go back and tweak the effing fenced code disaster area so that it will break fenced code out into its own div and consume html & footnotes at the top level.

2026-09-25 04:10:00 +03:00 · 2026-05-31 14:12:03 -07:00
parent ff09a191cd
commit 39be61ef2c
6 changed files with 105 additions and 98 deletions
@@ -22,6 +22,7 @@ Pptype(int typ)
    switch (typ) {
    case WHITESPACE: return "whitespace";
    case CODE      : return "code";
+    case FENCEDCODE: return "fenced code";
    case QUOTE     : return "quote";
    case MARKUP    : return "markup";
    case HTML      : return "html";
@@ -1867,11 +1867,7 @@ printblock(Paragraph *pp, MMIOT *f)

    Qstring(Begin[align], f);
    do {
-	if ( t->is_fenced ) {
-	    text(f);
-	    t = printfenced(t, f);
-	}
-	else if ( S(t->text) ) {
+	if ( S(t->text) ) {
 	    if ( t->next && S(t->text) > 2
 			 && T(t->text)[S(t->text)-2] == ' '
 			 && T(t->text)[S(t->text)-1] == ' ' ) {
@@ -2060,6 +2056,10 @@ display(Paragraph *p, MMIOT *f)
 	printhtml(p->text, f);
 	break;

+    case FENCEDCODE:
+	printfenced(p->text, f);
+	break;
+
    case CODE:
 	printcode(p->text, p->lang, f);
 	break;
@@ -632,8 +632,10 @@ codeblock(Paragraph *p)


 static int
-iscodefence(Line *r, int size, line_type kind, mkd_flag_t *flags)
+iscodefence(Line *r, Line *orig, mkd_flag_t *flags)
 {
+    int kind = 0, size = 2, dle = 0;
+    
    if ( !is_flag_set(flags, MKD_FENCEDCODE) || is_flag_set(flags, MKD_STRICT) )
 	return 0;

@@ -644,6 +646,15 @@ iscodefence(Line *r, int size, line_type kind, mkd_flag_t *flags)
 	return 0;


+    if ( orig ) {
+	kind = orig->kind;
+	size = orig->count;
+	dle  = orig->dle;
+    }
+    
+    if ( (dle>>2) != (r->dle>>2) )
+	return 0;
+
    if ( kind )
 	return (r->kind == kind) && (r->count >= size);
    else
@@ -652,19 +663,20 @@ iscodefence(Line *r, int size, line_type kind, mkd_flag_t *flags)


 static Line *
-fencedcodechunk(Line *first, mkd_flag_t *flags)
+fencedcodechunk(Line *first, mkd_flag_t *flags, int *yes)
 {
    Line *r, *q;

    /* don't allow zero-length code fences
    */
-    if ( (first->next == 0) || iscodefence(first->next, first->count, first->kind, flags) ) {
+    if ( (first->next == 0) || iscodefence(first->next, first, flags) ) {
 	first->kind = chk_text;
 	first->is_fenced = 0;
 	if ( first->next) {
 	    first->next->kind = chk_text;
 	    first->next->is_fenced = 0;
 	}
+	*yes = 0;
 	return first->next;
    }

@@ -672,7 +684,7 @@ fencedcodechunk(Line *first, mkd_flag_t *flags)
     * past the closing fence
     */
    for ( r = first->next; r; r = r->next ) {
-	if ( iscodefence(r, first->count, first->kind, flags) ) {
+	if ( iscodefence(r, first, flags) ) {
 	    if (S(first->text) - first->count > 0) {
 		char *lang_attr = T(first->text) + first->count;

@@ -689,9 +701,13 @@ fencedcodechunk(Line *first, mkd_flag_t *flags)
 	    S(first->text) = S(r->text) = 0;
 	    r->is_fenced = 0;

-	    return r->next;
+	    *yes = 1;
+	    q = r->next;
+	    r->next = 0;
+	    return q;
 	}
    }
+    *yes = 0;
    first->is_fenced = 0;
    first->kind = chk_text;
    return first;
@@ -742,9 +758,7 @@ textblock(Paragraph *p, int toplevel, mkd_flag_t *flags)
    Line *t, *next;

    for ( t = p->text; t ; t = next ) {
-	if ( iscodefence(t, t->count, 0, flags) )
-	    next = fencedcodechunk(t, flags);
-	else if ( ((next = t->next) == 0) || endoftextblock(next, toplevel, flags) ) {
+	if ( ((next = t->next) == 0) || endoftextblock(next, toplevel, flags) ) {
 	    p->align = centered(p->text, t);
 	    t->next = 0;
 	    return next;
@@ -1296,24 +1310,31 @@ compile_document(Line *ptr, MMIOT *f)
 	    ptr = consume(addfootnote(ptr, f), &eaten);
 	    previous_was_break = 1;
 	}
-	else if (iscodefence(ptr, 2, 0, &(f->flags))) {
-	    Line *more = ptr;
-
+	else if ( previous_was_break && iscodefence(ptr, 0, &(f->flags)) ) {
 	    /* scoop up _everything_ in a fenced code block, including html & footnotes
 	     */
+	    int yes = 0;
+	    Line *last = fencedcodechunk(ptr, &(f->flags), &yes );

-	    ATTACH(source,more);
+	    if ( yes ) {
+		int dummy;

-	    {   Cache checkpoint = source;
-		while ( (more = more->next) && !iscodefence(more, ptr->count, ptr->kind, &(f->flags)) )
-		    ATTACH(source,more);
+		/* dump out any previously cached text
+		 */
+		uncache(&source, &d, f);

-		if ( more )
-		    ptr = more;
-		else {
-		    source = checkpoint;
-		    ptr = ptr->next;
+		p = Pp(&d, ptr, FENCEDCODE);
+
+		ptr = consume(last, &dummy);
+		previous_was_break = 1;
 	    }
+	    else {
+		/* unterminated fenced code; treat as regular markdown & cache until the next
+		 * fenced code, htl, or style block
+		 */
+		ATTACH(source, ptr);
+		previous_was_break = 0;
+		ptr = ptr->next;
 	    }
 	}
 	else {
@@ -1474,15 +1495,26 @@ compile(Line *ptr, int toplevel, MMIOT *f)
 		ptr = htmlblock(p, tag, &unclosed);
 		if ( ! unclosed ) {
 		    p->typ = HTML;
+		    continue;
 		}
 	    }
-	    if ( unclosed ) {
+		
+	    if ( iscodefence(p->text, 0, &(f->flags)) ) {
+		int yes = 0;
+
+		ptr = fencedcodechunk(p->text, &(f->flags), &yes);
+
+		if ( yes ) {
+		    p->typ = FENCEDCODE;
+		    continue;
+		}
+	    }
+		
 	    ptr = textblock(p, toplevel, &(f->flags) );
 	    /* tables are a special kind of paragraph */
 	    if ( actually_a_table(f, p->text) )
 		p->typ = TABLE;
 	}
-	}
 	if ( (para||toplevel) && !p->align )
 	    p->align = PARA;

@@ -102,7 +102,7 @@ typedef struct paragraph {
    char *label;		/* toc label, uniqued */
    char *ident;		/* %id% tag for QUOTE */
    char *lang;			/* lang attribute for CODE */
-    enum { WHITESPACE=0, CODE, QUOTE, MARKUP,
+    enum { WHITESPACE=0, CODE, FENCEDCODE, QUOTE, MARKUP,
 	   HTML, STYLE, DL, UL, OL, AL, LISTITEM,
 	   HDR, HR, TABLE, SOURCE } typ;
    enum { IMPLICIT=0, PARA, CENTER} align;
@@ -50,19 +50,17 @@ code!

 still code!
 ~~~' \
-    '<p><pre><code>code!
+    '<pre><code>code!

 still code!
-</code></pre>
-</p>'
+</code></pre>'

 try  -ffencedcode 'fenced code block' \
 '~~~
 code!
 ~~~' \
-    '<p><pre><code>code!
-</code></pre>
-</p>'
+    '<pre><code>code!
+</code></pre>'

 try  -ffencedcode 'fenced code block in list' \
 '1. ~~~
@@ -78,9 +76,9 @@ try  -ffencedcode 'fenced code block in blockquote' \
 '>~~~
 code
 ~~~' \
-'<blockquote><p><pre><code>code
+'<blockquote><pre><code>code
 </code></pre>
-</p></blockquote>'
+</blockquote>'

 try  -ffencedcode 'unterminated fenced code block' \
 '~~~
@@ -94,11 +92,10 @@ try  -ffencedcode 'fenced code block with tildes' \
 code with tildes
 ~~~
 ~~~~~' \
-'<p><pre><code>~~~
+'<pre><code>~~~
 code with tildes
 ~~~
-</code></pre>
-</p>'
+</code></pre>'

 try  -ffencedcode 'paragraph with trailing fenced block' \
 'text text text
@@ -108,17 +105,16 @@ code code code?
 ~~~' \
 '<p>text text text
 text text text
-<pre><code>code code code?
-</code></pre>
-</p>'
+~~~
+code code code?
+~~~</p>'

 try  -ffencedcode 'fenced code blocks with backtick delimiters' \
 '```
 code
 ```' \
-'<p><pre><code>code
-</code></pre>
-</p>'
+'<pre><code>code
+</code></pre>'

 try  -ffencedcode 'fenced code block with mismatched delimters' \
 '```
@@ -132,57 +128,50 @@ try  -ffencedcode 'fenced code block with lang attribute' \
 '```lang
 code
 ```' \
-'<p><pre><code class="lang">code
-</code></pre>
-</p>'
+'<pre><code class="lang">code
+</code></pre>'

 try  -ffencedcode 'fenced code block with lang-name attribute' \
 '```lang-name
 code
 ```' \
-'<p><pre><code class="lang-name">code
-</code></pre>
-</p>'
+'<pre><code class="lang-name">code
+</code></pre>'

 try  -ffencedcode 'fenced code block with lang_name attribute' \
 '```lang_name
 code
 ```' \
-'<p><pre><code class="lang_name">code
-</code></pre>
-</p>'
+'<pre><code class="lang_name">code
+</code></pre>'

 try  -ffencedcode 'fenced code block with lang attribute and space' \
 '``` lang
 code
 ```' \
-'<p><pre><code class="lang">code
-</code></pre>
-</p>'
+'<pre><code class="lang">code
+</code></pre>'

 try  -ffencedcode 'fenced code block with lang attribute and multiple spaces' \
 '```       lang
 code
 ```' \
-'<p><pre><code class="lang">code
-</code></pre>
-</p>'
+'<pre><code class="lang">code
+</code></pre>'

 try  -ffencedcode 'fenced code block with lang-name attribute and space' \
 '``` lang-name
 code
 ```' \
-'<p><pre><code class="lang-name">code
-</code></pre>
-</p>'
+'<pre><code class="lang-name">code
+</code></pre>'

 try  -ffencedcode 'fenced code block with lang_name attribute and space' \
 '``` lang_name
 code
 ```' \
-'<p><pre><code class="lang_name">code
-</code></pre>
-</p>'
+'<pre><code class="lang_name">code
+</code></pre>'

 try -ffencedcode 'fenced code block with blank line in the middle' \
 '```
@@ -190,20 +179,18 @@ hello

 sailor
 ```' \
-'<p><pre><code>hello
+'<pre><code>hello

 sailor
-</code></pre>
-</p>'
+</code></pre>'


 try -ffencedcode 'fenced code block with html in the middle' \
 '~~~~
 <h1>hello, sailor</h1>
 ~~~~' \
-'<p><pre><code>&lt;h1&gt;hello, sailor&lt;/h1&gt;
-</code></pre>
-</p>'
+'<pre><code>&lt;h1&gt;hello, sailor&lt;/h1&gt;
+</code></pre>'

 try -ffencedcode 'fenced code block with trailing spaces in list item' \
 '1.  ~~~~    
@@ -229,9 +216,8 @@ bar
 ~~~~' \
 '<p>foo</p>

-<p><pre><code>bar
-</code></pre>
-</p>'
+<pre><code>bar
+</code></pre>'


 try -ffencedcode 'checkline misparse as fenced code' \
@@ -241,10 +227,9 @@ content
 ```
 ' \
 '<p><a href="#code"><code>label</code></a>
-<pre><code class="class">content
-</code></pre>
-</p>'
-
+<code>class
+content
+</code></p>'

 summary $0
 exit $rc
@@ -76,21 +76,10 @@ RESULT='<ul>

 ./rep '#header' ' ' 400 |  try -T 'a header with a bunch of trailing spaces' -heredoc "$RESULT"

-RESULT='<table>
-<thead>
-<tr>
-<th>: Y:</th>
-</tr>
-</thead>
-<tbody>
-<tr>
-<td></td>
-</tr>
-<tr>
-<td></td>
-</tr>
-</tbody>
-</table>'
+RESULT='<p>: Y:|
+<code>|
+|
+</code>|</p>'

 cat << \EOF | try '-F 0x03000000' 'random input that looks like a table' -heredoc "$RESULT"
 : Y:|