Merge pull request #80 from thewilkybarkid/source-indices

sourceLocations is broken when tags are implied

Author: Scrum

package-lock.json

@@ -39,7 +39,7 @@
     "posthtmltree"
   ],
   "dependencies": {
-    "htmlparser2": "^6.0.0"
+    "htmlparser2": "^7.1.1"
   },
   "devDependencies": {
     "@antfu/eslint-config-ts": "^0.4.3",

package.json

@@ -44,6 +44,7 @@ export const parser = (html: string, options: Options = {}): Node[] => {
   const locationTracker = new LocationTracker(html);
   const bufArray: Node[] = [];
   const results: Node[] = [];
+  let lastOpenTagEndIndex = 0;
 
   function bufferArrayLast(): Node {
     return bufArray[bufArray.length - 1];
@@ -122,14 +123,14 @@ export const parser = (html: string, options: Options = {}): Node[] => {
   }
 
   function onopentag(tag: string, attrs: Attributes) {
-    const start = locationTracker.getPosition(parser.startIndex);
     const buf: NodeTag = { tag };
 
     if (options.sourceLocations) {
       buf.location = {
-        start,
-        end: start
+        start: locationTracker.getPosition(parser.startIndex),
+        end: locationTracker.getPosition(parser.endIndex)
       };
+      lastOpenTagEndIndex = parser.endIndex;
     }
 
     if (Object.keys(attrs).length > 0) {
@@ -139,11 +140,15 @@ export const parser = (html: string, options: Options = {}): Node[] => {
     bufArray.push(buf);
   }
 
-  function onclosetag() {
+  function onclosetag(name: string, isImplied: boolean) {
     const buf: Node | undefined = bufArray.pop();
 
     if (buf && typeof buf === 'object' && buf.location && parser.endIndex !== null) {
-      buf.location.end = locationTracker.getPosition(parser.endIndex);
+      if (!isImplied) {
+        buf.location.end = locationTracker.getPosition(parser.endIndex);
+      } else if (lastOpenTagEndIndex < parser.startIndex) {
+        buf.location.end = locationTracker.getPosition(parser.startIndex - 1);
+      }
     }
 
     if (buf) {

src/index.ts

@@ -243,7 +243,7 @@ test('should be not converting html entity name', t => {
 });
 
 test('should parse with source locations', t => {
-  const html = '<h1>Test</h1>\n<p><b>Foo</b></p>';
+  const html = '<h1>Test</h1>\n<p><b>Foo</b><hr></p><p>Bar\n<hr>';
   const tree = parser(html, { sourceLocations: true });
   const expected = [
     {
@@ -284,11 +284,66 @@ test('should parse with source locations', t => {
           line: 2,
           column: 1
         },
+        end: {
+          line: 2,
+          column: 13
+        }
+      }
+    },
+    {
+      tag: 'hr',
+      location: {
+        start: {
+          line: 2,
+          column: 14
+        },
         end: {
           line: 2,
           column: 17
         }
       }
+    },
+    {
+      tag: 'p',
+      location: {
+        start: {
+          line: 2,
+          column: 18
+        },
+        end: {
+          line: 2,
+          column: 21
+        }
+      }
+    },
+    {
+      tag: 'p',
+      content: [
+        'Bar\n'
+      ],
+      location: {
+        start: {
+          line: 2,
+          column: 22
+        },
+        end: {
+          line: 2,
+          column: 28
+        }
+      }
+    },
+    {
+      tag: 'hr',
+      location: {
+        start: {
+          line: 3,
+          column: 1
+        },
+        end: {
+          line: 3,
+          column: 4
+        }
+      }
     }
   ];
   t.deepEqual(tree, expected);