refactor: Adjust inlay rendering and improve completion logic

[lkk214]

Description

• Single-line completion sometimes does not work
• There is a blank area after the cursor when completing multiple lines
• Incompatible line breaks cause accept to fail

Screenshots

Testing

[netlify]

✅ Deploy Preview for continuedev canceled.

Name	Link
🔨 Latest commit	79736c3
🔍 Latest deploy log	https://app.netlify.com/sites/continuedev/deploys/67741b3a79f1ff00089f9d2d

[sestinj]

This works great for me, I'm just curious if the url -> path was the solution to something that was previously broken

[sestinj]

why change url to path?

[sestinj]

(was this the solution to a clear problem?)

[lkk214]

I was just about to comment on it. I didn't pay attention to #3247, need to revert it here.

Because using File(URI(filepath)) will cause an error in Windows, I made temporary changes locally and submitted this by mistake.

[lkk214]

Uh, maybe it's better to fix it than revert it.

[lkk214]

Also, add directory checks and optimize token usage.

• Added check to skip directories when reading file content.
• Optimized token usage by removing unnecessary \r characters.

[sestinj]

Looks good! @Patrick-Erichsen for a second pass, then good to merge

[Patrick-Erichsen]

@RomneyDa any thoughts here? I'm wondering if .removeSuffix("/") is going to not work properly for non-macOS users

[Patrick-Erichsen]

LGTM 🫡

[AfterStories]

@lkk214 @Patrick-Erichsen @sestinj
I still found new similar problems, I think it is also caused by the same bug, but it has not been fixed:
When I select a piece of code and use the CTRL+I shortcut key, as shown in the screenshot, you can see that the model name is not displayed, only show a null

And at the same time, the log in the Run Extension window:

Error handling message: {"messageType":"config/getSerializedProfileInfo","data":{"result":{"config":{"allowAnonymousTelemetry":false,"models":[{"provider":"anthropic","model":"Gosu Model","title":"Gosu Model"... ....(After here, it's my config. json content) 
java.lang.NullPointerException: null cannot be cast to non-null type kotlin.collections.Map<kotlin.String, kotlin.Any>
Registered handler for editor

[lkk214]

From the log you provided, it seems that it is a response parsing error. It is not the same issue.
I have not updated the code, so it works normally locally. The relevant code for response parsing:

continue/extensions/intellij/src/main/kotlin/com/github/continuedev/continueintellijextension/editor/InlineEditAction.kt

Lines 131 to 135 in 7af1090

	continuePluginService.coreMessenger?.request("config/getSerializedProfileInfo", null, null) { response ->
	val config = response as Map<String, Any>
	val models = (config["config"] as Map<String, Any>)["models"] as List<Map<String, Any>>
	modelTitles.addAll(models.map { it["title"] as String })
	}

[AfterStories]

From the log you provided, it seems that it is a response parsing error. It is not the same issue. I have not updated the code, so it works normally locally. The relevant code for response parsing:

continue/extensions/intellij/src/main/kotlin/com/github/continuedev/continueintellijextension/editor/InlineEditAction.kt

Lines 131 to 135 in 7af1090

	continuePluginService.coreMessenger?.request("config/getSerializedProfileInfo", null, null) { response ->
	val config = response as Map<String, Any>
	val models = (config["config"] as Map<String, Any>)["models"] as List<Map<String, Any>>
	modelTitles.addAll(models.map { it["title"] as String })
	}

Yes, I also found this code, and I saw that it made a "config/getSerializedProfileInfo" request to get the content of the config,
but maybe there was a problem in the kt code post message to the “core” code ，so failed to get the correct response.
Or the core code failed to parse the config.json file and return its content，I am not understand the message manager part of the kt code to core layer code