From b3c260a14529c5c797df50eeae0dacaa931436a8 Mon Sep 17 00:00:00 2001 From: Mario Date: Wed, 20 Mar 2024 09:31:17 +0100 Subject: fix reshares from streams loosing image --- include/html2bbcode.php | 30 +++++++++++++++--------------- tests/unit/includes/BBCodeTest.php | 4 ++++ vendor/composer/installed.php | 4 ++-- 3 files changed, 21 insertions(+), 17 deletions(-) diff --git a/include/html2bbcode.php b/include/html2bbcode.php index 009c051e6..03e09cd62 100644 --- a/include/html2bbcode.php +++ b/include/html2bbcode.php @@ -187,25 +187,24 @@ function html2bbcode($message) node2bbcode($doc, 'u', array(), '[u]', '[/u]'); node2bbcode($doc, 's', array(), '[s]', '[/s]'); node2bbcode($doc, 'mark', array(), '[mark]', '[/mark]'); + node2bbcode($doc, 'span', array(), "", ""); node2bbcode($doc, 'big', array(), "[size=large]", "[/size]"); node2bbcode($doc, 'small', array(), "[size=small]", "[/size]"); - node2bbcode($doc, 'blockquote', array(), '[quote]', '[/quote]'); - // Use a temporary tag to keep line breaks node2bbcode($doc, 'br', array(), '[br]', ''); node2bbcode($doc, 'a', array('href'=>'/(.+)/'), '[url=$1]', '[/url]'); - node2bbcode($doc, 'p', array('class'=>'MsoNormal'), "\n", ""); - node2bbcode($doc, 'div', array('class'=>'MsoNormal'), "\r", ""); + node2bbcode($doc, 'img', array('src'=>'/(.+)/', 'width'=>'/(\d+)/', 'height'=>'/(\d+)/'), '[img=$2x$3]$1', '[/img]'); + node2bbcode($doc, 'img', array('src'=>'/(.+)/', 'alt'=>'/(.+)/'), '[img=$1]$2', '[/img]'); + node2bbcode($doc, 'img', array('src'=>'/(.+)/'), '[img]$1', '[/img]'); - node2bbcode($doc, 'span', array(), "", ""); + node2bbcode($doc, 'video', array('src'=>'/(.+)/'), '[video]$1', '[/video]'); + node2bbcode($doc, 'audio', array('src'=>'/(.+)/'), '[audio]$1', '[/audio]'); +// node2bbcode($doc, 'iframe', array('src'=>'/(.+)/'), '[iframe]$1', '[/iframe]'); - node2bbcode($doc, 'pre', array(), "", ""); - node2bbcode($doc, 'div', array(), "\r", "\r"); - node2bbcode($doc, 'p', array(), "\n", "\n"); node2bbcode($doc, 'ul', array(), "[list]", "[/list]"); node2bbcode($doc, 'ol', array(), "[list=1]", "[/list]"); @@ -229,18 +228,19 @@ function html2bbcode($message) node2bbcode($doc, 'h5', array(), "[h5]", "[/h5]"); node2bbcode($doc, 'h6', array(), "[h6]", "[/h6]"); - node2bbcode($doc, 'img', array('src'=>'/(.+)/', 'width'=>'/(\d+)/', 'height'=>'/(\d+)/'), '[img=$2x$3]$1', '[/img]'); - node2bbcode($doc, 'img', array('src'=>'/(.+)/', 'alt'=>'/(.+)/'), '[img=$1]$2', '[/img]'); - node2bbcode($doc, 'img', array('src'=>'/(.+)/'), '[img]$1', '[/img]'); - - node2bbcode($doc, 'video', array('src'=>'/(.+)/'), '[video]$1', '[/video]'); - node2bbcode($doc, 'audio', array('src'=>'/(.+)/'), '[audio]$1', '[/audio]'); -// node2bbcode($doc, 'iframe', array('src'=>'/(.+)/'), '[iframe]$1', '[/iframe]'); + node2bbcode($doc, 'blockquote', array(), '[quote]', '[/quote]'); + node2bbcode($doc, 'pre', array(), "", ""); node2bbcode($doc, 'code', array('class'=>'/(.+)/'), '[code=$1]', '[/code]'); node2bbcode($doc, 'code', array(), '[code]', '[/code]'); + node2bbcode($doc, 'p', array('class'=>'MsoNormal'), "\n", ""); + node2bbcode($doc, 'p', array(), "\n", "\n"); + + node2bbcode($doc, 'div', array('class'=>'MsoNormal'), "\r", ""); + node2bbcode($doc, 'div', array(), "\r", "\r"); + $message = $doc->saveHTML(); // I'm removing the UTF-8 encoding of a NO-BREAK SPACE codepoint diff --git a/tests/unit/includes/BBCodeTest.php b/tests/unit/includes/BBCodeTest.php index 0da50f77a..6f79c45f4 100644 --- a/tests/unit/includes/BBCodeTest.php +++ b/tests/unit/includes/BBCodeTest.php @@ -198,6 +198,10 @@ class BBCodeTest extends UnitTestCase { 'nested tags with ampersand and new line' => [ "\nfoo & bar", '[b] [i]foo & bar[/i][/b]' + ], + 'html reshares from streams' => [ + '
image/photo shared something
something
', + '[url=https://example.com][img=https://example.com/image.jpg]image/photo[/img][/url] shared something' . "\n" . 'something' ] ]; } diff --git a/vendor/composer/installed.php b/vendor/composer/installed.php index 17113d857..9c1c3eb6a 100644 --- a/vendor/composer/installed.php +++ b/vendor/composer/installed.php @@ -3,7 +3,7 @@ 'name' => 'zotlabs/hubzilla', 'pretty_version' => 'dev-9.0RC', 'version' => 'dev-9.0RC', - 'reference' => 'fe018d646a4f25d3512059f6e8d520441e95967c', + 'reference' => '8754f72e63ea0fd9067c20104a2b068e5c518d0a', 'type' => 'application', 'install_path' => __DIR__ . '/../../', 'aliases' => array(), @@ -340,7 +340,7 @@ 'zotlabs/hubzilla' => array( 'pretty_version' => 'dev-9.0RC', 'version' => 'dev-9.0RC', - 'reference' => 'fe018d646a4f25d3512059f6e8d520441e95967c', + 'reference' => '8754f72e63ea0fd9067c20104a2b068e5c518d0a', 'type' => 'application', 'install_path' => __DIR__ . '/../../', 'aliases' => array(), -- cgit v1.2.3 From 3de8f5e7de424563cfe521848b2e2d11d100fcc5 Mon Sep 17 00:00:00 2001 From: Mario Date: Wed, 20 Mar 2024 14:35:43 +0100 Subject: deal with wihitespace issues in tables and lists and add tests --- include/bbcode.php | 47 +++++++++++++++++++++++++++++++------- tests/unit/includes/BBCodeTest.php | 12 ++++++++++ vendor/composer/installed.php | 4 ++-- 3 files changed, 53 insertions(+), 10 deletions(-) diff --git a/include/bbcode.php b/include/bbcode.php index db4147c1e..86944ba60 100644 --- a/include/bbcode.php +++ b/include/bbcode.php @@ -1054,6 +1054,15 @@ function bb_fixtable_lf($match) { } +function bb_fix_lf($match) { + // remove extraneous whitespace between element tags since newlines will all + // be converted to '
' and turn your neatly crafted tables into a whole lot of + // empty space. + + $new_content = str_replace(["\n\r", "\n", "\r"], '', $match[1]); + return str_replace($match[1], $new_content, $match[0]); +} + function bbtopoll($s) { $pl = []; @@ -1207,8 +1216,6 @@ function bbcode($text, $options = []) { } } - - $text = bb_format_attachdata($text); // If we find any event code, turn it into an event. @@ -1249,7 +1256,7 @@ function bbcode($text, $options = []) { $text = str_replace("<", "<", $text); $text = str_replace(">", ">", $text); - $text = preg_replace_callback("/\[table\](.*?)\[\/table\]/ism",'bb_fixtable_lf',$text); + $text = str_replace(array("\t", " "), array("    ", "  "), $text); // Check for [code] text here, before the linefeeds are messed with. @@ -1487,10 +1494,10 @@ function bbcode($text, $options = []) { // Check for list text - $text = preg_replace("/
\[\*\]/ism",'[*]',$text); - + $text = str_replace(["\r\n[*]", "\r[*]", "\n[*]"], "[*]", $text); $text = str_replace("[*]", "
  • ", $text); + // handle nested lists $endlessloop = 0; @@ -1499,16 +1506,35 @@ function bbcode($text, $options = []) { ((strpos($text, "[/ul]") !== false) && (strpos($text, "[ul]") !== false)) || ((strpos($text, "[/dl]") !== false) && (strpos($text, "[dl") !== false)) || ((strpos($text, "[/li]") !== false) && (strpos($text, "[li]") !== false))) && (++$endlessloop < 20)) { + + $text = preg_replace_callback("/\[list\](.*?)\[\/list\]/ism",'bb_fix_lf', $text); $text = preg_replace("/\[list\](.*?)\[\/list\]/ism", '
      $1
    ', $text); + + $text = preg_replace_callback("/\[list=\](.*?)\[\/list\]/ism",'bb_fix_lf', $text); $text = preg_replace("/\[list=\](.*?)\[\/list\]/ism", '
      $1
    ', $text); + + $text = preg_replace_callback("/\[list=1\](.*?)\[\/list\]/ism",'bb_fix_lf', $text); $text = preg_replace("/\[list=1\](.*?)\[\/list\]/ism", '
      $1
    ', $text); + + $text = preg_replace_callback("/\[list=((?-i)i)\](.*?)\[\/list\]/ism",'bb_fix_lf', $text); $text = preg_replace("/\[list=((?-i)i)\](.*?)\[\/list\]/ism",'
      $2
    ', $text); + + $text = preg_replace_callback("/\[list=((?-i)I)\](.*?)\[\/list\]/ism",'bb_fix_lf', $text); $text = preg_replace("/\[list=((?-i)I)\](.*?)\[\/list\]/ism", '
      $2
    ', $text); + + $text = preg_replace_callback("/\[list=((?-i)a)\](.*?)\[\/list\]/ism",'bb_fix_lf', $text); $text = preg_replace("/\[list=((?-i)a)\](.*?)\[\/list\]/ism", '
      $2
    ', $text); + + $text = preg_replace_callback("/\[list=((?-i)A)\](.*?)\[\/list\]/ism",'bb_fix_lf', $text); $text = preg_replace("/\[list=((?-i)A)\](.*?)\[\/list\]/ism", '
      $2
    ', $text); - $text = preg_replace("/\[ul\](.*?)\[\/ul\]/ism", '
      $1
    ', $text); + + $text = preg_replace_callback("/\[ol\](.*?)\[\/ol\]/ism",'bb_fix_lf', $text); $text = preg_replace("/\[ol\](.*?)\[\/ol\]/ism", '
      $1
    ', $text); - $text = preg_replace("/\[\/li\]
    \[li\]/ism",'[/li][li]',$text); + + $text = preg_replace_callback("/\[ul\](.*?)\[\/ul\]/ism",'bb_fix_lf', $text); + $text = preg_replace("/\[ul\](.*?)\[\/ul\]/ism", '
      $1
    ', $text); + + $text = preg_replace("/\[\/li\]
    \[li\]/ism",'[/li][li]', $text); $text = preg_replace("/\[li\](.*?)\[\/li\]/ism", '
  • $1
    • ', $text); // [dl] tags have an optional [dl terms="bi"] form where bold/italic/underline/mono/large @@ -1521,24 +1547,29 @@ function bbcode($text, $options = []) { } if (strpos($text,'[checklist]') !== false) { + $text = preg_replace_callback("/\[checklist\](.*?)\[\/checklist\]/ism",'bb_fix_lf', $text); $text = preg_replace_callback("/\[checklist\](.*?)\[\/checklist\]/ism", 'bb_checklist', $text); } if (strpos($text,'[th]') !== false) { $text = preg_replace("/\[th\](.*?)\[\/th\]/sm", '$1', $text); } + if (strpos($text,'[td]') !== false) { $text = preg_replace("/\[td\](.*?)\[\/td\]/sm", '$1', $text); } + if (strpos($text,'[tr]') !== false) { $text = preg_replace("/\[tr\](.*?)\[\/tr\]/sm", '$1', $text); } + if (strpos($text,'[/table]') !== false) { + $text = preg_replace_callback("/\[table\](.*?)\[\/table\]/ism",'bb_fix_lf',$text); $text = preg_replace("/\[table\](.*?)\[\/table\]/sm", '$1
    ', $text); $text = preg_replace("/\[table border=1\](.*?)\[\/table\]/sm", '$1
    ', $text); $text = preg_replace("/\[table border=0\](.*?)\[\/table\]/sm", '$1
    ', $text); } - $text = str_replace('
    ', "\n", $text); + $text = str_replace('[hr]', '
    ', $text); // This is actually executed in prepare_body() diff --git a/tests/unit/includes/BBCodeTest.php b/tests/unit/includes/BBCodeTest.php index 6f79c45f4..100d28fc4 100644 --- a/tests/unit/includes/BBCodeTest.php +++ b/tests/unit/includes/BBCodeTest.php @@ -107,6 +107,18 @@ class BBCodeTest extends UnitTestCase { "[code]\ntestvar = \"this is a test\"\necho \"the message is \$testvar\"\n[/code]", '
    testvar = "this is a test"
    echo "the message is $testvar"
    ', ], + 'list with linebreaks \n' => [ + "some text\n[list]\n[*] item1\n[*] item2\n[/list]\nsome more text", + 'some text

    some more text' + ], + 'list with linebreaks \r' => [ + "some text\r[list]\r[*] item1\r[*] item2\r[/list]\rsome more text", + 'some text

    some more text' + ], + 'list with linebreaks \r\n' => [ + "some text\r\n[list]\r\n[*] item1\r\n[*] item2\r\n[/list]\r\nsome more text", + 'some text

    some more text' + ] ]; } diff --git a/vendor/composer/installed.php b/vendor/composer/installed.php index 9c1c3eb6a..4386e70b5 100644 --- a/vendor/composer/installed.php +++ b/vendor/composer/installed.php @@ -3,7 +3,7 @@ 'name' => 'zotlabs/hubzilla', 'pretty_version' => 'dev-9.0RC', 'version' => 'dev-9.0RC', - 'reference' => '8754f72e63ea0fd9067c20104a2b068e5c518d0a', + 'reference' => 'a18e873d08e733225c70b0ace31c3cbb025ff906', 'type' => 'application', 'install_path' => __DIR__ . '/../../', 'aliases' => array(), @@ -340,7 +340,7 @@ 'zotlabs/hubzilla' => array( 'pretty_version' => 'dev-9.0RC', 'version' => 'dev-9.0RC', - 'reference' => '8754f72e63ea0fd9067c20104a2b068e5c518d0a', + 'reference' => 'a18e873d08e733225c70b0ace31c3cbb025ff906', 'type' => 'application', 'install_path' => __DIR__ . '/../../', 'aliases' => array(), -- cgit v1.2.3