Absolute URLs (#9)

* [CI] Add test workflow * [CI] use npm ci instead of install * [CI] Run on a matrix. OS=[Win,Linux,Mac]; Node=[12,14,16] * [README.md] add badge for test workflow * [CI] Rename job from build to test * [README.md] Add npm badge * [Tests] use t.same instead of deprecated t.deepEqual Run npm test > [email protected] test /home/runner/work/parse-apache-directory-index/parse-apache-directory-index > tap --no-check-coverage test/*.js && semistandard TAP version 13 (node:1830) DeprecationWarning: deepEqual() is deprecated, use same() instead * Handle absolute URLs
juliangruber · Jul 18, 2021 · c22cd11 · c22cd11
1 parent 147ff07
commit c22cd11
Show file tree

Hide file tree

Showing 3 changed files with 97 additions and 3 deletions.
diff --git a/index.js b/index.js
@@ -47,19 +47,22 @@ module.exports = src => {
       const $tds = $(tr).find('td');
       const getCol = label => fieldCols[label] === undefined ? null : $tds.eq(fieldCols[label]);
       const getColText = label => getCol(label) && getCol(label).text().trim();
-
-      const path = getCol('Name').children().eq(0).attr('href');
       const name = getColText('Name');
 
       // Ignore 'Parent Directory' row
       if (name === 'Parent Directory' || !name) return;
 
+      let path = getCol('Name').children().eq(0).attr('href');
+      if (!path.startsWith('http://') && !path.startsWith('https://')) {
+        path = join(dir, path);
+      }
+
       files.push({
         type: path.endsWith('/')
           ? 'directory'
           : 'file',
         name: name,
-        path: join(dir, path),
+        path: path,
         lastModified: getCol('Last modified') && new Date(getColText('Last modified')),
         size: getCol('Size') && bytes(getColText('Size')),
         description: getColText('Description')

diff --git a/test/fixture/apache-index-absolute-and-relative-urls.html b/test/fixture/apache-index-absolute-and-relative-urls.html
@@ -0,0 +1,48 @@
+<html>
+<head>
+    <title>Index of /foo/bar</title>
+    <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
+</head>
+<body>
+<h1>Index of /foo/bar</h1>
+<table cellspacing="10">
+    <tr>
+        <th align="left">Name</th>
+        <th>Last modified</th>
+        <th>Size</th>
+        <th>Description</th>
+    </tr>
+    <tr>
+        <td><a href="../">Parent Directory</a></td>
+    </tr>
+    <tr>
+        <td><a href="https://www.example.org/subdir-absolute/">subdir-absolute/</a></td>
+        <td>Wed Jun 30 22:24:50 UTC 2021</td>
+        <td align="right"></td>
+        <td> </td>
+    </tr>
+    <tr>
+        <td><a href="subdir-relative/">subdir-relative/</a></td>
+        <td>Wed Jun 30 22:24:50 UTC 2021</td>
+        <td align="right"></td>
+        <td> </td>
+    </tr>
+    <tr>
+        <td><a href="https://www.example.org/subfile-absolute.xml">subfile-absolute.xml</a></td>
+        <td>Sun Jul 04 16:22:15 UTC 2021</td>
+        <td align="right">
+            25
+        </td>
+        <td> </td>
+    </tr>
+    <tr>
+        <td><a href="subfile-relative.xml">subfile-relative.xml</a></td>
+        <td>Sun Jul 04 16:22:15 UTC 2021</td>
+        <td align="right">
+            25
+        </td>
+        <td></td>
+    </tr>
+</table>
+</body>
+</html>
diff --git a/test/index.js b/test/index.js
@@ -131,3 +131,46 @@ test('should parse apache directory index files (alt format)', t => {
   });
   t.end();
 });
+
+test('Should parse both absolute and relative URLs', t => {
+  const index = fs.readFileSync(join(__dirname, '/fixture/apache-index-absolute-and-relative-urls.html'), 'utf8');
+
+  t.same(parse(index), {
+    dir: '/foo/bar',
+    files: [
+      {
+        name: 'subdir-absolute/',
+        type: 'directory',
+        path: 'https://www.example.org/subdir-absolute/',
+        description: '',
+        size: null,
+        lastModified: new Date('2021-06-30T22:24:50.000Z')
+      },
+      {
+        name: 'subdir-relative/',
+        type: 'directory',
+        path: '/foo/bar/subdir-relative/',
+        size: null,
+        lastModified: new Date('2021-06-30T22:24:50.000Z'),
+        description: ''
+      },
+      {
+        name: 'subfile-absolute.xml',
+        type: 'file',
+        path: 'https://www.example.org/subfile-absolute.xml',
+        size: 25,
+        lastModified: new Date('2021-07-04T16:22:15.000Z'),
+        description: ''
+      },
+      {
+        name: 'subfile-relative.xml',
+        type: 'file',
+        path: '/foo/bar/subfile-relative.xml',
+        size: 25,
+        lastModified: new Date('2021-07-04T16:22:15.000Z'),
+        description: ''
+      }
+    ]
+  });
+  t.end();
+});