rustdoc: simplify handling of plain text output

[jsha]

Replace use of f.alternate() in format.rs, and offers instead Type::print_plain, and Path::print_plain.

Adopt Askama template for rendering of functions and methods.

Motivation:

In rustdoc we generally emit HTML, but there are a few places where we need to emit a plain text form of clean::Type and/or clean::Path:

• In the sidebar, under Trait Implementations, we render traits, e.g. AsRef<str>. These should be plain because we don't want to separately link AsRef and str to those respective types; we want a single link that goes to that trait implementation.
• When creating ids and internal hrefs for headings related to trait implementations, we generate the ids based on a plain text version of the trait (and the type it is implemented for).
• Similarly, when creating the datastructures used to render "Notable traits", we index those datastructures by a plain text version of the type that has notable trait implementations.
• When rendering a function or associated function, we decide whether or not to line-wrap the declaration based on whether the plain text form would exceed 80 characters (imitating rustfmt).

Originally, plain text output was controlled via Formatter's f.alternate() mechanism, so you could generate HTML by formatting with {} or plain text by formatting with {:#}. This avoids having to generate HTML only to strip it out. However, it leads to a lot of duplication and branching in format.rs. Since the only HTML we care about comes from within a single module, we can write a very simple HTML remover that strips all tags and only cares about a handful of entities.

This also lets us simplify calculating the width of a function header - we can simply pass a template and count its bytes.

This PR can be reviewed commit-by-commit.

Related: #108868

[rustbot]

r? @notriddle

(rustbot has picked a reviewer for you, use r? to override)

[notriddle]

@bors try @rust-timer queue

[bors]

⌛ Trying commit dde3cced5ea68bec455d2b47f49b54621c692c95 with merge 56fa9753c5bb8be807bdabf9018e03618dd75fbf...

[bors]

💔 Test failed - checks-actions

[notriddle]

Looks relevant: https://github.com/rust-lang-ci/rust/actions/runs/4469549308/jobs/7851815712#step:26:40193

thread 'rustc' panicked at 'called Result::unwrap() on an Err value: Error', src/librustdoc/html/format.rs:131:34

[jsha]

Yep! Will take a look in a bit. Thanks for digging up the log line.

[fmease]

Hey :) Does this PR fix #109488 perchance?

[jsha]

When I make the example from #109488 into a test, I get a compiler ICE; presumably it needs the code you're working on?

I have a pretty strong suspicion the issue is this line, which I didn't touch yet because it looked fishy and I couldn't figure out the intention:

rust/src/librustdoc/html/format.rs

Line 1158 in 4c0f500

_ => write!(f, "{}{:#}", assoc.name, assoc.args.print(cx))?,

{:#} is requesting alternate formatting (i.e. plaintext), but in this branch of things, alternate formatting hasn't been requested by the caller. So this is likely emitting < into an HTML context. If you change it to {} in your branch I'm guessing that will fix things.

I should also change it in this branch because {:#} is becoming meaningless.

[jsha]

@bors try @rust-timer queue

[bors]

⌛ Trying commit 236c18f with merge 9c4e97025c634ec4878a1580cc34680aa24ce17e...

[bors]

☀️ Try build successful - checks-actions
Build commit: 9c4e97025c634ec4878a1580cc34680aa24ce17e (9c4e97025c634ec4878a1580cc34680aa24ce17e)

[rust-timer]

Finished benchmarking commit (9c4e97025c634ec4878a1580cc34680aa24ce17e): comparison URL.

Overall result: ❌ regressions - ACTION NEEDED

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR may lead to changes in compiler perf.

Next Steps: If you can justify the regressions found in this try perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please fix the regressions and do another perf run. If the next run shows neutral or positive results, the label will be automatically removed.

@bors rollup=never
@rustbot label: -S-waiting-on-perf +perf-regression

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.6%	[0.6%, 5.1%]	15
Regressions ❌ (secondary)	3.2%	[0.8%, 7.3%]	4
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	2.6%	[0.6%, 5.1%]	15

Max RSS (memory usage)

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	3.6%	[3.2%, 4.0%]	2
Regressions ❌ (secondary)	1.3%	[0.5%, 2.2%]	2
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.6%	[-0.8%, -0.4%]	6
All ❌✅ (primary)	3.6%	[3.2%, 4.0%]	2

Cycles

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	2.6%	[2.0%, 4.0%]	6
Regressions ❌ (secondary)	1.2%	[0.4%, 4.5%]	10
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.5%	[-0.5%, -0.5%]	1
All ❌✅ (primary)	2.6%	[2.0%, 4.0%]	6

[jsha]

Bummer, looks like performance got worse. I think I can probably iterate and improve it though.

[Dylan-DPC]

@jsha any updates on this?

[jsha]

I'll close this for now and hope to come back to it.