diff --git a/src/markup/engine/__tests__/PhutilRemarkupEngineTestCase.php b/src/markup/engine/__tests__/PhutilRemarkupEngineTestCase.php
index 8f53ecc..74932d9 100644
--- a/src/markup/engine/__tests__/PhutilRemarkupEngineTestCase.php
+++ b/src/markup/engine/__tests__/PhutilRemarkupEngineTestCase.php
@@ -1,115 +1,121 @@
markupText($root.$file);
}
}
private function markupText($markup_file) {
$contents = Filesystem::readFile($markup_file);
$file = basename($markup_file);
$parts = explode("\n~~~~~~~~~~\n", $contents);
$this->assertEqual(3, count($parts), $markup_file);
list($input_remarkup, $expected_output, $expected_text) = $parts;
$engine = $this->buildNewTestEngine();
switch ($file) {
case 'raw-escape.txt':
// NOTE: Here, we want to test PhutilRemarkupEscapeRemarkupRule and
// PhutilRemarkupBlockStorage, which are triggered by "\1". In the
// test, "~" is used as a placeholder for "\1" since it's hard to type
// "\1".
$input_remarkup = str_replace('~', "\1", $input_remarkup);
$expected_output = str_replace('~', "\1", $expected_output);
$expected_text = str_replace('~', "\1", $expected_text);
break;
case 'toc.txt':
$engine->setConfig('header.generate-toc', true);
break;
+ case 'link-same-window.txt':
+ $engine->setConfig('uri.same-window', true);
+ break;
+ case 'link-square.txt':
+ $engine->setConfig('uri.base', 'http://www.example.com/');
+ $engine->setConfig('uri.here', 'http://www.example.com/page/');
+ break;
}
$actual_output = (string)$engine->markupText($input_remarkup);
switch ($file) {
case 'toc.txt':
$table_of_contents =
PhutilRemarkupHeaderBlockRule::renderTableOfContents($engine);
$actual_output = $table_of_contents."\n\n".$actual_output;
break;
}
$this->assertEqual(
$expected_output,
$actual_output,
pht("Failed to markup HTML in file '%s'.", $file));
$engine->setMode(PhutilRemarkupEngine::MODE_TEXT);
$actual_output = (string)$engine->markupText($input_remarkup);
$this->assertEqual(
$expected_text,
$actual_output,
pht("Failed to markup text in file '%s'.", $file));
}
private function buildNewTestEngine() {
$engine = new PhutilRemarkupEngine();
- $engine->setConfig('uri.prefix', 'http://www.example.com/');
$engine->setConfig(
'uri.allowed-protocols',
array(
'http' => true,
'mailto' => true,
'tel' => true,
));
$rules = array();
$rules[] = new PhutilRemarkupEscapeRemarkupRule();
$rules[] = new PhutilRemarkupMonospaceRule();
$rules[] = new PhutilRemarkupDocumentLinkRule();
$rules[] = new PhutilRemarkupHyperlinkRule();
$rules[] = new PhutilRemarkupBoldRule();
$rules[] = new PhutilRemarkupItalicRule();
$rules[] = new PhutilRemarkupDelRule();
$rules[] = new PhutilRemarkupUnderlineRule();
$rules[] = new PhutilRemarkupHighlightRule();
$blocks = array();
$blocks[] = new PhutilRemarkupQuotesBlockRule();
$blocks[] = new PhutilRemarkupReplyBlockRule();
$blocks[] = new PhutilRemarkupHeaderBlockRule();
$blocks[] = new PhutilRemarkupHorizontalRuleBlockRule();
$blocks[] = new PhutilRemarkupCodeBlockRule();
$blocks[] = new PhutilRemarkupLiteralBlockRule();
$blocks[] = new PhutilRemarkupNoteBlockRule();
$blocks[] = new PhutilRemarkupTableBlockRule();
$blocks[] = new PhutilRemarkupSimpleTableBlockRule();
$blocks[] = new PhutilRemarkupDefaultBlockRule();
$blocks[] = new PhutilRemarkupListBlockRule();
$blocks[] = new PhutilRemarkupInterpreterBlockRule();
foreach ($blocks as $block) {
if (!($block instanceof PhutilRemarkupCodeBlockRule)) {
$block->setMarkupRules($rules);
}
}
$engine->setBlockRules($blocks);
return $engine;
}
}
diff --git a/src/markup/engine/__tests__/remarkup/link-same-window.txt b/src/markup/engine/__tests__/remarkup/link-same-window.txt
new file mode 100644
index 0000000..937c83f
--- /dev/null
+++ b/src/markup/engine/__tests__/remarkup/link-same-window.txt
@@ -0,0 +1,11 @@
+[[http://www.example.com/]]
+
+http://www.example.com/
+~~~~~~~~~~
+
+
+
+~~~~~~~~~~
+http://www.example.com/
+
+http://www.example.com/
diff --git a/src/markup/engine/__tests__/remarkup/link-square.txt b/src/markup/engine/__tests__/remarkup/link-square.txt
index 91d85f3..2b24f3a 100644
--- a/src/markup/engine/__tests__/remarkup/link-square.txt
+++ b/src/markup/engine/__tests__/remarkup/link-square.txt
@@ -1,29 +1,29 @@
[[http://www.example.com/]]
[[http://www.example.com/ | example.com]]
[[/]]
[[#anchor]]
[[#anchor | Anchors ]]
~~~~~~~~~~
-
+
-
+
-
+
~~~~~~~~~~
http://www.example.com/
example.com
http://www.example.com/
-http://www.example.com/#anchor
+http://www.example.com/page/#anchor
-Anchors
+Anchors
diff --git a/src/markup/engine/remarkup/markuprule/PhutilRemarkupDocumentLinkRule.php b/src/markup/engine/remarkup/markuprule/PhutilRemarkupDocumentLinkRule.php
index ed85d72..f632b98 100644
--- a/src/markup/engine/remarkup/markuprule/PhutilRemarkupDocumentLinkRule.php
+++ b/src/markup/engine/remarkup/markuprule/PhutilRemarkupDocumentLinkRule.php
@@ -1,132 +1,137 @@
getEngine()->isTextMode()) {
- $text = $link;
- if (strncmp($link, '/', 1) == 0 || strncmp($link, '#', 1) == 0) {
- $base = $this->getEngine()->getConfig('uri.prefix');
- if (strncmp($link, '/', 1) == 0) {
- $base = rtrim($base, '/');
- }
- $text = $base.$text;
- }
+ $engine = $this->getEngine();
+
+ $is_anchor = false;
+ if (strncmp($link, '/', 1) == 0) {
+ $base = $engine->getConfig('uri.base');
+ $base = rtrim($base, '/');
+ $link = $base.$link;
+ } else if (strncmp($link, '#', 1) == 0) {
+ $here = $engine->getConfig('uri.here');
+ $link = $here.$link;
+
+ $is_anchor = true;
+ }
+ if ($engine->isTextMode()) {
// If present, strip off "mailto:" or "tel:".
- $text = preg_replace('/^(?:mailto|tel):/', '', $text);
+ $link = preg_replace('/^(?:mailto|tel):/', '', $link);
- if ($link == $name) {
- return $text;
+ if (!strlen($name)) {
+ return $link;
}
- return $name.' <'.$text.'>';
- } else if ($this->getEngine()->isHTMLMailMode()) {
- if (strncmp($link, '/', 1) == 0 || strncmp($link, '#', 1) == 0) {
- $base = $this->getEngine()->getConfig('uri.base');
- $text = $link;
- if (strncmp($link, '/', 1) == 0) {
- $base = rtrim($base, '/');
- }
- $link = $base.$text;
- }
- }
- // By default, we open links in a new window or tab. For anchors on the same
- // page, just jump normally.
- $target = '_blank';
- if (strncmp($link, '#', 1) == 0) {
- $target = null;
+ return $name.' <'.$link.'>';
}
- $name = preg_replace('/^(?:mailto|tel):/', '', $name);
+ if (!strlen($name)) {
+ $name = $link;
+ $name = preg_replace('/^(?:mailto|tel):/', '', $name);
+ }
- if ($this->getEngine()->getState('toc')) {
+ if ($engine->getState('toc')) {
return $name;
+ }
+
+ $same_window = $engine->getConfig('uri.same-window', false);
+ if ($same_window) {
+ $target = null;
} else {
- return phutil_tag(
- 'a',
- array(
- 'href' => $link,
- 'class' => 'remarkup-link',
- 'target' => $target,
- ),
- $name);
+ $target = '_blank';
}
+
+ // For anchors on the same page, always stay here.
+ if ($is_anchor) {
+ $target = null;
+ }
+
+ return phutil_tag(
+ 'a',
+ array(
+ 'href' => $link,
+ 'class' => 'remarkup-link',
+ 'target' => $target,
+ ),
+ $name);
}
public function markupAlternateLink(array $matches) {
$uri = trim($matches[2]);
// NOTE: We apply some special rules to avoid false positives here. The
// major concern is that we do not want to convert `x[0][1](y)` in a
// discussion about C source code into a link. To this end, we:
//
// - Don't match at word boundaries;
// - require the URI to contain a "/" character or "@" character; and
// - reject URIs which being with a quote character.
if ($uri[0] == '"' || $uri[0] == "'" || $uri[0] == '`') {
return $matches[0];
}
if (strpos($uri, '/') === false &&
strpos($uri, '@') === false &&
strncmp($uri, 'tel:', 4)) {
return $matches[0];
}
return $this->markupDocumentLink(
array(
$matches[0],
$matches[2],
$matches[1],
));
}
public function markupDocumentLink(array $matches) {
$uri = trim($matches[1]);
- $name = trim(idx($matches, 2, $uri));
+ $name = trim(idx($matches, 2));
// If whatever is being linked to begins with "/" or "#", or has "://",
// or is "mailto:" or "tel:", treat it as a URI instead of a wiki page.
$is_uri = preg_match('@(^/)|(://)|(^#)|(^(?:mailto|tel):)@', $uri);
if ($is_uri && strncmp('/', $uri, 1) && strncmp('#', $uri, 1)) {
$protocols = $this->getEngine()->getConfig(
'uri.allowed-protocols',
array());
$protocol = id(new PhutilURI($uri))->getProtocol();
if (!idx($protocols, $protocol)) {
// Don't treat this as a URI if it's not an allowed protocol.
$is_uri = false;
}
}
if (!$is_uri) {
return $matches[0];
}
return $this->getEngine()->storeText($this->renderHyperlink($uri, $name));
}
}
diff --git a/src/markup/engine/remarkup/markuprule/PhutilRemarkupHyperlinkRule.php b/src/markup/engine/remarkup/markuprule/PhutilRemarkupHyperlinkRule.php
index b9c51e5..6147dba 100644
--- a/src/markup/engine/remarkup/markuprule/PhutilRemarkupHyperlinkRule.php
+++ b/src/markup/engine/remarkup/markuprule/PhutilRemarkupHyperlinkRule.php
@@ -1,99 +1,108 @@
" around them get linked exactly, without
// the "<>". Angle brackets are basically special and mean "this is a URL
// with weird characters". This is assumed to be reasonable because they
// don't appear in normal text or normal URLs.
$text = preg_replace_callback(
'@<(\w{3,}://[^\s'.PhutilRemarkupBlockStorage::MAGIC_BYTE.']+?)>@',
array($this, 'markupHyperlink'),
$text);
// Anything else we match "ungreedily", which means we'll look for
// stuff that's probably puncutation or otherwise not part of the URL and
// not link it. This lets someone write "QuicK! Go to
// http://www.example.com/!". We also apply some paren balancing rules.
// NOTE: We're explicitly avoiding capturing stored blocks, so text like
// `http://www.example.com/[[x | y]]` doesn't get aggressively captured.
$text = preg_replace_callback(
'@(\w{3,}://[^\s'.PhutilRemarkupBlockStorage::MAGIC_BYTE.']+)@',
array($this, 'markupHyperlinkUngreedy'),
$text);
return $text;
}
protected function markupHyperlink(array $matches) {
$protocols = $this->getEngine()->getConfig(
'uri.allowed-protocols',
array());
$protocol = id(new PhutilURI($matches[1]))->getProtocol();
if (!idx($protocols, $protocol)) {
// If this URI doesn't use a whitelisted protocol, don't link it. This
// is primarily intended to prevent javascript:// silliness.
return $this->getEngine()->storeText($matches[1]);
}
return $this->storeRenderedHyperlink($matches[1]);
}
protected function storeRenderedHyperlink($link) {
return $this->getEngine()->storeText($this->renderHyperlink($link));
}
protected function renderHyperlink($link) {
- if ($this->getEngine()->isTextMode()) {
+ $engine = $this->getEngine();
+
+ if ($engine->isTextMode()) {
return $link;
}
- if ($this->getEngine()->getState('toc')) {
+ if ($engine->getState('toc')) {
return $link;
+ }
+
+ $same_window = $engine->getConfig('uri.same-window', false);
+ if ($same_window) {
+ $target = null;
} else {
- return phutil_tag(
- 'a',
- array(
- 'href' => $link,
- 'class' => 'remarkup-link',
- 'target' => '_blank',
- ),
- $link);
+ $target = '_blank';
}
+
+ return phutil_tag(
+ 'a',
+ array(
+ 'href' => $link,
+ 'class' => 'remarkup-link',
+ 'target' => $target,
+ ),
+ $link);
}
protected function markupHyperlinkUngreedy($matches) {
$match = $matches[1];
$tail = null;
$trailing = null;
if (preg_match('/[;,.:!?]+$/', $match, $trailing)) {
$tail = $trailing[0];
$match = substr($match, 0, -strlen($tail));
}
// If there's a closing paren at the end but no balancing open paren in
// the URL, don't link the close paren. This is an attempt to gracefully
// handle the two common paren cases, Wikipedia links and English language
// parentheticals, e.g.:
//
// http://en.wikipedia.org/wiki/Noun_(disambiguation)
// (see also http://www.example.com)
//
// We could apply a craftier heuristic here which tries to actually balance
// the parens, but this is probably sufficient.
if (preg_match('/\\)$/', $match) && !preg_match('/\\(/', $match)) {
$tail = ')'.$tail;
$match = substr($match, 0, -1);
}
return hsprintf('%s%s', $this->markupHyperlink(array(null, $match)), $tail);
}
}