feat: support Text as an allowed child of body

We have a few use cases where we are publishing bodies with text
content:
- Some articles have trailing text before the body tag closes (we were
  previously handling this in the from-bodyxml transformer, but it was a
  hack)
- FT Events publish a body without any tags i.e. just text
- CCC fallbacks and Clip transcripts currently publish a body with just
  text

There's not a strong reason not to support Text as a valid child of Body
(despite the fact Spark doesn't typically allow it), so let's allow it.

Author: adgad

README.md

@@ -105,6 +105,7 @@ type BodyBlock =
 	| Tweet
 	| Video
 	| YoutubeVideo
+	| Text
 ```
 
 `BodyBlock` nodes are the only things that are valid as the top level of a `Body`.

content-tree.d.ts

@@ -1,5 +1,5 @@
 export declare namespace ContentTree {
-    type BodyBlock = Paragraph | Heading | ImageSet | Flourish | BigNumber | CustomCodeComponent | Layout | List | Blockquote | Pullquote | ScrollyBlock | ThematicBreak | Table | Recommended | Tweet | Video | YoutubeVideo;
+    type BodyBlock = Paragraph | Heading | ImageSet | Flourish | BigNumber | CustomCodeComponent | Layout | List | Blockquote | Pullquote | ScrollyBlock | ThematicBreak | Table | Recommended | Tweet | Video | YoutubeVideo | Text;
     type LayoutWidth = "auto" | "in-line" | "inset-left" | "inset-right" | "full-bleed" | "full-grid" | "mid-grid" | "full-width";
     type Phrasing = Text | Break | Strong | Emphasis | Strikethrough | Link;
     interface Node {
@@ -272,7 +272,7 @@ export declare namespace ContentTree {
         attributes: CustomCodeComponentAttributes;
     }
     namespace full {
-        type BodyBlock = Paragraph | Heading | ImageSet | Flourish | BigNumber | CustomCodeComponent | Layout | List | Blockquote | Pullquote | ScrollyBlock | ThematicBreak | Table | Recommended | Tweet | Video | YoutubeVideo;
+        type BodyBlock = Paragraph | Heading | ImageSet | Flourish | BigNumber | CustomCodeComponent | Layout | List | Blockquote | Pullquote | ScrollyBlock | ThematicBreak | Table | Recommended | Tweet | Video | YoutubeVideo | Text;
         type LayoutWidth = "auto" | "in-line" | "inset-left" | "inset-right" | "full-bleed" | "full-grid" | "mid-grid" | "full-width";
         type Phrasing = Text | Break | Strong | Emphasis | Strikethrough | Link;
         interface Node {
@@ -546,7 +546,7 @@ export declare namespace ContentTree {
         }
     }
     namespace transit {
-        type BodyBlock = Paragraph | Heading | ImageSet | Flourish | BigNumber | CustomCodeComponent | Layout | List | Blockquote | Pullquote | ScrollyBlock | ThematicBreak | Table | Recommended | Tweet | Video | YoutubeVideo;
+        type BodyBlock = Paragraph | Heading | ImageSet | Flourish | BigNumber | CustomCodeComponent | Layout | List | Blockquote | Pullquote | ScrollyBlock | ThematicBreak | Table | Recommended | Tweet | Video | YoutubeVideo | Text;
         type LayoutWidth = "auto" | "in-line" | "inset-left" | "inset-right" | "full-bleed" | "full-grid" | "mid-grid" | "full-width";
         type Phrasing = Text | Break | Strong | Emphasis | Strikethrough | Link;
         interface Node {
@@ -805,7 +805,7 @@ export declare namespace ContentTree {
         }
     }
     namespace loose {
-        type BodyBlock = Paragraph | Heading | ImageSet | Flourish | BigNumber | CustomCodeComponent | Layout | List | Blockquote | Pullquote | ScrollyBlock | ThematicBreak | Table | Recommended | Tweet | Video | YoutubeVideo;
+        type BodyBlock = Paragraph | Heading | ImageSet | Flourish | BigNumber | CustomCodeComponent | Layout | List | Blockquote | Pullquote | ScrollyBlock | ThematicBreak | Table | Recommended | Tweet | Video | YoutubeVideo | Text;
         type LayoutWidth = "auto" | "in-line" | "inset-left" | "inset-right" | "full-bleed" | "full-grid" | "mid-grid" | "full-width";
         type Phrasing = Text | Break | Strong | Emphasis | Strikethrough | Link;
         interface Node {

libraries/from-bodyxml/index.js

@@ -448,19 +448,11 @@ export function fromXast(bodyxast, transformers = defaultTransformers) {
   })(bodyxast);
 }
 
-/**
- * Turns e.g. </p> </body> into </p></body>, which a lot of our content has
- * This prevents random strat text nodes being created in bodyTree
- * @param {string} xml
- * @returns {string}
- */
-function removeWhitespaceBeforeBodyTag(xml) {
-  return xml.replace(/<\/p>\s+<\/body>/gi, '</p></body>');
-}
+
 
 /** @param {string} bodyxml */
 export function fromXML(bodyxml) {
-  return fromXast(xastFromXml(removeWhitespaceBeforeBodyTag(bodyxml)));
+  return fromXast(xastFromXml(bodyxml));
 }
 
 export default fromXML;

schemas/body-tree.schema.json

@@ -119,6 +119,9 @@
                 },
                 {
                     "$ref": "#/definitions/ContentTree.transit.YoutubeVideo"
+                },
+                {
+                    "$ref": "#/definitions/ContentTree.transit.Text"
                 }
             ]
         },

schemas/content-tree.schema.json

@@ -144,6 +144,9 @@
                 },
                 {
                     "$ref": "#/definitions/ContentTree.full.YoutubeVideo"
+                },
+                {
+                    "$ref": "#/definitions/ContentTree.full.Text"
                 }
             ]
         },

schemas/transit-tree.schema.json

@@ -144,6 +144,9 @@
                 },
                 {
                     "$ref": "#/definitions/ContentTree.transit.YoutubeVideo"
+                },
+                {
+                    "$ref": "#/definitions/ContentTree.transit.Text"
                 }
             ]
         },

tests/bodyxml-to-content-tree/input/simple-old-post.xml

@@ -8,5 +8,5 @@
     <pull-quote-text><p>Maecenas ac ipsum in elit aliquam consectetur. Proin felis metus, efficitur et nulla eu, interdum malesuada diam.</p></pull-quote-text><pull-quote-image><ft-content data-embedded="true" url="http://api-t.ft.com/content/77c8a5b5-c9e3-4df2-ad5f-3ef35fe1d9d4" type="http://www.ft.com/ontology/content/ImageSet"></ft-content></pull-quote-image><pull-quote-source>Pellentesque habitant, morbi tristique</pull-quote-source>
 </pull-quote>
 <p>Donec id faucibus erat. Suspendisse tempor laoreet lorem, sit amet vehicula massa facilisis at. Nulla quis feugiat massa. Praesent viverra non lectus ut ullamcorper. Phasellus <ft-content url="http://api-t.ft.com/content/c71efed9-fe5a-488d-9f47-20c15d177153" type="http://www.ft.com/ontology/content/Article">porttitor neque</ft-content> at volutpat pulvinar.</p>
-<p>“Curabitur fermentum, dolor vel interdum varius, tellus justo dapibus velit, interdum sollicitudin dolor nibh varius velit.”</p> 
+<p>“Curabitur fermentum, dolor vel interdum varius, tellus justo dapibus velit, interdum sollicitudin dolor nibh varius velit.”</p>
 </body>

tests/schema/body-tree/valid/kitchen-snippet.json

@@ -1283,6 +1283,10 @@
           "value": "If this sounds all-too-familiar, stick with me. I have a proven treatment for Kitchen Sink Syndrome."
         }
       ]
+    },
+    {
+      "type": "text",
+      "value": "",
     }
   ]
 }