Merge pull request #70 from TangoMan75/master

shaarli · Jan 31, 2021 · 5b2eb8d · 5b2eb8d
2 parents 76ff8d7 + 7017391
commit 5b2eb8d
Show file tree

Hide file tree

Showing 5 changed files with 82 additions and 42 deletions.
diff --git a/.gitignore b/.gitignore
@@ -1,4 +1,10 @@
 composer.lock
 vendor
-logs/
 !.gitkeep
+
+/logs/
+
+###> squizlabs/php_codesniffer ###
+/.phpcs-cache
+/phpcs.xml
+###< squizlabs/php_codesniffer ###
diff --git a/NetscapeBookmarkParser.php b/NetscapeBookmarkParser.php
@@ -120,56 +120,66 @@ public function parseString(string $bookmarkString): array
 
         $lines = explode("\n", $this->sanitizeString($bookmarkString));
 
-        foreach ($lines as $line_no => $line) {
+        foreach ($lines as $lineNumber => $line) {
             $item = [];
 
-            $this->logger->info('PARSING LINE #' . $line_no);
-            $this->logger->debug('[#' . $line_no . '] Content: ' . $line);
-            if (preg_match('/^<h\d.*>(.*)<\/h\d>/i', $line, $m1)) {
+            $this->logger->info('PARSING LINE #' . $lineNumber);
+            $this->logger->debug('[#' . $lineNumber . '] Content: ' . $line);
+            if (preg_match('/^<h\d.*>(.*)<\/h\d>/i', $line, $header)) {
                 // a header is matched:
                 // - links may be grouped in a (sub-)folder
                 // - append the header's content to the folder tags
-                $tag = static::sanitizeTags($m1[1]);
+                $tag = static::sanitizeTags($header[1]);
 
                 $groupedFolderTags[] = $tag;
                 $folderTags = static::flattenTagsList($groupedFolderTags);
-                $this->logger->debug('[#' . $line_no . '] Header found: ' . implode(' ', $tag));
+                $this->logger->debug('[#' . $lineNumber . '] Header found: ' . implode(' ', $tag));
                 continue;
             } elseif (preg_match('/^<\/DL>/i', $line)) {
                 // </DL> matched: stop using header value
                 $tag = array_pop($groupedFolderTags);
                 $folderTags = static::flattenTagsList($groupedFolderTags);
-                $this->logger->debug('[#' . $line_no . '] Header ended: ' . implode(' ', $tag ?? []));
+                $this->logger->debug('[#' . $lineNumber . '] Header ended: ' . implode(' ', $tag ?? []));
                 continue;
             }
 
-            if (preg_match('/<a/i', $line, $m2)) {
-                $this->logger->debug('[#' . $line_no . '] Link found');
-                if (preg_match('/href="(.*?)"/i', $line, $m3)) {
-                    $item['uri'] = $m3[1];
-                    $this->logger->debug('[#' . $line_no . '] URL found: ' . $m3[1]);
+            if (preg_match('/<a/i', $line)) {
+                $this->logger->debug('[#' . $lineNumber . '] Link found');
+                if (preg_match('/href="(.*?)"/i', $line, $href)) {
+                    $item['uri'] = $href[1];
+                    $this->logger->debug('[#' . $lineNumber . '] URL found: ' . $href[1]);
                 } else {
                     $item['uri'] = '';
-                    $this->logger->debug('[#' . $line_no . '] Empty URL');
+                    $this->logger->debug('[#' . $lineNumber . '] Empty URL');
                 }
 
-                if (preg_match('/<a.*?[^br]>(.*?)<\/a>/i', $line, $m4)) {
-                    $item['title'] = $m4[1];
-                    $this->logger->debug('[#' . $line_no . '] Title found: ' . $m4[1]);
+                if (preg_match('/icon="(.*?)"/i', $line, $icon)) {
+                    $item['icon'] = $icon[1];
+                    $this->logger->debug('[#' . $lineNumber . '] ICON found: ' . $href[1]);
+                } else {
+                    $item['icon'] = '';
+                    $this->logger->debug('[#' . $lineNumber . '] Empty ICON');
+                }
+
+                if (preg_match('/<a.*?[^br]>(.*?)<\/a>/i', $line, $title)) {
+                    $item['title'] = $title[1];
+                    $this->logger->debug('[#' . $lineNumber . '] Title found: ' . $title[1]);
                 } else {
                     $item['title'] = 'untitled';
-                    $this->logger->debug('[#' . $line_no . '] Empty title');
+                    $this->logger->debug('[#' . $lineNumber . '] Empty title');
                 }
 
-                if (preg_match('/(description|note)="(.*?)"/i', $line, $m5)) {
-                    $item['note'] = $m5[2];
-                    $this->logger->debug('[#' . $line_no . '] Content found: ' . substr($m5[2], 0, 50) . '...');
-                } elseif (preg_match('/<dd>(.*?)$/i', $line, $m6)) {
-                    $item['note'] = str_replace('<br>', "\n", $m6[1]);
-                    $this->logger->debug('[#' . $line_no . '] Content found: ' . substr($m6[1], 0, 50) . '...');
+                if (preg_match('/(description|note)="(.*?)"/i', $line, $description)) {
+                    $item['note'] = $description[2];
+                    $this->logger->debug(
+                        '[#' . $lineNumber . '] Content found: ' . substr($description[2], 0, 50) . '...'
+                    );
+                } elseif (preg_match('/<dd>(.*?)$/i', $line, $note)) {
+                    $item['note'] = str_replace('<br>', "\n", $note[1]);
+                    $this->logger->debug('[#' . $lineNumber . '] Content found: ' . substr($note[1], 0, 50) . '...');
                 } else {
                     $item['note'] = '';
-                    $this->logger->debug('[#' . $line_no . '] Empty content');
+                    $this->logger->debug('[#' . $lineNumber . '] Empty content');
                 }
 
                 $tags = [];
@@ -180,32 +190,32 @@ public function parseString(string $bookmarkString): array
                     $tags = array_merge($tags, $folderTags);
                 }
 
-                if (preg_match('/(tags?|labels?|folders?)="(.*?)"/i', $line, $m7)) {
-                    $separator = strpos($m7[2], ',') !== false ? ',' : ' ';
+                if (preg_match('/(tags?|labels?|folders?)="(.*?)"/i', $line, $labels)) {
+                    $separator = strpos($labels[2], ',') !== false ? ',' : ' ';
                     $tags = array_merge(
                         $tags,
-                        static::splitTagString($m7[2], $separator)
+                        static::splitTagString($labels[2], $separator)
                     );
                 }
                 $item['tags'] = $tags;
-                $this->logger->debug('[#' . $line_no . '] Tag list: ' . implode(' ', $item['tags']));
+                $this->logger->debug('[#' . $lineNumber . '] Tag list: ' . implode(' ', $item['tags']));
 
-                if (preg_match('/add_date="(.*?)"/i', $line, $m8)) {
-                    $item['time'] = $this->parseDate($m8[1]);
+                if (preg_match('/add_date="(.*?)"/i', $line, $addDate)) {
+                    $item['time'] = $this->parseDate($addDate[1]);
                 } else {
                     $item['time'] = time();
                 }
-                $this->logger->debug('[#' . $line_no . '] Date: ' . $item['time']);
+                $this->logger->debug('[#' . $lineNumber . '] Date: ' . $item['time']);
 
-                if (preg_match('/(public|published|pub)="(.*?)"/i', $line, $m9)) {
-                    $item['pub'] = $this->parseBoolean($m9[2]) ? 1 : 0;
-                } elseif (preg_match('/(private|shared)="(.*?)"/i', $line, $m10)) {
-                    $item['pub'] = $this->parseBoolean($m10[2]) ? 0 : 1;
+                if (preg_match('/(public|published|pub)="(.*?)"/i', $line, $public)) {
+                    $item['pub'] = $this->parseBoolean($public[2]) ? 1 : 0;
+                } elseif (preg_match('/(private|shared)="(.*?)"/i', $line, $private)) {
+                    $item['pub'] = $this->parseBoolean($private[2]) ? 0 : 1;
                 } else {
                     $item['pub'] = $this->defaultPub;
                 }
                 $this->logger->debug(
-                    '[#' . $line_no . '] Visibility: ' . ($item['pub'] ? 'public' : 'private')
+                    '[#' . $lineNumber . '] Visibility: ' . ($item['pub'] ? 'public' : 'private')
                 );
 
                 $items[] = $item;

diff --git a/phpcs.xml.dist b/phpcs.xml.dist
@@ -1,14 +1,19 @@
 <?xml version="1.0"?>
 <ruleset xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" name="PHP_CodeSniffer" xsi:noNamespaceSchemaLocation="phpcs.xsd">
+
+    <!-- Documentation: https://github.com/squizlabs/PHP_CodeSniffer -->
+
     <arg name="basepath" value="."/>
     <arg name="cache" value=".phpcs-cache"/>
     <arg name="colors"/>
     <arg name="extensions" value="php"/>
 
+    <file>NetscapeBookmarkParser.php</file>
+    <file>tests</file>
+
     <rule ref="PSR12" />
+
     <rule ref="Generic.PHP.RequireStrictTypes.MissingDeclaration" />
-    <rule ref="Generic.Arrays.DisallowLongArraySyntax" />
+    <rule ref="Generic.Arrays.DisallowLongArraySyntax"/>
 
-    <file>NetscapeBookmarkParser.php</file>
-    <file>tests</file>
 </ruleset>
diff --git a/tests/ParseChromiumBookmarksTest.php b/tests/ParseChromiumBookmarksTest.php
@@ -19,6 +19,23 @@ protected function tearDown(): void
         @unlink(LoggerTestsUtils::getLogFile());
     }
 
+    public function testParseChromiumBookmarkShouldIncludeIcon()
+    {
+        $parser = new NetscapeBookmarkParser(false, null, '1');
+        $bkm = $parser->parseFile('tests/input/chromium_flat.htm');
+
+        // https://github.com/squizlabs/PHP_CodeSniffer/wiki/Advanced-Usage#ignoring-parts-of-a-file
+        // phpcs:disable Generic.Files.LineLength
+        $this->assertEquals('data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABAAAAAQCAYAAAAf8/9hAAACeElEQVQ4jX1TW0hUURRd+8yZuaPOdQxNiSiipAdjICISNhBFVIIRiBoYRP9BSSBFBXPnoyj666O+AoOEmMmpNBUKHSQRiTCDKYqeX6KBzuOa4zzu2X3M+NbW1zn77LX3WntzgI1gsFh1Z6Z1sTzkuggzgUgBAB4nSt1SqThRFADn33jzAvmEPU/jDXA6L1E6tUMxVFn3wg9rIen/TTSxtgitIAsQqf2BeBsT34bNfnlfeeFQZDIqSMhWCe7IZLjt51f3eM5mXiUAoIVtee/yYCgeORCarVrrbG9g5mxVKBbeYE5MAFD9PFpS22c+rHkZ688lhJftBXIN6nvjX04MzXVUPpkpXrQsAOL6vsQFTWLEoWkuAXULzAQML0tsJQsA0hbfTMBRXVpMo95XsRYQMR0eiDaRwh3Fsmm0UY9sspVVkz/UG6uRQnURc7vUiK6RxMXBk3oEYZYY9qvKunP27++6MjAMBSL2BNjx6TOyeb4YO03jx15Hr5OiG9QwGP0l55Oe3sZtSfj9BMNQx9/Mep02WafY6ibBuhDijFOlHwWPlk8BTDD85K1td+savxd2sJovcBW0BIMCPh8DTFn5cYyt9IAg4UGWt3Mi2xkcfvBnkWz4fCgpyBZKwRlqHoq+gA0Tz45sMdb5/w+aw7P3FKOC2t7GdrNl9ZBAyGXZ7yfT88lpAG6HoFSCCADcu8rU9NQ0KgDAVVRkqcwVZpxKWdxIAHB+ZHKnUtpdaYOHhC3DSoFZ5UZPAEGAAGYiAitpKf4gNefVzjrXFK1cU0P/t+KS8gqHjjmYJqDrLphmTrKuAyaAvwupVI93q7m04qXDJt91QzCLRfI/LI8pZq5PEwYAAAAASUVORK5CYII=', $bkm[0]['icon']);
+        // phpcs:enable
+
+        $parser = new NetscapeBookmarkParser(true, null, '1');
+        $bkm = $parser->parseFile('tests/input/chromium_nested.htm');
+        // phpcs:disable Generic.Files.LineLength
+        $this->assertEquals('data:image/png;base64,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', $bkm[1]['icon']);
+        // phpcs:enable
+    }
+
     /**
      * Parse flat Chromium bookmarks (no directories)
      */

diff --git a/tests/ParseShaarliBookmarksTest.php b/tests/ParseShaarliBookmarksTest.php
@@ -208,7 +208,8 @@ public function testParseLegacy()
                 'note' => 'simple on/off button',
                 'tags' => ['css'],
                 'time' => 1470640652,
-                'pub' => 0
+                'pub' => 0,
+                'icon' => ''
             ],
             $bkm[0]
         );
@@ -221,7 +222,8 @@ public function testParseLegacy()
                 'note' => '',
                 'tags' => ['apache'],
                 'time' => 1469950052,
-                'pub' => 1
+                'pub' => 1,
+                'icon' => ''
             ],
             $bkm[1]
         );