This issue fixes two cases where `{{__hugo_ctx` artifacts were left in the rendered output:
1. Inclusion when `.RenderShortcodes` is wrapped in HTML.
2. Inclusion of Markdown file without a trailing newline in some cases.
Closes#12854
Updates #12998
Note that this also adds a new `.ContentWithoutSummary` method, and to do that we had to unify the different summary types:
Both `auto` and `manual` now returns HTML. Before this commit, `auto` would return plain text. This could be considered to be a slightly breaking change, but for the better: Now you can treat the `.Summary` the same without thinking about where it comes from, and if you want plain text, pipe it into `{{ .Summary | plainify }}`.
Fixes#8680Fixes#12761Fixes#12778Fixes#716
* Make throwOnError=true the new default
* Handle JS errors as part of the RPC request/response flow
* Return a new Result type with .Err on it
This enables constructs on the form:
```handlebars
{{ with transform.ToMath "c = \\foo{a^2 + b^2}" }}
{{ with .Err }}
{{ warnf "error: %s" . }}
{{ else }}
{{ . }}
{{ end }}
{{ end }}
```
Note that the new `Result` type behaves like `template.HTML` (or a string if needed) when printed, but it will panic if in a error state.
Closes#12748
While very useful on its own (and combined with the passthrough render hooks), this also serves as a proof of concept of using WASI (WebAssembly System Interface) modules in Hugo.
This will be marked _experimental_ in the documentation. Not because it will be removed or changed in a dramatic way, but we need to think a little more how to best set up/configure similar services, define where these WASM files gets stored, maybe we can allow user provided WASM files plugins via Hugo Modules mounts etc.
See these issues for more context:
* https://github.com/gohugoio/hugo/issues/12736
* https://github.com/gohugoio/hugo/issues/12737
See #11927
This change allows to use permalink tokens in url front matter fields. This should be useful to target more specific pages instead of using a global permalink configuration. It's expected to be used with cascade.
Fixes#9714
Note that we only use this for change detection.
The previous implementation invoked `MD5FromReaderFast` that created a MD5 has from 8 64 bytes chunks in the file, which is obviously very fast. The new implementation creates the hash from the entire file and ... seems to be even more effective:
```
name old time/op new time/op delta
HashImage-10 9.45µs ±21% 10.89µs ± 1% ~ (p=0.343 n=4+4)
name old alloc/op new alloc/op delta
HashImage-10 144B ± 0% 8B ± 0% -94.44% (p=0.029 n=4+4)
name old allocs/op new allocs/op delta
HashImage-10 4.00 ± 0% 1.00 ± 0% -75.00% (p=0.029 n=4+4)
```
