fix most of the autoFix issues using TSLint 4.0

[rlasjunies]

Hi @egamma

Here is a proposal of fix for most of the autoFix issue identified since TSLint 4.0. #135

Actions done:

• convert the offset in position
• add a sub tslint-tests at the root of vscode-tslint to collect some tests cases used in these days

Remaining activities identified:

• no-var-keyword still not supported. This autoFix need more work in the alogrithm. This is the only one not working. => To Do : support multiple inner replacements
• during the tests I've got the 'innerText' from undefined error ... without the opportunity to reproduce.
• there is a tsc error in the server code but does not prohibit emit

if (configurationResult.error) {
   throw configurationResult.error;
}

I let you have trial on this proposal, it should help.

BR

[egamma]

@rlasjunies thanks for looking into this.

Regarding

there is a tsc error in the server code but does not prohibit emit

sigh, tslint broke the API between version 4.0.1 and 4.0.2 palantir/tslint#1789. I'll try to work around it.

[rlasjunies]

that pretty good, I've seen you have fixed the innerText and the "error".

For the no_var_keyword, I think we need time to fix it. I do not understand what is provided by tslint innerReplacements. Do you know what is expected when there are several several records in the innerReplacements array.

In all the case, I believe it better to publish the fixes as they support the other autoFix rules from tslint 4.0 ( even import-ordered ).

[egamma]

@rlasjunies I've manually merged the changes, thanks 🌷

The tslint-tests are good, we should look into automating these tests. I'll look into this.

• I've added an array-type test case to tslint-test. This is also an auto fix with multiple edits
• I changed the code to not offer an auto fix if the fix consists of more than 1 replacement. In this way the source code isn't damaged.

For the no_var_keyword, I think we need time to fix it. I do not understand what is provided by tslint innerReplacements. Do you know what is expected when there are several several records in the innerReplacements array.

From looking at how the replacements are constructed for some rules in tslint I assume that the idea is that the replacements are applied in sequence. This means we need support creating multiple edits in record actions. Applying multiple (non-overlapping) edits is supported by the editor.

I agree that we should publish this improved version. I'll do so today.

Closing this PR.