# Description
Previously when nushell failed to parse the content type header, it
would emit an error instead of returning the response. Now it will fall
back to `text/plain` (which, in turn, will trigger type detection based
on file extension).
May fix (potentially) nushell/nushell#11927
Refs:
https://discord.com/channels/601130461678272522/614593951969574961/1272895236489613366
Supercedes: #13609
# User-Facing Changes
It's now possible to fetch content even if the server returns an invalid
content type header. Users may need to parse the response manually, but
it's still better than not getting the response at all.
# Tests + Formatting
Added a test for the new behaviour.
# After Submitting
Intends to close#8920
This PR suggests a new flag for the `http` commands, `--redirect-mode`,
which enables users to choose between different redirect handling modes.
The current behaviour of letting ureq silently follow redirects remains
the default, but two new options are introduced here, following the lead
of [JavaScript's `fetch`
API](https://developer.mozilla.org/en-US/docs/Web/API/fetch#redirect):
"manual", where any 3xx response to a request is simply returned as the
command's result, and "error", where any 3xx response causes a network
error like those caused by 4xx and 5xx responses.
This PR is a draft. Tests have not been added or run, the flag is
currently only implemented for the `http get` command, and design tweaks
are likely to be appropriate.
Most notably, it's not obvious to me whether a single flag which can
take one of three values is the nicest solution here.
We might instead consider two binary flags (like
`--no-following-redirects` and `--disallow-redirects`, although I'm bad
at naming things so I need help with that anyway), or completely drop
the "error" option if it's not deemed useful enough. (I personally think
it has some merit, especially since 4xx and 5xx responses are already
treated as errors by default; So this would allow users to treat only
immediate 2xx responses as success)
# User-facing changes
New options for the `http [method]` commands. Behaviour remains
unchanged when the command line flag introduced here is not used.
# Description
As described in https://github.com/nushell/nushell/issues/9912, the
`http` command could display the request headers with the `--full` flag,
which could help in debugging the requests. This PR adds such
functionality.
# User-Facing Changes
If `http get` or other `http` command which supports the `--full` flag
is invoked with the flag, it used to display the `headers` key which
contained an table of response headers. Now this key contains two nested
keys: `response` and `request`, each of them being a table of the
response and request headers accordingly.
# Description
See also: #9743
Before:
`http <subcommand> -H` took a list in the form:
```nushell
[my-header-key-A my-header-value-A my-header-key-B my-header-value-B]
```
Now:
In addition to the old format, Records can be passed, For example,
```nushell
> let reqHeaders = {
Cookie: "acc=barfoo",
User-Agent: "Mozilla/7.0 (Windows NT 33.0; Win64; x64) AppleWebKit/1038.90 (KHTML, like Gecko)"
}
> http get -H $reqHeaders https://example.com
```
is now equivalent to
```nushell
http get -H [Cookie "acc=barfoo" User-Agent "Mozilla/7.0 (Windows NT 33.0; Win64; x64) AppleWebKit/1038.90 (KHTML, like Gecko)"] https://example.com
```
# User-Facing Changes
No breaking changes, but Records can now also be passed to `http
<subcommand> -H`.
# Tests + Formatting
# After Submitting
This PR adds tests to confirm that:
1. `http get` does the right thing (bail) when it encounters common SSL
errors
2. the `--insecure` flag works to ignore SSL errors
It's prompted by #8098, where `--insecure` stopped working and we didn't
notice until it was reported by a user.
## Deets + considerations
This PR uses [badssl.com](https://badssl.com/), a very handy website
affiliated with the Google Chrome team. The badssl authors mention that
stability is not guaranteed:
> Most subdomains are likely to have stable functionality, but anything
could change without notice.
I suspect that the badssl.com subdomains I've chosen will be stable
enough in practice. Can revisit this if the tests end up being flaky.
This PR does mean our tests are now making an external network call...
which I _think_ is OK. Our CI isn't exactly designed for offline
machines; test runners already have to download a bunch of crates etc. I
think the new tests are quick enough:
# Description
_(Thank you for improving Nushell. Please, check our [contributing
guide](../CONTRIBUTING.md) and talk to the core team before making major
changes.)_
This pull request removes `Reqwest` and replaces it with `Ureq` to
remove some of our dependencies, giving us faster compile times as well
as smaller binaries. `Ureq` does not have an async runtime included so
we do not need build heavy dependencies such as `Tokio`. From older
tests I had the number of build units be reduced from `430 -> 392`.
The default of `Ureq` uses `Rustls` but it has been configured to
instead use `native_tls` which should work exactly the same as the `tls`
works now.
I removed `content-length` from the http commands as after refactoring i
did not see a reason to have it available, correct me if this is
something we should preserve.
In the medium, to long term, we should maybe consider changing to
`rustls` to have the same `tls` on all platforms.
# User-Facing Changes
_(List of all changes that impact the user experience here. This helps
us keep track of breaking changes.)_
# Tests + Formatting
Don't forget to add tests that cover your changes.
Make sure you've run and fixed any issues with these commands:
- `cargo fmt --all -- --check` to check standard code formatting (`cargo
fmt --all` applies these changes)
- `cargo clippy --workspace -- -D warnings -D clippy::unwrap_used -A
clippy::needless_collect` to check that you're using the standard code
style
- `cargo test --workspace` to check that all tests pass
# After Submitting
If your PR had any user-facing changes, update [the
documentation](https://github.com/nushell/nushell.github.io) after the
PR is merged, if necessary. This will help us keep the docs up to date.
# Description
Following the comment from @fdncred:
https://github.com/nushell/nushell/pull/8144#issuecomment-1442514338. I
added some unit tests for HTTP commands. The tests are not exhaustive,
but at least, this is a first good step IMO.
# User-Facing Changes
None.
# Tests + Formatting
Don't forget to add tests that cover your changes.
Make sure you've run and fixed any issues with these commands:
- `cargo fmt --all -- --check` to check standard code formatting (`cargo
fmt --all` applies these changes)
- `cargo clippy --workspace -- -D warnings -D clippy::unwrap_used -A
clippy::needless_collect` to check that you're using the standard code
style
- `cargo test --workspace` to check that all tests pass
# After Submitting
If your PR had any user-facing changes, update [the
documentation](https://github.com/nushell/nushell.github.io) after the
PR is merged, if necessary. This will help us keep the docs up to date.
