bluecloth 2.0.3 → 2.0.4

data/ChangeLog

@@ -1,3 +1,26 @@
+-- Tue, 30 Jun 2009 01:33:36 -0000 by deveiant (r124) -----
+   Changed: ext/markdown.c
+            ext/VERSION
+            ext/markdown.h
+            ext/generate.c
+
+    * Updated with Discount 1.4.4.
+
+
+-- Mon, 15 Jun 2009 13:15:54 -0000 by deveiant (r123) -----
+   Added:   spec/discount_spec.rb (new)
+   Changed: ext/cstring.h
+            ext/bluecloth.c
+            ext/markdown.c
+            spec/bluecloth/links_spec.rb
+            spec/lib/matchers.rb
+            ext/resource.c (and 10 other/s)
+
+    * Updated to Discount 1.4.2.
+    * Added some more tests for various Discount extensions.
+    * Added support for the MKD_AUTOLINK and MKD_SAFELINK.
+
+
 -- Tue, 26 May 2009 05:08:15 -0000 by deveiant (r121) -----
    Added:   ext/Csio.c (new)
             ext/xmlpage.c (new)

data/ext/Csio.c

@@ -26,9 +26,9 @@ Csprintf(Cstring *iot, char *fmt, ...)
     do {
 	RESERVE(*iot, siz);
 	va_start(ptr, fmt);
-	siz = vsnprintf(T(*iot)+S(*iot), ALL(*iot)-S(*iot), fmt, ptr);
+	siz = vsnprintf(T(*iot)+S(*iot), ALLOCATED(*iot)-S(*iot), fmt, ptr);
 	va_end(ptr);
-    } while ( siz > (ALL(*iot)-S(*iot)) );
+    } while ( siz > (ALLOCATED(*iot)-S(*iot)) );
 
     S(*iot) += siz;
     return siz;

data/ext/VERSION

@@ -1 +1 @@
-1.4.0
+1.4.4

data/ext/bluecloth.c

@@ -1,6 +1,6 @@
 /* 
  * BlueCloth -- a Ruby implementation of Markdown
- * $Id: bluecloth.c 117 2009-05-14 15:21:03Z deveiant $
+ * $Id: bluecloth.c 123 2009-06-15 13:15:54Z deveiant $
  * 
  * = Authors
  * 
@@ -358,6 +358,12 @@ void Init_bluecloth_ext( void ) {
 	/* MKD_NOLINKS|MKD_NOIMAGE|MKD_TAGTEXT */
 	rb_define_const( bluecloth_cBlueCloth, "MKD_EMBED",    INT2FIX(MKD_EMBED) );
 
+	/* Create links for inline URIs */
+	rb_define_const( bluecloth_cBlueCloth, "MKD_AUTOLINK", INT2FIX(MKD_AUTOLINK) );
+
+	/* Be paranoid about link protocols */
+	rb_define_const( bluecloth_cBlueCloth, "MKD_SAFELINK", INT2FIX(MKD_SAFELINK) );
+
 
 	/* don't process header blocks */
 	rb_define_const( bluecloth_cBlueCloth, "MKD_NOHEADER", INT2FIX(MKD_NOHEADER) );

data/ext/cstring.h

@@ -22,8 +22,8 @@
 			    : (T(x) = T(x) ? realloc(T(x), sizeof T(x)[0] * ((x).alloc += 100)) \
 					   : malloc(sizeof T(x)[0] * ((x).alloc += 100)) )]
 
-#define DELETE(x)	(x).alloc ? (free(T(x)), S(x) = (x).alloc = 0) \
-				  : ( S(x) = 0 )
+#define DELETE(x)	ALLOCATED(x) ? (free(T(x)), S(x) = (x).alloc = 0) \
+				     : ( S(x) = 0 )
 #define CLIP(t,i,sz)	\
 	    ( ((i) >= 0) && ((sz) > 0) && (((i)+(sz)) <= S(t)) ) ? \
 	    (memmove(&T(t)[i], &T(t)[i+sz], (S(t)-(i+sz)+1)*sizeof(T(t)[0])), \
@@ -50,7 +50,7 @@
  */
 #define T(x)		(x).text
 #define S(x)		(x).size
-#define ALL(x)		(x).alloc
+#define ALLOCATED(x)	(x).alloc
 
 /* abstract anchor type that defines a list base
  * with a function that attaches an element to

data/ext/generate.c

@@ -17,11 +17,6 @@
 #include "markdown.h"
 #include "amalloc.h"
 
-/* prefixes for <automatic links>
- */
-static char *autoprefix[] = { "http://", "https://", "ftp://", "news://" };
-#define SZAUTOPREFIX	(sizeof autoprefix / sizeof autoprefix[0])
-
 typedef int (*stfu)(const void*,const void*);
 
 
@@ -119,7 +114,7 @@ shift(MMIOT *f, int i)
 /* Qchar()
  */
 static void
-Qchar(char c, MMIOT *f)
+Qchar(int c, MMIOT *f)
 {
     block *cur;
     
@@ -328,18 +323,27 @@ ___mkd_reparse(char *bfr, int size, int flags, MMIOT *f)
  * write out a url, escaping problematic characters
  */
 static void
-puturl(char *s, int size, MMIOT *f)
+puturl(char *s, int size, MMIOT *f, int display)
 {
     unsigned char c;
 
     while ( size-- > 0 ) {
 	c = *s++;
 
+	if ( c == '\\' && size-- > 0 ) {
+	    c = *s++;
+
+	    if ( !( ispunct(c) || isspace(c) ) )
+		Qchar('\\', f);
+	}
+	
 	if ( c == '&' )
 	    Qstring("&amp;", f);
 	else if ( c == '<' )
 	    Qstring("&lt;", f);
-	else if ( isalnum(c) || ispunct(c) )
+	else if ( c == '"' )
+	    Qstring("%22", f);
+	else if ( isalnum(c) || ispunct(c) || (display && isspace(c)) )
 	    Qchar(c, f);
 	else
 	    Qprintf(f, "%%%02X", c);
@@ -372,186 +376,164 @@ parenthetical(int in, int out, MMIOT *f)
 	    return EOF;
 	else if ( c == in )
 	    ++indent;
+	else if ( (c == '\\') && (peek(f,1) == out) ) {
+	    ++size;
+	    pull(f);
+	}
 	else if ( c == out )
 	    --indent;
     }
-    return size-1;
+    return size ? (size-1) : 0;
 }
 
 
 /* extract a []-delimited label from the input stream.
  */
-static char *
-linkylabel(MMIOT *f, int *sizep)
+static int
+linkylabel(MMIOT *f, Cstring *res)
 {
     char *ptr = cursor(f);
+    int size;
 
-    if ( (*sizep = parenthetical('[',']',f)) != EOF )
-	return ptr;
+    if ( (size = parenthetical('[',']',f)) != EOF ) {
+	T(*res) = ptr;
+	S(*res) = size;
+	return 1;
+    }
     return 0;
 }
 
 
-/* extract a (-prefixed url from the input stream.
- * the label is either of the format `<link>`, where I
- * extract until I find a >, or it is of the format
- * `text`, where I extract until I reach a ')' or
- * whitespace.
+/* see if the quote-prefixed linky segment is actually a title.
  */
-static char*
-linkyurl(MMIOT *f, int *sizep)
+static int
+linkytitle(MMIOT *f, char quote, Footnote *ref)
 {
-    int size = 0;
-    char *ptr;
-    int c;
-
-    if ( (c = eatspace(f)) == EOF )
-	return 0;
+    int whence = mmiottell(f);
+    char *title = cursor(f);
+    char *e;
+    register int c;
 
-    ptr = cursor(f);
-
-    /* if I do (title:blah blah blah) embedded links, I need to subvert
-     * linkyurl to do a lookahead for the pseudo-protocol, then snarf
-     * up everything up to the terminating ')'
-     */
-
-    if ( c == '<' ) {
-	pull(f);
-	ptr++;
-	if ( (size = parenthetical('<', '>', f)) == EOF )
-	    return 0;
-    }
-    else {
-	for ( ; ((c=pull(f)) != ')') && !isspace(c); size++)
-	    if ( c == EOF ) return 0;
-	if ( c == ')' )
-	    shift(f, -1);
+    while ( (c = pull(f)) != EOF ) {
+	e = cursor(f);
+	if ( c == quote ) {
+	    if ( (c = eatspace(f)) == ')' ) {
+		T(ref->title) = 1+title;
+		S(ref->title) = (e-title)-2;
+		return 1;
+	    }
+	}
     }
-    *sizep = size;
-    return ptr;
+    mmiotseek(f, whence);
+    return 0;
 }
 
 
 /* extract a =HHHxWWW size from the input stream
  */
 static int
-linkysize(MMIOT *f, int *heightp, int *widthp)
+linkysize(MMIOT *f, Footnote *ref)
 {
     int height=0, width=0;
+    int whence = mmiottell(f);
     int c;
 
-    *heightp = 0;
-    *widthp = 0;
-
-    if ( (c = eatspace(f)) != '=' ) 
-	return (c != EOF);
-    pull(f);	/* eat '=' */
+    if ( isspace(peek(f,0)) ) {
+	pull(f);	/* eat '=' */
 
-    for ( c = pull(f); isdigit(c); c = pull(f))
-	width = (width * 10) + (c - '0');
-
-    if ( c == 'x' ) {
 	for ( c = pull(f); isdigit(c); c = pull(f))
-	    height = (height*10) + (c - '0');
-
-	if ( c != EOF ) {
-	    if ( !isspace(c) ) shift(f, -1);
-	    *heightp = height;
-	    *widthp = width;
-	    return 1;
-	}
-    }
-    return 0;
-}
-
+	    width = (width * 10) + (c - '0');
 
-/* extract a )-terminated title from the input stream.
- */
-static char*
-linkytitle(MMIOT *f, int *sizep)
-{
-    int countq=0, qc, c, size;
-    char *ret, *lastqc = 0;
-
-    eatspace(f);
-    if ( (qc=pull(f)) != '"' && qc != '\'' && qc != '(' )
-	return 0;
+	if ( c == 'x' ) {
+	    for ( c = pull(f); isdigit(c); c = pull(f))
+		height = (height*10) + (c - '0');
 
-    if ( qc == '(' ) qc = ')';
+	    if ( isspace(c) )
+		c = eatspace(f);
 
-    for ( ret = cursor(f); (c = pull(f)) != EOF;  ) {
-	if ( (c == ')') && countq ) {
-	    size = (lastqc ? lastqc : cursor(f)) - ret;
-	    *sizep = size-1;
-	    return ret;
-	}
-	else if ( c == qc ) {
-	    lastqc = cursor(f);
-	    countq++;
+	    if ( (c == ')') || ((c == '\'' || c == '"') && linkytitle(f, c, ref)) ) {
+		ref->height = height;
+		ref->width  = width;
+		return 1;
+	    }
 	}
     }
+    mmiotseek(f, whence);
     return 0;
 }
 
 
-/* look up (or construct) a footnote from the [xxx] link
- * at the head of the stream.
+/* extract a (-prefixed url from the input stream.
+ * the label is either of the format `<link>`, where I
+ * extract until I find a >, or it is of the format
+ * `text`, where I extract until I reach a ')', a quote,
+ * or (if image) a '='
  */
 static int
-linkykey(int image, Footnote *val, MMIOT *f)
+linkyurl(MMIOT *f, int image, Footnote *p)
 {
-    Footnote *ret;
-    Cstring mylabel;
-    int here;
-
-    memset(val, 0, sizeof *val);
+    int c;
+    int mayneedtotrim=0;
 
-    if ( (T(val->tag) = linkylabel(f, &S(val->tag))) == 0 )
+    if ( (c = eatspace(f)) == EOF )
 	return 0;
 
-    here = mmiottell(f);
-    eatspace(f);
-    switch ( pull(f) ) {
-    case '(':
-	/* embedded link */
-	if ( (T(val->link) = linkyurl(f,&S(val->link))) == 0 )
-	    return 0;
+    if ( c == '<' ) {
+	pull(f);
+	mayneedtotrim=1;
+    }
 
-	if ( image && !linkysize(f, &val->height, &val->width) )
+    T(p->link) = cursor(f);
+    for ( S(p->link)=0; (c = peek(f,1)) != ')'; ++S(p->link) ) {
+	if ( c == EOF )
 	    return 0;
+	else if ( (c == '"' || c == '\'') && linkytitle(f, c, p) )
+	    break;
+	else if ( image && (c == '=') && linkysize(f, p) )
+	    break;
+	else if ( (c == '\\') && ispunct(peek(f,2)) ) {
+	    ++S(p->link);
+	    pull(f);
+	}
+	pull(f);
+    }
+    if ( peek(f, 1) == ')' )
+	pull(f);
+	
+    ___mkd_tidy(&p->link);
+    
+    if ( mayneedtotrim && (T(p->link)[S(p->link)-1] == '>') )
+	--S(p->link);
+    
+    return 1;
+}
 
-	T(val->title) = linkytitle(f, &S(val->title));
 
-	return peek(f,0) == ')';
 
-    case '[':		/* footnote links /as defined in the standard/ */
-    default:		/* footnote links -- undocumented extension */
-	/* footnote link */
-	mylabel = val->tag;
-	if ( peek(f,0) == '[' ) {
-	    if ( (T(val->tag) = linkylabel(f, &S(val->tag))) == 0 )
-		return 0;
+/* prefixes for <automatic links>
+ */
+static struct {
+    char *name;
+    int   nlen;
+} protocol[] = { 
+#define _aprotocol(x)	{ x, (sizeof x)-1 }
+    _aprotocol( "http://" ), 
+    _aprotocol( "https://" ), 
+    _aprotocol( "ftp://" ), 
+    _aprotocol( "news://" ),
+#undef _aprotocol
+};
+#define NRPROTOCOLS	(sizeof protocol / sizeof protocol[0])
 
-	    if ( !S(val->tag) )
-		val->tag = mylabel;
-	}
-	else if ( f->flags & MKD_1_COMPAT )
-	    break;
-	else
-	    mmiotseek(f,here);
 
-	ret = bsearch(val, T(*f->footnotes), S(*f->footnotes),
-	               sizeof *val, (stfu)__mkd_footsort);
+static int
+isautoprefix(char *text)
+{
+    int i;
 
-	if ( ret ) {
-	    val->tag = mylabel;
-	    val->link = ret->link;
-	    val->title = ret->title;
-	    val->height = ret->height;
-	    val->width = ret->width;
+    for (i=0; i < NRPROTOCOLS; i++)
+	if ( strncasecmp(text, protocol[i].name, protocol[i].nlen) == 0 )
 	    return 1;
-	}
-    }
     return 0;
 }
 
@@ -569,12 +551,14 @@ typedef struct linkytype {
     char *text_pfx;	/* text prefix                  (eg: ">"           */
     char *text_sfx;	/* text suffix			(eg: "</a>"        */
     int      flags;	/* reparse flags */
+    int      kind;	/* tag is url or something else? */
+#define IS_URL	0x01
 } linkytype;
 
 static linkytype imaget = { 0, 0, "<img src=\"", "\"",
-			     1, " alt=\"", "\" />", DENY_IMG|INSIDE_TAG };
+			     1, " alt=\"", "\" />", DENY_IMG|INSIDE_TAG, IS_URL };
 static linkytype linkt  = { 0, 0, "<a href=\"", "\"",
-                             0, ">", "</a>", DENY_A };
+                             0, ">", "</a>", DENY_A, IS_URL };
 
 /*
  * pseudo-protocols for [][];
@@ -584,9 +568,10 @@ static linkytype linkt  = { 0, 0, "<a href=\"", "\"",
  * raw: just dump the link without any processing
  */
 static linkytype specials[] = {
-    { "id:", 3, "<a id=\"", "\"", 0, ">", "</a>", 0 },
-    { "class:", 6, "<span class=\"", "\"", 0, ">", "</span>", 0 },
-    { "raw:", 4, 0, 0, 0, 0, 0, 0 },
+    { "id:", 3, "<a id=\"", "\"", 0, ">", "</a>", 0, IS_URL },
+    { "class:", 6, "<span class=\"", "\"", 0, ">", "</span>", 0, 0 },
+    { "raw:", 4, 0, 0, 0, 0, 0, DENY_HTML, 0 },
+    { "abbr:", 5, "<abbr title=\"", "\"", 0, ">", "</abbr>", 0, 0 },
 } ;
 
 #define NR(x)	(sizeof x / sizeof x[0])
@@ -594,7 +579,7 @@ static linkytype specials[] = {
 /* see if t contains one of our pseudo-protocols.
  */
 static linkytype *
-extratag(Cstring t)
+pseudo(Cstring t)
 {
     int i;
     linkytype *r;
@@ -608,57 +593,122 @@ extratag(Cstring t)
 }
 
 
-/*
- * process embedded links and images
+/* print out a linky (or fail if it's Not Allowed)
  */
 static int
-linkylinky(int image, MMIOT *f)
+linkyformat(MMIOT *f, Cstring text, int image, Footnote *ref)
 {
-    int start = mmiottell(f);
-    Footnote link;
     linkytype *tag;
 
-    if ( !linkykey(image, &link, f) ) {
-	mmiotseek(f, start);
-	return 0;
-    }
-
     if ( image )
 	tag = &imaget;
-    else if ( (f->flags & NO_PSEUDO_PROTO) || (tag = extratag(link.link)) == 0 )
+    else if ( tag = pseudo(ref->link) ) {
+	if ( f->flags & (NO_PSEUDO_PROTO|SAFELINK) )
+	    return 0;
+    }
+    else if ( (f->flags & SAFELINK) && T(ref->link)
+				    && (T(ref->link)[0] != '/')
+				    && !isautoprefix(T(ref->link)) )
+	/* if SAFELINK, only accept links that are local or
+	 * a well-known protocol
+	 */
+	    return 0;
+    else
 	tag = &linkt;
 
-    if ( f->flags & tag-> flags ) {
-	mmiotseek(f, start);
+    if ( f->flags & tag->flags )
 	return 0;
-    }
 
     if ( tag->link_pfx ) {
 	Qstring(tag->link_pfx, f);
-	if ( f->base && (T(link.link)[tag->szpat] == '/') )
-	    puturl(f->base, strlen(f->base), f);
-	puturl(T(link.link) + tag->szpat, S(link.link) - tag->szpat, f);
+	
+	if ( tag->kind & IS_URL ) {
+	    if ( f->base && T(ref->link) && (T(ref->link)[tag->szpat] == '/') )
+		puturl(f->base, strlen(f->base), f, 0);
+	    puturl(T(ref->link) + tag->szpat, S(ref->link) - tag->szpat, f, 0);
+	}
+	else
+	    ___mkd_reparse(T(ref->link) + tag->szpat, S(ref->link) - tag->szpat, INSIDE_TAG, f);
+	
 	Qstring(tag->link_sfx, f);
 
-	if ( tag->WxH && link.height && link.width ) {
-	    Qprintf(f," height=\"%d\"", link.height);
-	    Qprintf(f, " width=\"%d\"", link.width);
+	if ( tag->WxH && ref->height && ref->width ) {
+	    Qprintf(f," height=\"%d\"", ref->height);
+	    Qprintf(f, " width=\"%d\"", ref->width);
 	}
 
-	if ( S(link.title) ) {
+	if ( S(ref->title) ) {
 	    Qstring(" title=\"", f);
-	    ___mkd_reparse(T(link.title), S(link.title), INSIDE_TAG, f);
+	    ___mkd_reparse(T(ref->title), S(ref->title), INSIDE_TAG, f);
 	    Qchar('"', f);
 	}
 
 	Qstring(tag->text_pfx, f);
-	___mkd_reparse(T(link.tag), S(link.tag), tag->flags, f);
+	___mkd_reparse(T(text), S(text), tag->flags, f);
 	Qstring(tag->text_sfx, f);
     }
     else
-	Qwrite(T(link.link) + tag->szpat, S(link.link) - tag->szpat, f);
+	Qwrite(T(ref->link) + tag->szpat, S(ref->link) - tag->szpat, f);
 
     return 1;
+} /* linkyformat */
+
+
+/*
+ * process embedded links and images
+ */
+static int
+linkylinky(int image, MMIOT *f)
+{
+    int start = mmiottell(f);
+    int implicit_mark;
+    Cstring name;
+    Footnote key, *ref;
+		
+    int status = 0;
+
+    CREATE(name);
+    bzero(&key, sizeof key);
+
+    if ( linkylabel(f, &name) ) {
+	implicit_mark = mmiottell(f);
+	eatspace(f);
+
+	switch ( pull(f) ) {
+	case '(':  /* embedded link */
+	    if ( linkyurl(f, image, &key) )
+		status = linkyformat(f, name, image, &key);
+	    break;
+	
+	case '[':/* footnote link */
+	default: /* (undocumented) implicit link */
+	    if ( peek(f, 0) != '[' ) {
+		mmiotseek(f, implicit_mark);
+		if ( f->flags & MKD_1_COMPAT )
+		    break;
+	    }
+	    else if ( !linkylabel(f, &key.tag) )
+		break;
+	    
+	    if ( !S(key.tag) ) {
+		DELETE(key.tag);
+		T(key.tag) = T(name);
+		S(key.tag) = S(name);
+	    }
+
+	    if ( ref = bsearch(&key, T(*f->footnotes), S(*f->footnotes),
+				      sizeof key, (stfu)__mkd_footsort) )
+		status = linkyformat(f, name, image, ref);
+	}
+    }
+
+    DELETE(name);
+    ___mkd_freefootnote(&key);
+
+    if ( status == 0 )
+	mmiotseek(f, start);
+
+    return status;
 }
 
 
@@ -711,6 +761,80 @@ forbidden_tag(MMIOT *f)
 }
 
 
+/* Check a string to see if it looks like a mail address
+ * "looks like a mail address" means alphanumeric + some
+ * specials, then a `@`, then alphanumeric + some specials,
+ * but with a `.`
+ */
+static int
+maybe_address(char *p, int size)
+{
+    int ok = 0;
+    
+    for ( ;size && (isalnum(*p) || strchr("._-+*", *p)); ++p, --size)
+	;
+
+    if ( ! (size && *p == '@') )
+	return 0;
+    
+    --size, ++p;
+
+    if ( size && *p == '.' ) return 0;
+    
+    for ( ;size && (isalnum(*p) || strchr("._-+", *p)); ++p, --size )
+	if ( *p == '.' && size > 1 ) ok = 1;
+
+    return size ? 0 : ok;
+}
+
+
+/* The size-length token at cursor(f) is either a mailto:, an
+ * implicit mailto:, one of the approved url protocols, or just
+ * plain old text.   If it's a mailto: or an approved protocol,
+ * linkify it, otherwise say "no"
+ */
+static int
+process_possible_link(MMIOT *f, int size)
+{
+    int address= 0;
+    int mailto = 0;
+    char *text = cursor(f);
+    
+    if ( f->flags & DENY_A ) return 0;
+
+    if ( (size > 7) && strncasecmp(text, "mailto:", 7) == 0 ) {
+	/* if it says it's a mailto, it's a mailto -- who am
+	 * I to second-guess the user?
+	 */
+	address = 1;
+	mailto = 7; 	/* 7 is the length of "mailto:"; we need this */
+    }
+    else 
+	address = maybe_address(text, size);
+
+    if ( address ) { 
+	Qstring("<a href=\"", f);
+	if ( !mailto ) {
+	    /* supply a mailto: protocol if one wasn't attached */
+	    mangle("mailto:", 7, f);
+	}
+	mangle(text, size, f);
+	Qstring("\">", f);
+	mangle(text+mailto, size-mailto, f);
+	Qstring("</a>", f);
+	return 1;
+    }
+    else if ( isautoprefix(text) ) {
+	Qstring("<a href=\"", f);
+	puturl(text,size,f, 0);
+	Qstring("\">", f);
+	puturl(text,size,f, 1);
+	Qstring("</a>", f);
+	return 1;
+    }
+    return 0;
+} /* process_possible_link */
+
 
 /* a < may be just a regular character, the start of an embedded html
  * tag, or the start of an <automatic link>.    If it's an automatic
@@ -721,68 +845,69 @@ forbidden_tag(MMIOT *f)
 static int
 maybe_tag_or_link(MMIOT *f)
 {
-    char *text;
-    int c, size, i;
-    int maybetag=1, maybeaddress=0;
-    int mailto;
+    int c, size;
+    int maybetag = 1;
 
     if ( f->flags & INSIDE_TAG )
 	return 0;
 
-    for ( size=0; ((c = peek(f,size+1)) != '>') && !isspace(c); size++ ) {
-	if ( ! (c == '/' || isalnum(c) || c == '~') )
-	    maybetag=0;
-	if ( c == '@' )
-	    maybeaddress=1;
-	else if ( c == EOF )
+    for ( size=0; (c = peek(f, size+1)) != '>'; size++) {
+	if ( c == EOF )
 	    return 0;
+	else if ( c == '\\' ) {
+	    maybetag=0;
+	    if ( peek(f, size+2) != EOF )
+		size++;
+	}
+	else if ( isspace(c) )
+	    break;
+	else if ( ! (c == '/' || isalnum(c) ) )
+	    maybetag=0;
     }
 
-    if ( size == 0 )
-	return 0;
-
-    if ( maybetag  || (size >= 3 && strncmp(cursor(f), "!--", 3) == 0) ) {
-	Qstring(forbidden_tag(f) ? "&lt;" : "<", f);
-	while ( ((c = peek(f, 1)) != EOF) && (c != '>') )
-	    cputc(pull(f), f);
-	return 1;
+    if ( size ) {
+	if ( maybetag || (size >= 3 && strncmp(cursor(f), "!--", 3) == 0) ) {
+	    Qstring(forbidden_tag(f) ? "&lt;" : "<", f);
+	    while ( ((c = peek(f, 1)) != EOF) && (c != '>') )
+		cputc(pull(f), f);
+	    return 1;
+	}
+	else if ( !isspace(c) && process_possible_link(f, size) ) {
+	    shift(f, size+1);
+	    return 1;
+	}
     }
+    
+    return 0;
+}
 
-    if ( f->flags & DENY_A ) return 0;
-
-    text = cursor(f);
-    shift(f, size+1);
 
-    for ( i=0; i < SZAUTOPREFIX; i++ )
-	if ( strncasecmp(text, autoprefix[i], strlen(autoprefix[i])) == 0 ) {
-	    Qstring("<a href=\"", f);
-	    puturl(text,size,f);
-	    Qstring("\">", f);
-	    puturl(text,size,f);
-	    Qstring("</a>", f);
-	    return 1;
-	}
-    if ( maybeaddress ) {
+/* autolinking means that all inline html is <a href'ified>.   A
+ * autolink url is alphanumerics, slashes, periods, underscores,
+ * the at sign, colon, and the % character.
+ */
+static int
+maybe_autolink(MMIOT *f)
+{
+    register int c;
+    int size;
 
-	Qstring("<a href=\"", f);
-	if ( (size > 7) && strncasecmp(text, "mailto:", 7) == 0 )
-	    mailto = 7;
-	else {
-	    mailto = 0;
-	    /* supply a mailto: protocol if one wasn't attached */
-	    mangle("mailto:", 7, f);
+    /* greedily scan forward for the end of a legitimate link.
+     */
+    for ( size=0; (c=peek(f, size+1)) != EOF; size++ )
+	if ( c == '\\' ) {
+	     if ( peek(f, size+2) != EOF )
+		++size;
 	}
+	else if ( isspace(c) || strchr("'\"()[]{}<>`", c) )
+	    break;
 
-	mangle(text, size, f);
-	Qstring("\">", f);
-	mangle(text+mailto, size-mailto, f);
-	Qstring("</a>", f);
+    if ( (size > 1) && process_possible_link(f, size) ) {
+	shift(f, size);
 	return 1;
     }
-
-    shift(f, -(size+1));
     return 0;
-} /* maybe_tag_or_link */
+}
 
 
 /* smartyquote code that's common for single and double quotes
@@ -868,7 +993,7 @@ smartypants(int c, int *flags, MMIOT *f)
 {
     int i;
 
-    if ( f->flags & DENY_SMARTY )
+    if ( f->flags & (DENY_SMARTY|INSIDE_TAG) )
 	return 0;
 
     for ( i=0; i < NRSMART; i++)
@@ -922,12 +1047,23 @@ text(MMIOT *f)
     int rep;
     int smartyflags = 0;
 
-    while ( (c = pull(f)) != EOF ) {
+    while (1) {
+        if ( (f->flags & AUTOLINK) && isalpha(peek(f,1)) )
+	    maybe_autolink(f);
+
+        c = pull(f);
+
+        if (c == EOF)
+          break;
+
 	if ( smartypants(c, &smartyflags, f) )
 	    continue;
 	switch (c) {
 	case 0:     break;
 
+	case 3:     Qstring("<br/>", f);
+		    break;
+
 	case '>':   if ( tag_text(f) )
 			Qstring("&gt;", f);
 		    else
@@ -971,15 +1107,24 @@ text(MMIOT *f)
 	case '_':
 #if RELAXED_EMPHASIS
 	/* Underscores don't count if they're in the middle of a word */
-		    if ( (!(f->flags & STRICT))
-			     && ((isthisspace(f,-1) && isthisspace(f,1))
-			      || (isthisalnum(f,-1) && isthisalnum(f,1))) ){
+		    if ( !(f->flags & STRICT) && isthisalnum(f,-1)
+					      && isthisalnum(f,1) ) {
+			Qchar(c, f);
+			break;
+		    }
+#endif
+	case '*':
+#if RELAXED_EMPHASIS
+	/* Underscores & stars don't count if they're out in the middle
+	 * of whitespace */
+		    if ( !(f->flags & STRICT) && isthisspace(f,-1)
+					      && isthisspace(f,1) ) {
 			Qchar(c, f);
 			break;
 		    }
 		    /* else fall into the regular old emphasis case */
 #endif
-	case '*':   if ( tag_text(f) )
+		    if ( tag_text(f) )
 			Qchar(c, f);
 		    else {
 			for (rep = 1; peek(f,1) == c; pull(f) )
@@ -1143,10 +1288,10 @@ printblock(Paragraph *pp, MMIOT *f)
 	    if ( S(t->text) > 2 && T(t->text)[S(t->text)-2] == ' '
 				&& T(t->text)[S(t->text)-1] == ' ') {
 		push(T(t->text), S(t->text)-2, f);
-		push("<br/>\n", 6, f);
+		push("\003\n", 2, f);
 	    }
 	    else {
-		___mkd_tidy(t);
+		___mkd_tidy(&t->text);
 		push(T(t->text), S(t->text), f);
 		if ( t->next )
 		    push("\n", 1, f);