rdiscount 2.2.0.1 → 2.2.7

data/ext/markdown.h

@@ -1,22 +1,22 @@
 #ifndef _MARKDOWN_D
 #define _MARKDOWN_D
 
+#include "config.h"
 #include "cstring.h"
 
-/* reference-style links (and images) are stored in an array
- * of footnotes.
+#ifdef HAVE_INTTYPES_H
+#   include <inttypes.h>
+#elif HAVE_STDINT_H
+#   include <stdint.h>
+#endif
+
+/* flags, captured into a named type
  */
-typedef struct footnote {
-    Cstring tag;		/* the tag for the reference link */
-    Cstring link;		/* what this footnote points to */
-    Cstring title;		/* what it's called (TITLE= attribute) */
-    int height, width;		/* dimensions (for image link) */
-    int dealloc;		/* deallocation needed? */
-    int refnumber;
-    int flags;
-#define EXTRA_BOOKMARK	0x01
-#define REFERENCED	0x02
-} Footnote;
+typedef DWORD mkd_flag_t;
+
+#define is_flag_set(flags, item)	((flags) & (item))
+#define set_flag(flags, item)		((flags) |= (item))
+#define clear_flag(flags, item)		((flags) &= ~(item))
 
 /* each input line is read into a Line, which contains the line,
  * the offset of the first non-space character [this assumes 
@@ -49,16 +49,38 @@ typedef struct paragraph {
     struct paragraph *down;	/* recompiled contents of this paragraph */
     struct line *text;		/* all the text in this paragraph */
     char *ident;		/* %id% tag for QUOTE */
-    char *lang;         /* lang attribute for CODE */
+    char *lang;			/* lang attribute for CODE */
     enum { WHITESPACE=0, CODE, QUOTE, MARKUP,
 	   HTML, STYLE, DL, UL, OL, AL, LISTITEM,
 	   HDR, HR, TABLE, SOURCE } typ;
     enum { IMPLICIT=0, PARA, CENTER} align;
     int hnumber;		/* <Hn> for typ == HDR */
+#if GITHUB_CHECKBOX
+    int flags;
+#define GITHUB_CHECK		0x01
+#define IS_CHECKED		0x02
+#endif
 } Paragraph;
 
 enum { ETX, SETEXT };	/* header types */
 
+/* reference-style links (and images) are stored in an array
+ * of footnotes.
+ */
+typedef struct footnote {
+    Cstring tag;		/* the tag for the reference link */
+    Cstring link;		/* what this footnote points to */
+    Cstring title;		/* what it's called (TITLE= attribute) */
+    Paragraph *text;		/* EXTRA_FOOTNOTE content */
+    
+    int height, width;		/* dimensions (for image link) */
+    int dealloc;		/* deallocation needed? */
+    int refnumber;
+    int flags;
+#define EXTRA_FOOTNOTE	0x01
+#define REFERENCED	0x02
+} Footnote;
+
 
 typedef struct block {
     enum { bTEXT, bSTAR, bUNDER } b_type;
@@ -78,7 +100,9 @@ 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_callback_t e_anchor;	/* callback for anchor types */
     mkd_free_t e_free;		/* edit/flags callback memory deallocator */
+    mkd_callback_t e_codefmt;	/* codeblock formatter (for highlighting) */
 } Callback_data;
 
 
@@ -101,11 +125,12 @@ typedef struct mmiot {
     Cstring out;
     Cstring in;
     Qblock Q;
+    char last;	/* last text character added to out */
     int isp;
     struct escaped *esc;
     char *ref_prefix;
     struct footnote_list *footnotes;
-    DWORD flags;
+    mkd_flag_t flags;
 #define MKD_NOLINKS	0x00000001
 #define MKD_NOIMAGE	0x00000002
 #define MKD_NOPANTS	0x00000004
@@ -136,14 +161,16 @@ typedef struct mmiot {
 #define MKD_GITHUBTAGS	0x08000000
 #define MKD_URLENCODEDANCHOR 0x10000000
 #define IS_LABEL	0x20000000
-#define USER_FLAGS	0x3FFFFFFF
+#define MKD_LATEX	0x40000000
+#define MKD_EXPLICITLIST	0x80000000
+#define USER_FLAGS	0xFFFFFFFF
 #define INPUT_MASK	(MKD_NOHEADER|MKD_TABSTOP)
 
     Callback_data *cb;
 } MMIOT;
 
 
-#define MKD_EOLN	3
+#define MKD_EOLN	'\r'
 
 
 /*
@@ -162,6 +189,7 @@ typedef struct document {
     ANCHOR(Line) content;	/* uncompiled text, not valid after compile() */
     Paragraph *code;		/* intermediate code generated by compile() */
     int compiled;		/* set after mkd_compile() */
+    int dirty;			/* flags or callbacks changed */
     int html;			/* set after (internal) htmlify() */
     int tabstop;		/* for properly expanding tabs (ick) */
     char *ref_prefix;
@@ -181,7 +209,7 @@ struct string_stream {
 
 
 extern int  mkd_firstnonblank(Line *);
-extern int  mkd_compile(Document *, DWORD);
+extern int  mkd_compile(Document *, mkd_flag_t);
 extern int  mkd_document(Document *, char **);
 extern int  mkd_generatehtml(Document *, FILE *);
 extern int  mkd_css(Document *, char **);
@@ -190,19 +218,19 @@ extern int  mkd_generatecss(Document *, FILE *);
 extern int  mkd_xml(char *, int , char **);
 extern int  mkd_generatexml(char *, int, FILE *);
 extern void mkd_cleanup(Document *);
-extern int  mkd_line(char *, int, char **, DWORD);
-extern int  mkd_generateline(char *, int, FILE*, DWORD);
+extern int  mkd_line(char *, int, char **, mkd_flag_t);
+extern int  mkd_generateline(char *, int, FILE*, mkd_flag_t);
 #define mkd_text mkd_generateline
 extern void mkd_basename(Document*, char *);
 
 typedef int (*mkd_sta_function_t)(const int,const void*);
-extern void mkd_string_to_anchor(char*,int, mkd_sta_function_t, void*, int, DWORD);
+extern void mkd_string_to_anchor(char*,int, mkd_sta_function_t, void*, int, MMIOT *);
 
-extern Document *mkd_in(FILE *, DWORD);
-extern Document *mkd_string(const char*,int, DWORD);
+extern Document *mkd_in(FILE *, mkd_flag_t);
+extern Document *mkd_string(const char*, int, mkd_flag_t);
 
-extern Document *gfm_in(FILE *, DWORD);
-extern Document *gfm_string(const char*,int, DWORD);
+extern Document *gfm_in(FILE *, mkd_flag_t);
+extern Document *gfm_string(const char*,int, mkd_flag_t);
 
 extern void mkd_initialize();
 extern void mkd_shlib_destructor();
@@ -220,13 +248,13 @@ extern void ___mkd_initmmiot(MMIOT *, void *);
 extern void ___mkd_freemmiot(MMIOT *, void *);
 extern void ___mkd_freeLineRange(Line *, Line *);
 extern void ___mkd_xml(char *, int, FILE *);
-extern void ___mkd_reparse(char *, int, int, MMIOT*, char*);
+extern void ___mkd_reparse(char *, int, mkd_flag_t, MMIOT*, char*);
 extern void ___mkd_emblock(MMIOT*);
 extern void ___mkd_tidy(Cstring *);
 
 extern Document *__mkd_new_Document();
 extern void __mkd_enqueue(Document*, Cstring *);
-extern void __mkd_header_dle(Line *);
+extern void __mkd_trim_line(Line *, int);
 
 extern int  __mkd_io_strget(struct string_stream *);
 

data/ext/mkdio.c

@@ -74,13 +74,19 @@ __mkd_enqueue(Document* a, Cstring *line)
 }
 
 
-/* trim leading blanks from a header line
+/* trim leading characters from a line, then adjust the dle.
  */
 void
-__mkd_header_dle(Line *p)
+__mkd_trim_line(Line *p, int clip)
 {
-    CLIP(p->text, 0, 1);
-    p->dle = mkd_firstnonblank(p);
+    if ( clip >= S(p->text) ) {
+	S(p->text) = p->dle = 0;
+	T(p->text)[0] = 0;
+    }
+    else if ( clip > 0 ) {
+	CLIP(p->text, 0, clip);
+	p->dle = mkd_firstnonblank(p);
+    }
 }
 
 
@@ -89,7 +95,7 @@ __mkd_header_dle(Line *p)
 typedef int (*getc_func)(void*);
 
 Document *
-populate(getc_func getc, void* ctx, int flags)
+populate(getc_func getc, void* ctx, mkd_flag_t flags)
 {
     Cstring line;
     Document *a = __mkd_new_Document();
@@ -98,7 +104,7 @@ populate(getc_func getc, void* ctx, int flags)
 
     if ( !a ) return 0;
 
-    a->tabstop = (flags & MKD_TABSTOP) ? 4 : TABSTOP;
+    a->tabstop = is_flag_set(flags, MKD_TABSTOP) ? 4 : TABSTOP;
 
     CREATE(line);
 
@@ -122,16 +128,16 @@ populate(getc_func getc, void* ctx, int flags)
 
     DELETE(line);
 
-    if ( (pandoc == 3) && !(flags & (MKD_NOHEADER|MKD_STRICT)) ) {
+    if ( (pandoc == 3) && !(is_flag_set(flags, MKD_NOHEADER) || is_flag_set(flags, MKD_STRICT)) ) {
 	/* the first three lines started with %, so we have a header.
 	 * clip the first three lines out of content and hang them
 	 * off header.
 	 */
 	Line *headers = T(a->content);
 
-	a->title = headers;             __mkd_header_dle(a->title);
-	a->author= headers->next;       __mkd_header_dle(a->author);
-	a->date  = headers->next->next; __mkd_header_dle(a->date);
+	a->title = headers;             __mkd_trim_line(a->title, 1);
+	a->author= headers->next;       __mkd_trim_line(a->author, 1);
+	a->date  = headers->next->next; __mkd_trim_line(a->date, 1);
 
 	T(a->content) = headers->next->next->next;
     }
@@ -143,7 +149,7 @@ populate(getc_func getc, void* ctx, int flags)
 /* convert a file into a linked list
  */
 Document *
-mkd_in(FILE *f, DWORD flags)
+mkd_in(FILE *f, mkd_flag_t flags)
 {
     return populate((getc_func)fgetc, f, flags & INPUT_MASK);
 }
@@ -165,7 +171,7 @@ __mkd_io_strget(struct string_stream *in)
 /* convert a block of text into a linked list
  */
 Document *
-mkd_string(const char *buf, int len, DWORD flags)
+mkd_string(const char *buf, int len, mkd_flag_t flags)
 {
     struct string_stream about;
 
@@ -185,7 +191,7 @@ mkd_generatehtml(Document *p, FILE *output)
     int szdoc;
 
     DO_OR_DIE( szdoc = mkd_document(p,&doc) );
-    if ( p->ctx->flags & MKD_CDATA )
+    if ( is_flag_set(p->ctx->flags, MKD_CDATA) )
 	DO_OR_DIE( mkd_generatexml(doc, szdoc, output) );
     else if ( fwrite(doc, szdoc, 1, output) != 1 )
 	return EOF;
@@ -197,7 +203,7 @@ mkd_generatehtml(Document *p, FILE *output)
 /* convert some markdown text to html
  */
 int
-markdown(Document *document, FILE *out, int flags)
+markdown(Document *document, FILE *out, mkd_flag_t flags)
 {
     if ( mkd_compile(document, flags) ) {
 	mkd_generatehtml(document, out);
@@ -208,51 +214,103 @@ markdown(Document *document, FILE *out, int flags)
 }
 
 
-/* write out a Cstring, mangled into a form suitable for `<a href=` or `<a id=`
+/* anchor_format a string, returning the formatted string in malloc()ed space
+ * MKD_URLENCODEDANCHOR is now perverted to being a html5 anchor
+ *
+ * !labelformat:  print all characters
+ * labelformat && h4anchor: prefix nonalpha label with L,
+ *                          expand all nonalnum, _, ':', '.' to hex
+ *                          except space which maps to -
+ * labelformat && !h4anchor:expand space to -, other isspace() & '%' to hex
  */
-void
-mkd_string_to_anchor(char *s, int len, mkd_sta_function_t outchar,
-				       void *out, int labelformat,
-				       DWORD flags)
+static char *
+mkd_anchor_format(char *s, int len, int labelformat, mkd_flag_t flags)
 {
-    static const unsigned char hexchars[] = "0123456789abcdef";
+    char *res;
     unsigned char c;
+    int i, needed, out = 0;
+    int h4anchor = !is_flag_set(flags, MKD_URLENCODEDANCHOR);
+    static const unsigned char hexchars[] = "0123456789abcdef";
 
-    int i, size;
-    char *line;
+    needed = (labelformat ? (4*len) : len) + 2 /* 'L', trailing null */;
 
-    size = mkd_line(s, len, &line, IS_LABEL);
+    if ( (res = malloc(needed)) == NULL )
+	return NULL;
 
-    if ( !(flags & MKD_URLENCODEDANCHOR)
-	 && labelformat
-	 && (size>0) && !isalpha(line[0]) )
-	(*outchar)('L',out);
-    for ( i=0; i < size ; i++ ) {
-	c = line[i];
+    if ( h4anchor && labelformat && !isalpha(s[0]) )
+	res[out++] = 'L';
+	
+    
+    for ( i=0; i < len ; i++ ) {
+	c = s[i];
 	if ( labelformat ) {
-	    if ( isalnum(c) || (c == '_') || (c == ':') || (c == '-') || (c == '.' ) )
-		(*outchar)(c, out);
-	    else if ( flags & MKD_URLENCODEDANCHOR ) {
-		(*outchar)('%', out);
-		(*outchar)(hexchars[c >> 4 & 0xf], out);
-		(*outchar)(hexchars[c      & 0xf], out);
+	    if ( h4anchor
+		    ? (isalnum(c) || (c == '_') || (c == ':') || (c == '.' ) )
+		    : !(isspace(c) || c == '%') )
+		res[out++] = c;
+	    else if ( c == ' ' )
+		res[out++] = '-';
+	    else {
+		    res[out++] = h4anchor ? '-' : '%';
+		    res[out++] = hexchars[c >> 4 & 0xf];
+		    res[out++] = hexchars[c      & 0xf];
+		    if ( h4anchor )
+			res[out++] = '-';
 	    }
-	    else
-		(*outchar)('.', out);
 	}
 	else
-	    (*outchar)(c,out);
+	    res[out++] = c;
     }
-	
-    if (line)
-	free(line);
+    
+    res[out++] = 0;
+    return res;
+} /* mkd_anchor_format */
+
+
+/* write out a Cstring, mangled into a form suitable for `<a href=` or `<a id=`
+ */
+void
+mkd_string_to_anchor(char *s, int len, mkd_sta_function_t outchar,
+				       void *out, int labelformat,
+				       MMIOT *f)
+{
+    char *res;
+    char *line;
+    int size;
+
+    int i;
+
+    size = mkd_line(s, len, &line, IS_LABEL);
+
+    if ( !line )
+	return;
+
+    if ( f->cb->e_anchor )
+	res = (*(f->cb->e_anchor))(line, size, f->cb->e_data);
+    else
+	res = mkd_anchor_format(line, size, labelformat, f->flags);
+
+    free(line);
+
+    if ( !res )
+	return;
+
+    for ( i=0; res[i]; i++ )
+	(*outchar)(res[i], out);
+
+    if ( f->cb->e_anchor ) {
+	if ( f->cb->e_free )
+	    (*(f->cb->e_free))(res, f->cb->e_data);
+    }
+    else 
+	free(res);
 }
 
 
 /*  ___mkd_reparse() a line
  */
 static void
-mkd_parse_line(char *bfr, int size, MMIOT *f, int flags)
+mkd_parse_line(char *bfr, int size, MMIOT *f, mkd_flag_t flags)
 {
     ___mkd_initmmiot(f, 0);
     f->flags = flags & USER_FLAGS;
@@ -264,7 +322,7 @@ mkd_parse_line(char *bfr, int size, MMIOT *f, int flags)
 /* ___mkd_reparse() a line, returning it in malloc()ed memory
  */
 int
-mkd_line(char *bfr, int size, char **res, DWORD flags)
+mkd_line(char *bfr, int size, char **res, mkd_flag_t flags)
 {
     MMIOT f;
     int len;
@@ -272,15 +330,14 @@ mkd_line(char *bfr, int size, char **res, DWORD flags)
     mkd_parse_line(bfr, size, &f, flags);
 
     if ( len = S(f.out) ) {
-	/* kludge alert;  we know that T(f.out) is malloced memory,
-	 * so we can just steal it away.   This is awful -- there
-	 * should be an opaque method that transparently moves 
-	 * the pointer out of the embedded Cstring.
-	 */
 	EXPAND(f.out) = 0;
-	*res = T(f.out);
-	T(f.out) = 0;
-	S(f.out) = ALLOCATED(f.out) = 0;
+	/* strdup() doesn't use amalloc(), so in an amalloc()ed
+	 * build this copies the string safely out of our memory
+	 * paranoia arena.  In a non-amalloc world, it's a spurious
+	 * memory allocation, but it avoids unintentional hilarity
+	 * with amalloc()
+	 */
+	*res = strdup(T(f.out));
     }
     else {
 	 *res = 0;
@@ -294,13 +351,13 @@ mkd_line(char *bfr, int size, char **res, DWORD flags)
 /* ___mkd_reparse() a line, writing it to a FILE
  */
 int
-mkd_generateline(char *bfr, int size, FILE *output, DWORD flags)
+mkd_generateline(char *bfr, int size, FILE *output, mkd_flag_t flags)
 {
     MMIOT f;
     int status;
 
     mkd_parse_line(bfr, size, &f, flags);
-    if ( flags & MKD_CDATA )
+    if ( is_flag_set(flags, MKD_CDATA) )
 	status = mkd_generatexml(T(f.out), S(f.out), output) != EOF;
     else
 	status = fwrite(T(f.out), S(f.out), 1, output) == S(f.out);
@@ -315,8 +372,11 @@ mkd_generateline(char *bfr, int size, FILE *output, DWORD flags)
 void
 mkd_e_url(Document *f, mkd_callback_t edit)
 {
-    if ( f )
+    if ( f ) {
+	if ( f->cb.e_url != edit )
+	    f->dirty = 1;
 	f->cb.e_url = edit;
+    }
 }
 
 
@@ -325,8 +385,24 @@ mkd_e_url(Document *f, mkd_callback_t edit)
 void
 mkd_e_flags(Document *f, mkd_callback_t edit)
 {
-    if ( f )
+    if ( f ) {
+	if ( f->cb.e_flags != edit )
+	    f->dirty = 1;
 	f->cb.e_flags = edit;
+    }
+}
+
+
+/* set the anchor formatter
+ */
+void
+mkd_e_anchor(Document *f, mkd_callback_t format)
+{
+    if ( f ) {
+	if ( f->cb.e_anchor != format )
+	    f->dirty = 1;
+	f->cb.e_anchor = format;
+    }
 }
 
 
@@ -335,8 +411,11 @@ mkd_e_flags(Document *f, mkd_callback_t edit)
 void
 mkd_e_free(Document *f, mkd_free_t dealloc)
 {
-    if ( f )
+    if ( f ) {
+	if ( f->cb.e_free != dealloc )
+	    f->dirty = 1;
 	f->cb.e_free = dealloc;
+    }
 }
 
 
@@ -345,8 +424,23 @@ mkd_e_free(Document *f, mkd_free_t dealloc)
 void
 mkd_e_data(Document *f, void *data)
 {
-    if ( f )
+    if ( f ) {
+	if ( f->cb.e_data != data )
+	    f->dirty = 1;
 	f->cb.e_data = data;
+    }
+}
+
+
+/* set the code block display callback
+ */
+void
+mkd_e_code_format(Document *f, mkd_callback_t codefmt)
+{
+    if ( f && (f->cb.e_codefmt != codefmt) ) {
+	f->dirty = 1;
+	f->cb.e_codefmt = codefmt;
+    }
 }
 
 
@@ -355,6 +449,9 @@ mkd_e_data(Document *f, void *data)
 void
 mkd_ref_prefix(Document *f, char *data)
 {
-    if ( f )
+    if ( f ) {
+	if ( f->ref_prefix != data )
+	    f->dirty = 1;
 	f->ref_prefix = data;
+    }
 }

data/ext/mkdio.h

@@ -3,9 +3,11 @@
 
 #include <stdio.h>
 
+#include <inttypes.h>
+
 typedef void MMIOT;
 
-typedef unsigned int mkd_flag_t;
+typedef uint32_t mkd_flag_t;
 
 /* line builder for markdown()
  */
@@ -30,12 +32,10 @@ void mkd_cleanup(MMIOT*);
 
 /* markup functions
  */
-int mkd_dump(MMIOT*, FILE*, int, char*);
+int mkd_dump(MMIOT*, FILE*, mkd_flag_t, char*);
 int markdown(MMIOT*, FILE*, mkd_flag_t);
 int mkd_line(char *, int, char **, mkd_flag_t);
-typedef int (*mkd_sta_function_t)(const int,const void*);
-void mkd_string_to_anchor(char *, int, mkd_sta_function_t, void*, int);
-int mkd_xhtmlpage(MMIOT*,int,FILE*);
+int mkd_xhtmlpage(MMIOT*,mkd_flag_t,FILE*);
 
 /* header block access
  */
@@ -67,6 +67,8 @@ typedef void   (*mkd_free_t)(char*, void*);
 
 void mkd_e_url(void *, mkd_callback_t);
 void mkd_e_flags(void *, mkd_callback_t);
+void mkd_e_anchor(void *, mkd_callback_t);
+void mkd_e_code_format(void*, mkd_callback_t);
 void mkd_e_free(void *, mkd_free_t );
 void mkd_e_data(void *, void *);
 
@@ -112,6 +114,8 @@ void mkd_ref_prefix(MMIOT*, char*);
 #define MKD_IDANCHOR	0x04000000	/* use id= anchors for TOC links */
 #define MKD_GITHUBTAGS	0x08000000	/* allow dash and underscore in element names */
 #define MKD_URLENCODEDANCHOR 0x10000000 /* urlencode non-identifier chars instead of replacing with dots */
+#define MKD_LATEX	0x40000000	/* handle embedded LaTeX escapes */
+#define MKD_EXPLICITLIST 0x80000000	/* don't combine numbered/bulletted lists */
 
 #define MKD_EMBED	MKD_NOLINKS|MKD_NOIMAGE|MKD_TAGTEXT
 

data/ext/mktags.c

@@ -3,6 +3,7 @@
 #include <stdio.h>
 
 #define __WITHOUT_AMALLOC 1
+#include "config.h"
 #include "cstring.h"
 #include "tags.h"
 
@@ -41,6 +42,7 @@ typedef int (*stfu)(const void*,const void*);
 
 /* load in the standard collection of html tags that markdown supports
  */
+int
 main()
 {
     int i;
@@ -65,6 +67,7 @@ main()
     KW("H6");
     KW("LISTING");
     KW("NOBR");
+    KW("FORM");
     KW("UL");
     KW("P");
     KW("OL");

data/ext/notspecial.c

@@ -0,0 +1,44 @@
+/*
+ * check a filename to see if it's a (fifo|character special|socket) object
+ * (if a stat() function doesn't exist, we can't stat so we'll just return
+ *  true no matter what.)
+ */
+
+#include "config.h"
+
+#if HAVE_STAT && HAS_ISCHR && HAS_ISFIFO && HAS_ISSOCK
+#include <sys/stat.h>
+
+int
+notspecial(char *file)
+{
+    struct stat info;
+
+    if ( stat(file, &info) != 0 )
+	return 1;
+    
+    return !( S_ISCHR(info.st_mode) || S_ISFIFO(info.st_mode) || S_ISSOCK(info.st_mode) );
+}
+#else
+int
+notspecial(char *file)
+{
+    return 1;
+}
+#endif
+
+
+#if DEBUG
+
+#include <stdio.h>
+
+int
+main(argc, argv)
+char **argv;
+{
+    int i;
+
+    for ( i=1; i < argc; i++ )
+	printf("%s is %sspecial\n", argv[i], notspecial(argv[i]) ? "not " : "");
+}
+#endif