feat(http): Add 'format' to http backend

[thejcannon]

(From discussion #6951)

Summary

This PR adds a format config to http backend, to help in corner-cases where the format isn't given (or is wrong).

Testing

Added a new E2E test (those are awesome!!)

Note: We could add more tests:

• one for compressed binary without extensions format = "gz"
• one for "completely wrong extension"
  if you wanna upload more binaries!

---

Note

Adds a format option (with platform-specific overrides) to force archive type detection when URLs lack or misstate extensions, with docs and e2e tests.

• Backend:
  • Support explicit format (incl. platform-specific) in src/backend/http.rs by appending the given format to the downloaded file path for correct archive detection.
  • Adjust extraction logic to handle compressed binaries/raw files using the updated path/extension.
• Docs:
  • Document format option and platform-specific formats in docs/dev-tools/backends/http.md.
• Tests:
  • Add e2e/backend/test_http_format covering explicit format and per-platform format.

^{Written by Cursor Bugbot for commit 3c978d7. This will update automatically on new commits. Configure here.}

[jdx]

bugbot run

[cursor]

✅ Bugbot reviewed your changes and found no bugs!

---

[thejcannon]

Whoops, forgot I had some local changes, let me revert that 😅

[Copilot]

Pull Request Overview

This PR adds a format configuration option to the HTTP backend, allowing users to explicitly specify archive formats when URLs lack proper file extensions or have incorrect ones.

Key Changes:

• Added format configuration support with platform-specific overrides
• Modified archive format detection to use the specified format when provided
• Added documentation and E2E tests for the new feature

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
src/backend/http.rs	Implements format detection override by appending the specified format as an extension to the file path for proper archive type detection
e2e/backend/test_http_format	Adds E2E tests covering both explicit format specification and platform-specific format configuration
docs/dev-tools/backends/http.md	Documents the new format option with examples for both global and platform-specific usage

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The logic for combining extensions may produce incorrect results. If the file already has a valid extension like 'tar.gz' and the user specifies format='tar.gz', this would result in 'tar.gz.tar.gz'. Consider validating that the specified format isn't already present in the extension before appending it.

Suggested change

	let current_ext = file_path.extension().and_then(|e| e.to_str()).unwrap_or("");
	let new_ext = if current_ext.is_empty() {
	added_extension.clone()
	} else {
	format!("{}.{}", current_ext, added_extension)
	};
	file_path.set_extension(new_ext);
	let file_name = file_path.file_name().and_then(|n| n.to_str()).unwrap_or("");
	// Only append the extension if it's not already present at the end of the file name
	if !file_name.ends_with(&format!(".{}", added_extension)) {
	let current_ext = file_path.extension().and_then(|e| e.to_str()).unwrap_or("");
	let new_ext = if current_ext.is_empty() {
	added_extension.clone()
	} else {
	format!("{}.{}", current_ext, added_extension)
	};
	file_path.set_extension(new_ext);
	}

[thejcannon]

This was arguably intentional for simplicity sake. Trying to "maybe replace" the extension is error-prone (since files like foo-1.0.0 have the extension .0)

[cursor]

Bug: Format Omission Corrupts Cache Integrity

The cache key generation doesn't include the format option, which can cause cache collisions when the same file is used with different format specifications. If two tools use identical files but specify different format values (like tar.gz vs zip vs no format), they'll share the same cache entry even though they require different extraction methods, resulting in incorrect installations.

src/backend/http.rs#L43-L60

mise/src/backend/http.rs

Lines 43 to 60 in 3c978d7

	/// Generate a cache key based on the actual file content (checksum) and extraction options
	fn get_file_based_cache_key(
	&self,
	file_path: &Path,
	opts: &ToolVersionOptions,
	) -> Result<String> {
	let checksum = hash::file_hash_blake3(file_path, None)?;
	// Include extraction options in cache key to handle different extraction needs
	let mut cache_key_parts = vec![checksum.clone()];
	if let Some(strip_components) = opts.get("strip_components") {
	cache_key_parts.push(format!("strip_{strip_components}"));
	}
	let cache_key = cache_key_parts.join("_");
	debug!("Using file-based checksum as cache key: {}", cache_key);
	Ok(cache_key)

 

---

[thejcannon]

@jdx what do you think of that Cursor Bugbot comment? I actually think it may have legs...

[jdx]

it might, I'm not sure