redcarpet 1.3.3 → 1.5.0

data/ext/markdown.c

@@ -56,6 +56,8 @@ struct render {
 	struct array refs;
 	char_trigger active_char[256];
 	struct parray work;
+	unsigned int ext_flags;
+	size_t max_nesting;
 };
 
 /* html_tag • structure for quick HTML tag search (inspired from discount) */
@@ -238,7 +240,7 @@ parse_inline(struct buf *ob, struct render *rndr, char *data, size_t size)
 	char_trigger action = 0;
 	struct buf work = { 0, 0, 0, 0, 0 };
 
-	if (rndr->work.size > rndr->make.parser_options.recursion_depth)
+	if (rndr->work.size > rndr->max_nesting)
 		return;
 
 	while (i < size) {
@@ -249,7 +251,7 @@ parse_inline(struct buf *ob, struct render *rndr, char *data, size_t size)
 		if (rndr->make.normal_text) {
 			work.data = data + i;
 			work.size = end - i;
-			rndr->make.normal_text(ob, &work, &rndr->make.render_options);
+			rndr->make.normal_text(ob, &work, rndr->make.opaque);
 		}
 		else
 			bufput(ob, data + i, end - i);
@@ -346,7 +348,7 @@ parse_emph1(struct buf *ob, struct render *rndr, char *data, size_t size, char c
 
 		if (data[i] == c && !isspace(data[i - 1])) {
 
-			if ((rndr->make.parser_options.flags & MKD_LAX_EMPHASIS) == 0) {
+			if ((rndr->ext_flags & MKDEXT_LAX_EMPHASIS) == 0) {
 				if (!(i + 1 == size || isspace(data[i + 1]) || ispunct(data[i + 1])))
 					continue;
 			}
@@ -360,7 +362,7 @@ parse_emph1(struct buf *ob, struct render *rndr, char *data, size_t size, char c
 			}
 
 			parse_inline(work, rndr, data, i);
-			r = rndr->make.emphasis(ob, work, c, &rndr->make.render_options);
+			r = rndr->make.emphasis(ob, work, c, rndr->make.opaque);
 			rndr->work.size -= 1;
 			return r ? i + 1 : 0;
 		}
@@ -395,7 +397,7 @@ parse_emph2(struct buf *ob, struct render *rndr, char *data, size_t size, char c
 			}
 
 			parse_inline(work, rndr, data, i);
-			r = rndr->make.double_emphasis(ob, work, c, &rndr->make.render_options);
+			r = rndr->make.double_emphasis(ob, work, c, rndr->make.opaque);
 			rndr->work.size -= 1;
 			return r ? i + 2 : 0;
 		}
@@ -433,7 +435,7 @@ parse_emph3(struct buf *ob, struct render *rndr, char *data, size_t size, char c
 			}
 
 			parse_inline(work, rndr, data, i);
-			r = rndr->make.triple_emphasis(ob, work, c, &rndr->make.render_options);
+			r = rndr->make.triple_emphasis(ob, work, c, rndr->make.opaque);
 			rndr->work.size -= 1;
 			return r ? i + 3 : 0;
 
@@ -497,7 +499,7 @@ char_linebreak(struct buf *ob, struct render *rndr, char *data, size_t offset, s
 	while (ob->size && ob->data[ob->size - 1] == ' ')
 		ob->size--;
 
-	return rndr->make.linebreak(ob, &rndr->make.render_options) ? 1 : 0;
+	return rndr->make.linebreak(ob, rndr->make.opaque) ? 1 : 0;
 }
 
 
@@ -533,10 +535,10 @@ char_codespan(struct buf *ob, struct render *rndr, char *data, size_t offset, si
 	/* real code span */
 	if (f_begin < f_end) {
 		struct buf work = { data + f_begin, f_end - f_begin, 0, 0, 0 };
-		if (!rndr->make.codespan(ob, &work, &rndr->make.render_options))
+		if (!rndr->make.codespan(ob, &work, rndr->make.opaque))
 			end = 0;
 	} else {
-		if (!rndr->make.codespan(ob, 0, &rndr->make.render_options))
+		if (!rndr->make.codespan(ob, 0, rndr->make.opaque))
 			end = 0;
 	}
 
@@ -554,7 +556,7 @@ char_escape(struct buf *ob, struct render *rndr, char *data, size_t offset, size
 		if (rndr->make.normal_text) {
 			work.data = data + 1;
 			work.size = 1;
-			rndr->make.normal_text(ob, &work, &rndr->make.render_options);
+			rndr->make.normal_text(ob, &work, rndr->make.opaque);
 		}
 		else bufputc(ob, data[1]);
 	}
@@ -584,7 +586,7 @@ char_entity(struct buf *ob, struct render *rndr, char *data, size_t offset, size
 	if (rndr->make.entity) {
 		work.data = data;
 		work.size = end;
-		rndr->make.entity(ob, &work, &rndr->make.render_options);
+		rndr->make.entity(ob, &work, rndr->make.opaque);
 	}
 	else bufput(ob, data, end);
 
@@ -604,10 +606,10 @@ char_langle_tag(struct buf *ob, struct render *rndr, char *data, size_t offset,
 		if (rndr->make.autolink && altype != MKDA_NOT_AUTOLINK) {
 			work.data = data + 1;
 			work.size = end - 2;
-			ret = rndr->make.autolink(ob, &work, altype, &rndr->make.render_options);
+			ret = rndr->make.autolink(ob, &work, altype, rndr->make.opaque);
 		}
 		else if (rndr->make.raw_html_tag)
-			ret = rndr->make.raw_html_tag(ob, &work, &rndr->make.render_options);
+			ret = rndr->make.raw_html_tag(ob, &work, rndr->make.opaque);
 	}
 
 	if (!ret) return 0;
@@ -847,9 +849,9 @@ char_link(struct buf *ob, struct render *rndr, char *data, size_t offset, size_t
 		if (ob->size && ob->data[ob->size - 1] == '!')
 			ob->size -= 1;
 
-		ret = rndr->make.image(ob, link, title, content, &rndr->make.render_options);
+		ret = rndr->make.image(ob, link, title, content, rndr->make.opaque);
 	} else
-		ret = rndr->make.link(ob, link, title, content, &rndr->make.render_options);
+		ret = rndr->make.link(ob, link, title, content, rndr->make.opaque);
 
 	/* cleanup */
 cleanup:
@@ -902,6 +904,34 @@ is_hrule(char *data, size_t size)
 	return n >= 3;
 }
 
+/* check if a line is a code fence; return its size if it is */
+static size_t
+is_codefence(char *data, size_t size)
+{
+	size_t i = 0, n = 0;
+	char c;
+
+	/* skipping initial spaces */
+	if (size < 3) return 0;
+	if (data[0] == ' ') { i += 1;
+	if (data[1] == ' ') { i += 1;
+	if (data[2] == ' ') { i += 1; } } }
+
+	/* looking at the hrule char */
+	if (i + 2 >= size || data[i] != '~')
+		return 0;
+
+	/* the whole line must be the char or whitespace */
+	while (i < size && data[i] != '\n') {
+		if (data[i] == '~') n++;
+		else if (data[i] != ' ' && data[i] != '\t')
+			return 0;
+		i++;
+	}
+
+	return n >= 3 ? i + 1 : 0;
+}
+
 /* is_headerline • returns whether the line is a setext-style hdr underline */
 static int
 is_headerline(char *data, size_t size)
@@ -1027,7 +1057,7 @@ parse_blockquote(struct buf *ob, struct render *rndr, char *data, size_t size)
 
 	parse_block(out, rndr, work_data, work_size);
 	if (rndr->make.blockquote)
-		rndr->make.blockquote(ob, out, &rndr->make.render_options);
+		rndr->make.blockquote(ob, out, rndr->make.opaque);
 	rndr->work.size -= 1;
 	return end;
 }
@@ -1084,7 +1114,7 @@ parse_paragraph(struct buf *ob, struct render *rndr, char *data, size_t size)
 
 		parse_inline(tmp, rndr, work.data, work.size);
 		if (rndr->make.paragraph)
-			rndr->make.paragraph(ob, tmp, &rndr->make.render_options);
+			rndr->make.paragraph(ob, tmp, rndr->make.opaque);
 		rndr->work.size--;
 
 	} else {
@@ -1114,7 +1144,7 @@ parse_paragraph(struct buf *ob, struct render *rndr, char *data, size_t size)
 				parse_inline(tmp, rndr, work.data, work.size);
 
 				if (rndr->make.paragraph)
-					rndr->make.paragraph(ob, tmp, &rndr->make.render_options);
+					rndr->make.paragraph(ob, tmp, rndr->make.opaque);
 
 				rndr->work.size -= 1;
 				work.data += beg;
@@ -1124,13 +1154,61 @@ parse_paragraph(struct buf *ob, struct render *rndr, char *data, size_t size)
 		}
 
 		if (rndr->make.header)
-			rndr->make.header(ob, &work, level, &rndr->make.render_options);
+			rndr->make.header(ob, &work, level, rndr->make.opaque);
 	}
 
 	return end;
 }
 
 /* parse_blockquote • hanldes parsing of a block-level code fragment */
+static size_t
+parse_fencedcode(struct buf *ob, struct render *rndr, char *data, size_t size)
+{
+	size_t beg, end;
+	struct buf *work = 0;
+
+	beg = is_codefence(data, size);
+	if (beg == 0) return 0;
+
+	if (rndr->work.size < rndr->work.asize) {
+		work = rndr->work.item[rndr->work.size ++];
+		work->size = 0;
+	} else {
+		work = bufnew(WORK_UNIT);
+		parr_push(&rndr->work, work);
+	}
+
+	while (beg < size) {
+		size_t fence_end;
+
+		fence_end = is_codefence(data + beg, size - beg);
+		if (fence_end != 0) {
+			beg += fence_end;
+			break;
+		}
+
+		for (end = beg + 1; end < size && data[end - 1] != '\n'; end += 1);
+
+		if (beg < end) {
+			/* verbatim copy to the working buffer,
+				escaping entities */
+			if (is_empty(data + beg, end - beg))
+				bufputc(work, '\n');
+			else bufput(work, data + beg, end - beg);
+		}
+		beg = end;
+	}
+
+	if (work->size && work->data[work->size - 1] != '\n')
+		bufputc(work, '\n');
+
+	if (rndr->make.blockcode)
+		rndr->make.blockcode(ob, work, rndr->make.opaque);
+
+	rndr->work.size -= 1;
+	return beg;
+}
+
 static size_t
 parse_blockcode(struct buf *ob, struct render *rndr, char *data, size_t size)
 {
@@ -1165,7 +1243,7 @@ parse_blockcode(struct buf *ob, struct render *rndr, char *data, size_t size)
 		work->size -= 1;
 	bufputc(work, '\n');
 	if (rndr->make.blockcode)
-		rndr->make.blockcode(ob, work, &rndr->make.render_options);
+		rndr->make.blockcode(ob, work, rndr->make.opaque);
 	rndr->work.size -= 1;
 	return beg;
 }
@@ -1292,7 +1370,7 @@ parse_listitem(struct buf *ob, struct render *rndr, char *data, size_t size, int
 
 	/* render of li itself */
 	if (rndr->make.listitem)
-		rndr->make.listitem(ob, inter, *flags, &rndr->make.render_options);
+		rndr->make.listitem(ob, inter, *flags, rndr->make.opaque);
 
 	rndr->work.size -= 2;
 	return beg;
@@ -1322,7 +1400,7 @@ parse_list(struct buf *ob, struct render *rndr, char *data, size_t size, int fla
 	}
 
 	if (rndr->make.list)
-		rndr->make.list(ob, work, flags, &rndr->make.render_options);
+		rndr->make.list(ob, work, flags, rndr->make.opaque);
 	rndr->work.size -= 1;
 	return i;
 }
@@ -1358,7 +1436,7 @@ parse_atxheader(struct buf *ob, struct render *rndr, char *data, size_t size)
 		work.size = end - i;
 
 		if (rndr->make.header)
-			rndr->make.header(ob, &work, (int)level, &rndr->make.render_options);
+			rndr->make.header(ob, &work, (int)level, rndr->make.opaque);
 
 	}
 
@@ -1433,7 +1511,7 @@ parse_htmlblock(struct buf *ob, struct render *rndr, char *data, size_t size, in
 			if (j) {
 				work.size = i + j;
 				if (do_render && rndr->make.blockhtml)
-					rndr->make.blockhtml(ob, &work, &rndr->make.render_options);
+					rndr->make.blockhtml(ob, &work, rndr->make.opaque);
 				return work.size;
 			} 
 		}
@@ -1450,7 +1528,7 @@ parse_htmlblock(struct buf *ob, struct render *rndr, char *data, size_t size, in
 				if (j) {
 					work.size = i + j;
 					if (do_render && rndr->make.blockhtml)
-						rndr->make.blockhtml(ob, &work, &rndr->make.render_options);
+						rndr->make.blockhtml(ob, &work, rndr->make.opaque);
 					return work.size;
 				}
 			} 
@@ -1505,11 +1583,192 @@ parse_htmlblock(struct buf *ob, struct render *rndr, char *data, size_t size, in
 	/* the end of the block has been found */
 	work.size = i;
 	if (do_render && rndr->make.blockhtml)
-		rndr->make.blockhtml(ob, &work, &rndr->make.render_options);
+		rndr->make.blockhtml(ob, &work, rndr->make.opaque);
 
 	return i;
 }
 
+static void
+parse_table_row(struct buf *ob, struct render *rndr, char *data, size_t size, size_t columns, int *col_data)
+{
+	size_t i = 0, col;
+	struct buf *row_work = 0;
+
+	if (rndr->work.size < rndr->work.asize) {
+		row_work = rndr->work.item[rndr->work.size ++];
+		row_work->size = 0;
+	} else {
+		row_work = bufnew(WORK_UNIT);
+		parr_push(&rndr->work, row_work);
+	}
+
+	if (i < size && data[i] == '|')
+		i++;
+
+	for (col = 0; col < columns && i < size; ++col) {
+		size_t cell_start, cell_end;
+		struct buf *cell_work;
+
+		if (rndr->work.size < rndr->work.asize) {
+			cell_work = rndr->work.item[rndr->work.size ++];
+			cell_work->size = 0;
+		} else {
+			cell_work = bufnew(WORK_UNIT);
+			parr_push(&rndr->work, cell_work);
+		}
+
+		while (i < size && isspace(data[i]))
+			i++;
+
+		cell_start = i;
+
+		while (i < size && data[i] != '|')
+			i++;
+
+		cell_end = i - 1;
+
+		while (cell_end > cell_start && isspace(data[cell_end]))
+			cell_end--;
+
+		parse_inline(cell_work, rndr, data + cell_start, 1 + cell_end - cell_start);
+		if (rndr->make.table_cell)
+			rndr->make.table_cell(row_work, cell_work, col_data ? col_data[col] : 0, rndr->make.opaque);
+
+		rndr->work.size -= 1;
+		i++;
+	}
+
+	for (; col < columns; ++col) {
+		struct buf empty_cell = {0, 0, 0, 0, 0};
+		if (rndr->make.table_cell)
+			rndr->make.table_cell(row_work, &empty_cell, col_data ? col_data[col] : 0, rndr->make.opaque);
+	}
+
+	if (rndr->make.table_row)
+		rndr->make.table_row(ob, row_work, rndr->make.opaque);
+
+	rndr->work.size -= 1;
+}
+
+static size_t
+parse_table_header(struct buf *ob, struct render *rndr, char *data, size_t size, size_t *columns, int **column_data)
+{
+	int pipes;
+	size_t i = 0, col, header_end, under_end;
+
+	pipes = 0;
+	while (i < size && data[i] != '\n')
+		if (data[i++] == '|')
+			pipes++;
+
+	if (i == size || pipes == 0)
+		return 0;
+
+	header_end = i;
+
+	if (data[0] == '|')
+		pipes--;
+
+	if (i > 2 && data[i - 1] == '|')
+		pipes--;
+
+	*columns = pipes + 1;
+	*column_data = calloc(*columns, sizeof(int));
+
+	/* Parse the header underline */
+	i++;
+	if (i < size && data[i] == '|')
+		i++;
+
+	under_end = i;
+	while (under_end < size && data[under_end] != '\n')
+		under_end++;
+
+	for (col = 0; col < *columns && i < under_end; ++col) {
+		size_t cell_start, cell_end;
+
+		if (data[i] == ':') {
+			i++; (*column_data)[col] |= MKD_TABLE_ALIGN_L;
+		}
+
+		while (i < under_end && data[i] == '-')
+			i++;
+
+		if (i < under_end && data[i] == ':') {
+			i++; (*column_data)[col] |= MKD_TABLE_ALIGN_R;
+		}
+
+		if (i < under_end && data[i] != '|')
+			break;
+
+		i++;
+	}
+
+	if (col < *columns)
+		return 0;
+
+	parse_table_row(ob, rndr, data, header_end, *columns, *column_data);
+	return under_end + 1;
+}
+
+static size_t
+parse_table(struct buf *ob, struct render *rndr, char *data, size_t size)
+{
+	size_t i;
+
+	struct buf *header_work = 0;
+	struct buf *body_work = 0;
+
+	size_t columns;
+	int *col_data = NULL;
+
+	if (rndr->work.size < rndr->work.asize) {
+		header_work = rndr->work.item[rndr->work.size ++];
+		header_work->size = 0;
+	} else {
+		header_work = bufnew(WORK_UNIT);
+		parr_push(&rndr->work, header_work);
+	}
+
+	if (rndr->work.size < rndr->work.asize) {
+		body_work = rndr->work.item[rndr->work.size ++];
+		body_work->size = 0;
+	} else {
+		body_work = bufnew(WORK_UNIT);
+		parr_push(&rndr->work, body_work);
+	}
+
+	i = parse_table_header(header_work, rndr, data, size, &columns, &col_data);
+	if (i > 0) {
+
+		while (i < size) {
+			size_t row_len;
+			size_t row_start;
+			int pipes = 0;
+
+			row_start = i;
+
+			while (i < size && data[i] != '\n')
+				if (data[i++] == '|')
+					pipes++;
+
+			if (pipes == 0 || i == size) {
+				i = row_start;
+				break;
+			}
+
+			parse_table_row(body_work, rndr, data + row_start, i - row_start, columns, col_data);
+			i++;
+		}
+
+		if (rndr->make.table)
+			rndr->make.table(ob, header_work, body_work, rndr->make.opaque);
+	}
+
+	free(col_data);
+	rndr->work.size -= 2;
+	return i;
+}
 
 /* parse_block • parsing of one block, returning next char to parse */
 static void
@@ -1519,7 +1778,7 @@ parse_block(struct buf *ob, struct render *rndr, char *data, size_t size)
 	char *txt_data;
 	beg = 0;
 
-	if (rndr->work.size > rndr->make.parser_options.recursion_depth)
+	if (rndr->work.size > rndr->max_nesting)
 		return;
 
 	while (beg < size) {
@@ -1529,7 +1788,8 @@ parse_block(struct buf *ob, struct render *rndr, char *data, size_t size)
 		if (data[beg] == '#')
 			beg += parse_atxheader(ob, rndr, txt_data, end);
 
-		else if (data[beg] == '<' && rndr->make.blockhtml && (i = parse_htmlblock(ob, rndr, txt_data, end, 1)) != 0)
+		else if (data[beg] == '<' && rndr->make.blockhtml &&
+				(i = parse_htmlblock(ob, rndr, txt_data, end, 1)) != 0)
 			beg += i;
 
 		else if ((i = is_empty(txt_data, end)) != 0)
@@ -1537,7 +1797,7 @@ parse_block(struct buf *ob, struct render *rndr, char *data, size_t size)
 
 		else if (is_hrule(txt_data, end)) {
 			if (rndr->make.hrule)
-				rndr->make.hrule(ob, &rndr->make.render_options);
+				rndr->make.hrule(ob, rndr->make.opaque);
 
 			while (beg < size && data[beg] != '\n')
 				beg++;
@@ -1545,6 +1805,14 @@ parse_block(struct buf *ob, struct render *rndr, char *data, size_t size)
 			beg++;
 		}
 
+		else if ((rndr->ext_flags & MKDEXT_FENCED_CODE) != 0 &&
+			(i = parse_fencedcode(ob, rndr, txt_data, end)) != 0)
+			beg += i;
+
+		else if ((rndr->ext_flags & MKDEXT_TABLES) != 0 &&
+			(i = parse_table(ob, rndr, txt_data, end)) != 0)
+			beg += i;
+
 		else if (prefix_quote(txt_data, end))
 			beg += parse_blockquote(ob, rndr, txt_data, end);
 
@@ -1702,7 +1970,7 @@ static void expand_tabs(struct buf *ob, const char *line, size_t size)
 
 /* markdown • parses the input buffer and renders it into the output buffer */
 void
-markdown(struct buf *ob, struct buf *ib, const struct mkd_renderer *rndrer) {
+markdown(struct buf *ob, struct buf *ib, const struct mkd_renderer *rndrer, unsigned int extensions) {
 	struct link_ref *lr;
 	struct buf *text = bufnew(TEXT_UNIT);
 	size_t i, beg, end;
@@ -1738,6 +2006,10 @@ markdown(struct buf *ob, struct buf *ib, const struct mkd_renderer *rndrer) {
 	rndr.active_char['\\'] = char_escape;
 	rndr.active_char['&'] = char_entity;
 
+	/* Extension data */
+	rndr.ext_flags = extensions;
+	rndr.max_nesting = 16;
+
 	/* first pass: looking for references, copying everything else */
 	beg = 0;
 	while (beg < ib->size) /* iterating over lines */
@@ -1775,12 +2047,12 @@ markdown(struct buf *ob, struct buf *ib, const struct mkd_renderer *rndrer) {
 
 	/* second pass: actual rendering */
 	if (rndr.make.doc_header)
-		rndr.make.doc_header(ob, &rndr.make.render_options);
+		rndr.make.doc_header(ob, rndr.make.opaque);
 
 	parse_block(ob, &rndr, text->data, text->size);
 
 	if (rndr.make.doc_footer)
-		rndr.make.doc_footer(ob, &rndr.make.render_options);
+		rndr.make.doc_footer(ob, rndr.make.opaque);
 
 	/* clean-up */
 	bufrelease(text);