Backport MessageContext parser changes from fluent 0.6.0 #148
Conversation
| @@ -13,6 +13,7 @@ export default { | |||
| ], | |||
| 'plugins': [ | |||
| 'external-helpers', | |||
| 'babel-plugin-optimize-starts-with', | |||
There was a problem hiding this comment.
I wanted to keep the code as close to fluent@0.6.0 as possible. We started using startsWith('-') in 0.6.0 to tell terms and messages apart. This plugin converts it to charCodeAt(0) == 45.
| "tag1" | ||
| ] | ||
| } | ||
| "qux": "Qux\n#tag1" |
There was a problem hiding this comment.
Tags are no longer parsed by the runtime parser so this simply becomes a multiline message value.
There was a problem hiding this comment.
I think it'd be good to unsupport tags in the tests, and probably relnotes.
Removing the tags from the ftl sources feels more discoverable of the change actually happening.
There was a problem hiding this comment.
I cherry picked ca5f604 as the first commit in the PR.
|
See also: all the changes in the |
stasm
force-pushed
the
0.4.x-update-parser
branch
from
da152ec to
9277844
Compare
| "tag1" | ||
| ] | ||
| } | ||
| "qux": "Qux\n#tag1" |
There was a problem hiding this comment.
I think it'd be good to unsupport tags in the tests, and probably relnotes.
Removing the tags from the ftl sources feels more discoverable of the change actually happening.
| "tag2" | ||
| ] | ||
| }, | ||
| "key4": "Value\n\n\n#tag1\n\n\n#tag2", |
| @@ -72,7 +74,8 @@ class RuntimeParser { | |||
| const ch = this._source[this._index]; | |||
|
|
|||
| // We don't care about comments or sections at runtime | |||
| if (ch === '/') { | |||
| if (ch === '/' || | |||
| (ch === '#' && [' ', '#'].includes(this._source[this._index + 1]))) { | |||
There was a problem hiding this comment.
You may need to skip also if ch === '#' && ch2 === '\n'.
There was a problem hiding this comment.
This has to be fixed also on master in #149. I'll backport it together with other changes when master becomes 0.6.1.
(cherry picked from commit ca5f604)
In order to make it easier to upgrade to the new syntax, we'll release fluent
0.4.3 which will keep the API compatibility with 0.4.2 and will be able to read
Fluent Syntax 0.5 files (just like fluent 0.6.0).
These changes make fluent 0.4.x able to read both Syntax 0.4 and Syntax 0.5 files:
- Supported Syntax 0.4 features:
- `//` comments.
- Sections (`[[ Foo ]]`).
- Message definitions may omit the `=` after the identifier if they don't
have a value.
- Supported Syntax 0.5 features:
- `#`, `##` and `###` comments.
- Identifiers starting with a dash (`-`) are parsed as terms.
The only feature of the old Syntax 0.4 which isn't supported anymore are Tags.
In practice we didn't see any of the current consumers of `fluent` 0.4.2 take
advantage of tags. In Syntax 0.5, [tags were replaced by attribute defined on
terms][terms].
[terms]: https://github.com/projectfluent/fluent/blob/master/spec/CHANGELOG.md#050-january-31-2018
stasm
force-pushed
the
0.4.x-update-parser
branch
from
9277844 to
b468110
Compare
In order to make it easier to upgrade to the new syntax, we'll release
fluent0.4.3 which will keep the API compatibility with 0.4.2 and will be able to read Fluent Syntax 0.5 files (just likefluent0.6.0).These changes make
fluent0.4.x able to read both Syntax 0.4 and Syntax 0.5 files:Supported Syntax 0.4 features:
//comments.[[ Foo ]]).=after the identifier if they don't have a value.Supported Syntax 0.5 features:
#,##and###comments.-) are parsed as terms.The only feature of the old Syntax 0.4 which isn't supported anymore are Tags. In practice we didn't see any of the current consumers of
fluent0.4.2 take advantage of tags. In Syntax 0.5, tags were replaced by attribute defined on terms.