document: Implement MATH extension

2 files changed, 69 insertions, 5 deletions

diff --git a/src/document.c b/src/document.c
index 8ae8eb9..8f35c98 100644
--- a/src/document.c
+++ b/src/document.c
@@ -78,6 +78,7 @@ static size_t char_autolink_email(hoedown_buffer *ob, hoedown_document *doc, uin
 static size_t char_autolink_www(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_t offset, size_t size);
 static size_t char_link(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_t offset, size_t size);
 static size_t char_superscript(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_t offset, size_t size);
+static size_t char_math(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_t offset, size_t size);
 
 enum markdown_char_t {
 	MD_CHAR_NONE = 0,
@@ -92,7 +93,8 @@ enum markdown_char_t {
 	MD_CHAR_AUTOLINK_EMAIL,
 	MD_CHAR_AUTOLINK_WWW,
 	MD_CHAR_SUPERSCRIPT,
-	MD_CHAR_QUOTE
+	MD_CHAR_QUOTE,
+	MD_CHAR_MATH
 };
 
 static char_trigger markdown_char_ptrs[] = {
@@ -108,7 +110,8 @@ static char_trigger markdown_char_ptrs[] = {
 	&char_autolink_email,
 	&char_autolink_www,
 	&char_superscript,
-	&char_quote
+	&char_quote,
+	&char_math
 };
 
 /* render • structure containing state for a parser instance */
@@ -685,6 +688,37 @@ parse_emph3(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_t siz
 	return 0;
 }
 
+/* parse_math • parses a math span until the given ending delimiter */
+static size_t
+parse_math(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_t offset, size_t size, const char *end, size_t delimsz, int displaymode)
+{
+	size_t i = delimsz;
+	if (!doc->md.math) return 0;
+
+	/* find ending delimiter */
+	while (1) {
+		while (i < size && data[i] != (uint8_t)end[0]) i++;
+		if (i >= size) return 0;
+
+		if (!is_escaped(data, i) && !(i + delimsz > size)
+			&& memcmp(data + i, end, delimsz) == 0)
+			break;
+		i++;
+	}
+
+	/* enforce spacing around the span */
+	if (offset && !_isspace(data[-1])) return 0;
+	if (i + delimsz < size && !_isspace(data[i + delimsz])) return 0;
+
+	/* prepare buffers */
+	hoedown_buffer text = { data + delimsz, i - delimsz, 0, 0, NULL, NULL, NULL };
+
+	/* call callback */
+	if (doc->md.math(ob, &text, displaymode, doc->md.opaque))
+		return i + delimsz;
+	return 0;
+}
+
 /* char_emphasis • single and double emphasis parsing */
 static size_t
 char_emphasis(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_t offset, size_t size)
@@ -832,10 +866,18 @@ char_quote(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_t offs
 static size_t
 char_escape(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_t offset, size_t size)
 {
-	static const char *escape_chars = "\\`*_{}[]()#+-.!:|&<>^~=\"";
+	static const char *escape_chars = "\\`*_{}[]()#+-.!:|&<>^~=\"$";
 	hoedown_buffer work = { 0, 0, 0, 0, NULL, NULL, NULL };
+	size_t w;
 
 	if (size > 1) {
+		if (data[1] == '\\' && (doc->ext_flags & HOEDOWN_EXT_MATH) &&
+			size > 2 && (data[2] == '(' || data[2] == '[')) {
+			const char *end = (data[2] == '[') ? "\\\\]" : "\\\\)";
+			w = parse_math(ob, doc, data, offset, size, end, 3, data[2] == '[');
+			if (w) return w;
+		}
+
 		if (strchr(escape_chars, data[1]) == NULL)
 			return 0;
 
@@ -1289,6 +1331,20 @@ char_superscript(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_
 	return (sup_start == 2) ? sup_len + 1 : sup_len;
 }
 
+static size_t
+char_math(hoedown_buffer *ob, hoedown_document *doc, uint8_t *data, size_t offset, size_t size)
+{
+	/* double dollar */
+	if (size > 1 && data[1] == '$')
+		return parse_math(ob, doc, data, offset, size, "$$", 2, doc->ext_flags & HOEDOWN_EXT_MATH_DOLLAR);
+
+	/* single dollar allowed only at with flag */
+	if (doc->ext_flags & HOEDOWN_EXT_MATH_DOLLAR)
+		return parse_math(ob, doc, data, offset, size, "$", 1, 0);
+
+	return 0;
+}
+
 /*********************************
  * BLOCK-LEVEL PARSING FUNCTIONS *
  *********************************/
@@ -2735,6 +2791,9 @@ hoedown_document_new(
 	if (extensions & HOEDOWN_EXT_QUOTE)
 		doc->active_char['"'] = MD_CHAR_QUOTE;
 
+	if (extensions & HOEDOWN_EXT_MATH)
+		doc->active_char['$'] = MD_CHAR_MATH;
+
 	/* Extension data */
 	doc->ext_flags = extensions;
 	doc->max_nesting = max_nesting;
diff --git a/src/document.h b/src/document.h
index 8fec835..23e850c 100644
--- a/src/document.h
+++ b/src/document.h
@@ -27,11 +27,13 @@ enum hoedown_extensions {
 	HOEDOWN_EXT_HIGHLIGHT = (1 << 6),
 	HOEDOWN_EXT_QUOTE = (1 << 7),
 	HOEDOWN_EXT_SUPERSCRIPT = (1 << 8),
+	HOEDOWN_EXT_MATH = (1 << 13),
 
 	/* other flags */
 	HOEDOWN_EXT_LAX_SPACING = (1 << 9),
 	HOEDOWN_EXT_NO_INTRA_EMPHASIS = (1 << 10),
 	HOEDOWN_EXT_SPACE_HEADERS = (1 << 11),
+	HOEDOWN_EXT_MATH_DOLLAR = (1 << 14),
 
 	/* negative flags */
 	HOEDOWN_EXT_DISABLE_INDENTED_CODE = (1 << 12)
@@ -48,12 +50,14 @@ enum hoedown_extensions {
 	HOEDOWN_EXT_UNDERLINE |\
 	HOEDOWN_EXT_HIGHLIGHT |\
 	HOEDOWN_EXT_QUOTE |\
-	HOEDOWN_EXT_SUPERSCRIPT )
+	HOEDOWN_EXT_SUPERSCRIPT |\
+	HOEDOWN_EXT_MATH )
 
 #define HOEDOWN_EXT_FLAGS (\
 	HOEDOWN_EXT_LAX_SPACING |\
 	HOEDOWN_EXT_NO_INTRA_EMPHASIS |\
-	HOEDOWN_EXT_SPACE_HEADERS )
+	HOEDOWN_EXT_SPACE_HEADERS |\
+	HOEDOWN_EXT_MATH_DOLLAR )
 
 #define HOEDOWN_EXT_NEGATIVE (\
 	HOEDOWN_EXT_DISABLE_INDENTED_CODE )
@@ -119,6 +123,7 @@ struct hoedown_renderer {
 	int (*strikethrough)(hoedown_buffer *ob, const hoedown_buffer *text, void *opaque);
 	int (*superscript)(hoedown_buffer *ob, const hoedown_buffer *text, void *opaque);
 	int (*footnote_ref)(hoedown_buffer *ob, unsigned int num, void *opaque);
+	int (*math)(hoedown_buffer *ob, const hoedown_buffer *text, int displaymode, void *opaque);
 
 	/* low level callbacks - NULL copies input directly into the output */
 	void (*entity)(hoedown_buffer *ob, const hoedown_buffer *entity, void *opaque);