redcarpet 1.5.3 → 1.6.0

data/ext/markdown.c

@@ -100,8 +100,27 @@ static struct html_tag block_tags[] = {
 #define DEL_TAG (block_tags + 10)
 
 /***************************
- * STATIC HELPER FUNCTIONS *
+ * HELPER FUNCTIONS *
  ***************************/
+int
+is_safe_link(const char *link, size_t link_len)
+{
+	static const size_t valid_uris_count = 4;
+	static const char *valid_uris[] = {
+		"http://", "https://", "ftp://", "mailto://"
+	};
+
+	size_t i;
+
+	for (i = 0; i < valid_uris_count; ++i) {
+		size_t len = strlen(valid_uris[i]);
+
+		if (link_len > len && memcmp(link, valid_uris[i], len) == 0)
+			return 1;
+	}
+
+	return 0;
+}
 
 /* cmp_link_ref • comparison function for link_ref sorted arrays */
 static int
@@ -245,8 +264,9 @@ parse_inline(struct buf *ob, struct render *rndr, char *data, size_t size)
 
 	while (i < size) {
 		/* copying inactive chars into the output */
-		while (end < size && (action = rndr->active_char[(unsigned char)data[end]]) == 0)
+		while (end < size && (action = rndr->active_char[(unsigned char)data[end]]) == 0) {
 			end++;
+		}
 
 		if (rndr->make.normal_text) {
 			work.data = data + i;
@@ -616,6 +636,25 @@ char_langle_tag(struct buf *ob, struct render *rndr, char *data, size_t offset,
 	else return end;
 }
 
+static size_t
+char_autolink(struct buf *ob, struct render *rndr, char *data, size_t offset, size_t size)
+{
+	struct buf work = { data, 0, 0, 0, 0 };
+
+	if (offset > 0 && !isspace(data[-1]))
+		return 0;
+
+	if (!is_safe_link(data, size))
+		return 0;
+
+	while (work.size < size && !isspace(data[work.size]))
+		work.size++;
+
+	if (rndr->make.autolink)
+		rndr->make.autolink(ob, &work, MKDA_NORMAL, rndr->make.opaque);
+
+	return work.size;
+}
 
 /* char_link • '[': parsing a link or an image */
 static size_t
@@ -2029,6 +2068,12 @@ markdown(struct buf *ob, struct buf *ib, const struct mkd_renderer *rndrer, unsi
 	rndr.active_char['\\'] = char_escape;
 	rndr.active_char['&'] = char_entity;
 
+	if (extensions & MKDEXT_AUTOLINK) {
+		rndr.active_char['h'] = char_autolink; // http, https
+		rndr.active_char['f'] = char_autolink; // ftp
+		rndr.active_char['m'] = char_autolink; // mailto
+	}
+
 	/* Extension data */
 	rndr.ext_flags = extensions;
 	rndr.max_nesting = 16;

data/ext/markdown.h

@@ -40,6 +40,7 @@ enum mkd_extensions {
 	MKDEXT_LAX_EMPHASIS = (1 << 0),
 	MKDEXT_TABLES = (1 << 1),
 	MKDEXT_FENCED_CODE = (1 << 2),
+	MKDEXT_AUTOLINK = (1 << 3),
 };
 
 /* mkd_renderer • functions for rendering parsed data */
@@ -94,6 +95,12 @@ struct mkd_renderer {
 #define MKD_TABLE_ALIGN_R (1 << 1)
 #define MKD_TABLE_ALIGN_CENTER (MKD_TABLE_ALIGN_L | MKD_TABLE_ALIGN_R)
 
+/*******************
+ * Auxiliar methods
+ *******************/
+int
+is_safe_link(const char *link, size_t link_len);
+
 /**********************
  * EXPORTED FUNCTIONS *
  **********************/

data/ext/redcarpet.c

@@ -39,10 +39,6 @@ static void rb_redcarpet__get_flags(VALUE ruby_obj,
 	if (rb_funcall(ruby_obj, rb_intern("filter_styles"), 0) == Qtrue)
 		render_flags |= XHTML_SKIP_STYLE;
 
-	/* autolink */
-	if (rb_funcall(ruby_obj, rb_intern("autolink"), 0) == Qtrue)
-		render_flags |= XHTML_AUTOLINK;
-
 	/* safelink */
 	if (rb_funcall(ruby_obj, rb_intern("safelink"), 0) == Qtrue)
 		render_flags |= XHTML_SAFELINK;
@@ -62,6 +58,9 @@ static void rb_redcarpet__get_flags(VALUE ruby_obj,
 	if (rb_funcall(ruby_obj, rb_intern("fenced_code"), 0) == Qtrue)
 		extensions |= MKDEXT_FENCED_CODE;
 
+	if (rb_funcall(ruby_obj, rb_intern("autolink"), 0) == Qtrue)
+		extensions |= MKDEXT_AUTOLINK;
+
 	if (rb_funcall(ruby_obj, rb_intern("strikethrough"), 0) == Qtrue)
 		render_flags |= XHTML_STRIKETHROUGH;
 

data/ext/xhtml.c

@@ -31,26 +31,6 @@ struct xhtml_renderopt {
 	unsigned int flags;
 };
 
-static int
-is_safe_link(const char *link, size_t link_len)
-{
-	static const size_t valid_uris_count = 4;
-	static const char *valid_uris[] = {
-		"http:", "https:", "ftp:", "mailto:"
-	};
-
-	size_t i;
-
-	for (i = 0; i < valid_uris_count; ++i) {
-		size_t len = strlen(valid_uris[i]);
-
-		if (link_len > len && memcmp(link, valid_uris[i], len) == 0)
-			return 1;
-	}
-
-	return 0;
-}
-
 static inline int
 put_scaped_char(struct buf *ob, char c)
 {
@@ -103,20 +83,6 @@ is_html_tag(struct buf *tag, const char *tagname)
 /********************
  * GENERIC RENDERER *
  ********************/
-
-static void
-rndr_autolink2(struct buf *ob, const char *link, size_t link_size, enum mkd_autolink type)
-{
-	BUFPUTSL(ob, "<a href=\"");
-	if (type == MKDA_IMPLICIT_EMAIL) BUFPUTSL(ob, "mailto:");
-	lus_attr_escape(ob, link, link_size);
-	BUFPUTSL(ob, "\">");
-	if (type == MKDA_EXPLICIT_EMAIL && link_size > 7)
-		lus_attr_escape(ob, link + 7, link_size - 7);
-	else	lus_attr_escape(ob, link, link_size);
-	BUFPUTSL(ob, "</a>");
-}
-
 static int
 rndr_autolink(struct buf *ob, struct buf *link, enum mkd_autolink type, void *opaque)
 {
@@ -128,7 +94,19 @@ rndr_autolink(struct buf *ob, struct buf *link, enum mkd_autolink type, void *op
 	if ((options->flags & XHTML_SAFELINK) != 0 && !is_safe_link(link->data, link->size))
 		return 0;
 
-	rndr_autolink2(ob, link->data, link->size, type);
+	BUFPUTSL(ob, "<a href=\"");
+	if (type == MKDA_IMPLICIT_EMAIL)
+		BUFPUTSL(ob, "mailto:");
+	bufput(ob, link->data, link->size);
+	BUFPUTSL(ob, "\">");
+
+	if (type == MKDA_EXPLICIT_EMAIL && link->size > 7)
+		lus_attr_escape(ob, link->data + 7, link->size - 7);
+	else
+		lus_attr_escape(ob, link->data, link->size);
+
+	BUFPUTSL(ob, "</a>");
+
 	return 1;
 }
 
@@ -492,140 +470,114 @@ smartypants_quotes(struct buf *ob, struct buf *text, size_t i, int is_open)
 
 static void
 rndr_normal_text(struct buf *ob, struct buf *text, void *opaque)
+{
+	if (text)
+		lus_attr_escape(ob, text->data, text->size);
+}
+
+static void
+rndr_smartypants(struct buf *ob, struct buf *text, void *opaque)
 {
 	size_t i;
 	int open_single = 0, open_double = 0, open_tag = 0;
-	struct xhtml_renderopt *options = opaque;	
-
-	int autolink = (options->flags & XHTML_AUTOLINK);
-	int smartypants = (options->flags & XHTML_SMARTYPANTS);
 
 	if (!text)
 		return;
 
-	if (!autolink && !smartypants) {
-		lus_attr_escape(ob, text->data, text->size);
-		return;
-	}
-
 	for (i = 0; i < text->size; ++i) {
 		size_t sub;
 		char c = text->data[i];
 
-		/*
-		 * Autolinking
-		 */
-		if (autolink) {
-			/* Autolinking is not standarized in the Markdown spec.
-			 * We only check for links immediately after a space  */
-			if ((i == 0 || isspace(text->data[i - 1])) &&
-				is_safe_link(text->data + i, text->size - i)) {
-				size_t j = i;
-
-				while (j < text->size && !isspace(text->data[j]))
-					j++;
-
-				rndr_autolink2(ob, &text->data[i], j - i, MKDA_NORMAL);
-				i = j - 1;
-				continue;
-			}
-		}
-
-		/*
-		 * Smartypants subsitutions
-		 */
-		if (smartypants) {
-			for (sub = 0; sub < SUBS_COUNT; ++sub) {
-				if (c == smartypants_subs[sub].c0 &&
-					smartypants_cmpsub(text, i, smartypants_subs[sub].pattern)) {
+		for (sub = 0; sub < SUBS_COUNT; ++sub) {
+			if (c == smartypants_subs[sub].c0 &&
+				smartypants_cmpsub(text, i, smartypants_subs[sub].pattern)) {
 
-					if (smartypants_subs[sub].entity)
-						bufputs(ob, smartypants_subs[sub].entity);
+				if (smartypants_subs[sub].entity)
+					bufputs(ob, smartypants_subs[sub].entity);
 
-					i += smartypants_subs[sub].skip;
-					break;
-				}
+				i += smartypants_subs[sub].skip;
+				break;
 			}
+		}
 
-			if (sub < SUBS_COUNT)
-				continue;
+		if (sub < SUBS_COUNT)
+			continue;
 
-			switch (c) {
-			case '<':
-				open_tag = 1;
-				break;
+		switch (c) {
+		case '<':
+			open_tag = 1;
+			break;
 
-			case '>':
-				open_tag = 0;
-				break;
+		case '>':
+			open_tag = 0;
+			break;
 
 #if 0
-			/*
-			 * FIXME: this is bongos.
-			 *
-			 * The markdown spec defines that code blocks can be delimited
-			 * by more than one backtick, e.g.
-			 *
-			 *		``There is a literal backtick (`) here.``
-			 *		<p><code>There is a literal backtick (`) here.</code></p>
-			 *
-			 * Obviously, there's no way to differentiate between the start
-			 * of a code block and the start of a quoted string for smartypants
-			 *
-			 * Look at this piece of Python code:
-			 *
-			 *		``result = ''.join(['this', 'is', 'bongos'])``
-			 *
-			 * This MD expression is clearly ambiguous since it can be parsed as:
-			 *
-			 *		<p>&ldquo;result = &rdquo;.join ...</p>
-			 *
-			 * Or also as:
-			 *
-			 *		<p><code>result = ''.join(['this', 'is', 'bongos'])</code></p>
-			 *
-			 * Fuck everything about this. This is temporarily disabled, because at GitHub
-			 * it's probably smarter to prioritize code blocks than pretty cutesy punctuation.
-			 *
-			 * The equivalent closing tag for the (``), ('') has also been disabled, because
-			 * it makes no sense to have closing tags without opening tags.
-			 */
-			case '`':
-				if (open_tag == 0) {
-					if (i + 1 < text->size && text->data[i + 1] == '`') {
-						BUFPUTSL(ob, "&ldquo;"); i++;
-						continue;
-					}
+		/*
+		 * FIXME: this is bongos.
+		 *
+		 * The markdown spec defines that code blocks can be delimited
+		 * by more than one backtick, e.g.
+		 *
+		 *		``There is a literal backtick (`) here.``
+		 *		<p><code>There is a literal backtick (`) here.</code></p>
+		 *
+		 * Obviously, there's no way to differentiate between the start
+		 * of a code block and the start of a quoted string for smartypants
+		 *
+		 * Look at this piece of Python code:
+		 *
+		 *		``result = ''.join(['this', 'is', 'bongos'])``
+		 *
+		 * This MD expression is clearly ambiguous since it can be parsed as:
+		 *
+		 *		<p>&ldquo;result = &rdquo;.join ...</p>
+		 *
+		 * Or also as:
+		 *
+		 *		<p><code>result = ''.join(['this', 'is', 'bongos'])</code></p>
+		 *
+		 * Fuck everything about this. This is temporarily disabled, because at GitHub
+		 * it's probably smarter to prioritize code blocks than pretty cutesy punctuation.
+		 *
+		 * The equivalent closing tag for the (``), ('') has also been disabled, because
+		 * it makes no sense to have closing tags without opening tags.
+		 */
+		case '`':
+			if (open_tag == 0) {
+				if (i + 1 < text->size && text->data[i + 1] == '`') {
+					BUFPUTSL(ob, "&ldquo;"); i++;
+					continue;
 				}
-				break;
+			}
+			break;
 #endif
 
-			case '\"':
-				if (open_tag == 0) {
-					if (smartypants_quotes(ob, text, i, open_double)) {
-						open_double = !open_double;
-						continue;
-					}
+		case '\"':
+			if (open_tag == 0) {
+				if (smartypants_quotes(ob, text, i, open_double)) {
+					open_double = !open_double;
+					continue;
 				}
-				break;
+			}
+			break;
 
-			case '\'':
-				if (open_tag == 0) {
+		case '\'':
+			if (open_tag == 0) {
 
 #if 0 /* temporarily disabled, see previous comment */
-					if (i + 1 < text->size && text->data[i + 1] == '\'') {
-						BUFPUTSL(ob, "&rdquo;"); i++;
-						continue;
-					} 
+				if (i + 1 < text->size && text->data[i + 1] == '\'') {
+					BUFPUTSL(ob, "&rdquo;"); i++;
+					continue;
+				} 
 #endif
 
-					if (smartypants_quotes(ob, text, i, open_single)) {
-						open_single = !open_single;
-						continue;
-					}
+				if (smartypants_quotes(ob, text, i, open_single)) {
+					open_single = !open_single;
+					continue;
 				}
-				break;
 			}
+			break;
 		}
 
 		/*
@@ -769,6 +721,9 @@ init_xhtml_renderer(struct mkd_renderer *renderer, unsigned int render_flags)
 		renderer->link = NULL;
 		renderer->autolink = NULL;
 	}
+
+	if (render_flags & XHTML_SMARTYPANTS)
+		renderer->normal_text = rndr_smartypants;
 }
 
 void

data/ext/xhtml.h

@@ -24,7 +24,6 @@ typedef enum {
 	XHTML_SKIP_LINKS = (1 << 3),
 	XHTML_SMARTYPANTS = (1 << 4),
 	XHTML_EXPAND_TABS = (1 << 5),
-	XHTML_AUTOLINK = (1 << 6),
 	XHTML_SAFELINK = (1 << 7),
 	XHTML_TOC = (1 << 8),
 	XHTML_STRIKETHROUGH = (1 << 10),

data/lib/redcarpet.rb

@@ -26,7 +26,7 @@
 #   end
 #
 class Redcarpet
-  VERSION = '1.5.3'
+  VERSION = '1.6.0'
 
   # Original Markdown formatted text.
   attr_reader :text
@@ -49,15 +49,15 @@ class Redcarpet
   # Disable superscript and relaxed emphasis processing.
   attr_accessor :strict
 
-  # Convert URL in links, even if they aren't encased in <tt><></tt>
-  attr_accessor :autolink
-
   # Don't make hyperlinks from <tt>[][]</tt> links that have unknown URL types.
   attr_accessor :safelink
 
   # Add TOC anchors to every header
   attr_accessor :generate_toc
 
+  # Enable the Autolinking extension
+  attr_accessor :autolink
+
   # Enable PHP-Markdown tables extension
   attr_accessor :tables
 

data/redcarpet.gemspec

@@ -1,6 +1,6 @@
 Gem::Specification.new do |s|
   s.name = 'redcarpet'
-  s.version = '1.5.3'
+  s.version = '1.6.0'
   s.summary = "Ruby bindings for libupskirt"
   s.date = '2011-04-12'
   s.email = 'vicent@github.com'

data/test/redcarpet_test.rb

@@ -199,8 +199,15 @@ EOS
   end
 
   def test_that_headers_are_linkable
-    markdown = Markdown.new('### Hello [GitHub](http://github.com)')
+    markdown = Redcarpet.new('### Hello [GitHub](http://github.com)')
     assert_equal "<h3>Hello <a href=\"http://github.com\">GitHub</a></h3>", markdown.to_html.strip
   end
+
+  def test_autolinking_with_ent_chars
+    markdown = Redcarpet.new(<<text, :autolink)
+This a stupid link: https://github.com/rtomayko/tilt/issues?milestone=1&state=open
+text
+    assert_equal "<p>This a stupid link: <a href=\"https://github.com/rtomayko/tilt/issues?milestone=1&state=open\">https://github.com/rtomayko/tilt/issues?milestone=1&amp;state=open</a></p>\n", markdown.to_html
+  end
   
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification 
 name: redcarpet
 version: !ruby/object:Gem::Version 
-  hash: 5
+  hash: 15
   prerelease: false
   segments: 
   - 1
-  - 5
-  - 3
-  version: 1.5.3
+  - 6
+  - 0
+  version: 1.6.0
 platform: ruby
 authors: 
 - "Natacha Port\xC3\xA9"