Python language server didn't work properly when editing '__init__.py'

[d3r3kk]

@CamberLoid commented on Wed Jan 30 2019

Issue Type: Bug

Python language server raised an exception and caused problems when handling this import (modules),(),
After double-checking on my Windows, I am sure this is because of the extra comma in the import expression in __init__.py ,

Using Anaconda distribution. Have not tried to reproduce on other OS.

Environment data

• Version: 1.30.2 (user setup)
• Commit: 61122f88f0bf01e2ac16bdb9e1bc4571755f5bd8
• Date: 2019-01-07T22:54:13.295Z
• Electron: 2.0.12
• Chrome: 61.0.3163.100
• Node.js: 8.9.3
• V8: 6.1.534.41
• OS: Windows_NT x64 10.0.18323

Actual behaviour

Cannot access origin libraries by Ctrl+(Modules), and the exceptions below.
It acts like this

Expected behaviour

Access origin libraries by Ctrl+(Modules)
and no Exceptions

Logs

[Error - 1:18:14 AM] System.ArgumentOutOfRangeException: Index was out of range. Must be non-negative and less than the size of the collection.
Parameter name: index
   at System.SZArrayHelper.get_Item[T](Int32 index)
   at Microsoft.PythonTools.Analysis.Analyzer.DDG.ImportModule(ImportStatement& node, IModule& module, DottedName& moduleImportExpression, NameExpression& asNameExpression)
   at Microsoft.PythonTools.Analysis.Analyzer.DDG.Walk(ImportStatement node)
   at Microsoft.PythonTools.Parsing.Ast.ImportStatement.Walk(PythonWalker walker)
   at Microsoft.PythonTools.Analysis.Analyzer.DDG.Walk(SuiteStatement node)
   at Microsoft.PythonTools.Parsing.Ast.SuiteStatement.Walk(PythonWalker walker)
   at Microsoft.PythonTools.Parsing.Ast.PythonAst.Walk(PythonWalker walker)
   at Microsoft.PythonTools.Analysis.AnalysisUnit.AnalyzeWorker(DDG ddg, CancellationToken cancel)
   at Microsoft.PythonTools.Analysis.Analyzer.DDG.Analyze(Deque`1 queue, CancellationToken cancel, Action`1 reportQueueSize, Int32 reportQueueInterval)
   at Microsoft.PythonTools.Analysis.PythonAnalyzer.AnalyzeQueuedEntries(CancellationToken cancel)
   at Microsoft.PythonTools.Intellisense.AnalysisQueue.GroupAnalysis.Analyze(CancellationToken cancel)
   at Microsoft.PythonTools.Intellisense.AnalysisQueue.HandleAnalyzable(IAnalyzable item, AnalysisPriority priority, CancellationToken cancellationToken)
   at Microsoft.PythonTools.Intellisense.AnalysisQueue.ConsumerLoop()

Extension version: 2019.1.0
VS Code version: Code 1.30.2 (61122f88f0bf01e2ac16bdb9e1bc4571755f5bd8, 2019-01-07T22:54:13.295Z)
OS version: Windows_NT x64 10.0.18323

System Info

Item	Value
CPUs	Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz (12 x 2208)
GPU Status	2d_canvas: enabled checker_imaging: disabled_off flash_3d: enabled flash_stage3d: enabled flash_stage3d_baseline: enabled gpu_compositing: enabled multiple_raster_threads: enabled_on native_gpu_memory_buffers: disabled_software rasterization: enabled video_decode: enabled video_encode: enabled webgl: enabled webgl2: enabled
Memory (System)	15.87GB (9.36GB free)
Process Argv
Screen Reader	no
VM	0%

---

@d3r3kk commented on Thu Jan 31 2019

Verified, here's another gif that might help figure out the issue:

This is using Microsoft Python Language Server version 0.1.78.0, Miniconda3 on Windows with Python 3.7.2

[d3r3kk]

Output found in the Output:Log(Window)

[2019-01-31 14:43:51.417] [renderer1] [error] name must not be falsy: Error: name must not be falsy
	at Function.e.validate (c:\Program Files (x86)\Microsoft VS Code\resources\app\out\vs\workbench\node\extensionHostProcess.js:357:869)
	at new e (c:\Program Files (x86)\Microsoft VS Code\resources\app\out\vs\workbench\node\extensionHostProcess.js:357:807)
	at R (C:\Users\derek\.vscode\extensions\ms-python.python-2019.1.0\out\client\extension.js:83:93748)
	at Array.map (<anonymous>)
	at Object.asDocumentSymbols (C:\Users\derek\.vscode\extensions\ms-python.python-2019.1.0\out\client\extension.js:83:97402)
	at module.exports.R.registerLanguageProvider.t.sendRequest.then.e (C:\Users\derek\.vscode\extensions\ms-python.python-2019.1.0\out\client\extension.js:1:142452)
	at <anonymous>

Problems Window

[jakebailey]

Aha, so that's what that stack trace is.

The diagnostics not going away are likely due to the LS exiting and leaving them there.

[jakebailey]

For reference, this was likely a bug beforehand too, it's just that the latest release now will actually report the exception and exit rather than swallowing it and then silently refusing any future requests.

[MikhailArkhipov]

Verify with new LS.

[MikhailArkhipov]

Verified, there is a test for that in #602.
Fix is not in public release yet, it will be some time before it gets to public since we are in a major rework of the LS.