bluecloth 2.0.7 → 2.0.9

data/ext/html5.c

@@ -0,0 +1,24 @@
+/* block-level tags for passing html5 blocks through the blender
+ */
+#include "tags.h"
+
+void
+mkd_with_html5_tags()
+{
+    static int populated = 0;
+
+    if ( populated ) return;
+    populated = 1;
+
+    mkd_prepare_tags();
+				 
+    mkd_define_tag("ASIDE", 0);
+    mkd_define_tag("FOOTER", 0);
+    mkd_define_tag("HEADER", 0);
+    mkd_define_tag("HGROUP", 0);
+    mkd_define_tag("NAV", 0);
+    mkd_define_tag("SECTION", 0);
+    mkd_define_tag("ARTICLE", 0);
+
+    mkd_sort_tags();
+}

data/ext/markdown.c

@@ -4,6 +4,8 @@
  * The redistribution terms are provided in the COPYRIGHT file that must
  * be distributed with this source code.
  */
+#include "config.h"
+
 #include <stdio.h>
 #include <string.h>
 #include <stdarg.h>
@@ -11,51 +13,15 @@
 #include <time.h>
 #include <ctype.h>
 
-#include "config.h"
-
 #include "cstring.h"
 #include "markdown.h"
 #include "amalloc.h"
-
-/* block-level tags for passing html blocks through the blender
- */
-struct kw {
-    char *id;
-    int  size;
-    int  selfclose;
-} ;
-
-#define KW(x)	{ x, sizeof(x)-1, 0 }
-#define SC(x)	{ x, sizeof(x)-1, 1 }
-
-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("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("IFRAME"), KW("MAP") };
-#define SZTAGS	(sizeof blocktags / sizeof blocktags[0])
-#define MAXTAG	11 /* sizeof "BLOCKQUOTE" */
+#include "tags.h"
 
 typedef int (*stfu)(const void*,const void*);
 
 typedef ANCHOR(Paragraph) ParagraphRoot;
 
-
-/* case insensitive string sort (for qsort() and bsearch() of block tags)
- */
-static int
-casort(struct kw *a, struct kw *b)
-{
-    if ( a->size != b->size )
-	return a->size - b->size;
-    return strncasecmp(a->id, b->id, b->size);
-}
-
-
 /* case insensitive string sort for Footnote tags.
  */
 int
@@ -135,19 +101,28 @@ ___mkd_tidy(Cstring *t)
 }
 
 
+static struct kw comment = { "!--", 3, 0 };
+
 static struct kw *
 isopentag(Line *p)
 {
     int i=0, len;
-    struct kw key, *ret;
+    char *line;
 
     if ( !p ) return 0;
 
+    line = T(p->text);
     len = S(p->text);
 
-    if ( len < 3 || T(p->text)[0] != '<' )
+    if ( len < 3 || line[0] != '<' )
 	return 0;
 
+    if ( line[1] == '!' && line[2] == '-' && line[3] == '-' )
+	/* comments need special case handling, because
+	 * the !-- doesn't need to end in a whitespace
+	 */
+	return &comment;
+    
     /* find how long the tag is so we can check to see if
      * it's a block-level tag
      */
@@ -156,13 +131,8 @@ isopentag(Line *p)
 		       && !isspace(T(p->text)[i]); ++i )
 	;
 
-    key.id = T(p->text)+1;
-    key.size = i-1;
-    
-    if ( ret = bsearch(&key, blocktags, SZTAGS, sizeof key, (stfu)casort))
-	return ret;
 
-    return 0;
+    return mkd_search_tags(T(p->text)+1, i-1);
 }
 
 
@@ -171,6 +141,8 @@ typedef struct _flo {
     int i;
 } FLO;
 
+#define floindex(x) (x.i)
+
 
 static int
 flogetc(FLO *f)
@@ -186,6 +158,41 @@ flogetc(FLO *f)
 }
 
 
+static void
+splitline(Line *t, int cutpoint)
+{
+    if ( t && (cutpoint < S(t->text)) ) {
+	Line *tmp = calloc(1, sizeof *tmp);
+
+	tmp->next = t->next;
+	t->next = tmp;
+
+	tmp->dle = t->dle;
+	SUFFIX(tmp->text, T(t->text)+cutpoint, S(t->text)-cutpoint);
+	S(t->text) = cutpoint;
+    }
+}
+
+
+static Line *
+commentblock(Paragraph *p)
+{
+    Line *t, *ret;
+    char *end;
+
+    for ( t = p->text; t ; t = t->next) {
+	if ( end = strstr(T(t->text), "-->") ) {
+	    splitline(t, 3 + (end - T(t->text)) );
+	    ret = t->next;
+	    t->next = 0;
+	    return ret;
+	}
+    }
+    return t;
+
+}
+
+
 static Line *
 htmlblock(Paragraph *p, struct kw *tag)
 {
@@ -194,7 +201,10 @@ htmlblock(Paragraph *p, struct kw *tag)
     int c;
     int i, closing, depth=0;
 
-    if ( tag->selfclose || (tag->size >= MAXTAG) ) {
+    if ( tag == &comment )
+	return commentblock(p);
+    
+    if ( tag->selfclose ) {
 	ret = f.t->next;
 	f.t->next = 0;
 	return ret;
@@ -232,6 +242,7 @@ htmlblock(Paragraph *p, struct kw *tag)
 			}
 			if ( !f.t )
 			    return 0;
+			splitline(f.t, floindex(f));
 			ret = f.t->next;
 			f.t->next = 0;
 			return ret;
@@ -244,23 +255,6 @@ htmlblock(Paragraph *p, struct kw *tag)
 }
 
 
-static Line *
-comment(Paragraph *p)
-{
-    Line *t, *ret;
-
-    for ( t = p->text; t ; t = t->next) {
-	if ( strstr(T(t->text), "-->") ) {
-	    ret = t->next;
-	    t->next = 0;
-	    return ret;
-	}
-    }
-    return t;
-
-}
-
-
 /* tables look like
  *   header|header{|header}
  *   ------|------{|......}
@@ -305,8 +299,8 @@ isfootnote(Line *t)
     for ( ++i; i < S(t->text) ; ++i ) {
 	if ( T(t->text)[i] == '[' )
 	    return 0;
-	else if ( T(t->text)[i] == ']' && T(t->text)[i+1] == ':' )
-	    return 1;
+	else if ( T(t->text)[i] == ']' )
+	    return ( T(t->text)[i+1] == ':' ) ;
     }
     return 0;
 }
@@ -315,7 +309,14 @@ isfootnote(Line *t)
 static int
 isquote(Line *t)
 {
-    return ( T(t->text)[0] == '>' );
+    int j;
+
+    for ( j=0; j < 4; j++ )
+	if ( T(t->text)[j] == '>' )
+	    return 1;
+	else if ( !isspace(T(t->text)[j]) )
+	    return 0;
+    return 0;
 }
 
 
@@ -355,6 +356,34 @@ ishr(Line *t)
 }
 
 
+static int
+issetext(Line *t, int *htyp)
+{
+    int i;
+    /* then check for setext-style HEADER
+     *                             ======
+     */
+
+    if ( t->next ) {
+	char *q = T(t->next->text);
+	int last = S(t->next->text);
+
+	if ( (*q == '=') || (*q == '-') ) {
+	    /* ignore trailing whitespace */
+	    while ( (last > 1) && isspace(q[last-1]) )
+		--last;
+
+	    for (i=1; i < last; i++)
+		if ( q[0] != q[i] )
+		    return 0;
+	    *htyp = SETEXT;
+	    return 1;
+	}
+    }
+    return 0;
+}
+
+
 static int
 ishdr(Line *t, int *htyp)
 {
@@ -376,22 +405,7 @@ ishdr(Line *t, int *htyp)
 	return 1;
     }
 
-    /* then check for setext-style HEADER
-     *                             ======
-     */
-
-    if ( t->next ) {
-	char *q = T(t->next->text);
-
-	if ( (*q == '=') || (*q == '-') ) {
-	    for (i=1; i < S(t->next->text); i++)
-		if ( q[0] != q[i] )
-		    return 0;
-	    *htyp = SETEXT;
-	    return 1;
-	}
-    }
-    return 0;
+    return issetext(t, htyp);
 }
 
 
@@ -475,7 +489,8 @@ headerblock(Paragraph *pp, int htyp)
 	     * the leading and trailing `#`'s
 	     */
 
-	    for (i=0; (T(p->text)[i] == T(p->text)[0]) && (i < S(p->text)-1); i++)
+	    for (i=0; (T(p->text)[i] == T(p->text)[0]) && (i < S(p->text)-1)
+						       && (i < 6); i++)
 		;
 
 	    pp->hnumber = i;
@@ -629,7 +644,16 @@ quoteblock(Paragraph *p)
 
     for ( t = p->text; t ; t = q ) {
 	if ( isquote(t) ) {
-	    qp = (T(t->text)[1] == ' ') ? 2 : 1;
+	    /* clip leading spaces */
+	    for (qp = 0; T(t->text)[qp] != '>'; qp ++)
+		/* assert: the first nonblank character on this line
+		 * will be a >
+		 */;
+	    /* clip '>' */
+	    qp++;
+	    /* clip next space, if any */
+	    if ( T(t->text)[qp] == ' ' )
+		qp++;
 	    CLIP(t->text, 0, qp);
 	    t->dle = mkd_firstnonblank(t);
 	}
@@ -709,8 +733,9 @@ listitem(Paragraph *p, int indent)
 	}
 
 	/* after a blank line, the next block needs to start with a line
-	 * that's indented 4 spaces, but after that the line doesn't
-	 * need any indentation
+	 * that's indented 4(? -- reference implementation allows a 1
+	 * character indent, but that has unfortunate side effects here)
+	 * spaces, but after that the line doesn't need any indentation
 	 */
 	if ( q != t->next ) {
 	    if (q->dle < indent) {
@@ -718,10 +743,12 @@ listitem(Paragraph *p, int indent)
 		t->next = 0;
 		return q;
 	    }
-	    indent = 4;
+	    /* indent at least 2, and at most as
+	     * as far as the initial line was indented. */
+	    indent = clip ? clip : 2;
 	}
 
-	if ( (q->dle < indent) && (ishr(q) || islist(q,&z)) && !ishdr(q,&z) ) {
+	if ( (q->dle < indent) && (ishr(q) || islist(q,&z)) && !issetext(q,&z) ) {
 	    q = t->next;
 	    t->next = 0;
 	    return q;
@@ -922,10 +949,7 @@ compile_document(Line *ptr, MMIOT *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);
+	    ptr = htmlblock(p, tag);
 	}
 	else if ( isfootnote(ptr) ) {
 	    /* footnotes, like cats, sleep anywhere; pull them
@@ -1028,15 +1052,15 @@ compile(Line *ptr, int toplevel, MMIOT *f)
 }
 
 
-static void
-initialize()
+void
+mkd_initialize()
 {
     static int first = 1;
 
     if ( first-- > 0 ) {
 	first = 0;
 	INITRNG(time(0));
-	qsort(blocktags, SZTAGS, sizeof blocktags[0], (stfu)casort);
+	mkd_prepare_tags();
     }
 }
 
@@ -1060,13 +1084,13 @@ mkd_compile(Document *doc, int flags)
 
     doc->compiled = 1;
     memset(doc->ctx, 0, sizeof(MMIOT) );
-    doc->ctx->flags = flags & USER_FLAGS;
-    doc->ctx->base = doc->base;
+    doc->ctx->cb        = &(doc->cb);
+    doc->ctx->flags     = flags & USER_FLAGS;
     CREATE(doc->ctx->in);
     doc->ctx->footnotes = malloc(sizeof doc->ctx->footnotes[0]);
     CREATE(*doc->ctx->footnotes);
 
-    initialize();
+    mkd_initialize();
 
     doc->code = compile_document(T(doc->content), doc->ctx);
     qsort(T(*doc->ctx->footnotes), S(*doc->ctx->footnotes),

data/ext/markdown.h

@@ -56,6 +56,17 @@ typedef struct block {
 typedef STRING(block) Qblock;
 
 
+typedef char* (*mkd_callback_t)(const char*, const int, void*);
+typedef void  (*mkd_free_t)(char*, void*);
+
+typedef struct callback_data {
+    void *e_data;		/* private data for callbacks */
+    mkd_callback_t e_url;	/* url edit callback */
+    mkd_callback_t e_flags;	/* extra href flags callback */
+    mkd_free_t e_free;		/* edit/flags callback memory deallocator */
+} Callback_data;
+
+
 /* a magic markdown io thing holds all the data structures needed to
  * do the backend processing of a markdown document
  */
@@ -75,13 +86,14 @@ typedef struct mmiot {
 #define NO_PSEUDO_PROTO	0x0040
 #define CDATA_OUTPUT	0x0080
 #define NOTABLES	0x0400
+#define NOSTRIKETHROUGH	0x0800
 #define TOC		0x1000
 #define MKD_1_COMPAT	0x2000
 #define AUTOLINK	0x4000
 #define SAFELINK	0x8000
 #define USER_FLAGS	0xFCFF
 #define EMBEDDED	DENY_A|DENY_IMG|NO_PSEUDO_PROTO|CDATA_OUTPUT
-    char *base;
+    Callback_data *cb;
 } MMIOT;
 
 
@@ -93,6 +105,8 @@ typedef struct mmiot {
  * root of the linked list of Lines.
  */
 typedef struct document {
+    int magic;			/* "I AM VALID" magic number */
+#define VALID_DOCUMENT		0x19600731
     Line *headers;		/* title -> author(s) -> date */
     ANCHOR(Line) content;	/* uncompiled text, not valid after compile() */
     Paragraph *code;		/* intermediate code generated by compile() */
@@ -100,7 +114,7 @@ typedef struct document {
     int html;			/* set after (internal) htmlify() */
     int tabstop;		/* for properly expanding tabs (ick) */
     MMIOT *ctx;			/* backend buffers, flags, and structures */
-    char *base;			/* url basename for url fragments */
+    Callback_data cb;		/* callback functions & private data */
 } Document;