Commonmark zero width annotations

packages/contenttype-commonmark/src/index.ts

@@ -15,23 +15,30 @@ const TAG_MAP: { [tagName: string]: string } = {
 
 const HANDLERS = {
   'hardbreak': (state, token) => {
+    state.content += '\n';
     state.annotations.push({
       type: token.tag,
-      start: state.content.length,
+      start: state.content.length - 1,
       end: state.content.length,
       attributes: {}
     });
-    state.content += '\n';
   },
 
   'softbreak': (state, token) => {
     state.content += '\n';
+    state.annotations.push({
+      type: 'soft-break',
+      start: state.content.length - 1,
+      end: state.content.length,
+      attributes: {}
+    });
   },
 
   'hr': (state, token) => {
+    state.content += '\uFFFC';
     state.annotations.push({
       type: token.tag,
-      start: state.content.length,
+      start: state.content.length - 1,
       end: state.content.length,
       attributes: {}
     });
@@ -132,9 +139,10 @@ const HANDLERS = {
 
     attributes.alt = fetchText(token);
 
+    state.content += '\uFFFC';
     state.annotations.push({
       type: token.tag,
-      start: state.content.length,
+      start: state.content.length - 1,
       end: state.content.length,
       attributes
     });

packages/contenttype-commonmark/test/atjson-contenttype-markdown-test.ts

@@ -141,11 +141,11 @@ describe('markdown -> atjson', function () {
     let parser = new Parser(markdown);
     let atjson = parser.parse();
 
-    expect(atjson.content).toBe('\n');
+    expect(atjson.content).toBe('\uFFFC\n');
 
     let expectedAnnotations = [
-      { type: 'paragraph', start: 0, end: 0, attributes: {} },
-      { type: 'image', start: 0, end: 0, attributes: { src: '/url', title: 'title', alt: 'foo' } }
+      { type: 'paragraph', start: 0, end: 1, attributes: {} },
+      { type: 'image', start: 0, end: 1, attributes: { src: '/url', title: 'title', alt: 'foo' } }
     ];
 
     expect(atjson.annotations).toEqual(expectedAnnotations);
@@ -184,5 +184,21 @@ describe('markdown -> atjson', function () {
 
     expect(atjson.annotations).toEqual(expectedAnnotations);
   });
+
+  it('edge-case images from commonmark tests', function () {
+    let markdown = '![foo]\n[]\n\n[foo]: /url "title"\n';
+
+    let parser = new Parser(markdown);
+    let atjson = parser.parse();
+
+    expect(atjson.content).toBe('\uFFFC\n[]\n');
+
+    let expectedAnnotations = [
+      { type: 'paragraph', start: 0, end: atjson.content.length - 1, attributes: {} },
+      { type: 'image', start: 0, end: 1, attributes: { src: '/url', alt: 'foo', title: 'title' } }
+    ];
+
+    expect(atjson.annotations).toEqual(expectedAnnotations);
+  });
 });
 

packages/hir/test/hir-test.ts

@@ -65,6 +65,14 @@ function paragraph(...children: node[]) {
   };
 }
 
+function image(attributes: object, ...children: node[]) {
+  return {
+    type: 'image',
+    attributes: attributes,
+    children
+  };
+}
+
 describe('@atjson/hir', function () {
 
   /**
@@ -264,5 +272,23 @@ describe('@atjson/hir', function () {
 
       expect(new HIR(zerolength).toJSON()).toEqual(expected);
     });
-});
+
+    // n.b. this case is somewhat ambiguous – the image should be explicitly
+    // assigned to an object replacement character in the text
+    it('regression test from commonmark Images tests', function () {
+      let atjson = new AtJSON({
+        content: '\uFFFC\n[]\n',
+        annotations: [
+          { type: 'paragraph', start: 0, end: 5 }
+          { type: 'image', start: 0, end: 1, attributes: { src: '/url', alt: 'foo', title: 'title' } },
+        ]
+      });
+
+      let expected = root(
+        paragraph(image({ alt: 'foo', src: '/url', title: 'title' }), '\n[]\n')
+      );
+
+      expect(new HIR(atjson).toJSON()).toEqual(expected);
+    });
+  });
 });