dataview DQL block cannot parse tsv data when the file's extension is .tsv, but can do so when the same data is in a .csv file

[vivvienne]

What happened?

As stated in the title, when tsv data is stored in a .tsv file, the function csv() (for importing csv data) in a dataview DQL block cannot parse that file, but when the same data is stored in a .csv file, it can parse that data without any problem.

It would be great if csv() can also accept .tsv files

DQL

The following is one minimal example. Create two files named test.csv and test.tsv, both files containing the same following text:

header_1	header_2	header_3
a1	a2	a3
b1	b2	b3
c1	c2	c3

(In particular, test.csv will be a file with csv-extension but tsv data.) Put both files at the root folder of your vault.

Now create a note, also at the root folder of your vault, with the following code block inside.

```dataview
TABLE WITHOUT ID header_1, header_2, header_3
FROM csv("test.tsv")
```

This will render as an empty space ("Dataview: No results to show for table query."). But if you parse the csv-extension file instead,

```dataview
TABLE WITHOUT ID header_1, header_2, header_3
FROM csv("test.csv")
```

then the table shows up without any problem.

JS

Dataview Version

0.5.67

Obsidian Version

1.8.9

OS

Windows

[holroy]

There is a hard coded requirement that the csv file has to end with .csv in

obsidian-dataview/src/data-index/index.ts

Line 293 in 9ea88b9

return path.toLowerCase().endsWith(".csv");

Whether that is sensible, or not, that is another matter. I need to look a little more into that aspect of it, but for now you'll need to rename the file extension to .csv.

Is there any particular reason you're using .tsv?

[vivvienne]

Thank you for the clarification about the hard coded check.

For me, it's not really a strict requirement I'm using the .tsv extension for tsv data files, but conceptually it's nicer to have file extensions align with their content. For example if I keep tsv data in a .csv file, other programs or plugins that handle .csv files very often start with assuming the data is in csv form (inferring from the extension), so it can become a chore having to correct the situation, especially in some cases where this cannot be done in a once and for all manner.

Seeing dataview can already parse tsv data, if the bottleneck (in my view) is only the extension restriction, it'd be really nice if it can accept .tsv extension files (with tsv data inside) too.

[holroy]

Without promising to do this, it do seem to be an easy fix, but I also read somewhere that there are vital differences in how csv vs tsv handles escaping of text, and I can see that Dataview is doing some indexing on csv files which possibly need to be addressed too.