Merge pull request #612 from virtualtam/refactor/bookmark-parser

Refactor bookmark import using a generic Netscape parser

10 files changed, 779 insertions, 122 deletions

diff --git a/application/NetscapeBookmarkUtils.php b/application/NetscapeBookmarkUtils.php
index fdbb0ad7..b99a432e 100644
--- a/application/NetscapeBookmarkUtils.php
+++ b/application/NetscapeBookmarkUtils.php
@@ -51,4 +51,146 @@ class NetscapeBookmarkUtils
 
         return $bookmarkLinks;
     }
+
+    /**
+     * Generates an import status summary
+     *
+     * @param string $filename       name of the file to import
+     * @param int    $filesize       size of the file to import
+     * @param int    $importCount    how many links were imported
+     * @param int    $overwriteCount how many links were overwritten
+     * @param int    $skipCount      how many links were skipped
+     *
+     * @return string Summary of the bookmark import status
+     */
+    private static function importStatus(
+        $filename,
+        $filesize,
+        $importCount=0,
+        $overwriteCount=0,
+        $skipCount=0
+    )
+    {
+        $status = 'File '.$filename.' ('.$filesize.' bytes) ';
+        if ($importCount == 0 && $overwriteCount == 0 && $skipCount == 0) {
+            $status .= 'has an unknown file format. Nothing was imported.';
+        } else {
+            $status .= 'was successfully processed: '.$importCount.' links imported, ';
+            $status .= $overwriteCount.' links overwritten, ';
+            $status .= $skipCount.' links skipped.';
+        }
+        return $status;
+    }
+
+    /**
+     * Imports Web bookmarks from an uploaded Netscape bookmark dump
+     *
+     * @param array  $post      Server $_POST parameters
+     * @param array  $file      Server $_FILES parameters
+     * @param LinkDB $linkDb    Loaded LinkDB instance
+     * @param string $pagecache Page cache
+     *
+     * @return string Summary of the bookmark import status
+     */
+    public static function import($post, $files, $linkDb, $pagecache)
+    {
+        $filename = $files['filetoupload']['name'];
+        $filesize = $files['filetoupload']['size'];
+        $data = file_get_contents($files['filetoupload']['tmp_name']);
+
+        // Sniff file type
+        if (! startsWith($data, '<!DOCTYPE NETSCAPE-Bookmark-file-1>')) {
+            return self::importStatus($filename, $filesize);
+        }
+
+        // Overwrite existing links?
+        $overwrite = ! empty($post['overwrite']);
+
+        // Add tags to all imported links?
+        if (empty($post['default_tags'])) {
+            $defaultTags = array();
+        } else {
+            $defaultTags = preg_split(
+                '/[\s,]+/',
+                escape($post['default_tags'])
+            );
+        }
+
+        // links are imported as public by default
+        $defaultPrivacy = 0;
+
+        $parser = new NetscapeBookmarkParser(
+            true,                       // nested tag support
+            $defaultTags,               // additional user-specified tags
+            strval(1 - $defaultPrivacy) // defaultPub = 1 - defaultPrivacy
+        );
+        $bookmarks = $parser->parseString($data);
+
+        $importCount = 0;
+        $overwriteCount = 0;
+        $skipCount = 0;
+
+        foreach ($bookmarks as $bkm) {
+            $private = $defaultPrivacy;
+            if (empty($post['privacy']) || $post['privacy'] == 'default') {
+                // use value from the imported file
+                $private = $bkm['pub'] == '1' ? 0 : 1;
+            } else if ($post['privacy'] == 'private') {
+                // all imported links are private
+                $private = 1;
+            } else if ($post['privacy'] == 'public') {
+                // all imported links are public
+                $private = 0;
+            }                
+
+            $newLink = array(
+                'title' => $bkm['title'],
+                'url' => $bkm['uri'],
+                'description' => $bkm['note'],
+                'private' => $private,
+                'linkdate'=> '',
+                'tags' => $bkm['tags']
+            );
+
+            $existingLink = $linkDb->getLinkFromUrl($bkm['uri']);
+
+            if ($existingLink !== false) {
+                if ($overwrite === false) {
+                    // Do not overwrite an existing link
+                    $skipCount++;
+                    continue;
+                }
+
+                // Overwrite an existing link, keep its date
+                $newLink['linkdate'] = $existingLink['linkdate'];
+                $linkDb[$existingLink['linkdate']] = $newLink;
+                $importCount++;
+                $overwriteCount++;
+                continue;
+            }
+
+            // Add a new link
+            $newLinkDate = new DateTime('@'.strval($bkm['time']));
+            while (!empty($linkDb[$newLinkDate->format(LinkDB::LINK_DATE_FORMAT)])) {
+                // Ensure the date/time is not already used
+                // - this hack is necessary as the date/time acts as a primary key
+                // - apply 1 second increments until an unused index is found
+                // See https://github.com/shaarli/Shaarli/issues/351
+                $newLinkDate->add(new DateInterval('PT1S'));
+            }
+            $linkDbDate = $newLinkDate->format(LinkDB::LINK_DATE_FORMAT);
+            $newLink['linkdate'] = $linkDbDate;
+            $linkDb[$linkDbDate] = $newLink;
+            $importCount++;
+        }
+
+        $linkDb->savedb($pagecache);
+        return self::importStatus(
+            $filename,
+            $filesize,
+            $importCount,
+            $overwriteCount,
+            $skipCount
+        );
+    }
 }
diff --git a/index.php b/index.php
index 9ae798ba..1f148d78 100644
--- a/index.php
+++ b/index.php
@@ -44,6 +44,10 @@ error_reporting(E_ALL^E_WARNING);
 //error_reporting(-1);
 
 
+// 3rd-party libraries
+require_once 'inc/rain.tpl.class.php';
+require_once __DIR__ . '/vendor/autoload.php';
+
 // Shaarli library
 require_once 'application/ApplicationUtils.php';
 require_once 'application/Cache.php';
@@ -65,7 +69,6 @@ require_once 'application/Utils.php';
 require_once 'application/PluginManager.php';
 require_once 'application/Router.php';
 require_once 'application/Updater.php';
-require_once 'inc/rain.tpl.class.php';
 
 // Ensure the PHP version is supported
 try {
@@ -1468,26 +1471,37 @@ function renderPage($conf, $pluginManager)
         exit;
     }
 
-    // -------- User is uploading a file for import
-    if (isset($_SERVER['QUERY_STRING']) && startsWith($_SERVER['QUERY_STRING'], 'do=upload'))
-    {
-        // If file is too big, some form field may be missing.
-        if (!isset($_POST['token']) || (!isset($_FILES)) || (isset($_FILES['filetoupload']['size']) && $_FILES['filetoupload']['size']==0))
-        {
-            $returnurl = ( empty($_SERVER['HTTP_REFERER']) ? '?' : $_SERVER['HTTP_REFERER'] );
-            echo '<script>alert("The file you are trying to upload is probably bigger than what this webserver can accept ('.getMaxFileSize().' bytes). Please upload in smaller chunks.");document.location=\''.escape($returnurl).'\';</script>';
+    if ($targetPage == Router::$PAGE_IMPORT) {
+        // Upload a Netscape bookmark dump to import its contents
+
+        if (! isset($_POST['token']) || ! isset($_FILES['filetoupload'])) {
+            // Show import dialog
+            $PAGE->assign('maxfilesize', getMaxFileSize());
+            $PAGE->renderPage('import');
             exit;
         }
-        if (!tokenOk($_POST['token'])) die('Wrong token.');
-        importFile($LINKSDB);
-        exit;
-    }
 
-    // -------- Show upload/import dialog:
-    if ($targetPage == Router::$PAGE_IMPORT)
-    {
-        $PAGE->assign('maxfilesize',getMaxFileSize());
-        $PAGE->renderPage('import');
+        // Import bookmarks from an uploaded file
+        if (isset($_FILES['filetoupload']['size']) && $_FILES['filetoupload']['size'] == 0) {
+            // The file is too big or some form field may be missing.
+            echo '<script>alert("The file you are trying to upload is probably'
+                .' bigger than what this webserver can accept ('
+                .getMaxFileSize().' bytes).'
+                .' Please upload in smaller chunks.");document.location=\'?do='
+                .Router::$PAGE_IMPORT .'\';</script>';
+            exit;
+        }
+        if (! tokenOk($_POST['token'])) {
+            die('Wrong token.');
+        }
+        $status = NetscapeBookmarkUtils::import(
+            $_POST,
+            $_FILES,
+            $LINKSDB,
+            $conf->get('resource.page_cache')
+        );
+        echo '<script>alert("'.$status.'");document.location=\'?do='
+             .Router::$PAGE_IMPORT .'\';</script>';
         exit;
     }
 
@@ -1545,95 +1559,6 @@ function renderPage($conf, $pluginManager)
 }
 
 /**
- * Process the import file form.
- *
- * @param LinkDB        $LINKSDB Loaded LinkDB instance.
- * @param ConfigManager $conf    Configuration Manager instance.
- */
-function importFile($LINKSDB, $conf)
-{
-    if (!isLoggedIn()) { die('Not allowed.'); }
-
-    $filename=$_FILES['filetoupload']['name'];
-    $filesize=$_FILES['filetoupload']['size'];
-    $data=file_get_contents($_FILES['filetoupload']['tmp_name']);
-    $private = (empty($_POST['private']) ? 0 : 1); // Should the links be imported as private?
-    $overwrite = !empty($_POST['overwrite']) ; // Should the imported links overwrite existing ones?
-    $import_count=0;
-
-    // Sniff file type:
-    $type='unknown';
-    if (startsWith($data,'<!DOCTYPE NETSCAPE-Bookmark-file-1>')) $type='netscape'; // Netscape bookmark file (aka Firefox).
-
-    // Then import the bookmarks.
-    if ($type=='netscape')
-    {
-        // This is a standard Netscape-style bookmark file.
-        // This format is supported by all browsers (except IE, of course), also Delicious, Diigo and others.
-        foreach(explode('<DT>',$data) as $html) // explode is very fast
-        {
-            $link = array('linkdate'=>'','title'=>'','url'=>'','description'=>'','tags'=>'','private'=>0);
-            $d = explode('<DD>',$html);
-            if (startsWith($d[0], '<A '))
-            {
-                $link['description'] = (isset($d[1]) ? html_entity_decode(trim($d[1]),ENT_QUOTES,'UTF-8') : '');  // Get description (optional)
-                preg_match('!<A .*?>(.*?)</A>!i',$d[0],$matches); $link['title'] = (isset($matches[1]) ? trim($matches[1]) : '');  // Get title
-                $link['title'] = html_entity_decode($link['title'],ENT_QUOTES,'UTF-8');
-                preg_match_all('! ([A-Z_]+)=\"(.*?)"!i',$html,$matches,PREG_SET_ORDER);  // Get all other attributes
-                $raw_add_date=0;
-                foreach($matches as $m)
-                {
-                    $attr=$m[1]; $value=$m[2];
-                    if ($attr=='HREF') $link['url']=html_entity_decode($value,ENT_QUOTES,'UTF-8');
-                    elseif ($attr=='ADD_DATE')
-                    {
-                        $raw_add_date=intval($value);
-                        if ($raw_add_date>30000000000) $raw_add_date/=1000;     //If larger than year 2920, then was likely stored in milliseconds instead of seconds
-                    }
-                    elseif ($attr=='PRIVATE') $link['private']=($value=='0'?0:1);
-                    elseif ($attr=='TAGS') $link['tags']=html_entity_decode(str_replace(',',' ',$value),ENT_QUOTES,'UTF-8');
-                }
-                if ($link['url']!='')
-                {
-                    if ($private==1) $link['private']=1;
-                    $dblink = $LINKSDB->getLinkFromUrl($link['url']); // See if the link is already in database.
-                    if ($dblink==false)
-                    {  // Link not in database, let's import it...
-                       if (empty($raw_add_date)) $raw_add_date=time(); // In case of shitty bookmark file with no ADD_DATE
-
-                       // Make sure date/time is not already used by another link.
-                       // (Some bookmark files have several different links with the same ADD_DATE)
-                       // We increment date by 1 second until we find a date which is not used in DB.
-                       // (so that links that have the same date/time are more or less kept grouped by date, but do not conflict.)
-                       while (!empty($LINKSDB[date('Ymd_His',$raw_add_date)])) { $raw_add_date++; }// Yes, I know it's ugly.
-                       $link['linkdate']=date('Ymd_His',$raw_add_date);
-                       $LINKSDB[$link['linkdate']] = $link;
-                       $import_count++;
-                    }
-                    else // Link already present in database.
-                    {
-                        if ($overwrite)
-                        {   // If overwrite is required, we import link data, except date/time.
-                            $link['linkdate']=$dblink['linkdate'];
-                            $LINKSDB[$link['linkdate']] = $link;
-                            $import_count++;
-                        }
-                    }
-
-                }
-            }
-        }
-        $LINKSDB->savedb($conf->get('resource.page_cache'));
-
-        echo '<script>alert("File '.json_encode($filename).' ('.$filesize.' bytes) was successfully processed: '.$import_count.' links imported.");document.location=\'?\';</script>';
-    }
-    else
-    {
-        echo '<script>alert("File '.json_encode($filename).' ('.$filesize.' bytes) has an unknown file format. Nothing was imported.");document.location=\'?\';</script>';
-    }
-}
-
-/**
  * Template for the list of links (<div id="linklist">)
  * This function fills all the necessary fields in the $PAGE for the template 'linklist.html'
  *
diff --git a/tests/NetscapeBookmarkUtilsTest.php b/tests/NetscapeBookmarkUtils/BookmarkExportTest.php
index 41e6d84c..cc54ab9f 100644
--- a/tests/NetscapeBookmarkUtilsTest.php
+++ b/tests/NetscapeBookmarkUtils/BookmarkExportTest.php
@@ -3,9 +3,9 @@
 require_once 'application/NetscapeBookmarkUtils.php';
 
 /**
- * Netscape bookmark import and export
+ * Netscape bookmark export
  */
-class NetscapeBookmarkUtilsTest extends PHPUnit_Framework_TestCase
+class BookmarkExportTest extends PHPUnit_Framework_TestCase
 {
     /**
      * @var string datastore to test write operations
diff --git a/tests/NetscapeBookmarkUtils/BookmarkImportTest.php b/tests/NetscapeBookmarkUtils/BookmarkImportTest.php
new file mode 100644
index 00000000..2d4e7557
--- /dev/null
+++ b/tests/NetscapeBookmarkUtils/BookmarkImportTest.php
@@ -0,0 +1,518 @@
+<?php
+
+require_once 'application/NetscapeBookmarkUtils.php';
+
+
+/**
+ * Utility function to load a file's metadata in a $_FILES-like array
+ *
+ * @param string $filename Basename of the file
+ *
+ * @return array A $_FILES-like array
+ */
+function file2array($filename)
+{
+    return array(
+        'filetoupload' => array(
+            'name'     => $filename,
+            'tmp_name' => __DIR__ . '/input/' . $filename,
+            'size'     => filesize(__DIR__ . '/input/' . $filename)
+        )
+    );
+}
+
+
+/**
+ * Netscape bookmark import
+ */
+class BookmarkImportTest extends PHPUnit_Framework_TestCase
+{
+    /**
+     * @var string datastore to test write operations
+     */
+    protected static $testDatastore = 'sandbox/datastore.php';
+
+    /**
+     * @var LinkDB private LinkDB instance
+     */
+    protected $linkDb = null;
+
+    /**
+     * @var string Dummy page cache
+     */
+    protected $pagecache = 'tests';
+
+    /**
+     * Resets test data before each test
+     */
+    protected function setUp()
+    {
+        if (file_exists(self::$testDatastore)) {
+            unlink(self::$testDatastore);
+        }
+        // start with an empty datastore
+        file_put_contents(self::$testDatastore, '<?php /* S7QysKquBQA= */ ?>');
+        $this->linkDb = new LinkDB(self::$testDatastore, true, false);
+    }
+
+    /**
+     * Attempt to import bookmarks from an empty file
+     */
+    public function testImportEmptyData()
+    {
+        $files = file2array('empty.htm');
+        $this->assertEquals(
+            'File empty.htm (0 bytes) has an unknown file format.'
+            .' Nothing was imported.',
+            NetscapeBookmarkUtils::import(NULL, $files, NULL, NULL)
+        );
+        $this->assertEquals(0, count($this->linkDb));
+    }
+
+    /**
+     * Attempt to import bookmarks from a file with no Doctype
+     */
+    public function testImportNoDoctype()
+    {
+        $files = file2array('no_doctype.htm');
+        $this->assertEquals(
+            'File no_doctype.htm (350 bytes) has an unknown file format. Nothing was imported.',
+            NetscapeBookmarkUtils::import(NULL, $files, NULL, NULL)
+        );
+        $this->assertEquals(0, count($this->linkDb));
+    }
+
+    /**
+     * Import bookmarks nested in a folder hierarchy
+     */
+    public function testImportNested()
+    {
+        $files = file2array('netscape_nested.htm');
+        $this->assertEquals(
+            'File netscape_nested.htm (1337 bytes) was successfully processed:'
+            .' 8 links imported, 0 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import(array(), $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(8, count($this->linkDb));
+        $this->assertEquals(2, count_private($this->linkDb));
+
+        $this->assertEquals(
+            array(
+                'linkdate' => '20160225_205541',
+                'title' => 'Nested 1',
+                'url' => 'http://nest.ed/1',
+                'description' => '',
+                'private' => 0,
+                'tags' => 'tag1 tag2'
+            ),
+            $this->linkDb->getLinkFromUrl('http://nest.ed/1')
+        );
+        $this->assertEquals(
+            array(
+                'linkdate' => '20160225_205542',
+                'title' => 'Nested 1-1',
+                'url' => 'http://nest.ed/1-1',
+                'description' => '',
+                'private' => 0,
+                'tags' => 'folder1 tag1 tag2'
+            ),
+            $this->linkDb->getLinkFromUrl('http://nest.ed/1-1')
+        );
+        $this->assertEquals(
+            array(
+                'linkdate' => '20160225_205547',
+                'title' => 'Nested 1-2',
+                'url' => 'http://nest.ed/1-2',
+                'description' => '',
+                'private' => 0,
+                'tags' => 'folder1 tag3 tag4'
+            ),
+            $this->linkDb->getLinkFromUrl('http://nest.ed/1-2')
+        );
+        $this->assertEquals(
+            array(
+                'linkdate' => '20160202_172222',
+                'title' => 'Nested 2-1',
+                'url' => 'http://nest.ed/2-1',
+                'description' => 'First link of the second section',
+                'private' => 1,
+                'tags' => 'folder2'
+            ),
+            $this->linkDb->getLinkFromUrl('http://nest.ed/2-1')
+        );
+        $this->assertEquals(
+            array(
+                'linkdate' => '20160119_200227',
+                'title' => 'Nested 2-2',
+                'url' => 'http://nest.ed/2-2',
+                'description' => 'Second link of the second section',
+                'private' => 1,
+                'tags' => 'folder2'
+            ),
+            $this->linkDb->getLinkFromUrl('http://nest.ed/2-2')
+        );
+        $this->assertEquals(
+            array(
+                'linkdate' => '20160202_172223',
+                'title' => 'Nested 3-1',
+                'url' => 'http://nest.ed/3-1',
+                'description' => '',
+                'private' => 0,
+                'tags' => 'folder3 folder3-1 tag3'
+            ),
+            $this->linkDb->getLinkFromUrl('http://nest.ed/3-1')
+        );
+        $this->assertEquals(
+            array(
+                'linkdate' => '20160119_200228',
+                'title' => 'Nested 3-2',
+                'url' => 'http://nest.ed/3-2',
+                'description' => '',
+                'private' => 0,
+                'tags' => 'folder3 folder3-1'
+            ),
+            $this->linkDb->getLinkFromUrl('http://nest.ed/3-2')
+        );
+        $this->assertEquals(
+            array(
+                'linkdate' => '20160229_081541',
+                'title' => 'Nested 2',
+                'url' => 'http://nest.ed/2',
+                'description' => '',
+                'private' => 0,
+                'tags' => 'tag4'
+            ),
+            $this->linkDb->getLinkFromUrl('http://nest.ed/2')
+        );
+    }
+
+    /**
+     * Import bookmarks with the default privacy setting (reuse from file)
+     *
+     * The $_POST array is not set.
+     */
+    public function testImportDefaultPrivacyNoPost()
+    {
+        $files = file2array('netscape_basic.htm');
+        $this->assertEquals(
+            'File netscape_basic.htm (482 bytes) was successfully processed:'
+            .' 2 links imported, 0 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import(array(), $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(2, count($this->linkDb));
+        $this->assertEquals(1, count_private($this->linkDb));
+
+        $this->assertEquals(
+            array(
+                'linkdate' => '20001010_105536',
+                'title' => 'Secret stuff',
+                'url' => 'https://private.tld',
+                'description' => "Super-secret stuff you're not supposed to know about",
+                'private' => 1,
+                'tags' => 'private secret'
+            ),
+            $this->linkDb->getLinkFromUrl('https://private.tld')
+        );
+        $this->assertEquals(
+            array(
+                'linkdate' => '20160225_205548',
+                'title' => 'Public stuff',
+                'url' => 'http://public.tld',
+                'description' => '',
+                'private' => 0,
+                'tags' => 'public hello world'
+            ),
+            $this->linkDb->getLinkFromUrl('http://public.tld')
+        );
+    }
+
+    /**
+     * Import bookmarks with the default privacy setting (reuse from file)
+     */
+    public function testImportKeepPrivacy()
+    {
+        $post = array('privacy' => 'default');
+        $files = file2array('netscape_basic.htm');
+        $this->assertEquals(
+            'File netscape_basic.htm (482 bytes) was successfully processed:'
+            .' 2 links imported, 0 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import($post, $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(2, count($this->linkDb));
+        $this->assertEquals(1, count_private($this->linkDb));
+
+        $this->assertEquals(
+            array(
+                'linkdate' => '20001010_105536',
+                'title' => 'Secret stuff',
+                'url' => 'https://private.tld',
+                'description' => "Super-secret stuff you're not supposed to know about",
+                'private' => 1,
+                'tags' => 'private secret'
+            ),
+            $this->linkDb->getLinkFromUrl('https://private.tld')
+        );
+        $this->assertEquals(
+            array(
+                'linkdate' => '20160225_205548',
+                'title' => 'Public stuff',
+                'url' => 'http://public.tld',
+                'description' => '',
+                'private' => 0,
+                'tags' => 'public hello world'
+            ),
+            $this->linkDb->getLinkFromUrl('http://public.tld')
+        );
+    }
+
+    /**
+     * Import links as public
+     */
+    public function testImportAsPublic()
+    {
+        $post = array('privacy' => 'public');
+        $files = file2array('netscape_basic.htm');
+        $this->assertEquals(
+            'File netscape_basic.htm (482 bytes) was successfully processed:'
+            .' 2 links imported, 0 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import($post, $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(2, count($this->linkDb));
+        $this->assertEquals(0, count_private($this->linkDb));
+        $this->assertEquals(
+            0,
+            $this->linkDb['20001010_105536']['private']
+        );
+        $this->assertEquals(
+            0,
+            $this->linkDb['20160225_205548']['private']
+        );
+    }
+
+    /**
+     * Import links as private
+     */
+    public function testImportAsPrivate()
+    {
+        $post = array('privacy' => 'private');
+        $files = file2array('netscape_basic.htm');
+        $this->assertEquals(
+            'File netscape_basic.htm (482 bytes) was successfully processed:'
+            .' 2 links imported, 0 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import($post, $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(2, count($this->linkDb));
+        $this->assertEquals(2, count_private($this->linkDb));
+        $this->assertEquals(
+            1,
+            $this->linkDb['20001010_105536']['private']
+        );
+        $this->assertEquals(
+            1,
+            $this->linkDb['20160225_205548']['private']
+        );
+    }
+
+    /**
+     * Overwrite private links so they become public
+     */
+    public function testOverwriteAsPublic()
+    {
+        $files = file2array('netscape_basic.htm');
+
+        // import links as private
+        $post = array('privacy' => 'private');
+        $this->assertEquals(
+            'File netscape_basic.htm (482 bytes) was successfully processed:'
+            .' 2 links imported, 0 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import($post, $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(2, count($this->linkDb));
+        $this->assertEquals(2, count_private($this->linkDb));
+        $this->assertEquals(
+            1,
+            $this->linkDb['20001010_105536']['private']
+        );
+        $this->assertEquals(
+            1,
+            $this->linkDb['20160225_205548']['private']
+        );
+
+        // re-import as public, enable overwriting
+        $post = array(
+            'privacy' => 'public',
+            'overwrite' => 'true'
+        );
+        $this->assertEquals(
+            'File netscape_basic.htm (482 bytes) was successfully processed:'
+            .' 2 links imported, 2 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import($post, $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(2, count($this->linkDb));
+        $this->assertEquals(0, count_private($this->linkDb));
+        $this->assertEquals(
+            0,
+            $this->linkDb['20001010_105536']['private']
+        );
+        $this->assertEquals(
+            0,
+            $this->linkDb['20160225_205548']['private']
+        );
+    }
+
+    /**
+     * Overwrite public links so they become private
+     */
+    public function testOverwriteAsPrivate()
+    {
+        $files = file2array('netscape_basic.htm');
+
+        // import links as public
+        $post = array('privacy' => 'public');
+        $this->assertEquals(
+            'File netscape_basic.htm (482 bytes) was successfully processed:'
+            .' 2 links imported, 0 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import($post, $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(2, count($this->linkDb));
+        $this->assertEquals(0, count_private($this->linkDb));
+        $this->assertEquals(
+            0,
+            $this->linkDb['20001010_105536']['private']
+        );
+        $this->assertEquals(
+            0,
+            $this->linkDb['20160225_205548']['private']
+        );
+
+        // re-import as private, enable overwriting
+        $post = array(
+            'privacy' => 'private',
+            'overwrite' => 'true'
+        );
+        $this->assertEquals(
+            'File netscape_basic.htm (482 bytes) was successfully processed:'
+            .' 2 links imported, 2 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import($post, $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(2, count($this->linkDb));
+        $this->assertEquals(2, count_private($this->linkDb));
+        $this->assertEquals(
+            1,
+            $this->linkDb['20001010_105536']['private']
+        );
+        $this->assertEquals(
+            1,
+            $this->linkDb['20160225_205548']['private']
+        );
+    }
+
+    /**
+     * Attept to import the same links twice without enabling overwriting
+     */
+    public function testSkipOverwrite()
+    {
+        $post = array('privacy' => 'public');
+        $files = file2array('netscape_basic.htm');
+        $this->assertEquals(
+            'File netscape_basic.htm (482 bytes) was successfully processed:'
+            .' 2 links imported, 0 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import($post, $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(2, count($this->linkDb));
+        $this->assertEquals(0, count_private($this->linkDb));
+
+        // re-import as private, DO NOT enable overwriting
+        $post = array('privacy' => 'private');
+        $this->assertEquals(
+            'File netscape_basic.htm (482 bytes) was successfully processed:'
+            .' 0 links imported, 0 links overwritten, 2 links skipped.',
+            NetscapeBookmarkUtils::import($post, $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(2, count($this->linkDb));
+        $this->assertEquals(0, count_private($this->linkDb));
+    }
+
+    /**
+     * Add user-specified tags to all imported bookmarks
+     */
+    public function testSetDefaultTags()
+    {
+        $post = array(
+            'privacy' => 'public',
+            'default_tags' => 'tag1,tag2 tag3'
+        );
+        $files = file2array('netscape_basic.htm');
+        $this->assertEquals(
+            'File netscape_basic.htm (482 bytes) was successfully processed:'
+            .' 2 links imported, 0 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import($post, $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(2, count($this->linkDb));
+        $this->assertEquals(0, count_private($this->linkDb));
+        $this->assertEquals(
+            'tag1 tag2 tag3 private secret',
+            $this->linkDb['20001010_105536']['tags']
+        );
+        $this->assertEquals(
+            'tag1 tag2 tag3 public hello world',
+            $this->linkDb['20160225_205548']['tags']
+        );
+    }
+
+    /**
+     * The user-specified tags contain characters to be escaped
+     */
+    public function testSanitizeDefaultTags()
+    {
+        $post = array(
+            'privacy' => 'public',
+            'default_tags' => 'tag1&,tag2 "tag3"'
+        );
+        $files = file2array('netscape_basic.htm');
+        $this->assertEquals(
+            'File netscape_basic.htm (482 bytes) was successfully processed:'
+            .' 2 links imported, 0 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import($post, $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(2, count($this->linkDb));
+        $this->assertEquals(0, count_private($this->linkDb));
+        $this->assertEquals(
+            'tag1&amp; tag2 &quot;tag3&quot; private secret',
+            $this->linkDb['20001010_105536']['tags']
+        );
+        $this->assertEquals(
+            'tag1&amp; tag2 &quot;tag3&quot; public hello world',
+            $this->linkDb['20160225_205548']['tags']
+        );
+    }
+
+    /**
+     * Ensure each imported bookmark has a unique linkdate
+     *
+     * See https://github.com/shaarli/Shaarli/issues/351
+     */
+    public function testImportSameDate()
+    {
+        $files = file2array('same_date.htm');
+        $this->assertEquals(
+            'File same_date.htm (453 bytes) was successfully processed:'
+            .' 3 links imported, 0 links overwritten, 0 links skipped.',
+            NetscapeBookmarkUtils::import(array(), $files, $this->linkDb, $this->pagecache)
+        );
+        $this->assertEquals(3, count($this->linkDb));
+        $this->assertEquals(0, count_private($this->linkDb));
+        $this->assertEquals(
+            '20160225_205548',
+            $this->linkDb['20160225_205548']['linkdate']
+        );
+        $this->assertEquals(
+            '20160225_205549',
+            $this->linkDb['20160225_205549']['linkdate']
+        );
+        $this->assertEquals(
+            '20160225_205550',
+            $this->linkDb['20160225_205550']['linkdate']
+        );
+    }
+}
diff --git a/tests/NetscapeBookmarkUtils/input/empty.htm b/tests/NetscapeBookmarkUtils/input/empty.htm
new file mode 100644
index 00000000..e69de29b
--- /dev/null
+++ b/tests/NetscapeBookmarkUtils/input/empty.htm
diff --git a/tests/NetscapeBookmarkUtils/input/netscape_basic.htm b/tests/NetscapeBookmarkUtils/input/netscape_basic.htm
new file mode 100644
index 00000000..affe0cf8
--- /dev/null
+++ b/tests/NetscapeBookmarkUtils/input/netscape_basic.htm
@@ -0,0 +1,11 @@
+<!DOCTYPE NETSCAPE-Bookmark-file-1>
+<!-- This is an automatically generated file.
+It will be read and overwritten.
+Do Not Edit! -->
+<TITLE>Bookmarks</TITLE>
+<H1>Bookmarks</H1>
+<DL><p>
+<DT><A HREF="https://private.tld" ADD_DATE="10/Oct/2000:13:55:36 +0300" PRIVATE="1" TAGS="private secret">Secret stuff</A>
+<DD>Super-secret stuff you're not supposed to know about
+<DT><A HREF="http://public.tld" ADD_DATE="1456433748" PRIVATE="0" TAGS="public hello world">Public stuff</A>
+</DL><p>
diff --git a/tests/NetscapeBookmarkUtils/input/netscape_nested.htm b/tests/NetscapeBookmarkUtils/input/netscape_nested.htm
new file mode 100644
index 00000000..b486fe18
--- /dev/null
+++ b/tests/NetscapeBookmarkUtils/input/netscape_nested.htm
@@ -0,0 +1,31 @@
+<!DOCTYPE NETSCAPE-Bookmark-file-1>
+<!-- This is an automatically generated file.
+It will be read and overwritten.
+Do Not Edit! -->
+<TITLE>Bookmarks</TITLE>
+<H1>Bookmarks</H1>
+<DL><p>
+  <DT><A HREF="http://nest.ed/1" ADD_DATE="1456433741" PRIVATE="0" TAGS="tag1,tag2">Nested 1</A>
+  <DT><H3 ADD_DATE="1456433722" LAST_MODIFIED="1456433739">Folder1</H3>
+  <DL><p>
+    <DT><A HREF="http://nest.ed/1-1" ADD_DATE="1456433742" PRIVATE="0" TAGS="tag1,tag2">Nested 1-1</A>
+    <DT><A HREF="http://nest.ed/1-2" ADD_DATE="1456433747" PRIVATE="0" TAGS="tag3,tag4">Nested 1-2</A>
+  </DL><p>
+  <DT><H3 ADD_DATE="1456433722">Folder2</H3>
+  <DD>This second folder contains wonderful links!
+  <DL><p>
+    <DT><A HREF="http://nest.ed/2-1" ADD_DATE="1454433742" PRIVATE="1">Nested 2-1</A>
+    <DD>First link of the second section
+    <DT><A HREF="http://nest.ed/2-2" ADD_DATE="1453233747" PRIVATE="1">Nested 2-2</A>
+    <DD>Second link of the second section
+  </DL><p>
+  <DT><H3>Folder3</H3>
+  <DL><p>
+    <DT><H3>Folder3-1</H3>
+    <DL><p>
+      <DT><A HREF="http://nest.ed/3-1" ADD_DATE="1454433742" PRIVATE="0" TAGS="tag3">Nested 3-1</A>
+      <DT><A HREF="http://nest.ed/3-2" ADD_DATE="1453233747" PRIVATE="0">Nested 3-2</A>
+    </DL><p>
+  </DL><p>
+  <DT><A HREF="http://nest.ed/2" ADD_DATE="1456733741" PRIVATE="0" TAGS="tag4">Nested 2</A>
+</DL><p>
diff --git a/tests/NetscapeBookmarkUtils/input/no_doctype.htm b/tests/NetscapeBookmarkUtils/input/no_doctype.htm
new file mode 100644
index 00000000..766d398b
--- /dev/null
+++ b/tests/NetscapeBookmarkUtils/input/no_doctype.htm
@@ -0,0 +1,7 @@
+<TITLE>Bookmarks</TITLE>
+<H1>Bookmarks</H1>
+<DL><p>
+<DT><A HREF="https://private.tld" ADD_DATE="10/Oct/2000:13:55:36 +0300" PRIVATE="1" TAGS="private secret">Secret stuff</A>
+<DD>Super-secret stuff you're not supposed to know about
+<DT><A HREF="http://public.tld" ADD_DATE="1456433748" PRIVATE="0" TAGS="public hello world">Public stuff</A>
+</DL><p>
diff --git a/tests/NetscapeBookmarkUtils/input/same_date.htm b/tests/NetscapeBookmarkUtils/input/same_date.htm
new file mode 100644
index 00000000..9d58a582
--- /dev/null
+++ b/tests/NetscapeBookmarkUtils/input/same_date.htm
@@ -0,0 +1,11 @@
+<!DOCTYPE NETSCAPE-Bookmark-file-1>
+<!-- This is an automatically generated file.
+It will be read and overwritten.
+Do Not Edit! -->
+<TITLE>Bookmarks</TITLE>
+<H1>Bookmarks</H1>
+<DL><p>
+<DT><A HREF="https://fir.st" ADD_DATE="1456433748" PRIVATE="0">Today's first link</A>
+<DT><A HREF="https://seco.nd" ADD_DATE="1456433748" PRIVATE="0">Today's second link</A>
+<DT><A HREF="https://thi.rd" ADD_DATE="1456433748" PRIVATE="0">Today's third link</A>
+</DL><p>
diff --git a/tpl/import.html b/tpl/import.html
index 6c4f9421..071e1160 100644
--- a/tpl/import.html
+++ b/tpl/import.html
@@ -3,19 +3,31 @@
 <head>{include="includes"}</head>
 <body onload="document.uploadform.filetoupload.focus();">
 <div id="pageheader">
-        {include="page.header"}
-        <div id="uploaddiv">
-        Import Netscape HTML bookmarks (as exported from Firefox/Chrome/Opera/Delicious/Diigo...) (Max: {$maxfilesize} bytes).
-        <form method="POST" action="?do=upload" enctype="multipart/form-data" name="uploadform" id="uploadform">
-            <input type="hidden" name="token" value="{$token}">
-            <input type="file" name="filetoupload">
-            <input type="hidden" name="MAX_FILE_SIZE" value="{$maxfilesize}">
-            <input type="submit" name="import_file" value="Import" class="bigbutton"><br>
-            <input type="checkbox" name="private" id="private"><label for="private">&nbsp;Import all links as private</label><br>
-            <input type="checkbox" name="overwrite" id="overwrite"><label for="overwrite">&nbsp;Overwrite existing links</label>
-        </form>
-        </div>
+  {include="page.header"}
+  <div id="uploaddiv">
+    Import Netscape HTML bookmarks (as exported from Firefox/Chrome/Opera/Delicious/Diigo...) (Max: {$maxfilesize} bytes).
+    <form method="POST" action="?do=import" enctype="multipart/form-data"
+          name="uploadform" id="uploadform">
+      <input type="hidden" name="token" value="{$token}">
+      <input type="hidden" name="MAX_FILE_SIZE" value="{$maxfilesize}">
+      <input type="file" name="filetoupload">
+      <input type="submit" name="import_file" value="Import" class="bigbutton"><br>
+
+      <label for="privacy">&nbsp;Visibility:</label><br>
+      <input type="radio" name="privacy" value="default" checked="true">
+      &nbsp;Use values from the imported file, default to public<br>
+      <input type="radio" name="privacy" value="private">
+      &nbsp;Import all bookmarks as private<br>
+      <input type="radio" name="privacy" value="public">
+      &nbsp;Import all bookmarks as public<br>
+
+      <input type="checkbox" name="overwrite" id="overwrite">
+      <label for="overwrite">&nbsp;Overwrite existing bookmarks</label><br>
+      <label for="default_tags">&nbsp;Add default tags</label>
+      <input type="text" name="default_tags" id="default_tags">
+    </form>
+  </div>
 </div>
 {include="page.footer"}
 </body>
-</html>
\ No newline at end of file
+</html>