python2
|
2 |
"""
|
3 |
location.py - Library to get source location info from nodes.
|
4 |
|
5 |
This makes syntax errors nicer.
|
6 |
"""
|
7 |
from __future__ import print_function
|
8 |
|
9 |
from _devbuild.gen.syntax_asdl import (
|
10 |
loc,
|
11 |
loc_t,
|
12 |
loc_e,
|
13 |
command,
|
14 |
command_e,
|
15 |
command_t,
|
16 |
sh_lhs_expr,
|
17 |
sh_lhs_expr_e,
|
18 |
sh_lhs_expr_t,
|
19 |
word,
|
20 |
word_e,
|
21 |
word_t,
|
22 |
word_part,
|
23 |
word_part_e,
|
24 |
word_part_t,
|
25 |
CompoundWord,
|
26 |
SimpleVarSub,
|
27 |
Token,
|
28 |
ShArrayLiteral,
|
29 |
SingleQuoted,
|
30 |
DoubleQuoted,
|
31 |
CommandSub,
|
32 |
BracedVarSub,
|
33 |
BraceGroup,
|
34 |
arith_expr,
|
35 |
arith_expr_e,
|
36 |
arith_expr_t,
|
37 |
)
|
38 |
from _devbuild.gen.runtime_asdl import lvalue
|
39 |
from mycpp.mylib import log
|
40 |
from mycpp.mylib import tagswitch
|
41 |
|
42 |
_ = log
|
43 |
|
44 |
from typing import cast, Optional
|
45 |
|
46 |
|
47 |
def LName(name):
|
48 |
# type: (str) -> lvalue.Named
|
49 |
"""Wrapper for lvalue.Named() with location.
|
50 |
|
51 |
TODO: add locations and remove
|
52 |
this.
|
53 |
"""
|
54 |
return lvalue.Named(name, loc.Missing)
|
55 |
|
56 |
|
57 |
def TokenFor(loc_):
|
58 |
# type: (loc_t) -> Optional[Token]
|
59 |
"""Given a location, get a Token.
|
60 |
|
61 |
This is useful because a Token points to a single line.
|
62 |
"""
|
63 |
UP_location = loc_
|
64 |
with tagswitch(loc_) as case:
|
65 |
if case(loc_e.Missing):
|
66 |
return None
|
67 |
|
68 |
elif case(loc_e.Token):
|
69 |
tok = cast(Token, UP_location)
|
70 |
if tok:
|
71 |
return tok
|
72 |
else:
|
73 |
return None
|
74 |
|
75 |
elif case(loc_e.ArgWord):
|
76 |
w = cast(CompoundWord, UP_location)
|
77 |
return LeftTokenForWord(w)
|
78 |
|
79 |
elif case(loc_e.WordPart):
|
80 |
loc_ = cast(loc.WordPart, UP_location)
|
81 |
if loc_.p:
|
82 |
return LeftTokenForWordPart(loc_.p)
|
83 |
else:
|
84 |
return None
|
85 |
|
86 |
elif case(loc_e.Word):
|
87 |
loc_ = cast(loc.Word, UP_location)
|
88 |
if loc_.w:
|
89 |
return LeftTokenForWord(loc_.w)
|
90 |
else:
|
91 |
return None
|
92 |
|
93 |
elif case(loc_e.Command):
|
94 |
loc_ = cast(loc.Command, UP_location)
|
95 |
if loc_.c:
|
96 |
return TokenForCommand(loc_.c)
|
97 |
else:
|
98 |
return None
|
99 |
|
100 |
elif case(loc_e.Arith):
|
101 |
loc_ = cast(loc.Arith, UP_location)
|
102 |
if loc_.a:
|
103 |
return TokenForArith(loc_.a)
|
104 |
else:
|
105 |
return None
|
106 |
|
107 |
else:
|
108 |
raise AssertionError()
|
109 |
|
110 |
raise AssertionError()
|
111 |
|
112 |
|
113 |
def TokenForCommand(node):
|
114 |
# type: (command_t) -> Optional[Token]
|
115 |
"""Used directly in _CheckStatus()"""
|
116 |
UP_node = node # type: command_t
|
117 |
tag = node.tag()
|
118 |
|
119 |
if tag == command_e.Sentence:
|
120 |
node = cast(command.Sentence, UP_node)
|
121 |
#log("node.child %s", node.child)
|
122 |
return node.terminator # & or ;
|
123 |
|
124 |
if tag == command_e.Simple:
|
125 |
node = cast(command.Simple, UP_node)
|
126 |
return node.blame_tok
|
127 |
|
128 |
if tag == command_e.ShAssignment:
|
129 |
node = cast(command.ShAssignment, UP_node)
|
130 |
return node.left
|
131 |
|
132 |
if tag == command_e.Pipeline:
|
133 |
node = cast(command.Pipeline, UP_node)
|
134 |
if len(node.ops):
|
135 |
return node.ops[0] # first | or |&
|
136 |
else:
|
137 |
assert node.negated is not None
|
138 |
return node.negated # ! false
|
139 |
|
140 |
if tag == command_e.AndOr:
|
141 |
node = cast(command.AndOr, UP_node)
|
142 |
return node.ops[0] # first && or ||
|
143 |
|
144 |
if tag == command_e.DoGroup:
|
145 |
node = cast(command.DoGroup, UP_node)
|
146 |
return node.left # 'do' token
|
147 |
if tag == command_e.BraceGroup:
|
148 |
node = cast(BraceGroup, UP_node)
|
149 |
return node.left # { token
|
150 |
if tag == command_e.Subshell:
|
151 |
node = cast(command.Subshell, UP_node)
|
152 |
return node.left # ( token
|
153 |
|
154 |
if tag == command_e.WhileUntil:
|
155 |
node = cast(command.WhileUntil, UP_node)
|
156 |
return node.keyword # while
|
157 |
if tag == command_e.If:
|
158 |
node = cast(command.If, UP_node)
|
159 |
return node.if_kw
|
160 |
if tag == command_e.Case:
|
161 |
node = cast(command.Case, UP_node)
|
162 |
return node.case_kw
|
163 |
if tag == command_e.TimeBlock:
|
164 |
node = cast(command.TimeBlock, UP_node)
|
165 |
return node.keyword
|
166 |
|
167 |
# We never have this case?
|
168 |
#if node.tag == command_e.CommandList:
|
169 |
# pass
|
170 |
|
171 |
return None
|
172 |
|
173 |
|
174 |
def TokenForArith(node):
|
175 |
# type: (arith_expr_t) -> Optional[Token]
|
176 |
UP_node = node
|
177 |
with tagswitch(node) as case:
|
178 |
if case(arith_expr_e.VarSub):
|
179 |
vsub = cast(SimpleVarSub, UP_node)
|
180 |
# $(( x ))
|
181 |
return vsub.left
|
182 |
|
183 |
elif case(arith_expr_e.Word):
|
184 |
w = cast(CompoundWord, UP_node)
|
185 |
return LeftTokenForWord(w)
|
186 |
|
187 |
elif case(arith_expr_e.Unary):
|
188 |
node = cast(arith_expr.Unary, UP_node)
|
189 |
return TokenForArith(node.child)
|
190 |
|
191 |
elif case(arith_expr_e.Binary):
|
192 |
node = cast(arith_expr.Binary, UP_node)
|
193 |
|
194 |
# TODO: should blame op
|
195 |
# blaming left is arbitrary, but better than nothing
|
196 |
return TokenForArith(node.left)
|
197 |
|
198 |
elif case(arith_expr_e.TernaryOp):
|
199 |
node = cast(arith_expr.TernaryOp, UP_node)
|
200 |
|
201 |
# TODO: should blame op
|
202 |
# blaming cond is arbitrary, but better than nothing
|
203 |
return TokenForArith(node.cond)
|
204 |
|
205 |
return None
|
206 |
|
207 |
|
208 |
def LeftTokenForWordPart(part):
|
209 |
# type: (word_part_t) -> Optional[Token]
|
210 |
UP_part = part
|
211 |
with tagswitch(part) as case:
|
212 |
if case(word_part_e.ShArrayLiteral):
|
213 |
part = cast(ShArrayLiteral, UP_part)
|
214 |
return part.left
|
215 |
|
216 |
elif case(word_part_e.BashAssocLiteral):
|
217 |
part = cast(word_part.BashAssocLiteral, UP_part)
|
218 |
return part.left
|
219 |
|
220 |
elif case(word_part_e.Literal):
|
221 |
tok = cast(Token, UP_part)
|
222 |
return tok
|
223 |
|
224 |
elif case(word_part_e.EscapedLiteral):
|
225 |
part = cast(word_part.EscapedLiteral, UP_part)
|
226 |
return part.token
|
227 |
|
228 |
elif case(word_part_e.SingleQuoted):
|
229 |
part = cast(SingleQuoted, UP_part)
|
230 |
return part.left
|
231 |
|
232 |
elif case(word_part_e.DoubleQuoted):
|
233 |
part = cast(DoubleQuoted, UP_part)
|
234 |
return part.left
|
235 |
|
236 |
elif case(word_part_e.SimpleVarSub):
|
237 |
part = cast(SimpleVarSub, UP_part)
|
238 |
return part.left
|
239 |
|
240 |
elif case(word_part_e.BracedVarSub):
|
241 |
part = cast(BracedVarSub, UP_part)
|
242 |
return part.left
|
243 |
|
244 |
elif case(word_part_e.CommandSub):
|
245 |
part = cast(CommandSub, UP_part)
|
246 |
return part.left_token
|
247 |
|
248 |
elif case(word_part_e.TildeSub):
|
249 |
part = cast(word_part.TildeSub, UP_part)
|
250 |
return part.token
|
251 |
|
252 |
elif case(word_part_e.ArithSub):
|
253 |
part = cast(word_part.ArithSub, UP_part)
|
254 |
return part.left
|
255 |
|
256 |
elif case(word_part_e.ExtGlob):
|
257 |
part = cast(word_part.ExtGlob, UP_part)
|
258 |
return part.op
|
259 |
|
260 |
elif case(word_part_e.BracedRange):
|
261 |
part = cast(word_part.BracedRange, UP_part)
|
262 |
return part.blame_tok
|
263 |
|
264 |
elif case(word_part_e.BracedTuple):
|
265 |
part = cast(word_part.BracedTuple, UP_part)
|
266 |
# TODO: Derive token from part.words[0]
|
267 |
return None
|
268 |
|
269 |
elif case(word_part_e.Splice):
|
270 |
part = cast(word_part.Splice, UP_part)
|
271 |
return part.blame_tok
|
272 |
|
273 |
elif case(word_part_e.ExprSub):
|
274 |
part = cast(word_part.ExprSub, UP_part)
|
275 |
return part.left # $[
|
276 |
|
277 |
else:
|
278 |
raise AssertionError(part.tag())
|
279 |
|
280 |
|
281 |
def _RightTokenForWordPart(part):
|
282 |
# type: (word_part_t) -> Token
|
283 |
UP_part = part
|
284 |
with tagswitch(part) as case:
|
285 |
if case(word_part_e.ShArrayLiteral):
|
286 |
part = cast(ShArrayLiteral, UP_part)
|
287 |
return part.right
|
288 |
|
289 |
elif case(word_part_e.BashAssocLiteral):
|
290 |
part = cast(word_part.BashAssocLiteral, UP_part)
|
291 |
return part.right
|
292 |
|
293 |
elif case(word_part_e.Literal):
|
294 |
tok = cast(Token, UP_part)
|
295 |
# Just use the token
|
296 |
return tok
|
297 |
|
298 |
elif case(word_part_e.EscapedLiteral):
|
299 |
part = cast(word_part.EscapedLiteral, UP_part)
|
300 |
return part.token
|
301 |
|
302 |
elif case(word_part_e.SingleQuoted):
|
303 |
part = cast(SingleQuoted, UP_part)
|
304 |
return part.right # right '
|
305 |
|
306 |
elif case(word_part_e.DoubleQuoted):
|
307 |
part = cast(DoubleQuoted, UP_part)
|
308 |
return part.right # right "
|
309 |
|
310 |
elif case(word_part_e.SimpleVarSub):
|
311 |
part = cast(SimpleVarSub, UP_part)
|
312 |
# left and right are the same for $myvar
|
313 |
return part.left
|
314 |
|
315 |
elif case(word_part_e.BracedVarSub):
|
316 |
part = cast(BracedVarSub, UP_part)
|
317 |
return part.right
|
318 |
|
319 |
elif case(word_part_e.CommandSub):
|
320 |
part = cast(CommandSub, UP_part)
|
321 |
return part.right
|
322 |
|
323 |
elif case(word_part_e.TildeSub):
|
324 |
part = cast(word_part.TildeSub, UP_part)
|
325 |
return part.token
|
326 |
|
327 |
elif case(word_part_e.ArithSub):
|
328 |
part = cast(word_part.ArithSub, UP_part)
|
329 |
return part.right
|
330 |
|
331 |
elif case(word_part_e.ExtGlob):
|
332 |
part = cast(word_part.ExtGlob, UP_part)
|
333 |
return part.right
|
334 |
|
335 |
elif case(word_part_e.BracedRange):
|
336 |
part = cast(word_part.BracedRange, UP_part)
|
337 |
return part.blame_tok
|
338 |
|
339 |
elif case(word_part_e.BracedTuple):
|
340 |
part = cast(word_part.BracedTuple, UP_part)
|
341 |
# TODO: Derive token from part.words[0]
|
342 |
return None
|
343 |
|
344 |
elif case(word_part_e.Splice):
|
345 |
part = cast(word_part.Splice, UP_part)
|
346 |
return part.blame_tok
|
347 |
|
348 |
elif case(word_part_e.ExprSub):
|
349 |
part = cast(word_part.ExprSub, UP_part)
|
350 |
return part.right
|
351 |
|
352 |
else:
|
353 |
raise AssertionError(part.tag())
|
354 |
|
355 |
|
356 |
def LeftTokenForCompoundWord(w):
|
357 |
# type: (CompoundWord) -> Optional[Token]
|
358 |
if len(w.parts):
|
359 |
return LeftTokenForWordPart(w.parts[0])
|
360 |
else:
|
361 |
# This is possible for empty brace sub alternative {a,b,}
|
362 |
return None
|
363 |
|
364 |
|
365 |
def LeftTokenForWord(w):
|
366 |
# type: (word_t) -> Optional[Token]
|
367 |
if w is None:
|
368 |
return None # e.g. builtin_bracket word.String() EOF
|
369 |
|
370 |
UP_w = w
|
371 |
with tagswitch(w) as case:
|
372 |
if case(word_e.Compound):
|
373 |
w = cast(CompoundWord, UP_w)
|
374 |
return LeftTokenForCompoundWord(w)
|
375 |
|
376 |
elif case(word_e.Operator):
|
377 |
tok = cast(Token, UP_w)
|
378 |
return tok
|
379 |
|
380 |
elif case(word_e.BracedTree):
|
381 |
w = cast(word.BracedTree, UP_w)
|
382 |
# This should always have one part?
|
383 |
return LeftTokenForWordPart(w.parts[0])
|
384 |
|
385 |
elif case(word_e.String):
|
386 |
w = cast(word.String, UP_w)
|
387 |
# See _StringWordEmitter in osh/builtin_bracket.py
|
388 |
return LeftTokenForWord(w.blame_loc)
|
389 |
|
390 |
else:
|
391 |
raise AssertionError(w.tag())
|
392 |
|
393 |
raise AssertionError('for -Wreturn-type in C++')
|
394 |
|
395 |
|
396 |
def RightTokenForWord(w):
|
397 |
# type: (word_t) -> Token
|
398 |
"""Used for alias expansion and history substitution.
|
399 |
|
400 |
and here doc delimiters?
|
401 |
"""
|
402 |
UP_w = w
|
403 |
with tagswitch(w) as case:
|
404 |
if case(word_e.Compound):
|
405 |
w = cast(CompoundWord, UP_w)
|
406 |
if len(w.parts):
|
407 |
end = w.parts[-1]
|
408 |
return _RightTokenForWordPart(end)
|
409 |
else:
|
410 |
# This is possible for empty brace sub alternative {a,b,}
|
411 |
return None
|
412 |
|
413 |
elif case(word_e.Operator):
|
414 |
tok = cast(Token, UP_w)
|
415 |
return tok
|
416 |
|
417 |
elif case(word_e.BracedTree):
|
418 |
w = cast(word.BracedTree, UP_w)
|
419 |
# Note: this case may be unused
|
420 |
return _RightTokenForWordPart(w.parts[-1])
|
421 |
|
422 |
elif case(word_e.String):
|
423 |
w = cast(word.String, UP_w)
|
424 |
# Note: this case may be unused
|
425 |
return RightTokenForWord(w.blame_loc)
|
426 |
|
427 |
else:
|
428 |
raise AssertionError(w.tag())
|
429 |
|
430 |
raise AssertionError('for -Wreturn-type in C++')
|
431 |
|
432 |
|
433 |
def TokenForLhsExpr(node):
|
434 |
# type: (sh_lhs_expr_t) -> Token
|
435 |
"""Currently unused?
|
436 |
|
437 |
Will be useful for translating YSH assignment
|
438 |
"""
|
439 |
# This switch is annoying but we don't have inheritance from the sum type
|
440 |
# (because of diamond issue). We might change the schema later, which maeks
|
441 |
# it moot. See the comment in frontend/syntax.asdl.
|
442 |
UP_node = node
|
443 |
with tagswitch(node) as case:
|
444 |
if case(sh_lhs_expr_e.Name):
|
445 |
node = cast(sh_lhs_expr.Name, UP_node)
|
446 |
return node.left
|
447 |
elif case(sh_lhs_expr_e.IndexedName):
|
448 |
node = cast(sh_lhs_expr.IndexedName, UP_node)
|
449 |
return node.left
|
450 |
else:
|
451 |
# Should not see UnparsedIndex
|
452 |
raise AssertionError()
|
453 |
|
454 |
raise AssertionError()
|