Bug: ENOTDIR, not a directory

[baruchiro]

This is a bug report from the app interface.

• Open in this repo (currently opened in the old one)
• Typo in the title
• Add OS and App version to the bug report

Detailes

I installed the app, then I configured my bank accounts.
After that, I configured the JsonExporter to the full path: C:\Users\BaruchR\source\repos\transactions\data\transactions.json and the SheetsExporter I configured to create a new spreadsheet (because it has no access to exists spreadsheets).
I didn't change to Config.

Now I pressed on the RUN button and almost immediately I got the error: ENOTDIR.

The problem is also the message is not descriptive, I don't know where it comes from, and there are no logs about that.

You can see לאומי don't have an import error. Don't know if it points to something.

Log

[2020-12-23 21:50:26.651] [info] Welcome to budget-tracking log
[2020-12-23 21:50:26.661] [info] Version: 0.1.3-beta
[2020-12-23 21:50:38.471] [info] The renderer process got the logger
[2020-12-23 21:57:33.357] [info] Welcome to budget-tracking log
[2020-12-23 21:57:33.376] [info] Version: 0.1.3-beta
[2020-12-23 21:57:45.838] [info] The renderer process got the logger
[2020-12-24 09:30:18.589] [info] Welcome to budget-tracking log
[2020-12-24 09:30:18.605] [info] Version: 0.1.3-beta
[2020-12-24 09:30:22.235] [info] The renderer process got the logger

Reference: download-chromium#download

[baruchiro]

Update, maybe another bug:

There is an error with leumi but no "alert" sign in לאומי

[baruchiro]

This is about ChromeDownloader, I'm pretty sure.

[baruchiro]

@brafdlog We are using ChromeDownloader during the scrape method, and it's mean that it will run for each bank account.
In the first run, it will try to download the Chromium driver multiple times in parallel. I think it downloads to a temp location, then copy to the target, and there it getting an error.

https://github.com/juliangruber/download-chromium/blob/c5bf61eb9186914008614bf30d2d07dc418485fb/index.js#L93-L104

I see some options here:

1. Get the chrome path from the config. It's mean that the scraping does not care about the chrome downloading. It sounds logical since it reduces one more responsibility from the scraping.
2. Download the chrome some levels up from the scraping, before the parallel, and pass the path down to the scraping.
3. Hide the chrome downloader behind a custom class that will always return the same promise (if exists).

WDYT?

Edit:

I'm not sure this is the bug. Maybe the bug is just in the destination path. But maybe all that is right in any way.
I will check it next time.

[baruchiro]

Reopen- still want to fix the user report problem.