fix(JATS): Workaround for xml-js dropping spaces between elements

See elifesciences/enhanced-preprints-issues#680
stencila · Jun 21, 2023 · a34678c · a34678c
1 parent 4234ead
commit a34678c
Show file tree

Hide file tree

Showing 9 changed files with 102 additions and 7 deletions.
diff --git a/src/codecs/jats/__file_snapshots__/48c60452-6c66-1014-adf9-c7b61873ecd3.json b/src/codecs/jats/__file_snapshots__/48c60452-6c66-1014-adf9-c7b61873ecd3.json
@@ -2845,7 +2845,7 @@
           "value": "10.3120/0024-9637-56.3.209"
         }
       ],
-      "title": "PattersonB.2009. Systematics, Evolution, and Biogeography of Compositae, Madroño. doi:10.3120/0024-9637-56.3.209"
+      "title": "Patterson B. 2009. Systematics, Evolution, and Biogeography of Compositae, Madroño. doi:10.3120/0024-9637-56.3.209"
     },
     {
       "type": "Article",
@@ -6089,6 +6089,7 @@
             "10A;"
           ]
         },
+        " ",
         {
           "type": "Link",
           "relation": "table",

diff --git a/src/codecs/jats/__file_snapshots__/87253.json b/src/codecs/jats/__file_snapshots__/87253.json
@@ -7882,6 +7882,7 @@
                 "(A)"
               ]
             },
+            " ",
             {
               "type": "Emphasis",
               "content": [
@@ -8823,6 +8824,7 @@
                 "(D)"
               ]
             },
+            " ",
             {
               "type": "Emphasis",
               "content": [
@@ -9720,6 +9722,7 @@
                 "(F)"
               ]
             },
+            " ",
             {
               "type": "Emphasis",
               "content": [
@@ -11146,6 +11149,7 @@
                 "(A)"
               ]
             },
+            " ",
             {
               "type": "Emphasis",
               "content": [

diff --git a/src/codecs/jats/__file_snapshots__/8758d58a-6c3d-1014-87a3-bbc202f7d832.json b/src/codecs/jats/__file_snapshots__/8758d58a-6c3d-1014-87a3-bbc202f7d832.json
@@ -5629,7 +5629,7 @@
         "type": "Periodical",
         "name": "Annu. Rev. Immunol."
       },
-      "title": "A. C.Hayday, P.Vantourout, Annu. Rev. Immunol., in press, doi:10.1146/annurev-immunol-102819-023144.",
+      "title": "A. C. Hayday, P. Vantourout, Annu. Rev. Immunol., in press, doi:10.1146/annurev-immunol-102819-023144.",
       "meta": {
         "label": "46."
       }

diff --git a/src/codecs/jats/__file_snapshots__/ca7917ff-6cb0-1014-9b19-d05ef1e56b05.json b/src/codecs/jats/__file_snapshots__/ca7917ff-6cb0-1014-9b19-d05ef1e56b05.json
@@ -3538,7 +3538,7 @@
         },
         "volumeNumber": 29
       },
-      "title": "SteckK, WalkerSJ, ItskovPM, BaltazarC, MoreiraJ-M, RibeiroC.2018. Internal amino acid state modulates yeast taste neurons to support protein homeostasis in Drosophila29."
+      "title": "Steck K, Walker SJ, Itskov PM, Baltazar C, Moreira J-M, Ribeiro C. 2018. Internal amino acid state modulates yeast taste neurons to support protein homeostasis in Drosophila 29."
     },
     {
       "type": "Article",
@@ -6816,6 +6816,7 @@
                 }
               ]
             },
+            " ",
             {
               "type": "Emphasis",
               "content": [

diff --git a/src/codecs/jats/__file_snapshots__/spaces-around-marks.jats.xml b/src/codecs/jats/__file_snapshots__/spaces-around-marks.jats.xml
@@ -0,0 +1,28 @@
+<?xml version="1.0" encoding="utf-8"?>
+<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.1 20151215//EN" "JATS-archivearticle1.dtd">
+<article xmlns:xlink="http://www.w3.org/1999/xlink" article-type="research-article">
+    <front>
+        <article-meta>
+            <title-group>
+                <article-title/>
+            </title-group>
+            <contrib-group/>
+        </article-meta>
+    </front>
+    <body>
+        <sec>
+            <p>Test that spaces around marks are retained. A regression test based on a published example</p>
+            <list list-type="order">
+                <list-item>
+                    <p>
+                        <bold>
+                            <italic>Space after</italic>
+                        </bold> 
+                        <italic>space before and after</italic> 
+                    </p>
+                </list-item>
+            </list>
+        </sec>
+    </body>
+    <back/>
+</article>
diff --git a/src/codecs/jats/__file_snapshots__/spaces-around-marks.yaml b/src/codecs/jats/__file_snapshots__/spaces-around-marks.yaml
@@ -0,0 +1,26 @@
+type: Article
+content:
+  - type: Paragraph
+    content:
+      - >-
+        Test that spaces around marks are retained. A regression test based on a
+        published example
+  - type: List
+    items:
+      - type: ListItem
+        content:
+          - type: Paragraph
+            content:
+              - type: Strong
+                content:
+                  - type: Emphasis
+                    content:
+                      - Space after
+              - ' '
+              - type: Emphasis
+                content:
+                  - space before and after
+              - ' '
+    order: Ascending
+    meta:
+      listType: roman-lower
diff --git a/src/codecs/jats/__fixtures__/spaces-around-marks.xml b/src/codecs/jats/__fixtures__/spaces-around-marks.xml
@@ -0,0 +1,11 @@
+<?xml version="1.0"?>
+<article>
+    <body>
+        <p>Test that spaces around marks are retained. A regression test based on a published example</p>
+        <list list-type="roman-lower">
+            <list-item>
+                <p><bold><italic>Space after</italic></bold> <italic>space before and after</italic> </p>
+            </list-item>
+        </list>
+    </body>
+</article>
diff --git a/src/codecs/jats/index.ts b/src/codecs/jats/index.ts
@@ -61,6 +61,16 @@ type Content = stencila.InlineContent | stencila.BlockContent
 const DOCTYPE =
   'article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.1 20151215//EN" "JATS-archivearticle1.dtd"'
 
+/**
+ * The spacing character to use to workaround a bug
+ * in xml-js that strips whitespace between elements
+ * https://github.com/nashwaan/xml-js/issues/177
+ *
+ * This is an invisible character but xml-js does not treat as whitespace
+ * https://unicode-explorer.com/c/2800
+ */
+const SPACER = '\u2800'
+
 export class JatsCodec extends Codec implements Codec {
   /**
    * Media types that this codec will match.
@@ -110,8 +120,13 @@ export class JatsCodec extends Codec implements Codec {
   public readonly decode = async (
     file: vfile.VFile
   ): Promise<stencila.Article | Content[]> => {
-    const jats = await vfile.dump(file)
-    const doc = xml.load(jats)
+    let jats = await vfile.dump(file)
+
+    // See notes on SPACER for why this is needed
+    let fixed = jats.replaceAll('> <', `>${SPACER}<`)
+
+    const doc = xml.load(fixed)
+
     return decodeDocument(doc)
   }
 
@@ -1387,6 +1402,9 @@ export function decodeReference(
     title = textOrUndefined(elem)
   }
 
+  // Remove any space placeholders
+  title = title?.replaceAll(SPACER, ' ')
+
   let publisher: stencila.Organization | undefined
   const publisherName = textOrUndefined(child(elem, 'publisher-name'))
   if (publisherName !== undefined) {
@@ -1612,7 +1630,8 @@ function encodeBody(nodes: stencila.Node[], state: EncodeState): xml.Element {
  * Decode a JATS element to an array of Stencila `Node`s
  */
 function decodeElement(elem: xml.Element, state: DecodeState): stencila.Node[] {
-  if (elem.type === 'text') return [elem.text ?? '']
+  if (elem.type === 'text')
+    return [elem.text?.toString().replace(SPACER, ' ') ?? '']
   switch (elem.name) {
     case 'alternatives':
       return decodeAlternatives(elem, state)
@@ -1694,7 +1713,11 @@ function encodeNode(node: stencila.Node, state: EncodeState): xml.Element[] {
     case 'Strong':
       return encodeMark(node as stencila.Strong, state, 'bold')
     case 'NontextualAnnotation':
-      return encodeMark(node as stencila.NontextualAnnotation, state, 'underline')
+      return encodeMark(
+        node as stencila.NontextualAnnotation,
+        state,
+        'underline'
+      )
     case 'Superscript':
       return encodeMark(node as stencila.Superscript, state, 'sup')
     case 'Subscript':

diff --git a/src/codecs/jats/jats.test.ts b/src/codecs/jats/jats.test.ts
@@ -294,6 +294,7 @@ test.each([
   'nested-list.xml',
   'supplementary-material.xml',
   'labelled-list-items.xml',
+  'spaces-around-marks.xml',
   'elife-30274-v1',
   'elife-43154-v2',
   'elife-46472-v3',