From 38311a4a02ef5d16a0b10a953a345cd7f2da6336 Mon Sep 17 00:00:00 2001
From: Hypolite Petovan <hypolite@mrpetovan.com>
Date: Sat, 15 Jun 2019 22:24:51 -0400
Subject: [PATCH] Move link conversion to the end of BBCode::convert
- Addresses https://github.com/friendica/friendica/issues/7271
- Add related tests
---
src/Content/Text/BBCode.php | 253 +++++++++++++-------------
tests/src/Content/Text/BBCodeTest.php | 43 +++++
2 files changed, 170 insertions(+), 126 deletions(-)
diff --git a/src/Content/Text/BBCode.php b/src/Content/Text/BBCode.php
index 146dedf470..f51d7f27b3 100644
--- a/src/Content/Text/BBCode.php
+++ b/src/Content/Text/BBCode.php
@@ -1289,123 +1289,8 @@ class BBCode extends BaseObject
// Handle attached links or videos
$text = self::convertAttachment($text, $simple_html, $try_oembed);
- // if the HTML is used to generate plain text, then don't do this search, but replace all URL of that kind to text
- if (!$for_plaintext) {
- $text = preg_replace(Strings::autoLinkRegEx(), '[url]$1[/url]', $text);
- if (in_array($simple_html, [7, 9])) {
- $text = preg_replace_callback("/\[url\]([$URLSearchString]*)\[\/url\]/ism", 'self::convertUrlForOStatusCallback', $text);
- $text = preg_replace_callback("/\[url\=([$URLSearchString]*)\]([$URLSearchString]*)\[\/url\]/ism", 'self::convertUrlForOStatusCallback', $text);
- }
- } else {
- $text = preg_replace("(\[url\]([$URLSearchString]*)\[\/url\])ism", " $1 ", $text);
- $text = preg_replace_callback("&\[url=([^\[\]]*)\]\[img\](.*)\[\/img\]\[\/url\]&Usi", 'self::removePictureLinksCallback', $text);
- }
-
- $text = str_replace(["\r","\n"], ['<br />', '<br />'], $text);
-
- // Remove all hashtag addresses
- if ((!$try_oembed || $simple_html) && !in_array($simple_html, [3, 7, 9])) {
- $text = preg_replace("/([#@!])\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '$1$3', $text);
- } elseif ($simple_html == 3) {
- // The ! is converted to @ since Diaspora only understands the @
- $text = preg_replace("/([@!])\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism",
- '@<a href="$2">$3</a>',
- $text);
- } elseif (in_array($simple_html, [7, 9])) {
- $text = preg_replace("/([@!])\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism",
- '$1<span class="vcard"><a href="$2" class="url u-url mention" title="$3"><span class="fn nickname mention">$3</span></a></span>',
- $text);
- } elseif (!$simple_html) {
- $text = preg_replace("/([@!])\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism",
- '$1<a href="$2" class="userinfo mention" title="$3">$3</a>',
- $text);
- }
-
- // Bookmarks in red - will be converted to bookmarks in friendica
- $text = preg_replace("/#\^\[url\]([$URLSearchString]*)\[\/url\]/ism", '[bookmark=$1]$1[/bookmark]', $text);
- $text = preg_replace("/#\^\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '[bookmark=$1]$2[/bookmark]', $text);
- $text = preg_replace("/#\[url\=[$URLSearchString]*\]\^\[\/url\]\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/i",
- "[bookmark=$1]$2[/bookmark]", $text);
-
- if (in_array($simple_html, [2, 6, 7, 8])) {
- $text = preg_replace_callback("/([^#@!])\[url\=([^\]]*)\](.*?)\[\/url\]/ism", "self::expandLinksCallback", $text);
- //$Text = preg_replace("/[^#@!]\[url\=([^\]]*)\](.*?)\[\/url\]/ism", ' $2 [url]$1[/url]', $Text);
- $text = preg_replace("/\[bookmark\=([^\]]*)\](.*?)\[\/bookmark\]/ism", ' $2 [url]$1[/url]',$text);
- }
-
- if ($simple_html == 5) {
- $text = preg_replace("/[^#@!]\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '[url]$1[/url]', $text);
- }
-
- // Perform URL Search
- if ($try_oembed) {
- $text = preg_replace_callback("/\[bookmark\=([^\]]*)\](.*?)\[\/bookmark\]/ism", $try_oembed_callback, $text);
- }
-
- if ($simple_html == 5) {
- $text = preg_replace("/\[bookmark\=([^\]]*)\](.*?)\[\/bookmark\]/ism", '[url]$1[/url]', $text);
- } else {
- $text = preg_replace("/\[bookmark\=([^\]]*)\](.*?)\[\/bookmark\]/ism", '[url=$1]$2[/url]', $text);
- }
-
- // Handle Diaspora posts
- $text = preg_replace_callback(
- "&\[url=/?posts/([^\[\]]*)\](.*)\[\/url\]&Usi",
- function ($match) {
- return "[url=" . System::baseUrl() . "/display/" . $match[1] . "]" . $match[2] . "[/url]";
- }, $text
- );
-
- $text = preg_replace_callback(
- "&\[url=/people\?q\=(.*)\](.*)\[\/url\]&Usi",
- function ($match) {
- return "[url=" . System::baseUrl() . "/search?search=%40" . $match[1] . "]" . $match[2] . "[/url]";
- }, $text
- );
-
- // Server independent link to posts and comments
- // See issue: https://github.com/diaspora/diaspora_federation/issues/75
- $expression = "=diaspora://.*?/post/([0-9A-Za-z\-_@.:]{15,254}[0-9A-Za-z])=ism";
- $text = preg_replace($expression, System::baseUrl()."/display/$1", $text);
-
- /* Tag conversion
- * Supports:
- * - #[url=<anything>]<term>[/url]
- * - [url=<anything>]#<term>[/url]
- */
- $text = preg_replace_callback("/(?:#\[url\=[$URLSearchString]*\]|\[url\=[$URLSearchString]*\]#)(.*?)\[\/url\]/ism", function($matches) {
- return '#<a href="'
- . System::baseUrl() . '/search?tag=' . rawurlencode($matches[1])
- . '" class="tag" title="' . XML::escape($matches[1]) . '">'
- . XML::escape($matches[1])
- . '</a>';
- }, $text);
-
- // We need no target="_blank" for local links
- // convert links start with System::baseUrl() as local link without the target="_blank" attribute
- $escapedBaseUrl = preg_quote(System::baseUrl(), '/');
- $text = preg_replace("/\[url\](".$escapedBaseUrl."[$URLSearchString]*)\[\/url\]/ism", '<a href="$1">$1</a>', $text);
- $text = preg_replace("/\[url\=(".$escapedBaseUrl."[$URLSearchString]*)\](.*?)\[\/url\]/ism", '<a href="$1">$2</a>', $text);
-
- $text = preg_replace("/\[url\]([$URLSearchString]*)\[\/url\]/ism", '<a href="$1" target="_blank">$1</a>', $text);
- $text = preg_replace("/\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '<a href="$1" target="_blank">$2</a>', $text);
-
- // Red compatibility, though the link can't be authenticated on Friendica
- $text = preg_replace("/\[zrl\=([$URLSearchString]*)\](.*?)\[\/zrl\]/ism", '<a href="$1" target="_blank">$2</a>', $text);
-
-
- // we may need to restrict this further if it picks up too many strays
- // link acct:user@host to a webfinger profile redirector
-
- $text = preg_replace('/acct:([^@]+)@((?!\-)(?:[a-zA-Z\d\-]{0,62}[a-zA-Z\d]\.){1,126}(?!\d+)[a-zA-Z\d]{1,63})/', '<a href="' . System::baseUrl() . '/acctlink?addr=$1@$2" target="extlink">acct:$1@$2</a>', $text);
-
- // Perform MAIL Search
- $text = preg_replace("/\[mail\]([$MAILSearchString]*)\[\/mail\]/", '<a href="mailto:$1">$1</a>', $text);
- $text = preg_replace("/\[mail\=([$MAILSearchString]*)\](.*?)\[\/mail\]/", '<a href="mailto:$1">$2</a>', $text);
-
// leave open the posibility of [map=something]
// this is replaced in Item::prepareBody() which has knowledge of the item location
-
if (strpos($text, '[/map]') !== false) {
$text = preg_replace_callback(
"/\[map\](.*?)\[\/map\]/ism",
@@ -1415,6 +1300,7 @@ class BBCode extends BaseObject
$text
);
}
+
if (strpos($text, '[map=') !== false) {
$text = preg_replace_callback(
"/\[map=(.*?)\]/ism",
@@ -1424,6 +1310,7 @@ class BBCode extends BaseObject
$text
);
}
+
if (strpos($text, '[map]') !== false) {
$text = preg_replace("/\[map\]/", '<p class="map"></p>', $text);
}
@@ -1490,9 +1377,9 @@ class BBCode extends BaseObject
$endlessloop = 0;
while ((((strpos($text, "[/list]") !== false) && (strpos($text, "[list") !== false)) ||
- ((strpos($text, "[/ol]") !== false) && (strpos($text, "[ol]") !== false)) ||
- ((strpos($text, "[/ul]") !== false) && (strpos($text, "[ul]") !== false)) ||
- ((strpos($text, "[/li]") !== false) && (strpos($text, "[li]") !== false))) && (++$endlessloop < 20)) {
+ ((strpos($text, "[/ol]") !== false) && (strpos($text, "[ol]") !== false)) ||
+ ((strpos($text, "[/ul]") !== false) && (strpos($text, "[ul]") !== false)) ||
+ ((strpos($text, "[/li]") !== false) && (strpos($text, "[li]") !== false))) && (++$endlessloop < 20)) {
$text = preg_replace("/\[list\](.*?)\[\/list\]/ism", '<ul class="listbullet" style="list-style-type: circle;">$1</ul>', $text);
$text = preg_replace("/\[list=\](.*?)\[\/list\]/ism", '<ul class="listnone" style="list-style-type: none;">$1</ul>', $text);
$text = preg_replace("/\[list=1\](.*?)\[\/list\]/ism", '<ul class="listdecimal" style="list-style-type: decimal;">$1</ul>', $text);
@@ -1540,8 +1427,8 @@ class BBCode extends BaseObject
$endlessloop = 0;
while ((strpos($text, "[/spoiler]")!== false) && (strpos($text, "[spoiler=") !== false) && (++$endlessloop < 20)) {
$text = preg_replace("/\[spoiler=[\"\']*(.*?)[\"\']*\](.*?)\[\/spoiler\]/ism",
- "<br /><strong class=".'"spoiler"'.">" . $t_wrote . "</strong><blockquote class=".'"spoiler"'.">$2</blockquote>",
- $text);
+ "<br /><strong class=".'"spoiler"'.">" . $t_wrote . "</strong><blockquote class=".'"spoiler"'.">$2</blockquote>",
+ $text);
}
// Declare the format for [quote] layout
@@ -1562,8 +1449,8 @@ class BBCode extends BaseObject
$endlessloop = 0;
while ((strpos($text, "[/quote]")!== false) && (strpos($text, "[quote=") !== false) && (++$endlessloop < 20)) {
$text = preg_replace("/\[quote=[\"\']*(.*?)[\"\']*\](.*?)\[\/quote\]/ism",
- "<p><strong class=".'"author"'.">" . $t_wrote . "</strong></p><blockquote>$2</blockquote>",
- $text);
+ "<p><strong class=".'"author"'.">" . $t_wrote . "</strong></p><blockquote>$2</blockquote>",
+ $text);
}
@@ -1631,9 +1518,9 @@ class BBCode extends BaseObject
$text = preg_replace_callback("/\[audio\](.*?)\[\/audio\]/ism", $try_oembed_callback, $text);
} else {
$text = preg_replace("/\[video\](.*?)\[\/video\]/ism",
- '<a href="$1" target="_blank">$1</a>', $text);
+ '<a href="$1" target="_blank">$1</a>', $text);
$text = preg_replace("/\[audio\](.*?)\[\/audio\]/ism",
- '<a href="$1" target="_blank">$1</a>', $text);
+ '<a href="$1" target="_blank">$1</a>', $text);
}
// html5 video and audio
@@ -1660,7 +1547,7 @@ class BBCode extends BaseObject
$text = preg_replace("/\[youtube\]([A-Za-z0-9\-_=]+)(.*?)\[\/youtube\]/ism", '<iframe width="' . $a->videowidth . '" height="' . $a->videoheight . '" src="https://www.youtube.com/embed/$1" frameborder="0" ></iframe>', $text);
} else {
$text = preg_replace("/\[youtube\]([A-Za-z0-9\-_=]+)(.*?)\[\/youtube\]/ism",
- '<a href="https://www.youtube.com/watch?v=$1" target="_blank">https://www.youtube.com/watch?v=$1</a>', $text);
+ '<a href="https://www.youtube.com/watch?v=$1" target="_blank">https://www.youtube.com/watch?v=$1</a>', $text);
}
if ($try_oembed) {
@@ -1675,7 +1562,7 @@ class BBCode extends BaseObject
$text = preg_replace("/\[vimeo\]([0-9]+)(.*?)\[\/vimeo\]/ism", '<iframe width="' . $a->videowidth . '" height="' . $a->videoheight . '" src="https://player.vimeo.com/video/$1" frameborder="0" ></iframe>', $text);
} else {
$text = preg_replace("/\[vimeo\]([0-9]+)(.*?)\[\/vimeo\]/ism",
- '<a href="https://vimeo.com/$1" target="_blank">https://vimeo.com/$1</a>', $text);
+ '<a href="https://vimeo.com/$1" target="_blank">https://vimeo.com/$1</a>', $text);
}
// oembed tag
@@ -1706,6 +1593,120 @@ class BBCode extends BaseObject
$text = Smilies::replace($text);
}
+ // if the HTML is used to generate plain text, then don't do this search, but replace all URL of that kind to text
+ if (!$for_plaintext) {
+ $text = preg_replace(Strings::autoLinkRegEx(), '[url]$1[/url]', $text);
+ if (in_array($simple_html, [7, 9])) {
+ $text = preg_replace_callback("/\[url\]([$URLSearchString]*)\[\/url\]/ism", 'self::convertUrlForOStatusCallback', $text);
+ $text = preg_replace_callback("/\[url\=([$URLSearchString]*)\]([$URLSearchString]*)\[\/url\]/ism", 'self::convertUrlForOStatusCallback', $text);
+ }
+ } else {
+ $text = preg_replace("(\[url\]([$URLSearchString]*)\[\/url\])ism", " $1 ", $text);
+ $text = preg_replace_callback("&\[url=([^\[\]]*)\]\[img\](.*)\[\/img\]\[\/url\]&Usi", 'self::removePictureLinksCallback', $text);
+ }
+
+ $text = str_replace(["\r","\n"], ['<br />', '<br />'], $text);
+
+ // Remove all hashtag addresses
+ if ((!$try_oembed || $simple_html) && !in_array($simple_html, [3, 7, 9])) {
+ $text = preg_replace("/([#@!])\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '$1$3', $text);
+ } elseif ($simple_html == 3) {
+ // The ! is converted to @ since Diaspora only understands the @
+ $text = preg_replace("/([@!])\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism",
+ '@<a href="$2">$3</a>',
+ $text);
+ } elseif (in_array($simple_html, [7, 9])) {
+ $text = preg_replace("/([@!])\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism",
+ '$1<span class="vcard"><a href="$2" class="url u-url mention" title="$3"><span class="fn nickname mention">$3</span></a></span>',
+ $text);
+ } elseif (!$simple_html) {
+ $text = preg_replace("/([@!])\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism",
+ '$1<a href="$2" class="userinfo mention" title="$3">$3</a>',
+ $text);
+ }
+
+ // Bookmarks in red - will be converted to bookmarks in friendica
+ $text = preg_replace("/#\^\[url\]([$URLSearchString]*)\[\/url\]/ism", '[bookmark=$1]$1[/bookmark]', $text);
+ $text = preg_replace("/#\^\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '[bookmark=$1]$2[/bookmark]', $text);
+ $text = preg_replace("/#\[url\=[$URLSearchString]*\]\^\[\/url\]\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/i",
+ "[bookmark=$1]$2[/bookmark]", $text);
+
+ if (in_array($simple_html, [2, 6, 7, 8])) {
+ $text = preg_replace_callback("/([^#@!])\[url\=([^\]]*)\](.*?)\[\/url\]/ism", "self::expandLinksCallback", $text);
+ //$Text = preg_replace("/[^#@!]\[url\=([^\]]*)\](.*?)\[\/url\]/ism", ' $2 [url]$1[/url]', $Text);
+ $text = preg_replace("/\[bookmark\=([^\]]*)\](.*?)\[\/bookmark\]/ism", ' $2 [url]$1[/url]',$text);
+ }
+
+ if ($simple_html == 5) {
+ $text = preg_replace("/[^#@!]\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '[url]$1[/url]', $text);
+ }
+
+ // Perform URL Search
+ if ($try_oembed) {
+ $text = preg_replace_callback("/\[bookmark\=([^\]]*)\](.*?)\[\/bookmark\]/ism", $try_oembed_callback, $text);
+ }
+
+ if ($simple_html == 5) {
+ $text = preg_replace("/\[bookmark\=([^\]]*)\](.*?)\[\/bookmark\]/ism", '[url]$1[/url]', $text);
+ } else {
+ $text = preg_replace("/\[bookmark\=([^\]]*)\](.*?)\[\/bookmark\]/ism", '[url=$1]$2[/url]', $text);
+ }
+
+ // Handle Diaspora posts
+ $text = preg_replace_callback(
+ "&\[url=/?posts/([^\[\]]*)\](.*)\[\/url\]&Usi",
+ function ($match) {
+ return "[url=" . System::baseUrl() . "/display/" . $match[1] . "]" . $match[2] . "[/url]";
+ }, $text
+ );
+
+ $text = preg_replace_callback(
+ "&\[url=/people\?q\=(.*)\](.*)\[\/url\]&Usi",
+ function ($match) {
+ return "[url=" . System::baseUrl() . "/search?search=%40" . $match[1] . "]" . $match[2] . "[/url]";
+ }, $text
+ );
+
+ // Server independent link to posts and comments
+ // See issue: https://github.com/diaspora/diaspora_federation/issues/75
+ $expression = "=diaspora://.*?/post/([0-9A-Za-z\-_@.:]{15,254}[0-9A-Za-z])=ism";
+ $text = preg_replace($expression, System::baseUrl()."/display/$1", $text);
+
+ /* Tag conversion
+ * Supports:
+ * - #[url=<anything>]<term>[/url]
+ * - [url=<anything>]#<term>[/url]
+ */
+ $text = preg_replace_callback("/(?:#\[url\=[$URLSearchString]*\]|\[url\=[$URLSearchString]*\]#)(.*?)\[\/url\]/ism", function($matches) {
+ return '#<a href="'
+ . System::baseUrl() . '/search?tag=' . rawurlencode($matches[1])
+ . '" class="tag" title="' . XML::escape($matches[1]) . '">'
+ . XML::escape($matches[1])
+ . '</a>';
+ }, $text);
+
+ // We need no target="_blank" for local links
+ // convert links start with System::baseUrl() as local link without the target="_blank" attribute
+ $escapedBaseUrl = preg_quote(System::baseUrl(), '/');
+ $text = preg_replace("/\[url\](".$escapedBaseUrl."[$URLSearchString]*)\[\/url\]/ism", '<a href="$1">$1</a>', $text);
+ $text = preg_replace("/\[url\=(".$escapedBaseUrl."[$URLSearchString]*)\](.*?)\[\/url\]/ism", '<a href="$1">$2</a>', $text);
+
+ $text = preg_replace("/\[url\]([$URLSearchString]*)\[\/url\]/ism", '<a href="$1" target="_blank">$1</a>', $text);
+ $text = preg_replace("/\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism", '<a href="$1" target="_blank">$2</a>', $text);
+
+ // Red compatibility, though the link can't be authenticated on Friendica
+ $text = preg_replace("/\[zrl\=([$URLSearchString]*)\](.*?)\[\/zrl\]/ism", '<a href="$1" target="_blank">$2</a>', $text);
+
+
+ // we may need to restrict this further if it picks up too many strays
+ // link acct:user@host to a webfinger profile redirector
+
+ $text = preg_replace('/acct:([^@]+)@((?!\-)(?:[a-zA-Z\d\-]{0,62}[a-zA-Z\d]\.){1,126}(?!\d+)[a-zA-Z\d]{1,63})/', '<a href="' . System::baseUrl() . '/acctlink?addr=$1@$2" target="extlink">acct:$1@$2</a>', $text);
+
+ // Perform MAIL Search
+ $text = preg_replace("/\[mail\]([$MAILSearchString]*)\[\/mail\]/", '<a href="mailto:$1">$1</a>', $text);
+ $text = preg_replace("/\[mail\=([$MAILSearchString]*)\](.*?)\[\/mail\]/", '<a href="mailto:$1">$2</a>', $text);
+
// Unhide all [noparse] contained bbtags unspacefying them
// and triming the [noparse] tag.
diff --git a/tests/src/Content/Text/BBCodeTest.php b/tests/src/Content/Text/BBCodeTest.php
index 864a3794d3..f74aebaab8 100644
--- a/tests/src/Content/Text/BBCodeTest.php
+++ b/tests/src/Content/Text/BBCodeTest.php
@@ -134,4 +134,47 @@ class BBCodeTest extends MockedTest
$this->assertNotEquals($assert, $output);
}
}
+
+ public function dataBBCodes()
+ {
+ return [
+ 'bug-7271-condensed-space' => [
+ 'expectedHtml' => '<ul class="listdecimal" style="list-style-type: decimal;"><li> <a href="http://example.com/" target="_blank">http://example.com/</a></li></ul>',
+ 'text' => '[ol][*] http://example.com/[/ol]',
+ ],
+ 'bug-7271-condensed-nospace' => [
+ 'expectedHtml' => '<ul class="listdecimal" style="list-style-type: decimal;"><li><a href="http://example.com/" target="_blank">http://example.com/</a></li></ul>',
+ 'text' => '[ol][*]http://example.com/[/ol]',
+ ],
+ 'bug-7271-indented-space' => [
+ 'expectedHtml' => '<ul class="listbullet" style="list-style-type: circle;"><li> <a href="http://example.com/" target="_blank">http://example.com/</a></li></ul>',
+ 'text' => '[ul]
+[*] http://example.com/
+[/ul]',
+ ],
+ 'bug-7271-indented-nospace' => [
+ 'expectedHtml' => '<ul class="listbullet" style="list-style-type: circle;"><li><a href="http://example.com/" target="_blank">http://example.com/</a></li></ul>',
+ 'text' => '[ul]
+[*]http://example.com/
+[/ul]',
+ ],
+ ];
+ }
+
+ /**
+ * Test convert bbcodes to HTML
+ * @dataProvider dataBBCodes
+ *
+ * @param string $expectedHtml Expected HTML output
+ * @param string $text BBCode text
+ * @param int $simpleHtml BBCode::convert method $simple_html parameter value, optional.
+ * @param bool $forPlaintext BBCode::convert method $for_plaintext parameter value, optional.
+ * @throws \Friendica\Network\HTTPException\InternalServerErrorException
+ */
+ public function testConvert($expectedHtml, $text, $simpleHtml = 0, $forPlaintext = false)
+ {
+ $actual = BBCode::convert($text, false, $simpleHtml, $forPlaintext);
+
+ $this->assertEquals($expectedHtml, $actual);
+ }
}
--
2.39.5