feat(c14n): add canonicalization functionality

[Tieske]

GitHub: fixGH-36

[kou]

Suggested change

[kou]

Wow! I didn't know that libxml2 provides the c14n API.

[Tieske]

@kou thx for the review. Updated the PR. please have another look.

[Tieske]

For the CI failures, not sure what happened. Since the failures are unrelated to this PR, it reminds me of the ffi-reloading issue. The test frameworks tend to isolate tests, by clearing the environment and run the next test clean. However the LuaJIT ffi is not suited for that and will occasionally segfault if you do.

See lunarmodules/busted#369 (comment)

[Tieske]

I reworked your review comments one-by-one, a commit per comment. Just the last commit (merge into document.lua) is a big one that obfuscates all other changes. So probably best to review the changes on a per commit basis.

[kou]

Could you add some tests for document:canonicalize?

[Tieske]

Could you add some tests for document:canonicalize?

yes it should, but my priorities are shifting, so might take a while.

[Tieske]

basic test added

[Tieske]

@kou any update on this?

[kou]

Ah, we don't need to test C14N behaviors implemented in libxml2. Because they must be tested in libxml2.

We only need to test our API. For example, document:canonicalize(function) case, document:canonicalize(nodes) case, document:canonicalize(..., options) case and so on.

[Tieske]

agreed, all I did was just grab some of them to verify it works as intended.
This tests with and without "with_comments" probably also needs a test with the prefix list

[kou]

Sorry for my late response...

How about the followings to proceed this?

1. Can we remove the PrefixToken format parsing feature from this PR and discuss it as a follow-up task? feat(c14n): add canonicalization functionality #37 (comment)
2. Can I write tests for document:canonicalize() and push to this branch? (Or could you update tests?) feat(c14n): add canonicalization functionality #37 (comment)
3. Can we merge this after 1. and 2.?

(We can write documents for this later.)

[Tieske]

Sorry for my late response...

How about the followings to proceed this?

1. Can we remove the `PrefixToken` format parsing feature from this PR and discuss it as a follow-up task? [feat(c14n): add canonicalization functionality #37 (comment)](https://github.com/clear-code/xmlua/pull/37#discussion_r1878857950)

Done

2. Can I write tests for `document:canonicalize()` and push to this branch? (Or could you update tests?) [feat(c14n): add canonicalization functionality #37 (comment)](https://github.com/clear-code/xmlua/pull/37#discussion_r1878862733)
3. Can we merge this after 1. and 2.?

if you handle item 2, then 3 is also yours to handle. So please go ahead.

[kou]

I've fixed test failures on macOS on master and rebased on master.

I've added tests. I also added some changes:

• Removed C14N_ prefix from mode value because it's redundant in the document:canonicalize() context.
• Fixed a bug that document:canonicalize({node}) doesn't work.
• Improved some names to adjust other codes.

I'll merge this if you don't object these changes.

[Tieske]

Lgtm!

[kou]

Merged!

Should we release the current master as a new version? Or should we release a new version after #40?

[Tieske]

no real opinion here. Since that would add a feature, it can be done at any later time as well.

[kou]

OK. I'll release a new version before #40.

[kou]

Released.