Ignore empty articles, and be more flexible in parsing articles (Gitb…

…ookIO#1741)

* Ignore empty articles, and be more flexible in parsing articles

* gitbook: Update markup-it^3.4.0

packages/gitbook/package.json

@@ -43,7 +43,7 @@
     "json-schema-defaults": "0.1.1",
     "jsonschema": "1.1.0",
     "juice": "2.0.0",
-    "markup-it": "^3.2.2",
+    "markup-it": "^3.4.0",
     "mkdirp": "0.5.1",
     "moment": "2.13.0",
     "npm": "3.10.9",

packages/gitbook/src/parse/__tests__/fixtures/summary/empty-items.yaml

@@ -21,5 +21,11 @@ nodes:
           - kind: block
             type: unstyled
             nodes:
-              - kind: text
-                text: ''
+              - kind: inline
+                type: link
+                data:
+                  href: 'art1.md'
+                nodes:
+                  - kind: text
+                    text: '' # No title so ignored
+

packages/gitbook/src/parse/__tests__/fixtures/summary/flexible.yaml

@@ -0,0 +1,115 @@
+nodes:
+  - kind: block
+    type: header_one
+    nodes:
+      - kind: text
+        text: Summary
+  - kind: block
+    type: unordered_list
+    nodes:
+      - kind: block
+        type: list_item
+        nodes:
+          - kind: block
+            type: unstyled
+            nodes:
+              - kind: text
+                text: 'ignored text'
+              - kind: inline
+                type: link
+                data:
+                  href: 'art1.md'
+                nodes:
+                  - kind: text
+                    text: 'Article 1'
+              - kind: text
+                text: ''
+
+      - kind: block
+        type: list_item
+        nodes:
+          - kind: block
+            type: paragraph
+            nodes:
+              - kind: text
+                text: 'Article 2'
+          - kind: block
+            type: paragraph
+            nodes:
+              - kind: text
+                text: 'Ignored'
+
+      - kind: block
+        type: list_item
+        nodes:
+          - kind: block
+            type: paragraph
+            nodes:
+              - kind: text
+                text: 'First paragraph. There is a link further, so should be ignored'
+
+          # Sublist, that is not after a link. Should be ignored
+          - kind: block
+            type: unordered_list
+            nodes:
+              - kind: block
+                type: list_item
+                nodes:
+                  - kind: block
+                    type: unstyled
+                    nodes:
+                      - kind: inline
+                        type: link
+                        data:
+                          href: 'ignored.md'
+                        nodes:
+                          - kind: text
+                            text: 'Ignored'
+
+          - kind: block
+            type: paragraph
+            nodes:
+              - kind: inline
+                type: link
+                data:
+                  href: 'art3.md'
+                nodes:
+                  - kind: text
+                    text: 'Article 3'
+
+          # First sublist that is after the link. Parsed as subarticles
+          - kind: block
+            type: unordered_list
+            nodes:
+              - kind: block
+                type: list_item
+                nodes:
+                  - kind: block
+                    type: unstyled
+                    nodes:
+                      - kind: inline
+                        type: link
+                        data:
+                          href: 'art2.1.md'
+                        nodes:
+                          - kind: text
+                            text: 'Article 2.1'
+
+          # Ignored because we already parsed the previous list
+          - kind: block
+            type: unordered_list
+            nodes:
+              - kind: block
+                type: list_item
+                nodes:
+                  - kind: block
+                    type: unstyled
+                    nodes:
+                      - kind: inline
+                        type: link
+                        data:
+                          href: 'ignored'
+                        nodes:
+                          - kind: text
+                            text: 'ignored'
+

packages/gitbook/src/parse/__tests__/summaryFromDocument.js

@@ -92,22 +92,13 @@ describe('summaryFromDocument', () => {
         ]);
     });
 
-    it('should parse empty items', () => {
+    it('should ignore empty items', () => {
         const summary = readSummary('summary/empty-items.yaml');
 
         expectParts(summary, [
             {
                 title: '',
-                articles: [
-                    {
-                        title: '',
-                        ref: ''
-                    },
-                    {
-                        title: '',
-                        ref: ''
-                    }
-                ]
+                articles: []
             }
         ]);
     });
@@ -210,4 +201,36 @@ describe('summaryFromDocument', () => {
         expectParts(summary, [
         ]);
     });
+
+    it('should be flexible', () => {
+        const summary = readSummary('summary/flexible.yaml');
+
+        expectParts(summary, [
+            {
+                title: '',
+                articles: [
+                    {
+                        title: 'Article 1',
+                        ref: 'art1.md'
+                    },
+                    {
+                        title: 'Article 2',
+                        articles: []
+                    },
+                    {
+                        title: 'Article 3',
+                        ref: 'art3.md',
+                        articles: [
+                            {
+                                title: 'Article 2.1',
+                                ref: 'art2.1.md',
+                                articles: []
+                            }
+                        ]
+                    }
+                ]
+            }
+        ]);
+    });
+
 });

packages/gitbook/src/parse/summaryFromDocument.js

@@ -9,17 +9,48 @@ const isLink = node => node.type === INLINES.LINK;
 /**
  * Create a summary article from a list item.
  * @param  {Block} item
- * @return {SummaryArticleLike} article
+ * @return {SummaryArticleLike | Null} article
  */
 function createArticleFromItem(item) {
     const { nodes } = item;
 
-    const titleParent = nodes.first();
-    const list = nodes.skip(1).find(isList);
+    // Find the link that represents the article's title
+    const linkParent = nodes
+        .filterNot(isList)
+        .find(node => node.findDescendant(isLink));
+
+    // Or find text that could act as title
+    const textParent = nodes.filterNot(node => isList(node) || node.isEmpty).first();
+
+    let title, ref, parent;
+    if (linkParent) {
+        const link = linkParent.findDescendant(isLink);
+
+        if (!link.isEmpty) {
+            parent = linkParent;
+            title = link.text;
+            ref = link.data.get('href');
+        }
+    }
+
+    if (!parent) {
+        // Could not find a proper link
+
+        if (textParent) {
+            parent = textParent;
+            title = textParent.text;
+            ref = null;
+        } else {
+            // This item has no proper title or link
+            return null;
+        }
+    }
+
+    const list = nodes
+        // Skip until after the article's title or link
+        .skipUntil(node => node === parent).skip(1)
+        .find(isList);
     const articles = list ? listArticles(list) : [];
-    const title = titleParent.text;
-    const link = titleParent.findDescendant(isLink);
-    const ref = link ? link.data.get('href') : null;
 
     return {
         title,
@@ -35,7 +66,9 @@ function createArticleFromItem(item) {
  */
 function listArticles(list) {
     const { nodes } = list;
-    return nodes.map(item => createArticleFromItem(item));
+    return nodes
+        .map(item => createArticleFromItem(item))
+        .filter(article => Boolean(article));
 }
 
 /**