Fix paste attributes after #1902

[ellatrix]

No description provided.

[ellatrix]

@aduth Should all of these attributes have defaults set?

[aduth]

Alternatively, I'd wondered if we just need to feed in the attributes as source instead of defining their full type, since the full type is essentially a duplicate of what should already exist in the destination block type.

diff --git a/blocks/api/paste.js b/blocks/api/paste.js
index 84285728..76e53d3c 100644
--- a/blocks/api/paste.js
+++ b/blocks/api/paste.js
@@ -1,7 +1,7 @@
 /**
  * External dependencies
  */
-import { find, get, flowRight as compose } from 'lodash';
+import { find, get, flowRight as compose, mapValues } from 'lodash';
 
 /**
  * WordPress dependencies
@@ -96,7 +96,7 @@ export default function( nodes ) {
 
                        const attributes = getSourcedAttributes(
                                node.outerHTML,
-                               transform.attributes,
+                               mapValues( transform.attributes, ( source ) => ( { source } ) ),
                        );
 
                        return createBlock( blockType.name, attributes );

Can we add some unit tests for paste block transforms to capture these issues?

[codecov]

Codecov Report

Merging #2408 into master will increase coverage by 0.54%.
The diff coverage is 50%.

@@            Coverage Diff             @@
##           master    #2408      +/-   ##
==========================================
+ Coverage    25.9%   26.45%   +0.54%     
==========================================
  Files         157      157              
  Lines        4853     4858       +5     
  Branches      822      824       +2     
==========================================
+ Hits         1257     1285      +28     
+ Misses       3035     3017      -18     
+ Partials      561      556       -5

Impacted Files	Coverage Δ
blocks/library/image/index.js	15% <ø> (ø)	⬆️
blocks/library/paragraph/index.js	47.05% <ø> (ø)	⬆️
blocks/library/list/index.js	6.97% <0%> (ø)	⬆️
blocks/library/heading/index.js	23.8% <0%> (ø)	⬆️
blocks/api/paste.js	97.36% <100%> (+35.82%)	⬆️
blocks/color-palette/index.js	0% <0%> (ø)	⬆️
blocks/library/cover-text/index.js	36% <0%> (+1%)	⬆️
blocks/api/paste/remove-spans.js	77.77% <0%> (+66.66%)	⬆️
blocks/api/paste/strip-attributes.js	100% <0%> (+100%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 77620b2...c04c554. Read the comment docs.

[ellatrix]

What about cases where they don't map? I guess we should think of it when it pops up then.

[aduth]

Also, it should be responsibility of createBlock to fill in attribute defaults.

[ellatrix]

Why shouldn't we then just loop through getSourcedAttributes instead of source and see if it returns any attributes? That removes the need entirely for these raw transforms.

[ellatrix]

Maybe this could work if we add required, and then look if that's satisfied.

[ellatrix]

Okay, maybe not, there will always be strange cases we'll need to handle. Let's just add what you suggested.

[aduth]

Since the block registry is global, we should make sure to reset to its original state after our tests are done.

Example from serializer tests:

gutenberg/blocks/api/test/serializer.js

Lines 20 to 24 in 77620b2

	afterEach( () => {
	getBlockTypes().forEach( block => {
	unregisterBlockType( block.name );
	} );
	} );

[aduth]

Why shouldn't we then just loop through getSourcedAttributes instead of source and see if it returns any attributes? That removes the need entirely for these raw transforms.

Yeah, in the case of the raw transforms, if we know that the original content is a match, we shouldn't need to redefine attributes and instead just trigger the default parsing of getBlockAttributes.

I think I might have been a little overreaching with the renaming of matcher: to source:. In the case of raw transforms, I think matcher: makes a little more sense for what it's doing (returning boolean, maybe also match / isMatch). What do you think?

[ellatrix]

Sounds good. Changed it.

[aduth]

Minor: Conventionally setup and teardown occurs together before any test cases.

https://facebook.github.io/jest/docs/setup-teardown.html