no intelliSense for python namedtuple

[chen19901225]

Environment data

• VS Code version: 1.25.1
• Extension version (available under the Extensions sidebar): python2018.7.1
• OS and version: windows10
• Python version (& distribution if applicable, e.g. Anaconda): python3.6.5
• Type of virtual environment used (N/A | venv | virtualenv | conda | ...): N/A

with config "python.jediEnabled": false

Actual behavior

from collections import namedtuple
RequestStartLine = namedtuple("RequestStartLine", "method path version".split())
start_line = RequestStartLine("GET", "/query", "HTTP/1.1")
print(start_line)
start_line.  # <- no suggestion

[purpleP]

@chen19901225 This was discussed in my issue #2259

[brettcannon]

It should be noted that supporting namedtuple for IntelliSense requires special handling due to how the module dynamically generates code.

[purpleP]

@brettcannon well, there's eval in namedtuple but it's harmless. If you're simulating code execution why exactly you can't simulate that eval also? You know, I get why pycharm haven't had namedtuple autocompletion for a long time. They only use static analysis, so they really have to use special handling here otherwise they would need basically emulate the whole interpreter to understand what this eval will do at runtime. But you claim that you have some simulation going on. So maybe you could just simulate code execution for namedtuple and inspect its output?

[MikhailArkhipov]

@purpleP - code in libraries goes through lighter weight evaluation (as compared to user code) since otherwise performance suffers. In VS there used to be long process of building the library database. BTW, PyCharm does eval too ('building index...') but it also has a number of pre-cooked modules/heuristics for some common things.

We just need to improve the eval.