chaptastic-rdiscount 1.4.1

data/ext/cstring.h

@@ -0,0 +1,74 @@
+/* two template types:  STRING(t) which defines a pascal-style string
+ * of element (t) [STRING(char) is the closest to the pascal string],
+ * and ANCHOR(t) which defines a baseplate that a linked list can be
+ * built up from.   [The linked list /must/ contain a ->next pointer
+ * for linking the list together with.]
+ */
+#ifndef _CSTRING_D
+#define _CSTRING_D
+
+#include <string.h>
+#include <stdlib.h>
+
+#include "amalloc.h"
+
+/* expandable Pascal-style string.
+ */
+#define STRING(type)	struct { type *text; int size, alloc; }
+
+#define CREATE(x)	T(x) = (void*)(S(x) = (x).alloc = 0)
+#define EXPAND(x)	(S(x)++)[(S(x) < (x).alloc) \
+			    ? (T(x)) \
+			    : (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)	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])), \
+		S(t) -= (sz)) : -1
+
+#define RESERVE(x, sz)	T(x) = ((x).alloc > S(x) + (sz) \
+			    ? T(x) \
+			    : T(x) \
+				? realloc(T(x), sizeof T(x)[0] * ((x).alloc = 100+(sz)+S(x))) \
+				: malloc(sizeof T(x)[0] * ((x).alloc = 100+(sz)+S(x))))
+#define SUFFIX(t,p,sz)	\
+	    memcpy(((S(t) += (sz)) - (sz)) + \
+		    (T(t) = T(t) ? realloc(T(t), sizeof T(t)[0] * ((t).alloc += sz)) \
+				 : malloc(sizeof T(t)[0] * ((t).alloc += sz))), \
+		    (p), sizeof(T(t)[0])*(sz))
+
+#define PREFIX(t,p,sz)	\
+	    RESERVE( (t), (sz) ); \
+	    if ( S(t) ) { memmove(T(t)+(sz), T(t), S(t)); } \
+	    memcpy( T(t), (p), (sz) ); \
+	    S(t) += (sz)
+
+/* reference-style links (and images) are stored in an array
+ */
+#define T(x)		(x).text
+#define S(x)		(x).size
+#define ALLOCATED(x)	(x).alloc
+
+/* abstract anchor type that defines a list base
+ * with a function that attaches an element to
+ * the end of the list.
+ *
+ * the list base field is named .text so that the T()
+ * macro will work with it.
+ */
+#define ANCHOR(t)	struct { t *text, *end; }
+
+#define ATTACH(t, p)	( (t).text ?( ((t).end->next = (p)), ((t).end = (p)) ) \
+				   :( ((t).text = (t).end = (p)) ) )
+
+typedef STRING(char) Cstring;
+
+extern void Csputc(int, Cstring *);
+extern int Csprintf(Cstring *, char *, ...);
+extern int Cswrite(Cstring *, char *, int);
+extern void Csreparse(Cstring *, char *, int, int);
+
+#endif/*_CSTRING_D*/

data/ext/docheader.c

@@ -0,0 +1,43 @@
+/*
+ * docheader -- get values from the document header
+ *
+ * Copyright (C) 2007 David L Parsons.
+ * The redistribution terms are provided in the COPYRIGHT file that must
+ * be distributed with this source code.
+ */
+#include "config.h"
+#include <stdio.h>
+#include <stdlib.h>
+#include <ctype.h>
+
+#include "cstring.h"
+#include "markdown.h"
+#include "amalloc.h"
+
+#define afterdle(t)	(T((t)->text) + (t)->dle)
+
+char *
+mkd_doc_title(Document *doc)
+{
+    if ( doc && doc->headers )
+	return afterdle(doc->headers);
+    return 0;
+}
+
+
+char *
+mkd_doc_author(Document *doc)
+{
+    if ( doc && doc->headers && doc->headers->next )
+	return afterdle(doc->headers->next);
+    return 0;
+}
+
+
+char *
+mkd_doc_date(Document *doc)
+{
+    if ( doc && doc->headers && doc->headers->next && doc->headers->next->next )
+	return afterdle(doc->headers->next->next);
+    return 0;
+}

data/ext/dumptree.c

@@ -0,0 +1,147 @@
+/* markdown: a C implementation of John Gruber's Markdown markup language.
+ *
+ * Copyright (C) 2007 David L Parsons.
+ * The redistribution terms are provided in the COPYRIGHT file that must
+ * be distributed with this source code.
+ */
+#include <stdio.h>
+#include "markdown.h"
+#include "cstring.h"
+#include "amalloc.h"
+
+struct frame {
+    int indent;
+    char c;
+};
+
+typedef STRING(struct frame) Stack;
+
+static char *
+Pptype(int typ)
+{
+    switch (typ) {
+    case WHITESPACE: return "whitespace";
+    case CODE      : return "code";
+    case QUOTE     : return "quote";
+    case MARKUP    : return "markup";
+    case HTML      : return "html";
+    case DL        : return "dl";
+    case UL        : return "ul";
+    case OL        : return "ol";
+    case LISTITEM  : return "item";
+    case HDR       : return "header";
+    case HR        : return "HR";
+    default        : return "mystery node!";
+    }
+}
+
+static void
+pushpfx(int indent, char c, Stack *sp)
+{
+    struct frame *q = &EXPAND(*sp);
+
+    q->indent = indent;
+    q->c = c;
+}
+
+
+static void
+poppfx(Stack *sp)
+{
+    S(*sp)--;
+}
+
+
+static void
+changepfx(Stack *sp, char c)
+{
+    char ch;
+
+    if ( !S(*sp) ) return;
+
+    ch = T(*sp)[S(*sp)-1].c;
+
+    if ( ch == '+' || ch == '|' )
+	T(*sp)[S(*sp)-1].c = c;
+}
+
+
+static void
+printpfx(Stack *sp, FILE *f)
+{
+    int i;
+    char c;
+
+    if ( !S(*sp) ) return;
+
+    c = T(*sp)[S(*sp)-1].c;
+
+    if ( c == '+' || c == '-' ) {
+	fprintf(f, "--%c", c);
+	T(*sp)[S(*sp)-1].c = (c == '-') ? ' ' : '|';
+    }
+    else
+	for ( i=0; i < S(*sp); i++ ) {
+	    if ( i )
+		fprintf(f, "  ");
+	    fprintf(f, "%*s%c", T(*sp)[i].indent + 2, " ", T(*sp)[i].c);
+	    if ( T(*sp)[i].c == '`' )
+		T(*sp)[i].c = ' ';
+	}
+    fprintf(f, "--");
+}
+
+
+static void
+dumptree(Paragraph *pp, Stack *sp, FILE *f)
+{
+    int count;
+    Line *p;
+    int d;
+    static char *Begin[] = { 0, "P", "center" };
+
+    while ( pp ) {
+	if ( !pp->next )
+	    changepfx(sp, '`');
+	printpfx(sp, f);
+
+	d = fprintf(f, "[%s", Pptype(pp->typ));
+	if ( pp->align )
+	    d += fprintf(f, ", <%s>", Begin[pp->align]);
+
+	for (count=0, p=pp->text; p; ++count, (p = p->next) )
+	    ;
+
+	if ( count )
+	    d += fprintf(f, ", %d line%s", count, (count==1)?"":"s");
+
+	d += fprintf(f, "]");
+
+	if ( pp->down ) {
+	    pushpfx(d, pp->down->next ? '+' : '-', sp);
+	    dumptree(pp->down, sp, f);
+	    poppfx(sp);
+	}
+	else fputc('\n', f);
+	pp = pp->next;
+    }
+}
+
+
+int
+mkd_dump(Document *doc, FILE *out, int flags, char *title)
+{
+    Stack stack;
+
+    if (mkd_compile(doc, flags) ) {
+
+	CREATE(stack);
+	pushpfx(fprintf(out, "%s", title), doc->code->next ? '+' : '-', &stack);
+	dumptree(doc->code, &stack, out);
+	DELETE(stack);
+
+	mkd_cleanup(doc);
+	return 0;
+    }
+    return -1;
+}

data/ext/extconf.rb

@@ -0,0 +1,8 @@
+require 'mkmf'
+
+dir_config('rdiscount')
+
+HAVE_RANDOM = have_func('random')
+HAVE_SRANDOM = have_func('srandom')
+
+create_makefile('rdiscount')

data/ext/generate.c

@@ -0,0 +1,1457 @@
+/* markdown: a C implementation of John Gruber's Markdown markup language.
+ *
+ * Copyright (C) 2007 David L Parsons.
+ * The redistribution terms are provided in the COPYRIGHT file that must
+ * be distributed with this source code.
+ */
+#include <stdio.h>
+#include <string.h>
+#include <stdarg.h>
+#include <stdlib.h>
+#include <time.h>
+#include <ctype.h>
+
+#include "config.h"
+
+#include "cstring.h"
+#include "markdown.h"
+#include "amalloc.h"
+
+typedef int (*stfu)(const void*,const void*);
+
+
+/* forward declarations */
+static int iscodeblock(MMIOT*);
+static void code(int, MMIOT*);
+static void text(MMIOT *f);
+static Paragraph *display(Paragraph*, MMIOT*);
+
+/* externals from markdown.c */
+int __mkd_footsort(Footnote *, Footnote *);
+
+/*
+ * push text into the generator input buffer
+ */
+static void
+push(char *bfr, int size, MMIOT *f)
+{
+    while ( size-- > 0 )
+	EXPAND(f->in) = *bfr++;
+}
+
+
+/* look <i> characters ahead of the cursor.
+ */
+static int
+peek(MMIOT *f, int i)
+{
+
+    i += (f->isp-1);
+
+    return (i >= 0) && (i < S(f->in)) ? T(f->in)[i] : EOF;
+}
+
+
+/* pull a byte from the input buffer
+ */
+static int
+pull(MMIOT *f)
+{
+    return ( f->isp < S(f->in) ) ? T(f->in)[f->isp++] : EOF;
+}
+
+
+/* return a pointer to the current position in the input buffer.
+ */
+static char*
+cursor(MMIOT *f)
+{
+    return T(f->in) + f->isp;
+}
+
+
+static int
+isthisspace(MMIOT *f, int i)
+{
+    int c = peek(f, i);
+
+    return isspace(c) || (c == EOF);
+}
+
+
+static int
+isthisalnum(MMIOT *f, int i)
+{
+    int c = peek(f, i);
+
+    return (c != EOF) && isalnum(c);
+}
+
+
+static int
+isthisnonword(MMIOT *f, int i)
+{
+    return isthisspace(f, i) || ispunct(peek(f,i));
+}
+
+
+/* return/set the current cursor position
+ */
+#define mmiotseek(f,x)	(f->isp = x)
+#define mmiottell(f)	(f->isp)
+
+
+/* move n characters forward ( or -n characters backward) in the input buffer.
+ */
+static void
+shift(MMIOT *f, int i)
+{
+    if (f->isp + i >= 0 )
+	f->isp += i;
+}
+
+
+/* Qchar()
+ */
+static void
+Qchar(int c, MMIOT *f)
+{
+    block *cur;
+    
+    if ( S(f->Q) == 0 ) {
+	cur = &EXPAND(f->Q);
+	memset(cur, 0, sizeof *cur);
+	cur->b_type = bTEXT;
+    }
+    else
+	cur = &T(f->Q)[S(f->Q)-1];
+
+    EXPAND(cur->b_text) = c;
+    
+}
+
+
+/* Qstring()
+ */
+static void
+Qstring(char *s, MMIOT *f)
+{
+    while (*s)
+	Qchar(*s++, f);
+}
+
+
+/* Qwrite()
+ */
+static void
+Qwrite(char *s, int size, MMIOT *f)
+{
+    while (size-- > 0)
+	Qchar(*s++, f);
+}
+
+
+/* Qprintf()
+ */
+static void
+Qprintf(MMIOT *f, char *fmt, ...)
+{
+    char bfr[80];
+    va_list ptr;
+
+    va_start(ptr,fmt);
+    vsnprintf(bfr, sizeof bfr, fmt, ptr);
+    va_end(ptr);
+    Qstring(bfr, f);
+}
+
+
+/* Qem()
+ */
+static void
+Qem(MMIOT *f, char c, int count)
+{
+    block *p = &EXPAND(f->Q);
+
+    memset(p, 0, sizeof *p);
+    p->b_type = (c == '*') ? bSTAR : bUNDER;
+    p->b_char = c;
+    p->b_count = count;
+
+    memset(&EXPAND(f->Q), 0, sizeof(block));
+}
+
+
+/* empair()
+ */
+static int
+empair(MMIOT *f, int go, int level)
+{
+    
+    int i;
+    block *begin, *p;
+
+    begin = &T(f->Q)[go];
+    for (i=go+1; i < S(f->Q); i++) {
+	p = &T(f->Q)[i];
+
+	if ( (p->b_type != bTEXT) && (p->b_count <= 0) )
+	    break;
+	
+	if ( p->b_type == begin->b_type ) {
+	    if ( p->b_count == level )	/* exact match */
+		return i-go;
+
+	    if ( p->b_count > 2 )	/* fuzzy match */
+		return i-go;
+	}
+    }
+    return EOF;
+}
+
+
+
+static struct emtags {
+    char open[10];
+    char close[10];
+    int size;
+} emtags[] = {  { "<em>" , "</em>", 5 }, { "<strong>", "</strong>", 9 } };
+
+
+static void
+emclose(Cstring *s, int level)
+{
+    PREFIX(*s, emtags[level-1].close, emtags[level-1].size);
+}
+
+
+static void
+emopen(Cstring *s, int level)
+{
+    SUFFIX(*s, emtags[level-1].open, emtags[level-1].size-1);
+}
+
+
+/* emmatch()
+ */
+static void
+emmatch(MMIOT *f, int go)
+{
+    block *start = &T(f->Q)[go], *end;
+    int e, e2, i, match;
+
+    while ( start->b_count ) {
+	switch (start->b_count) {
+	case 2: e = empair(f,go,match=2);
+		if ( e != EOF ) break;
+	case 1: e = empair(f,go,match=1); break;
+	default:
+	    e = empair(f,go,1);
+	    e2= empair(f,go,2);
+
+	    if ( e == EOF || ((e2 != EOF) && (e2 >= e)) ) {
+		e = e2;
+		match = 2;
+	    } 
+	    else
+		match = 1;
+	}
+	if ( e != EOF ) {
+	    end = &T(f->Q)[go+e];
+	    emclose(&end->b_post, match);
+	    emopen(&start->b_text, match);
+	    end->b_count -= match;
+	}
+	else {
+	    for (i=0; i < match; i++)
+		EXPAND(start->b_text) = start->b_char;
+	}
+	
+	start->b_count -= match;
+    }
+}
+
+
+/* ___mkd_emblock()
+ */
+void
+___mkd_emblock(MMIOT *f)
+{
+    int i;
+    block *p;
+
+    for (i=0; i < S(f->Q); i++) {
+	p = &T(f->Q)[i];
+	
+	if ( p->b_type != bTEXT ) emmatch(f, i);
+
+	if ( S(p->b_post) ) { SUFFIX(f->out, T(p->b_post), S(p->b_post));
+			      DELETE(p->b_post); }
+	if ( S(p->b_text) ) { SUFFIX(f->out, T(p->b_text), S(p->b_text));
+			      DELETE(p->b_text); }
+    }
+    S(f->Q) = 0;
+}
+
+
+/* generate html from a markup fragment
+ */
+void
+___mkd_reparse(char *bfr, int size, int flags, MMIOT *f)
+{
+    MMIOT sub;
+
+    ___mkd_initmmiot(&sub, f->footnotes);
+    
+    sub.flags = f->flags | flags;
+    sub.base = f->base;
+
+    push(bfr, size, &sub);
+    EXPAND(sub.in) = 0;
+    S(sub.in)--;
+    
+    text(&sub);
+    ___mkd_emblock(&sub);
+    
+    Qwrite(T(sub.out), S(sub.out), f);
+
+    ___mkd_freemmiot(&sub, f->footnotes);
+}
+
+
+/*
+ * write out a url, escaping problematic characters
+ */
+static void
+puturl(char *s, int size, MMIOT *f)
+{
+    unsigned char c;
+
+    while ( size-- > 0 ) {
+	c = *s++;
+
+	if ( c == '&' )
+	    Qstring("&amp;", f);
+	else if ( c == '<' )
+	    Qstring("&lt;", f);
+	else if ( isalnum(c) || ispunct(c) )
+	    Qchar(c, f);
+	else
+	    Qprintf(f, "%%%02X", c);
+    }
+}
+
+
+/* advance forward until the next character is not whitespace
+ */
+static int
+eatspace(MMIOT *f)
+{
+    int c;
+
+    for ( ; ((c=peek(f, 1)) != EOF) && isspace(c); pull(f) )
+	;
+    return c;
+}
+
+
+/* (match (a (nested (parenthetical (string.)))))
+ */
+static int
+parenthetical(int in, int out, MMIOT *f)
+{
+    int size, indent, c;
+
+    for ( indent=1,size=0; indent; size++ ) {
+	if ( (c = pull(f)) == EOF )
+	    return EOF;
+	else if ( c == in )
+	    ++indent;
+	else if ( (c == '\\') && (peek(f,1) == out) ) {
+	    ++size;
+	    pull(f);
+	}
+	else if ( c == out )
+	    --indent;
+    }
+    return size ? (size-1) : 0;
+}
+
+
+/* extract a []-delimited label from the input stream.
+ */
+static int
+linkylabel(MMIOT *f, Cstring *res)
+{
+    char *ptr = cursor(f);
+    int size;
+
+    if ( (size = parenthetical('[',']',f)) != EOF ) {
+	T(*res) = ptr;
+	S(*res) = size;
+	return 1;
+    }
+    return 0;
+}
+
+
+/* see if the quote-prefixed linky segment is actually a title.
+ */
+static int
+linkytitle(MMIOT *f, char quote, Footnote *ref)
+{
+    int whence = mmiottell(f);
+    char *title = cursor(f);
+    char *e;
+    register c;
+
+    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;
+	    }
+	}
+    }
+    mmiotseek(f, whence);
+    return 0;
+}
+
+
+/* extract a =HHHxWWW size from the input stream
+ */
+static int
+linkysize(MMIOT *f, Footnote *ref)
+{
+    int height=0, width=0;
+    int whence = mmiottell(f);
+    int c;
+
+    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 ( isspace(c) )
+		c = eatspace(f);
+
+	    if ( (c == ')') || ((c == '\'' || c == '"') && linkytitle(f, c, ref)) ) {
+		ref->height = height;
+		ref->width  = width;
+		return 1;
+	    }
+	}
+    }
+    mmiotseek(f, whence);
+    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 ')', a quote,
+ * or (if image) a '='
+ */
+static int
+linkyurl(MMIOT *f, int image, Footnote *p)
+{
+    int c;
+    int mayneedtotrim=0;
+
+    if ( (c = eatspace(f)) == EOF )
+	return 0;
+
+    if ( c == '<' ) {
+	pull(f);
+	mayneedtotrim=1;
+    }
+
+    while ( (c = peek(f,1)) != ')' ) {
+	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)) )
+	    pull(f);
+	EXPAND(p->link) = 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;
+}
+
+
+
+/* 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])
+
+
+static int
+isautoprefix(char *text)
+{
+    int i;
+
+    for (i=0; i < NRPROTOCOLS; i++)
+	if ( strncasecmp(text, protocol[i].name, protocol[i].nlen) == 0 )
+	    return 1;
+    return 0;
+}
+
+
+/*
+ * all the tag types that linkylinky can produce are
+ * defined by this structure. 
+ */
+typedef struct linkytype {
+    char      *pat;
+    int      szpat;
+    char *link_pfx;	/* tag prefix and link pointer  (eg: "<a href="\"" */
+    char *link_sfx;	/* link suffix			(eg: "\""          */
+    int        WxH;	/* this tag allows width x height arguments */
+    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, IS_URL };
+static linkytype linkt  = { 0, 0, "<a href=\"", "\"",
+                             0, ">", "</a>", DENY_A, IS_URL };
+
+/*
+ * pseudo-protocols for [][];
+ *
+ * id: generates <a id="link">tag</a>
+ * class: generates <span class="link">tag</span>
+ * raw: just dump the link without any processing
+ */
+static linkytype specials[] = {
+    { "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])
+
+/* see if t contains one of our pseudo-protocols.
+ */
+static linkytype *
+pseudo(Cstring t)
+{
+    int i;
+    linkytype *r;
+
+    for ( i=0; i < NR(specials); i++ ) {
+	r = &specials[i];
+	if ( (S(t) > r->szpat) && (strncasecmp(T(t), r->pat, r->szpat) == 0) )
+	    return r;
+    }
+    return 0;
+}
+
+
+/* print out a linky (or fail if it's Not Allowed)
+ */
+static int
+linkyformat(MMIOT *f, Cstring text, int image, Footnote *ref)
+{
+    linkytype *tag;
+
+    if ( image )
+	tag = &imaget;
+    else if ( (f->flags & NO_PSEUDO_PROTO) || (tag = pseudo(ref->link)) == 0 ) {
+	/* if SAFELINK, only accept links that are local or
+	 * a well-known protocol
+	 */
+	if ( (f->flags & SAFELINK) && T(ref->link)
+				   && (T(ref->link)[0] != '/')
+				   && !isautoprefix(T(ref->link)) )
+	    return 0;
+	tag = &linkt;
+    }
+
+    if ( f->flags & tag->flags )
+	return 0;
+
+    if ( tag->link_pfx ) {
+	Qstring(tag->link_pfx, f);
+	
+	if ( tag->kind & IS_URL ) {
+	    if ( f->base && T(ref->link) && (T(ref->link)[tag->szpat] == '/') )
+		puturl(f->base, strlen(f->base), f);
+	    puturl(T(ref->link) + tag->szpat, S(ref->link) - tag->szpat, f);
+	}
+	else
+	    ___mkd_reparse(T(ref->link) + tag->szpat, S(ref->link) - tag->szpat, INSIDE_TAG, f);
+	
+	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 ( S(ref->title) ) {
+	    Qstring(" title=\"", f);
+	    ___mkd_reparse(T(ref->title), S(ref->title), INSIDE_TAG, f);
+	    Qchar('"', f);
+	}
+
+	Qstring(tag->text_pfx, f);
+	___mkd_reparse(T(text), S(text), tag->flags, f);
+	Qstring(tag->text_sfx, f);
+    }
+    else
+	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;
+}
+
+
+/* write a character to output, doing text escapes ( & -> &amp;,
+ *                                          > -> &gt; < -> &lt; )
+ */
+static void
+cputc(int c, MMIOT *f)
+{
+    switch (c) {
+    case '&':   Qstring("&amp;", f); break;
+    case '>':   Qstring("&gt;", f); break;
+    case '<':   Qstring("&lt;", f); break;
+    default :   Qchar(c, f); break;
+    }
+}
+
+ 
+/*
+ * convert an email address to a string of nonsense
+ */
+static void
+mangle(char *s, int len, MMIOT *f)
+{
+    while ( len-- > 0 ) {
+	Qstring("&#", f);
+	Qprintf(f, COINTOSS() ? "x%02x;" : "%02d;", *((unsigned char*)(s++)) );
+    }
+}
+
+
+/* before letting a tag through, validate against
+ * DENY_A and DENY_IMG
+ */
+static int
+forbidden_tag(MMIOT *f)
+{
+    int c = toupper(peek(f, 1));
+
+    if ( f->flags & DENY_HTML )
+	return 1;
+
+    if ( c == 'A' && (f->flags & DENY_A) && !isthisalnum(f,2) )
+	return 1;
+    if ( c == 'I' && (f->flags & DENY_IMG)
+		  && strncasecmp(cursor(f)+1, "MG", 2) == 0
+		  && !isthisalnum(f,4) )
+	return 1;
+    return 0;
+}
+
+
+/* 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);
+	Qstring("\">", f);
+	puturl(text,size,f);
+	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
+ * link, we also need to know if it's an email address because if it
+ * is we need to mangle it in our futile attempt to cut down on the
+ * spaminess of the rendered page.
+ */
+static int
+maybe_tag_or_link(MMIOT *f)
+{
+    int c, size;
+    int maybetag = 1;
+
+    if ( f->flags & INSIDE_TAG )
+	return 0;
+
+    for ( size=0; (c = peek(f, size+1)) != '>'; size++) {
+	if ( c == EOF )
+	    return 0;
+	else if ( isspace(c) )
+	    break;
+	else if ( ! (c == '/' || isalnum(c) ) )
+	    maybetag=0;
+    }
+
+    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;
+}
+
+
+/* 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;
+
+    /* greedily scan forward for the end of a legitimate link.
+     */
+    for ( size=0; (c=peek(f, size+1)) != EOF; size++ )
+	if ( !(isalnum(c) || strchr("/:._%~@", c)) )
+	    break;
+
+    if ( (size > 1) && process_possible_link(f, size) ) {
+	shift(f, size);
+	return 1;
+    }
+    return 0;
+}
+
+
+/* smartyquote code that's common for single and double quotes
+ */
+static int
+smartyquote(int *flags, char typeofquote, MMIOT *f)
+{
+    int bit = (typeofquote == 's') ? 0x01 : 0x02;
+
+    if ( bit & (*flags) ) {
+	if ( isthisnonword(f,1) ) {
+	    Qprintf(f, "&r%cquo;", typeofquote);
+	    (*flags) &= ~bit;
+	    return 1;
+	}
+    }
+    else if ( isthisnonword(f,-1) && peek(f,1) != EOF ) {
+	Qprintf(f, "&l%cquo;", typeofquote);
+	(*flags) |= bit;
+	return 1;
+    }
+    return 0;
+}
+
+
+static int
+islike(MMIOT *f, char *s)
+{
+    int len;
+    int i;
+
+    if ( s[0] == '<' ) {
+	if ( !isthisnonword(f, -1) )
+	    return 0;
+       ++s;
+    }
+
+    if ( !(len = strlen(s)) )
+	return 0;
+
+    if ( s[len-1] == '>' ) {
+	if ( !isthisnonword(f,len-1) )
+	    return 0;
+	len--;
+    }
+
+    for (i=1; i < len; i++)
+	if (tolower(peek(f,i)) != s[i])
+	    return 0;
+    return 1;
+}
+
+
+static struct smarties {
+    char c0;
+    char *pat;
+    char *entity;
+    int shift;
+} smarties[] = {
+    { '\'', "'s>",      "rsquo",  0 },
+    { '\'', "'t>",      "rsquo",  0 },
+    { '-',  "--",       "mdash",  1 },
+    { '-',  "<->",      "ndash",  0 },
+    { '.',  "...",      "hellip", 2 },
+    { '.',  ". . .",    "hellip", 4 },
+    { '(',  "(c)",      "copy",   2 },
+    { '(',  "(r)",      "reg",    2 },
+    { '(',  "(tm)",     "trade",  3 },
+    { '3',  "<3/4>",    "frac34", 2 },
+    { '3',  "<3/4ths>", "frac34", 2 },
+    { '1',  "<1/2>",    "frac12", 2 },
+    { '1',  "<1/4>",    "frac14", 2 },
+    { '1',  "<1/4th>",  "frac14", 2 },
+    { '&',  "&#0;",      0,       3 },
+} ;
+#define NRSMART ( sizeof smarties / sizeof smarties[0] )
+
+
+/* Smarty-pants-style chrome for quotes, -, ellipses, and (r)(c)(tm)
+ */
+static int
+smartypants(int c, int *flags, MMIOT *f)
+{
+    int i;
+
+    if ( f->flags & (DENY_SMARTY|INSIDE_TAG) )
+	return 0;
+
+    for ( i=0; i < NRSMART; i++)
+	if ( (c == smarties[i].c0) && islike(f, smarties[i].pat) ) {
+	    if ( smarties[i].entity )
+		Qprintf(f, "&%s;", smarties[i].entity);
+	    shift(f, smarties[i].shift);
+	    return 1;
+	}
+
+    switch (c) {
+    case '<' :  return 0;
+    case '\'':  if ( smartyquote(flags, 's', f) ) return 1;
+		break;
+
+    case '"':	if ( smartyquote(flags, 'd', f) ) return 1;
+		break;
+
+    case '`':   if ( peek(f, 1) == '`' ) {
+		    int j = 2;
+
+		    while ( (c=peek(f,j)) != EOF ) {
+			if ( c == '\\' )
+			    j += 2;
+			else if ( c == '`' )
+			    break;
+			else if ( c == '\'' && peek(f, j+1) == '\'' ) {
+			    Qstring("&ldquo;", f);
+			    ___mkd_reparse(cursor(f)+1, j-2, 0, f);
+			    Qstring("&rdquo;", f);
+			    shift(f,j+1);
+			    return 1;
+			}
+			else ++j;
+		    }
+
+		}
+		break;
+    }
+    return 0;
+} /* smartypants */
+
+
+#define tag_text(f)	(f->flags & INSIDE_TAG)
+
+
+static void
+text(MMIOT *f)
+{
+    int c, j;
+    int rep;
+    int smartyflags = 0;
+
+    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
+			Qchar(c, f);
+		    break;
+
+	case '"':   if ( tag_text(f) )
+			Qstring("&quot;", f);
+		    else
+			Qchar(c, f);
+		    break;
+			
+	case '!':   if ( peek(f,1) == '[' ) {
+			pull(f);
+			if ( tag_text(f) || !linkylinky(1, f) )
+			    Qstring("![", f);
+		    }
+		    else
+			Qchar(c, f);
+		    break;
+	case '[':   if ( tag_text(f) || !linkylinky(0, f) )
+			Qchar(c, f);
+		    break;
+#if SUPERSCRIPT
+	/* A^B -> A<sup>B</sup> */
+	case '^':   if ( (f->flags & (STRICT|INSIDE_TAG)) || isthisspace(f,-1) || isthisspace(f,1) )
+			Qchar(c,f);
+		    else {
+			char *sup = cursor(f);
+			int len = 0;
+			Qstring("<sup>",f);
+			while ( !isthisspace(f,1+len) ) {
+			    ++len;
+			}
+			shift(f,len);
+			___mkd_reparse(sup, len, 0, f);
+			Qstring("</sup>", f);
+		    }
+		    break;
+#endif
+	case '_':
+#if RELAXED_EMPHASIS
+	/* Underscores don't count if they're in the middle of a word */
+		    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
+		    if ( tag_text(f) )
+			Qchar(c, f);
+		    else {
+			for (rep = 1; peek(f,1) == c; pull(f) )
+			    ++rep;
+			Qem(f,c,rep);
+		    }
+		    break;
+	
+	case '`':   if ( tag_text(f) || !iscodeblock(f) )
+			Qchar(c, f);
+		    else {
+			Qstring("<code>", f);
+			if ( peek(f, 1) == '`' ) {
+			    pull(f);
+			    code(2, f);
+			}
+			else
+			    code(1, f);
+			Qstring("</code>", f);
+		    }
+		    break;
+
+	case '\\':  switch ( c = pull(f) ) {
+		    case '&':   Qstring("&amp;", f);
+				break;
+		    case '<':   Qstring("&lt;", f);
+				break;
+		    case '\\':
+		    case '>': case '#': case '.': case '-':
+		    case '+': case '{': case '}': case ']':
+		    case '(': case ')': case '"': case '\'':
+		    case '!': case '[': case '*': case '_':
+		    case '`':	Qchar(c, f);
+				break;
+		    default:
+				Qchar('\\', f);
+				if ( c != EOF )
+				    shift(f,-1);
+				break;
+		    }
+		    break;
+
+	case '<':   if ( !maybe_tag_or_link(f) )
+			Qstring("&lt;", f);
+		    break;
+
+	case '&':   j = (peek(f,1) == '#' ) ? 2 : 1;
+		    while ( isthisalnum(f,j) )
+			++j;
+
+		    if ( peek(f,j) != ';' )
+			Qstring("&amp;", f);
+		    else
+			Qchar(c, f);
+		    break;
+
+	default:    Qchar(c, f);
+		    break;
+	}
+    }
+    /* truncate the input string after we've finished processing it */
+    S(f->in) = f->isp = 0;
+} /* text */
+
+
+static int
+iscodeblock(MMIOT *f)
+{
+    int i=1, single = 1, c;
+    
+    if ( peek(f,i) == '`' ) {
+	single=0;
+	i++;
+    }
+    while ( (c=peek(f,i)) != EOF ) {
+	if ( (c == '`') && (single || peek(f,i+1) == '`') )
+	    return 1;
+	else if ( c == '\\' )
+	    i++;
+	i++;
+    }
+    return 0;
+    
+}
+
+static int
+endofcode(int escape, int offset, MMIOT *f)
+{
+    switch (escape) {
+    case 2: if ( peek(f, offset+1) == '`' ) {
+		shift(f,1);
+    case 1:     shift(f,offset);
+		return 1;
+	    }
+    default:return 0;
+    }
+}
+
+
+/* the only characters that have special meaning in a code block are
+ * `<' and `&' , which are /always/ expanded to &lt; and &amp;
+ */
+static void
+code(int escape, MMIOT *f)
+{
+    int c;
+
+    if ( escape && (peek(f,1) == ' ') )
+	shift(f,1);
+
+    while ( (c = pull(f)) != EOF ) {
+	switch (c) {
+	case ' ':   if ( peek(f,1) == '`' && endofcode(escape, 1, f) )
+			return;
+		    Qchar(c, f);
+		    break;
+
+	case '`':   if ( endofcode(escape, 0, f) )
+			return;
+		    Qchar(c, f);
+		    break;
+
+	case '\\':  cputc(c, f);
+		    if ( peek(f,1) == '>' || (c = pull(f)) == EOF )
+			break;
+	
+	default:    cputc(c, f);
+		    break;
+	}
+    }
+} /* code */
+
+
+/* print a header block
+ */
+static void
+printheader(Paragraph *pp, MMIOT *f)
+{
+    Qprintf(f, "<h%d", pp->hnumber);
+    if ( f->flags & TOC ) {
+	Qprintf(f, " id=\"", pp->hnumber);
+	mkd_string_to_anchor(T(pp->text->text), S(pp->text->text), Qchar, f);
+	Qchar('"', f);
+    }
+    Qchar('>', f);
+    push(T(pp->text->text), S(pp->text->text), f);
+    text(f);
+    Qprintf(f, "</h%d>", pp->hnumber);
+}
+
+
+static int
+printblock(Paragraph *pp, MMIOT *f)
+{
+    Line *t = pp->text;
+    static char *Begin[] = { "", "<p>", "<center>"  };
+    static char *End[]   = { "", "</p>","</center>" };
+
+    while (t) {
+	if ( S(t->text) ) {
+	    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("\003\n", 2, f);
+	    }
+	    else {
+		___mkd_tidy(&t->text);
+		push(T(t->text), S(t->text), f);
+		if ( t->next )
+		    push("\n", 1, f);
+	    }
+	}
+	t = t->next;
+    }
+    Qstring(Begin[pp->align], f);
+    text(f);
+    Qstring(End[pp->align], f);
+    return 1;
+}
+
+
+static void
+printcode(Line *t, MMIOT *f)
+{
+    int blanks;
+
+    for ( blanks = 0; t ; t = t->next )
+	if ( S(t->text) > t->dle ) {
+	    while ( blanks ) {
+		push("\n", 1, f);
+		--blanks;
+	    }
+	    push(T(t->text), S(t->text), f);
+	    push("\n", 1, f);
+	}
+	else blanks++;
+
+    Qstring("<pre><code>", f);
+    code(0, f);
+    Qstring("</code></pre>", f);
+}
+
+
+static void
+printhtml(Line *t, MMIOT *f)
+{
+    int blanks;
+    
+    for ( blanks=0; t ; t = t->next )
+	if ( S(t->text) ) {
+	    for ( ; blanks; --blanks ) 
+		Qchar('\n', f);
+
+	    Qwrite(T(t->text), S(t->text), f);
+	    Qchar('\n', f);
+	}
+	else
+	    blanks++;
+}
+
+
+static void
+htmlify(Paragraph *p, char *block, char *arguments, MMIOT *f)
+{
+    ___mkd_emblock(f);
+    if ( block )
+	Qprintf(f, arguments ? "<%s %s>" : "<%s>", block, arguments);
+    ___mkd_emblock(f);
+
+    while (( p = display(p, f) )) {
+	___mkd_emblock(f);
+	Qstring("\n\n", f);
+    }
+
+    if ( block )
+	 Qprintf(f, "</%s>", block);
+    ___mkd_emblock(f);
+}
+
+
+#if DL_TAG_EXTENSION
+static void
+definitionlist(Paragraph *p, MMIOT *f)
+{
+    Line *tag;
+
+    if ( p ) {
+	Qstring("<dl>\n", f);
+
+	for ( ; p ; p = p->next) {
+	    for ( tag = p->text; tag; tag = tag->next ) {
+		Qstring("<dt>", f);
+		___mkd_reparse(T(tag->text), S(tag->text), 0, f);
+		Qstring("</dt>\n", f);
+	    }
+
+	    htmlify(p->down, "dd", p->ident, f);
+	}
+
+	Qstring("</dl>", f);
+    }
+}
+#endif
+
+
+static void
+listdisplay(int typ, Paragraph *p, MMIOT* f)
+{
+    if ( p ) {
+	Qprintf(f, "<%cl", (typ==UL)?'u':'o');
+	if ( typ == AL )
+	    Qprintf(f, " type=a");
+	Qprintf(f, ">\n");
+
+	for ( ; p ; p = p->next ) {
+	    htmlify(p->down, "li", p->ident, f);
+	    Qchar('\n', f);
+	}
+
+	Qprintf(f, "</%cl>\n", (typ==UL)?'u':'o');
+    }
+}
+
+
+/* dump out a Paragraph in the desired manner
+ */
+static Paragraph*
+display(Paragraph *p, MMIOT *f)
+{
+    if ( !p ) return 0;
+    
+    switch ( p->typ ) {
+    case STYLE:
+    case WHITESPACE:
+	break;
+
+    case HTML:
+	printhtml(p->text, f);
+	break;
+	
+    case CODE:
+	printcode(p->text, f);
+	break;
+	
+    case QUOTE:
+	htmlify(p->down, p->ident ? "div" : "blockquote", p->ident, f);
+	break;
+	
+    case UL:
+    case OL:
+    case AL:
+	listdisplay(p->typ, p->down, f);
+	break;
+
+#if DL_TAG_EXTENSION
+    case DL:
+	definitionlist(p->down, f);
+	break;
+#endif
+
+    case HR:
+	Qstring("<hr />", f);
+	break;
+
+    case HDR:
+	printheader(p, f);
+	break;
+
+    default:
+	printblock(p, f);
+	break;
+    }
+    return p->next;
+}
+
+
+/* return a pointer to the compiled markdown
+ * document.
+ */
+int
+mkd_document(Document *p, char **res)
+{
+    if ( p && p->compiled ) {
+	if ( ! p->html ) {
+	    htmlify(p->code, 0, 0, p->ctx);
+	    p->html = 1;
+	}
+
+	*res = T(p->ctx->out);
+	return S(p->ctx->out);
+    }
+    return EOF;
+}
+