Dave Halter
3a0008ea80
Simplification.
2016-07-14 08:40:32 +02:00
Dave Halter
cc953ffff0
Goto on trailers wasn't correct. Fixes #571 .
2016-07-13 19:15:28 +02:00
Dave Halter
927534a8d5
Strange unicode characters are error tokens. Fixes #587 .
2016-07-13 08:53:08 +02:00
Dave Halter
c26f740dcd
The import path can be a list as well as a tuple. Deal with it.
2016-07-13 08:33:57 +02:00
Dave Halter
e18c8200dd
Fixed an issue with error nodes and completion in more complex settings.
2016-07-11 17:32:00 +02:00
Dave Halter
72634a94b8
Try to use line numbers instead of offsets in the fast parser.
2016-07-11 08:56:30 +02:00
Dave Halter
3ad67a4ec7
Jedi raised an error when defined_names was called on empty functions, fixes #697 .
2016-07-10 18:15:06 +02:00
Dave Halter
1c0aa06c7d
PEP 3132 unpacking should not raise an error (it may yield wrong results though at the moment), fixes #707 .
2016-07-10 17:51:01 +02:00
Dave Halter
7d64069780
An empty path given to Jedi should not raise errors. Fixes #577 .
2016-07-09 17:27:57 +02:00
Dave Halter
e0cb1346e1
Add basic yield from type inference. References #647 .
2016-07-09 02:33:56 +02:00
Dave Halter
5280f567f9
The docstring of import completions was wrong.
...
This is fixed now. However, since this might massively decrease performance,
it's not enabled by default. You can enable it with `docstring(fast=False)`
(see test changes), but I wouldn't recommend it at this point.
Fixes #656 .
2016-07-09 01:04:15 +02:00
Dave Halter
baa745a6ac
A minor issue for getting the stack at a position, fixes #590 .
2016-07-08 08:39:36 +02:00
Dave Halter
e5f09e1c7d
Fix an issue with end_pos of a module.
2016-07-08 00:03:52 +02:00
Dave Halter
adcc1c2b51
Don't delete ErrorNode names. They are part of the parser now.
...
Fixes #594 and possibly also #590 and #579 .
2016-07-07 18:33:45 +02:00
Dave Halter
4a19376187
Fix issue https://github.com/DamnWidget/anaconda/issues/449 . Using self should not cause side effects in completion.
2016-07-06 18:31:47 +02:00
Dave Halter
3ad159b0aa
The import logic cannot assume that a file is always importable, fixes #716 .
2016-07-06 08:52:23 +02:00
Dave Halter
4243adf54b
Add param splitting test.
2016-07-06 08:30:27 +02:00
Dave Halter
074a154af3
Fix a small issue that coul dhappen e.g. in stdin.
2016-07-06 08:05:50 +02:00
Dave Halter
24cddda8e7
Remove the old interpreter logic.
2016-07-04 08:35:22 +02:00
Dave Halter
6a8138d185
Improve the compiled object generation caching, which was very wrong and is ok now, but still needs improvements.
2016-07-03 15:32:08 +02:00
Dave Halter
0223471237
Increase the parser pickling version.
2016-07-03 12:31:57 +02:00
Dave Halter
62786158da
Some more Python compatibility improvements.
2016-07-03 11:35:07 +02:00
Dave Halter
9245181a8c
Some python 2.7 (and 3.3) compatibility improvements.
2016-07-03 02:54:21 +02:00
Dave Halter
609965d07c
Finally fix all python 3.4 tests again.
2016-07-01 20:59:24 +02:00
Dave Halter
67a0f604a7
Fix an issue with interpreter exceptions in certain cases.
2016-07-01 19:32:05 +02:00
Dave Halter
2652666080
Remove the logic to not use getattr on instances in CompiledObjects.
2016-07-01 18:11:44 +02:00
Dave Halter
d5098ef096
Forgot to add parser/utils.py.
2016-06-30 19:36:48 +02:00
Dave Halter
f7278f5bf1
Some more bug fixes for MixedObject.
2016-06-30 19:36:21 +02:00
Dave Halter
6b41db96bf
Refactor something to use .type instead of isinstance.
2016-06-30 09:55:21 +02:00
Dave Halter
689284c615
Refactor Evaluator.wrap to use the types in a more consequent way.
2016-06-29 21:06:35 +02:00
Dave Halter
a3b263a599
REPL completion is working again partially. There's some progress at least.
2016-06-29 08:49:20 +02:00
Dave Halter
52c42c3392
Reenable call signature caching and move a lot of parser specific caching to the parser itself.
2016-06-28 08:46:29 +02:00
Dave Halter
746a513ef9
Return a TokenInfo object in the tokenizer instead of a tuple (it's a namedtuple now.)
2016-06-27 09:14:11 +02:00
Dave Halter
9b85080fb8
Finally removed the user_context. Goodbye old hacks...
2016-06-27 08:49:27 +02:00
Dave Halter
969100e471
Move the parsing away from user_context to api.py.
2016-06-27 08:48:36 +02:00
Dave Halter
0445d51d34
Remove the user_scope from the user_context module.
2016-06-27 08:35:24 +02:00
Dave Halter
bb4ab45131
Don't use UserContextParser.user_stmt anymore, since we can access it directly.
2016-06-23 16:36:12 +02:00
Dave Halter
73e71b3c1a
Finally able to remove the user_context. This is awesome!
...
Now we only use the syntax tree to understand where the user is doing something.
2016-06-23 16:26:28 +02:00
Dave Halter
672bf9908c
Remove path argument from completions, because it's really not needed anymore.
2016-06-23 16:11:23 +02:00
Dave Halter
9225db084a
user_context is not needed anymore for completions. yay!
2016-06-23 09:19:20 +02:00
Dave Halter
8f39a6e89d
'source' should not be public in the API. Move it to _source.
2016-06-23 08:53:34 +02:00
Dave Halter
77e66e01e3
Remove the inference module, it's unused code.
2016-06-23 08:49:26 +02:00
Dave Halter
1ab4eb3696
Exchange the completion trailer evaluation logic. It's way more consistent now.
2016-06-23 08:47:43 +02:00
Dave Halter
cbef4235ff
Remove needs_dot and settings.add_dot_after_module. Both are not really used anymore with context completions anymore.
...
Also the setting doesn't seem to be used anywhere as far as I can tell.
2016-06-22 22:52:10 +02:00
Dave Halter
1355ea01b3
Simplify completions further to eventually get rid of user_context.
2016-06-22 09:15:32 +02:00
Dave Halter
80aa9ad079
A small python 2 bugfix.
2016-06-22 00:45:14 +02:00
Dave Halter
57b1fdaa26
Remove code that's not used anymore.
2016-06-22 00:31:36 +02:00
Dave Halter
6fec29d778
All tests except the Interpreter tests are working again.
2016-06-22 00:27:21 +02:00
Dave Halter
8e67facecc
Refactoring: call_of_name -> call_of_leaf.
2016-06-21 18:42:20 +02:00
Dave Halter
d0eb8137e2
Remove old unused call_of_name madness.
2016-06-21 18:39:35 +02:00