Dave Halter
d9aa84f971
simplify the use of 'context'
2014-10-08 16:33:35 +02:00
Dave Halter
e54dac3777
Start setting up the parser.representation merge
2014-10-08 16:15:27 +02:00
Dave Halter
220ddc8a74
Probably the last big removals, there's not a lot left.
2014-10-08 15:35:45 +02:00
Dave Halter
a23ad3df10
Remove more and more from pytree.
2014-10-08 15:30:15 +02:00
Dave Halter
2781a4ac98
Remove comparison methods.
2014-10-08 15:22:44 +02:00
Dave Halter
1fb6e15750
Fix prefixes of Leafs.
2014-10-08 15:19:36 +02:00
Dave Halter
36368cd606
More removals and cleanups.
2014-10-08 15:13:43 +02:00
Dave Halter
07d9111c77
Remove some more pytree methods.
2014-10-08 15:05:29 +02:00
Dave Halter
a7fff54d7b
Remove some fixer related stuff (lib2to3 stuff).
2014-10-08 15:04:01 +02:00
Dave Halter
834172a3e9
Add a new parser, check it pgen2 would work. (already modified outside this repository)
2014-10-08 14:26:52 +02:00
Dave Halter
27444ed64d
Remove Import.alias_name_part, it was simply an alias for another lookup.
2014-09-26 16:32:36 +02:00
Dave Halter
03e01631cc
Remove NamePart from existance and rename it to Name.
2014-09-26 16:29:53 +02:00
Dave Halter
522c9eda90
Remove pr.Name completely.
2014-09-26 16:18:10 +02:00
Dave Halter
4d7db35340
Fix a few last tests, now Jedi's working again, tests are passing.
2014-09-26 16:02:03 +02:00
Dave Halter
6f29e802c2
Fix an issue with as_names.
2014-09-26 15:48:49 +02:00
Dave Halter
90842ce62d
Fixed global variables.
2014-09-26 11:58:11 +02:00
Dave Halter
47c4369d28
Fix remaining issues with sys path checks.
2014-09-25 18:28:03 +02:00
Dave Halter
f4c99259b5
Fix an issue with sys.path. Also moved the names closure for isinstance checks away (used for sys.path stuff) and use a get_code check instead, which is more flexible.
2014-09-25 12:35:53 +02:00
Dave Halter
9ecf3774a0
Import issues again.
2014-09-24 21:59:08 +02:00
Dave Halter
c43afae24a
Issues with imports.
2014-09-24 21:12:38 +02:00
Dave Halter
ff61c1d81c
Fix an extremely annoying bug that made pickling impossible.
2014-09-24 20:58:29 +02:00
Dave Halter
56243e10c6
The get_code generation of imports was buggy.
2014-09-24 17:48:11 +02:00
Dave Halter
db31536d78
Fix issue with descriptors.
2014-09-24 16:52:44 +02:00
Dave Halter
f300e63dae
A couple of changes:
...
- parser.representation now uses ArrayStmt and ExprStatement to be able to differentiate easily with isinstance.
- NameParts are temporarily allowed again as InstanceElements.
2014-09-24 15:21:56 +02:00
Dave Halter
0a65eea2cf
Start to change the logic for self.foo variables.
2014-09-24 14:33:15 +02:00
Dave Halter
d5fbc006e2
Add a names_dict to scopes. This is good for the future parser and now useful to process self.foo and other stuff.
2014-09-24 12:44:24 +02:00
Dave Halter
c61f79314b
Introduce get_names_dict to statements to actually fetch the calls out of a statement.
...
This is going to be the new default method to do dynamic stuff as well as self.foo resolutions for instances.
2014-09-24 12:06:41 +02:00
Dave Halter
5efa467449
Few import issues.
2014-09-22 23:45:48 +02:00
Dave Halter
1d71b25109
Previously forgot to add the NameParts to used_names. (which had worked before that)
2014-09-22 23:24:29 +02:00
Dave Halter
dae1a48d70
Remove a lot of the old Name.names usages in favor of a direct NamePart usage.
2014-09-22 22:34:33 +02:00
Dave Halter
04cf742973
Temporary parser implementation. Now we're pretty much done with pr.Name.
2014-09-22 17:05:23 +02:00
Dave Halter
c4e45916c6
Modules also use a NamePart as a name, now.
2014-09-22 12:52:48 +02:00
Dave Halter
779618c08b
First changes to eventually replace Name by NamePart.
2014-09-22 11:57:19 +02:00
Dave Halter
b2342c76be
Refactoring: Make Import.get_all_import_names return NameParts.
2014-09-19 01:40:09 +02:00
Dave Halter
83d2af5138
First imports are working with goto.
2014-09-19 01:21:17 +02:00
Dave Halter
1fb9b4bc6b
Completion now always takes a NamePart as input.
2014-09-10 18:59:08 +02:00
Dave Halter
110f130741
Make it possible to get previous statements of Calls.
2014-09-09 16:44:56 +02:00
Dave Halter
ff7680c15f
Generate the expression_list of a statement in any case.
...
This is a consequence of being able to have Calls as parents of Names. Otherwise we would have changing Name.parent values.
2014-09-09 12:54:46 +02:00
Dave Halter
2e7e2f0a29
Name parents are now Calls (once their statements have generated the Calls).
...
This makes the goto function more powerful. Also fixes an issue with the deep_ast_copy, that I tried to fix previously, but failed, because I hadn't tested it.
2014-09-06 12:19:07 +02:00
Dave Halter
cb84bc0829
Start using ExprStmt.
2014-09-06 10:46:59 +02:00
Dave Halter
f57d9ef675
Rename Name.get_parent_stmt to Name.get_definition, because it's not always a statement. Also start using it in the NameFinder.
2014-09-06 10:43:26 +02:00
Dave Halter
99116cdcb7
Add a Name.get_parent_stmt() function.
2014-09-05 22:26:55 +02:00
Dave Halter
6c07c7acfe
Create an ExprStatement class to replace the Statement class in the future and separate array parts of actual statements
2014-09-05 22:21:26 +02:00
Dave Halter
7b2e11d71b
Rewrote sys_path._paths_from_assignment.
2014-09-04 14:12:10 +02:00
Dave Halter
4180005893
Forgot to add the the params in the case of a class in the previous commit.
2014-09-04 12:53:42 +02:00
Dave Halter
06699993f1
Class inheritance definitions shouldn't be params. It's just statements.
2014-09-04 12:28:50 +02:00
Dave Halter
18204c4c19
By trying to get rid of search_name in usages, we had to fix an issue with imports:
...
If used like 'follow(is_goto)', it could return a ModuleWrapper instead of a Name, which is what we actually want.
2014-09-03 19:30:00 +02:00
Dave Halter
77673ba986
Add an optional param 'parent' to parser.representation.Simple, which simplifies some calls to that superclass.
2014-08-19 00:30:17 +02:00
Dave Halter
8006d6f190
Change implementation of StatementElement.
...
Instead of having both next and execution as attributes, we now only have next, because it's an execution if there's an array.
2014-08-18 22:25:55 +02:00
Dave Halter
00d15da143
refactor search_call_signatures. Now we don't need to set Call.next.parent in a strange way anymore and the whole thing seems to be more logical.
2014-08-18 14:51:38 +02:00
