bluecloth 2.0.5 → 2.0.6.pre120

data/ext/bluecloth.h

@@ -0,0 +1,19 @@
+/* 
+ * BlueCloth -- a Ruby implementation of Markdown
+ * $Id$
+ * 
+ */
+
+#ifndef BLUECLOTH_H
+#define BLUECLOTH_H
+
+#include "config.h"
+
+#include "mkdio.h"
+#include "ruby.h"
+
+#ifdef HAVE_RUBY_ENCODING_H
+#	include "ruby/encoding.h"
+#endif
+
+#endif

data/ext/config.h

@@ -6,6 +6,10 @@
 #ifndef CONFIG_H_RZLE3ADO
 #define CONFIG_H_RZLE3ADO
 
+#ifdef RUBY_EXTCONF_H
+#	include RUBY_EXTCONF_H
+#endif
+
 #ifdef HAVE_SRANDOM
 #	define INITRNG(x) srandom((unsigned int)x)
 #elif  HAVE_SRAND
@@ -14,6 +18,10 @@
 #	define INITRNG(x) (void)1
 #endif
 
+#ifndef HAVE_BZERO
+#	define bzero(s, n) (memset((void *)s, 0, (size_t)n))
+#endif
+
 #ifdef HAVE_STRCASECMP
 #elif  HAVE_STRICMP
 #	define strcasecmp stricmp

data/ext/cstring.h

@@ -60,9 +60,10 @@
  * macro will work with it.
  */
 #define ANCHOR(t)	struct { t *text, *end; }
+#define E(t)		((t).end)
 
-#define ATTACH(t, p)	( (t).text ?( ((t).end->next = (p)), ((t).end = (p)) ) \
-				   :( ((t).text = (t).end = (p)) ) )
+#define ATTACH(t, p)	( T(t) ? ( (E(t)->next = (p)), (E(t) = (p)) ) \
+			       : ( (T(t) = E(t) = (p)) ) )
 
 typedef STRING(char) Cstring;
 

data/ext/extconf.rb

@@ -19,7 +19,7 @@ $CPPFLAGS << %Q{ -DVERSION=\\"#{version}\\"}
 
 # Add my own debugging hooks if building for me
 if ENV['DEBUGGING_BUILD']
-	$CFLAGS << ' -ggdb' << ' -DDEBUG' 
+	$CFLAGS << ' -ggdb' << ' -DDEBUG'
 end
 
 def fail( *messages )
@@ -31,6 +31,9 @@ end
 have_func( "srand" ) || have_func( "srandom" )
 have_func( "random" ) || have_func( "rand" )
 
+# bzero() isn't ANSI C, so use memset() if it isn't defined
+have_func( "bzero", %w[string.h strings.h] )
+
 unless have_func( "strcasecmp" ) || have_func( "stricmp" )
 	fail( "This extension requires either strcasecmp() or stricmp()" )
 end
@@ -40,6 +43,10 @@ end
 
 have_header( 'mkdio.h' ) or fail( "missing mkdio.h" )
 
+# Check for 1.9.xish encoding header
+have_header( 'ruby/encoding.h' )
+
+create_header()
 create_makefile( 'bluecloth_ext' )
 
 FileUtils.rm_rf( 'conftest.dSYM' ) # MacOS X cleanup

data/ext/generate.c

@@ -632,9 +632,9 @@ linkyformat(MMIOT *f, Cstring text, int image, Footnote *ref)
 	
 	Qstring(tag->link_sfx, f);
 
-	if ( tag->WxH && ref->height && ref->width ) {
-	    Qprintf(f," height=\"%d\"", ref->height);
-	    Qprintf(f, " width=\"%d\"", ref->width);
+	if ( tag->WxH) {
+	    if ( ref->height) Qprintf(f," height=\"%d\"", ref->height);
+	    if ( ref->width) Qprintf(f, " width=\"%d\"", ref->width);
 	}
 
 	if ( S(ref->title) ) {
@@ -661,44 +661,46 @@ 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);
+    memset(&key, 0, sizeof key);
 
     if ( linkylabel(f, &name) ) {
-	implicit_mark = mmiottell(f);
-	eatspace(f);
-
-	switch ( pull(f) ) {
-	case '(':  /* embedded link */
+	if ( peek(f,1) == '(' ) {
+	    pull(f);
 	    if ( linkyurl(f, image, &key) )
 		status = linkyformat(f, name, image, &key);
-	    break;
+	}
+	else {
+	    int goodlink, implicit_mark = mmiottell(f);
 	
-	case '[':/* footnote link */
-	default: /* (undocumented) implicit link */
-	    if ( peek(f, 0) != '[' ) {
+	    if ( eatspace(f) == '[' ) {
+		pull(f);	/* consume leading '[' */
+		goodlink = linkylabel(f, &key.tag);
+	    }
+	    else {
+		/* new markdown implicit name syntax doesn't
+		 * require a second []
+		 */
 		mmiotseek(f, implicit_mark);
-		if ( f->flags & MKD_1_COMPAT )
-		    break;
+		goodlink = !(f->flags & MKD_1_COMPAT);
 	    }
-	    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 ( goodlink ) {
+		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);
+		if ( ref = bsearch(&key, T(*f->footnotes), S(*f->footnotes),
+					  sizeof key, (stfu)__mkd_footsort) )
+		    status = linkyformat(f, name, image, ref);
+	    }
 	}
     }
 
@@ -867,9 +869,19 @@ maybe_tag_or_link(MMIOT *f)
 
     if ( size ) {
 	if ( maybetag || (size >= 3 && strncmp(cursor(f), "!--", 3) == 0) ) {
+
+	    /* It is not a html tag unless we find the closing '>' in
+	     * the same block.
+	     */
+	    while ( (c = peek(f, size+1)) != '>' )
+		if ( c == EOF )
+		    return 0;
+		else
+		    size++;
+	    
 	    Qstring(forbidden_tag(f) ? "&lt;" : "<", f);
 	    while ( ((c = peek(f, 1)) != EOF) && (c != '>') )
-		cputc(pull(f), f);
+		Qchar(pull(f), f);
 	    return 1;
 	}
 	else if ( !isspace(c) && process_possible_link(f, size) ) {
@@ -1276,6 +1288,106 @@ printheader(Paragraph *pp, MMIOT *f)
 }
 
 
+enum e_alignments { a_NONE, a_CENTER, a_LEFT, a_RIGHT };
+
+static char* alignments[] = { "", " align=\"center\"", " align=\"left\"",
+				  " align=\"right\"" };
+
+typedef STRING(int) Istring;
+
+static int
+splat(Line *p, char *block, Istring align, int force, MMIOT *f)
+{
+    int first,
+	idx = 0,
+	colno = 0;
+
+    Qstring("<tr>\n", f);
+    while ( idx < S(p->text) ) {
+	first = idx;
+	if ( force && (colno >= S(align)-1) )
+	    idx = S(p->text);
+	else
+	    while ( (idx < S(p->text)) && (T(p->text)[idx] != '|') )
+		++idx;
+
+	Qprintf(f, "<%s%s>",
+		   block,
+		   alignments[ (colno < S(align)) ? T(align)[colno] : a_NONE ]);
+	___mkd_reparse(T(p->text)+first, idx-first, 0, f);
+	Qprintf(f, "</%s>\n", block);
+	idx++;
+	colno++;
+    }
+    if ( force )
+	while (colno < S(align) ) {
+	    Qprintf(f, "<%s></%s>\n", block, block);
+	    ++colno;
+	}
+    Qstring("</tr>\n", f);
+    return colno;
+}
+
+static int
+printtable(Paragraph *pp, MMIOT *f)
+{
+    /* header, dashes, then lines of content */
+
+    Line *hdr, *dash, *body;
+    Istring align;
+    int start;
+    int hcols;
+    char *p;
+
+    if ( !(pp->text && pp->text->next) )
+	return 0;
+
+    hdr = pp->text;
+    dash= hdr->next;
+    body= dash->next;
+
+    /* first figure out cell alignments */
+
+    CREATE(align);
+
+    for (p=T(dash->text), start=0; start < S(dash->text); ) {
+	char first, last;
+	int end;
+	
+	last=first=0;
+	for (end=start ; (end < S(dash->text)) && p[end] != '|'; ++ end ) {
+	    if ( !isspace(p[end]) ) {
+		if ( !first) first = p[end];
+		last = p[end];
+	    }
+	}
+	EXPAND(align) = ( first == ':' ) ? (( last == ':') ? a_CENTER : a_LEFT)
+					 : (( last == ':') ? a_RIGHT : a_NONE );
+	start = 1+end;
+    }
+
+    Qstring("<table>\n", f);
+    Qstring("<thead>\n", f);
+    hcols = splat(hdr, "th", align, 0, f);
+    Qstring("</thead>\n", f);
+
+    if ( hcols < S(align) )
+	S(align) = hcols;
+    else
+	while ( hcols > S(align) )
+	    EXPAND(align) = a_NONE;
+
+    Qstring("<tbody>\n", f);
+    for ( ; body; body = body->next)
+	splat(body, "td", align, 1, f);
+    Qstring("</tbody>\n", f);
+    Qstring("</table>\n", f);
+
+    DELETE(align);
+    return 1;
+}
+
+
 static int
 printblock(Paragraph *pp, MMIOT *f)
 {
@@ -1382,6 +1494,7 @@ definitionlist(Paragraph *p, MMIOT *f)
 	    }
 
 	    htmlify(p->down, "dd", p->ident, f);
+	    Qchar('\n', f);
 	}
 
 	Qstring("</dl>", f);
@@ -1453,6 +1566,14 @@ display(Paragraph *p, MMIOT *f)
 	printheader(p, f);
 	break;
 
+    case TABLE:
+	printtable(p, f);
+	break;
+
+    case SOURCE:
+	htmlify(p->down, 0, 0, f);
+	break;
+	
     default:
 	printblock(p, f);
 	break;

data/ext/markdown.c

@@ -32,11 +32,11 @@ static struct kw blocktags[] = { KW("!--"), KW("STYLE"), KW("SCRIPT"),
 				 KW("ADDRESS"), KW("BDO"), KW("BLOCKQUOTE"),
 				 KW("CENTER"), KW("DFN"), KW("DIV"), KW("H1"),
 				 KW("H2"), KW("H3"), KW("H4"), KW("H5"),
-				 KW("H6"), KW("IFRAME"), KW("LISTING"), KW("NOBR"),
+				 KW("H6"), KW("LISTING"), KW("NOBR"),
 				 KW("UL"), KW("P"), KW("OL"), KW("DL"),
 				 KW("PLAINTEXT"), KW("PRE"), KW("TABLE"),
 				 KW("WBR"), KW("XMP"), SC("HR"), SC("BR"),
-				 KW("MAP") };
+				 KW("IFRAME"), KW("MAP") };
 #define SZTAGS	(sizeof blocktags / sizeof blocktags[0])
 #define MAXTAG	11 /* sizeof "BLOCKQUOTE" */
 
@@ -230,6 +230,8 @@ htmlblock(Paragraph *p, struct kw *tag)
 			    /* consume trailing gunk in close tag */
 			    c = flogetc(&f);
 			}
+			if ( !f.t )
+			    return 0;
 			ret = f.t->next;
 			f.t->next = 0;
 			return ret;
@@ -259,6 +261,37 @@ comment(Paragraph *p)
 }
 
 
+/* tables look like
+ *   header|header{|header}
+ *   ------|------{|......}
+ *   {body lines}
+ */
+static int
+istable(Line *t)
+{
+    char *p;
+    Line *dashes = t->next;
+    int contains = 0;	/* found character bits; 0x01 is |, 0x02 is - */
+    
+    /* two lines, first must contain | */
+    if ( !(dashes && memchr(T(t->text), '|', S(t->text))) )
+	return 0;
+
+    /* second line must contain - or | and nothing
+     * else except for whitespace or :
+     */
+    for ( p = T(dashes->text)+S(dashes->text)-1; p >= T(dashes->text); --p)
+	if ( *p == '|' )
+	    contains |= 0x01;
+	else if ( *p == '-' )
+	    contains |= 0x02;
+	else if ( ! ((*p == ':') || isspace(*p)) )
+	    return 0;
+
+    return (contains & 0x03);
+}
+
+
 /* footnotes look like ^<whitespace>{0,3}[stuff]: <content>$
  */
 static int
@@ -554,17 +587,17 @@ szmarkerclass(char *p)
  * marker %[kind:]name%
  */
 static int
-isdivmarker(Line *p)
+isdivmarker(Line *p, int start)
 {
 #if DIV_QUOTE
     char *s = T(p->text);
     int len = S(p->text);
     int i;
 
-    if ( !(len && s[0] == '%' && s[len-1] == '%') ) return 0;
+    if ( !(len && s[start] == '%' && s[len-1] == '%') ) return 0;
 
-    i = szmarkerclass(s+1);
-    --len;
+    i = szmarkerclass(s+start+1)+start;
+    len -= start+1;
 
     while ( ++i < len )
 	if ( !isalnum(s[i]) )
@@ -601,13 +634,15 @@ quoteblock(Paragraph *p)
 	    t->dle = mkd_firstnonblank(t);
 	}
 
-	if ( !(q = skipempty(t->next)) || ((q != t->next) && !isquote(q)) ) {
+	q = skipempty(t->next);
+
+	if ( (q == 0) || ((q != t->next) && (!isquote(q) || isdivmarker(q,1))) ) {
 	    ___mkd_freeLineRange(t, q);
 	    t = q;
 	    break;
 	}
     }
-    if ( isdivmarker(p->text) ) {
+    if ( isdivmarker(p->text,0) ) {
 	char *prefix = "class";
 	int i;
 	
@@ -628,6 +663,25 @@ quoteblock(Paragraph *p)
 }
 
 
+/*
+ * A table block starts with a table header (see istable()), and continues
+ * until EOF or a line that /doesn't/ contain a |.
+ */
+static Line *
+tableblock(Paragraph *p)
+{
+    Line *t, *q;
+
+    for ( t = p->text; t && (q = t->next); t = t->next ) {
+	if ( !memchr(T(q->text), '|', S(q->text)) ) {
+	    t->next = 0;
+	    return q;
+	}
+    }
+    return 0;
+}
+
+
 static Paragraph *Pp(ParagraphRoot *, Line *, int);
 static Paragraph *compile(Line *, int, MMIOT *);
 
@@ -659,7 +713,7 @@ listitem(Paragraph *p, int indent)
 	 * need any indentation
 	 */
 	if ( q != t->next ) {
-	    if (q->dle < 4) {
+	    if (q->dle < indent) {
 		q = t->next;
 		t->next = 0;
 		return q;
@@ -684,9 +738,10 @@ listblock(Paragraph *top, int trim, MMIOT *f)
 {
     ParagraphRoot d = { 0, 0 };
     Paragraph *p;
-    Line *q = top->text, *text;
-    Line *label;
-    int para = 0;
+    Line *q = top->text, *text, *label;
+    int isdl = (top->typ == DL),
+	para = 0,
+	ltype;
 
     while (( text = q )) {
 	if ( top->typ == DL ) {
@@ -710,7 +765,8 @@ listblock(Paragraph *top, int trim, MMIOT *f)
 
 	if ( para && (top->typ != DL) && p->down ) p->down->align = PARA;
 
-	if ( !(q = skipempty(text)) || (islist(q, &trim) == 0) )
+	if ( !(q = skipempty(text)) || ((ltype = islist(q, &trim)) == 0)
+				    || (isdl != (ltype == DL)) )
 	    break;
 
 	if ( para = (q != text) ) {
@@ -840,6 +896,64 @@ consume(Line *ptr, int *eaten)
 }
 
 
+/*
+ * top-level compilation; break the document into
+ * style, html, and source blocks with footnote links
+ * weeded out.
+ */
+static Paragraph *
+compile_document(Line *ptr, MMIOT *f)
+{
+    ParagraphRoot d = { 0, 0 };
+    ANCHOR(Line) source = { 0, 0 };
+    Paragraph *p = 0;
+    struct kw *tag;
+    int eaten;
+
+    while ( ptr ) {
+	if ( !(f->flags & DENY_HTML) && (tag = isopentag(ptr)) ) {
+	    /* If we encounter a html/style block, compile and save all
+	     * of the cached source BEFORE processing the html/style.
+	     */
+	    if ( T(source) ) {
+		E(source)->next = 0;
+		p = Pp(&d, 0, SOURCE);
+		p->down = compile(T(source), 1, f);
+		T(source) = E(source) = 0;
+	    }
+	    p = Pp(&d, ptr, strcmp(tag->id, "STYLE") == 0 ? STYLE : HTML);
+	    if ( strcmp(tag->id, "!--") == 0 )
+		ptr = comment(p);
+	    else
+		ptr = htmlblock(p, tag);
+	}
+	else if ( isfootnote(ptr) ) {
+	    /* footnotes, like cats, sleep anywhere; pull them
+	     * out of the input stream and file them away for
+	     * later processing
+	     */
+	    ptr = consume(addfootnote(ptr, f), &eaten);
+	}
+	else {
+	    /* source; cache it up to wait for eof or the
+	     * next html/style block
+	     */
+	    ATTACH(source,ptr);
+	    ptr = ptr->next;
+	}
+    }
+    if ( T(source) ) {
+	/* if there's any cached source at EOF, compile
+	 * it now.
+	 */
+	E(source)->next = 0;
+	p = Pp(&d, 0, SOURCE);
+	p->down = compile(T(source), 1, f);
+    }
+    return T(d);
+}
+
+
 /*
  * break a collection of markdown input into
  * blocks of lists, code, html, and text to
@@ -850,22 +964,15 @@ compile(Line *ptr, int toplevel, MMIOT *f)
 {
     ParagraphRoot d = { 0, 0 };
     Paragraph *p = 0;
-    struct kw *tag;
     Line *r;
     int para = toplevel;
+    int blocks = 0;
     int hdr_type, list_type, indent;
 
     ptr = consume(ptr, &para);
 
     while ( ptr ) {
-	if ( toplevel && !(f->flags & DENY_HTML) && (tag = isopentag(ptr)) ) {
-	    p = Pp(&d, ptr, strcmp(tag->id, "STYLE") == 0 ? STYLE : HTML);
-	    if ( strcmp(tag->id, "!--") == 0 )
-		ptr = comment(p);
-	    else
-		ptr = htmlblock(p, tag);
-	}
-	else if ( iscode(ptr) ) {
+	if ( iscode(ptr) ) {
 	    p = Pp(&d, ptr, CODE);
 	    
 	    if ( f->flags & MKD_1_COMPAT) {
@@ -897,9 +1004,9 @@ compile(Line *ptr, int toplevel, MMIOT *f)
 	    p = Pp(&d, ptr, HDR);
 	    ptr = headerblock(p, hdr_type);
 	}
-	else if ( toplevel && (isfootnote(ptr)) ) {
-	    ptr = consume(addfootnote(ptr, f), &para);
-	    continue;
+	else if ( istable(ptr) && !(f->flags & (STRICT|NOTABLES)) ) {
+	    p = Pp(&d, ptr, TABLE);
+	    ptr = tableblock(p);
 	}
 	else {
 	    p = Pp(&d, ptr, MARKUP);
@@ -909,7 +1016,8 @@ compile(Line *ptr, int toplevel, MMIOT *f)
 	if ( (para||toplevel) && !p->align )
 	    p->align = PARA;
 
-	para = toplevel;
+	blocks++;
+	para = toplevel || (blocks > 1);
 	ptr = consume(ptr, &para);
 
 	if ( para && !p->align )
@@ -960,7 +1068,7 @@ mkd_compile(Document *doc, int flags)
 
     initialize();
 
-    doc->code = compile(T(doc->content), 1, doc->ctx);
+    doc->code = compile_document(T(doc->content), doc->ctx);
     qsort(T(*doc->ctx->footnotes), S(*doc->ctx->footnotes),
 		        sizeof T(*doc->ctx->footnotes)[0],
 			           (stfu)__mkd_footsort);